Tweet
I picked this up the other day for a decent price and it works just fine. It's not the last word in pulse delay sensitivity but I plan to use it in fast salt water (just behind where the 2 footers break). I believe the pulse delay is in the high twenties and I'm okay with that. I have stripped it down and removed just about everything. I'm guessing it only has seen one or two outings and is pristine. The O-rings were still greasy from the silicone lube and all cables are like new (have been treated with 303 Protectant already). This particular unit was from the far south of my state here in Western Australia so it was never subjected to hot temperatures as the area is cool and cloudy just about all the time. The coil is near flawless and importantly the coil cable is shiny and very pliable....no signs of any age related problems. I will probably swap headphones depending on how my other spare waterproof headphones sound. The PCB is out and although it didn't need it, I cleaned the whole board down with a specialist PCB cleaning spray. The voltage converter (charge pump) chip has been removed from the board (that's the bare spot on the board at the lower left) and will be replaced with a higher rated one to 12v (thanks to Geotech on that advice). I will replace the 6 electrolytic capacitors with new ones as well. A new snap connector power lead will also be soldered on.
I ran a magnifying glass over the whole board and nothing appears out of the ordinary as far as dis-coloured components is concerned.
Unfortunately I don't have the skills to actually mod this detector (faster delay) but I'm not too worried about that considering how I will be using it. All cable gland and O-rings will be replaced as well but don't appear to need it. I will also Plast-Dip the lower half of the coil for added insurance. These 950 Zero-Buoyancy coils are well regarded and are almost impossible to find in this condition....they are nice and heavy and require no effort to keep down.
If anyone else has any other ideas then feel free to comment. I know many of you can easily mod this detector.....having a pulse delay of around 20uS would be nice but I suspect my technical abilities is the weak point.
Thanks again.
***** I have included a Word doc that I generated from Findmall. I copy and pasted most of the relevant sections regarding the modification of the PI3000. It's quite dis-jointed but there are some ideas being thrown around. I'm not sure if there was a final outcome to all of this *****
This is the contents of the Word doc. I couldn't upload it separately. I think it is from around 2003-2004.
Hello!
I own a Whites PI3000 (the yellow one) that I'm succesfully using for beach/shore detecting.
Since I never found coins or rings deeper than a couple of inches, I'm asking myself if I can improve depth of detection by some sort of modification to its electronics.
I already replaced standard alkaline batteries (six) with a self built NiCD pack of 8 AA batteries.
I have the original Whites schematics (don't ask how I got them!) that report a possible Vin of 6.5 to 10.5 Volts at a current drain of 60 to 110 mA.
A NiCD pack of 8 AA 1,2V 800mA batteries can supply 9,6V for about 8 hours with a peak of 10.5 Volts (!) just after recharging.
I did not use NiMh batteries because I believe that power supply must be as much steady as possible expecially when a voltage regulator IS NOT employed in the electronics as is the case of my PI3000 and NiCds have the stadiest voltage supply of all.
It also appears to me that transmission power is fed to the coil directly from V+ that is derived from Vin minus 0.55 Volts circa. So my conclusion is that the higher the Voltage the higher the transmission power injected into the coil. Am I wrong?
Are there other mods to be made to the PI3000?
On the motherboard there is a variable resistor R46 100Kohm that looks to be able to trim gain of the first stage 5534 RX OpAmp. Is this a sensitivity control and what appens if I tweak it?
Happy Hunting
Luca
Being a newcomer to PI technology, it still seems as though you might be swinging the coil too quickly or perhaps, adjusting the sensitivity improperly: in Auto or too high (article below mentions retuning regularly as being necessary and Auto retuning reducing depth substantially)--which will actually reduce detection depth, since you won't be able to hear deeper signals-- Lastly, it is possible that there is something wrong with the coil or the detector itself.
Below is a link to an article about the White's PI 3000, which seems to have some good user tips.
Whell,'I'll air-test the unit and post results here. I'm using it correctly but I have no references on how deep it should go. One thing I noticed for sure is that it misses some gold, like small chains for example.
I was asking myself if someone made any modification or has someone tested it against other newer detectors.
I read tests of other brands were is claimed a detection depth of 20" on good size coins; that appears to me rocket science.
I can reach those depths only on very big targets like coke cans.
Thanbks all for the help....
Hello,
I follow my prec post with the following notes on Whites PI3000 sensitivity.
I air tested my PI3000 with some coins (euros) and a few silver,gold,copper rings and chains I found using a Whites Eagle Spectrum (very good machine) here in Italy.
The largest euro coin is the 2 Euros (worth today about 2.5 dollars) ; the PI3000 can detect it at about 10 inch while my Spectrum can go up to about 15 inch. These figures look good but gold is the real problem.
I tested a few gold rings and some (2 of them) are barely detected at coil contact!
One thin gold chain with small gold pendant and a thin silver chain are completely invisible to the PI3000.
I include a picture that shows my findings in order of detection sensitivity (right less or none detection, left maximum detection).
I'm worried I lose gold while detecting wet sand.
Did anyone test a PI3000 against other PI detectors?
Any idea?
Metal detectors see only one link of the gold chain.
This makes it hard to find gold chains. I have found only one gold chain in my 10 years of metal detecting and it was because there was a large pendant attached to the chain. It should be picking up gold rings though. You might want to send it in to Whites for a check up.
Good luck!
HH
Emmett
Luca
Lucia,
You might have to change time delay to do better on gold.Does the PI-3000, have these IC's 555 timer,4017 or 4069 IC. What part of Italy are you in.
Hello,
I'm in the center of Italy. I can reach both west and east coasts. West coasts (Rome) are heavily mineralized (black sand). Garret has recently modified its Infinium Pulse LS detector to better workout our western coasts . They call it Infinium Mediterraneo but I had no chances to try it.
So a pulse detector is a must here.
As for the PI3000 , in the schematic I see that the basic 30Khz clock is divided by 10 with a CD4017 decade divider. I don't know how long is the pulse in uS, I've no idea of how to calculate it from the schema. I'm good at digital circuits but bad in analogue.
The only thing I understood is that a pulse is transmitted every 3Khz (e.g. 30Khz div 10) but since I don't know how long the pulse is, I'm not able to determine sampling timing.
I found also the PI1000 schematic and it is basically equal to the PI3000. Same TX circuit, same RX circuit (amplifiers and everyting). Differences only in the threshold and audio so stating that the PI3000 is better than the PI1000 is simply wrong.
One idea could be to make a piggyback board to replace the CD4017 with a PIC16F84 to program timings at will (or even at the push of a button) and this is easy to do for me but first of all I have to understand all the original timings ....
Luca,
Sounds like you can repair the PI-3000, if you damage it.To lower time delay might be as simple as change-ing a Cap.My pi-1000 was modifyed by lowering a cap.white might have all ready made the change in your pi-3000.I need to see the print to know.If you have a Infinium PI, a WHITE XLT, Why bother with the PI-3000.Does it work OK in black sand.
I have a Garrett Infinium also,there is no black sand,only soft and hard packed beach sand where I live N-FLA East coast.to use the infinium here I have to adjust it back,low thershold, fast or slow,ground balance wont hold.Do you know what Garrett did to modify it.if they adjusted it back inside the detector,it might cut down on dept.
I have a Minelab SOV; nothing bothers it here on the beach.it dont go as deep,you dont dig all that trash.
Do you ever get down to Scily.My family came from there.my father's father had some kind of farm there,
they found Ceaser toom on his land a end of 1800.most likely Agustus Ceaser.The town has same name.You can research my name to locate home-site and land. The toom is in a out-cropeing of large rocks. use my E-Mail, [email protected]
I get back to you in a fue days. Frank.
Hello,
I have only a PI3000 a Whites Spectrum and a Whithes TM600 cache locator. Alla machines were made in the '90s.
Do you know where I can find the specs of the PI1000 modification ? I could check it against PI3000 schematics.
In one word I'm just worried of sensitivity to gold because I'm able to find a lot of coins but not much gold.
I'm preferring the PI3000 just because it lets me search where most of the other won't do: on wet sand and UW.
Cheers
Hello,
the PI3000 has a Clock generator made out of a CD4069 and a Clock divider (by ten) CD4017 to drive all the functions.
My opinion is that the PI3000 works really well on wet sand ; I've been able to retrieve an impressive quantity of coins, a watch and a silver ring,in a 3 days beach search last week. The only point here is to find ways to improve sensitivity to small gold. Pinpointing is good. I use always a scoop, since you can always find rusted nails to cut your fingers....
I tested pin TP2 (R12) of the clock generator.
Clock speed is 30Khz (32 in my machine) .
As obvious from the schematic, clock is fed in to Clk pin of CD4017 that acts as a Clock divider (by ten).
A TX pulse is generated by the CD4017 at Q1 pin 2. TX pulse length is 30uS.
Ater TX shutoff, sampling is started by CD4017 after a delay of 30uS (!).
Final sampling of background noise is taken 210uS after Tx shutoff.
So my conclusion is that this is a 30uS samplig detector and this well explains why it does't easily (or at all) detect small gold....
HH
Luca, 4069 is IC that will lower time delay. I modified my 1000,its not safe to tell you to make a change,with out compairing your print with my print to see if they are the same or close.Did you have any problem getting 13000 apart.make sure to use new O-rings and put water proof lub on them.
4069 IC mod; increased dept about 4".it was more sensitive to smaller non-ferrous metal and smaller gold rings.(Dont make the same mistake that I maked)I wonted a PI that would detect small nuggets.(my mistake)I lowered resastance to fast to get where I wonted to go.End results a transister got hot,detector off line now.can not get the trans;
If you lower R by putting another resister across it.Start off with a very high value resister, slowly come down to what you think it chould be.Good Luck.
Hi Luca,
Are you sure the TX pulse is only 30uS? This seems a bit short for that type of detector. Normally two or three of the 4017 outputs are OR'd together to give a TX pulse that is either 60 or 90uS wide. However, if this were done, it would not leave enough outputs to give the 210uS second delay.
If you wanted to improve the sensitivity to gold, I would double the clock frequency to give 15uS pulses out of the 4017. Then OR pins 2,4 and 7 with diodes into the transmitter drive to give a 45uS TX pulse. Take pin 1 as the signal sample pulse and pin 3 as the noise subtraction pulse. You are left with 60uS between the end of the signal sample and the start of the noise sample, which is adequate for most gold objects.
I'm not familiar with the circuitry of the PI3000 so there are bound to be other things that are not optimum, but the above is a good start on improving the sensitivity.
Eric.
Thanks for the suggestions!
I'm sure (as far as my understanding of the schematic is good)that TX pulse is 30uS.
The TX transistors are bound only to pin 2 (Q1) of CD4017 .
Signal pulse is bound only to pin 11 (Q3) of CD4017 and ground pulse to pin 7 (Q9).
So if I made it right, given a clock of 30Khz (I measured it) e.g. 30uS cycle , the cd4017 will pulse each gate for as long as 30uS.
So at Q1 we transmit (30uS), at Q3 we sample signal (after 30US of Q2) and at Q9 we measure ground signal , 7 cycles after TX pulse, so 210uS later.
My schematic of the PI3000 is very similar to the PI1000; they did't change much...
I hope I did not bother too much since this is a hold detector and , maybe, I should get a better one.
HH
Hello,
well removing the board worries me . Probably I can get it out without having to loose the front cables because it looks like there is enought free cable inside the case to let the board out.
If I undrestood your suggestion you mean I can double the clock speed working on the CD4069 and achieve a 15uS speed as Eric also suggested.
What I did not understand is how you "fried" the ZTX750 TX transistor.
My idea of a possible modification is to make a small board with a PIC16F84 processor to replace the CD4017 and program a diferent timing. This would leave the TX pulse width unchanged and not arm the TX transistor, but will the 4Mhz clock of the PIC influence the LM5534 RX first stage amplifier? Shielding required
Hi Luca
I follow your posts and have one more suggestion to adjust your timing circuit. I'm sure it works. Change/replace either R11 or R12 with a poti. One of them will work. By replacing the resistor with a poti you can adjust the clock circuit frequency.
The Clock circuit (CD4069) triggers your IC U4. With adjusting the poti you can "speed up" or "slow down" your clock circuit frequency. The clock "speed" adjustment works than like a discrimination and it gives you the opportunity to adjust your sample delay.
Chris
Hello Chris,
I know it will work, but a problem I see is that, in this way, I also shorten TX pulse lenght and this is undesiderable since , as Eric stated, the TX pulse lenght appears to be already too short (30uS); bringing it down to 15uS would limit detection depth in my opinion.
While it is a very easy modification( R12 can be replaced without disassembly of the whole detector) I still think that replacing the CD4017 would be better. I'm looking forward to see if a piggiback board for a PIC16F84 would fit.
I fried 750 by lowering R12 or R11 to fast.Not sure witch one, will check and let you know later.R11 AND R12 is same as R5-R6 on a 555 timmer.R12, start off with 470K then go lower.first thing you wont to do is lower a cap at 4069.if it has a 150 pf go to 130pf no lower.4069 is main timer.I HAD asked to see the print to compair.if the 1300 has been changed from 1000.you might frie something.I am at libary what I just said is from memery.(Large cap next to 4069 what size is it).
4017 I could be wrong,Eric siad to make a change to pin () () TO SPEAD up clock. dest-a durd-a
Frank-S
Hi Frank and Luca,
Changing the clock speed by half, up or down, should not fry anything. Whether you speed up, or slow down the clock, the transmitter current should remain within reasonable limits. That is because the pulse width and pulse rate change proportionally. In practice you will find that the TX current falls somewhat as the clock speed is raised. This is because the current is limited by the coil inductance. Assuming a coil inductance of two or three hundred uH, a 30us pulse is too short for the current to reach its maximum value, which would be limited by the circuit resistance with a much wider pulse. A 200uH coil with a total of 2 ohms series resistance, has a time constant of 100uS. To reach 95% of its final value, the TX pulse would need to be 300uS wide. A 30uS pulse will only reach about 25% of the final value current before it switches off. At this point it is still rising steeply and inducing eddy currents in the target which are in opposition to those which you want at switch off. Result; loss in detection range.
I suspect that in Frank
Hi Luca,
you can try to remove R13 from U4 PIN2 and connect the resistor to U4 PIN4 + the clock circuit mods. This mod will trigger your TX pulse later at a later stage.
Chris
Hello!
I designed, built, and tested my modification to the Whites PI3000 (veteran? toy? you name it) pulse detector.
Modification is based on the substitution of the main timing chip (CD4017) that gives the timings to the TX pulse, target sample pulse , and ground sample pulse of my PI3000.
I substitute the CD4017 with a small board that carries a PIC16F628 microcontroller at 4Mhz clock speed and some other support logic.
The board is connected inplace of the CD4017 by means of a flat cable.
The board fits nicely inside the PI3000 housing.
I have added a rotary switch so that I can select up to 10 different programmed timing combinations to eventually adapt to different ground conditions or even simulate the original timings.
Thanks to Eric Foster's suggestions I programmed a timing sequence that gives 2 more full inches of sensitivity on dimes (euro cents here).
The sequence is:
TX Pulse 60uS (originally 33uS)
RX Target delay 26uS for 15uS time (originally both 33uS)
RX Ground 50uS after RX Target and again for 15uS time (originally 165uS later ! for 33uS time).
All this to keep the original 3Khz sampling times even if I doubled the pulses and quadrupled the TX enery...
I programmed other timings as well but it's too long to list here.
I've not been able to move RX timing faster than 26uS because of the coil response in combination to the (nice) feature of automatic ground balance that the PI3000 provides.
I probably match an unmodified Whites Surf PI.
I post the schematic and I'll post a detailed "how to" only on heavy request!
The improvements reported are "on air tests" at the moment.....
Hello, it is impossible for me to make a real test here in my garage and I had no chance to try it in the open. Here I have iron everywhere.
I'll be at the beach during the weekend and I'll answer with a report next week.
The only thing I could test was that , under the same conditions ,so even inside the garage, by changing from the "original" mode to the new one I could gain about 2 inches of depth.
The improvement over the original mode is evident also with the other combinations I programmed, but only the beach will tell by how much.
There is also the possibility that the best combination for air tests won't be the best on wet sand.
I'll be back!
Greethings
Hi Luca,
Your mod; looks good. looks like you re-invented the wheel.2" increase not to good after all that work.I am sure that you will get it to go deeper.your mod is very impresive.
Reg,helped me with mine.took me about 5 mi to install a cap.with a 3" increase in dept on US nickel,Dont forget I later ruined PI-1000.dont have your skills.Now you need to find gold with it.
Been detecting over 40 yrs. might be able to help you there.Try what I call a military grid pattern on your beach. At low tide start about 70 ft above low water.walk in strate line down to water,keep coil flat about 1"from ground,keep it flat to end of your swing.at end of swing push coil forwards(still flat) half diameter of coil come back across with you swing.Do this both ways left and right.this is more than enough over lap for your PI.
The object is to locate all target.if you miss one you step on a mine and blow-up.
Now you need to walk in a strate line down to water and back up 70 ft. then back down to water until you cover the entire beach.I have watched a lot of people over the years, most of the can not walk a strate line.To over come this drag your foot in sand then look back,this will keep you strate.
Dont forget each time that you turn around on each approach, step over abut half length of your swing,and go back in strait line. good luck
Frank.
Hello Frank,
thanks for the suggestions.
Can I ask you which cap did you change on the PI1000?
Chris suggested to try to change the preamplifier opamp feedback resitor (1MOhm) to 1.2 or 1.5 MOhm (see PI1000 schematic below) to try to have a higher preamp gain.
That, in my opinion, should be changed with a potentiometer to adjust gain in accordance to ground conditions but this would mean to carry around wires that could pick up noise and get the situation worse
.
Anyway I did not try it yet, because I first want to evaluate my modification and play with it for a while.
I have to thank Eric Foster for his detailed articles about PIs that sparked my entusiasm and willing to share it with you all.
Hello,
I think that the 5.6 Ohm resistor in series with the coil has been put there by Whites to limit current flow in the Ztx750 transistor that has a maximum current rating of 2A. This is my opinion, maybe Eric has a better explanation....
<img src="/metal/html/confused.gif" border=0 width=15 height=22 alt=":?">
HH
Luca,
I am at public libary, might have told you about cap a month ago.this is from menery.Its the main cap at 4069 IC.might be 150pf dont go lower 120pf.You have all ready made a mager change.dont mess with this untill you finished your mod.
I think that Cris was talking about lowering the resistor at 4069 timer.its a 5 mi job.all you do is put one resistor on bottom(parall)of resistor.this will seedup timer and lower time delay.you can also put a pot ther for a non-ferrous ID.no need to remove resistors.
Frank.
Hi Luca,
Glad the technical info on this forum got your enthusiasm going for doing a bit of experimenting.
The 5534 in your schematic is running at a gain of 1000, which is more than I normally use for a front end. You need to look up the gain/bandwidth curves for the 5534 to see where you are sitting at the moment. Increasing the gain will reduce the bandwidth and may affect the earliest point that you can sample. The reason for this is that the amplifier takes longer to come out of saturation after the transmitter pulse. Using a 5534, I find that a front end gain of about 500 is about as high as I would want to go, to be able to sample at 15uS. If you were sampling later than that (which the White's did), then you can have more gain and a longer recovery.
Having leads and a pot to alter the front end gain is not a good idea. This will add capacitance to the feedback circuit which will result in even less bandwidth than you would expect from the curves.
I don't see the point of the 10K resistor from pin 2 to ground in the schematic.
The 5.6 ohm resistor limits the maximum coil current and improves the coil circuit time constant, so that the switch on current is not still rapidly growing at the point it is switched off.
Eric.
Hello Chris,
I have a 10 positions rotary switch to select one out of ten different timing programs.
I choose the following to start with:
N. TX Dly Target Dly Ground (all times uS)
0 33 33 33 165 33
1 33 30 33 165 33
2 33 28 33 165 33
3 33 26 33 165 33
4 66 33 33 165 33
5 66 30 33 165 33
6 66 28 33 165 33
7 66 26 33 165 33
8 33 26 15 50 15
9 66 26 15 50 15
Prog 0 is same as the unmodified PI3000.
1 to 3 just a sampling speedup.
4 to 7 double TX time
8 and 9 are special and also the most sensitive.
If you look at the numbers in prog. n. 8 and 9 I shortened the whole Tx/RX cycle by sampling for only 15uS and allowing a 50 uS only delay between Rx target sample and Rx ground sample. Apparently this should give HALF sensitivity but since this whole cycle is long HALF of the original one (e.g. prog.n. 0) this one is executed twice in the same time period.
In other words the sampling capacitor is charged always by the same time length but the signal is taken more toward the 26uS area(! <img src="/metal/html/wink.gif" border=0 width=15 height=15 alt="
"> ) where more signal is available .
I remind that I can't go faster than 26uS because I have a strange curve in the 20 to 26 uS that goes in the minus area (below ground!).
Probably this is an effect of the slow recovery speed of the 5534 Op amp as suggested by Eric. I'll try to lower it's gain to 500 times and see if this beaviour vanishes. In that case my programs will change again....
Sensitivity is really improved. I gain 2 inches on any target and I noticed a greater sensitivity to very small targets as well.
No doubts about it.
Cheers
Luca
I'll check the 5534 and half its gain to see what happens.
I admit my curve, measured at the output of 5534 is high and flat between 0 to 13us, decays to ground between 13 and 20uS, goes BELOW ground from 20 to 26-28 uS and is around ground from 28uS on.This prevents me to sample faster than 26uS.
Do you mean that I should try also to remove the 10K resistor to pin 2 and ground of 5534?
Hello,
it is not exactly what you asked, anyway I tested a solid 18K gold ring with gems in dry sand and I spot it at true 10 inches depth.
My modification does not make miracles but works!
Hi Luka,
thanks for the update. You should try to reduce the gain of your receiver like Eric suggested. This might help you to gain m
Hi Luca,
I am a bit puzzled when you say that the waveform is high and flat from 0 - 13uS. This is the period when the transmitter is on, and is usually more than 100uS for an underwater detector. At the end of the TX pulse, the waveform will swing negative due to the reverse emf generated in the coil at this time. You can see this emf if you put the scope directly on the coil. It appears as a high voltage spike at the end of the TX pulse. When you look at the coil, everything is reversed, compared to the amplifier output, as the amplifier inverts the signal.
If you try a lower gain in the 5534, you could increase the gain in the dc amplifier to compensate. Can you post a scope picture of the 5534 output waveform?
Take out the 10K from pin 2 to ground. Pin 2 is referenced to ground by the other resistors connected to it.
Eric.
Hello,
I explain...
I measure at the output of 5534, pin6.
All times are measured from TX pulse END.
After a 33uS TX pulse from 0uS to 13uS I see a flat high signal (diodes levelling?) that starts to drop in a curve between 13.8 and 20 uS. From 20 to 26uS signal goes below ground and levels at ground after 27uS.
I can't sample between 20 and 26uS because the PI3000 goes unstable, mute or almost unsensitive in that area.
I have two explanations:
1) the 5534 is unstable till 26-28uS, and this seems to be the case looking at the 5534 gain/bandwidth charts
2) I have a faulty head.
One way to check that is probably to look at the signal before and after the 5534.
I'll do it next week, as soon as I can.
As for the picture of the waveform, my problem is that I need a good digital camera. I'll get organized as soon as possible.
About the 10K resistor, could it be there to avoid overvoltage to pin 2 of 5534? I have no other ideas other than this stupid
Why not replace the 5534 with a National Semiconductor LM6365? If I'm not wrong noise figures are similar but LM6365 is as fast as 1uS at a gain of 1000 and they are pin compatible as far as I understand...
Hi Luca,
That is one amplifier I have not tried. There are a lot of other considerations when going to a faster amplifier and trying to sample earlier. I doubt if the coil on the PI1000 would be any good for delays less than about 20uS. Coil design becomes more critical and a good shield is essential. I have tried fast amplifiers such as the AD8055, but one problem is that the low frequency noise is often considerably worse. One thing about the 5534 is that it is nice and quiet.
Eric.
Hi Luca,
Pin 2 is a virtual ground and with an ideal opamp, remains at ground potential. The source drives current through the 1K resistor and the output of the opamp goes negative sufficiently to give the same current in the 1M feedback resistor. Hence the voltage gain. Just like a see-saw where pin 2 is the pivot, which is much nearer one end than the other. Connecting a 10K resistor from a virtual earth to ground does nothing. Protection diodes are necessary at this point because the coil back emf is much larger than the amplifier can handle and still look like an opamp. Some devices have the diodes built in, but I always fit two external ones in case I change to a type that doesn't.
Eric.
Hello,
I did get a picture of the waveforms on my Tek oscilloscope.
In the picture you see two waveforms:
Upper) taken at Pin2 of 5534 (Vin-)
Lower) taken at Pin6 of 5534 (Vin+)
reference is GND
Upper is 1V division, Lower is 5V division, DeltaT (12uS) is measured from X to X (see picture).
My idea is that my head is 12uS fast and that the 5534 is overloaded until 30uS (measuring from TX end), otherwise I do not understand why Lower) curve takes other 10uS and more to smooth down.
Any other explanation?
Hi Luca, I like your stubborn toilling in harness, while others wait for a copy. Throw the petrified 5534 to thrash, take some opamp with high slew rate and low noise, my favorite are OP37 and better LT1028. With the better speed you will choose new damping resistor. Also check grounding on the DPS - voltage supply, ground end of coil and the reference input of preamplifier should not be distant from each other, the best place of grounding is by the ground end of limitting diodes. Good luck. Sid
Thank you very much! I was looking for the AD767 (sounds like a jet plane!) that seems fast enough to handle 10uS signals at 1000 times gain and has a very (very!) low noise figure.
It appears to me that the real problem should be the overdrive recovery time of these op amps.
Even if they should be related to eachother, overdrive recovery time should be one thing and bandwidth another one I guess.
I'm building a small test circuit (I have a function generator) to evaluate some of them.
It will take a while.
Again I'll post my results here.
One question: How do you null OP Amp output offset? I have the example connections (a trimmer) . Should I ground Vin- and Vin+ and trim for zero output at Vout?
Hello,
The ad767 is a d/a converter. What is your opinion of the ne5539?
Hi Luca,
Good picture. I now see that it is inverted to what I am used to, as the TX is the other polarity. Both waveforms are quite normal. What looks like a sharp knee and straight line on the input, is not so when amplified to a high degree at the output. There is often a long low current tail to the TX switch off, which indicates that the coil may be over damped. This was often the case in earlier PI's where no attempt was made for ultra short sample delays. I remember when 75uS seemed short.
The longest delay between the initiation of the TX switch off and the start of sampling, is dissipating the energy stored in the coil field. This is my first line of approach, which affects both the design of the TX itself and also the coil. Having got this right, then you can use a faster amplifier if required. Beware though, that fast wide band am-lifiers, although having good noise performance at high frequencies, are often poor at the frequency components we are interested in. This will result in the detector having a rough audio threshold.
The 5534 can be used, with suitable coil and transmitter, for sample delays as short as 5uS at a gain of 500x, so I only use a faster amplifier if I need to go shorter than this. So far, this has only been for special industrial uses.
Eric.
Hi Sid,
Data on LT1028 looks good. Low noise at low frequencies too. I'll get one and try it.
Eric.
Hello,
I ordered today a couple of AD797 from RS Components Italy.
I decided to try the AD797 after some specs and literature reading.
The problem I'm trying to solve is the rather strange instability of the output of my 5534 in the 20uS-30uS area.
You suggested to check 5534 bandwidth at a factor of 1000 gain as the case is in my PI3000.
I did it. In fact,on the 5534 specs, at a gain of 1000 times, bandwidth seems to be gone for signals faster than 30uS.
Since I do want to respect original PI3000 design, I need to amplify coil signal at least 1000 times.
So my first goal is to try replace the 5534 with something faster.
You also mentioned noise, so I tried to find something that could satisfy both constraints.
The AD797 is a very low noise op amp (less than half of the 5534) and it's bandwidth curve is just above 100000 Khz (10uS) at G=1000 . This should warrantee a stable signal at 15uS delay. Is all that I am saying orribly WRONG?
As for the 500 times gain suggestion, I took it seriously and , in fact , I got a second Ad797 to eventually cascade it to the first one in a Inverting->Voltage follower chain to still obtain the required 1000 times gain (I choose gain 30 for the inverting op amp and 35 for the voltage follower).This configuration will have a badwidth in the Mhzs...
The LT1028 has a good noise figure but is a little slower than the AD797.
Look at the charts for a comparison...
I hope I'm not wrong, I'm new to this kind of electronics..
Hi Luca, you are right, but theory is grey and practise is green. Both these op amps are designed to other purpose and are tested in different conditions,eg. input resistance. The 797 gets his low noise from high biased input currentt mirror, so expect high energy consuption, high offset, low temperature stability, the more, it is about ten years old. The LTs have more progressed technology and wide spectra of properties, I could not get the 1222 and 1226, also promissing ones. Only experiments will solve these conditons of high saturation and its recovery. Good luck, Sid.
Thanks , in fact I did try to drop the 797 in the 5534 socket, having checked first to avoid connecting pins 8 and 1 . What I got is a very odd output curve.
I suspect this is due to the low input impedance requirements of the 797.... I'll try the LT1028 if I can find one.
In the meanwhile I replaced the original 5534N of my PI3000 with a 5534AN that has warranteed noise figures.
Hello,
I tried the 797 incircuit with some combinations of resistors. Lowering gain ( I tested it down to G=10) or lowering the surce resistor ( I tried 10Ohm instead of 1K) led to nothing. It looks to me that the 797 is not able to recover fast from overdrive.
It remains in ovedrive condition for about 30uS before reacting violently.
This is the only explanation my poor experience can give.
I hope I can get an LT1028 and try it...
Eric Foster has posted time and again that he has tested many OP AMPs with great specs but don't recover quickly. Looks good on paper but doesn't work out for this application. As for low noise, Eric has said many times also that the noise picked up by the coil far outweighs the small noise generated in the front end op amp and so a less noisy op amp isn't going to make a difference there. It will be interesting to see what the LT does for you.
Good luck,
FJ
Luca,
Have you tryed a LM6365 IC in place of 5534.
I am looking for a better CI to replace 5534 in my
PI detectot.one that will plug into same socket.
Hows your mod comeing on white PI-3000.
Frank Serra. in US.
Hello Frank,
I did try a AD797 opamp in place of the NE5534.
Unfortunately it is unstable.
I learned that I need to connect it correctly (compensation and decoupling) otherwise it oscillates.
My mod to the PI3000 works really well.
Recently I got water inside the coil so I'm waiting for a replacement one from Whites UK.
I had then the chance to open my old coil and count the windings : 36 windings. It looks a lot for a 11" coil.
Anyway I'm now sure that the long RX recovery time (25uS)my PI3000 takes after TX pulse is due to the NE5534 overload and NOT to coil inductance or lack of shielding.
The problem is that if you want the NE5534 to recover faster you need to reduce gain (1000 in the case of PI3000), also good power supply decoupling is advisable as I understand by studing theory.
I got all the components to build part of the PI3000 schematic and try different combinations of Opamps. When I'll have news , I'll post it!
Cheers
Luca
What do you think of LM6365 is it a waist of time.
Try winding your own coil, make it light-er.Mono coils are easy to make.wind it on a non-metal template so that you can test it and re-use wire.
36 turns of wire might not be to much.I went much higher with my PI-1000, Dont go lower 24 turns,This is from memory. Experment and be safe.
Frank.
Hello,
the LM6365 is one of the OpAmps I would have tried. It should recover very fast.
Anyway I still belive that we can obtain a very fast amplification by using two NE5534 instead of one by making amplification the square root of required one.
For example, I want to try to replace the NE5534 pre amp of the PI3000 working at a gain of 1000, with two NE5534 working at gain 33 each. Lowering amplification means grater bandwidth and faster recovery time!
Cheers
I ran a magnifying glass over the whole board and nothing appears out of the ordinary as far as dis-coloured components is concerned.
Unfortunately I don't have the skills to actually mod this detector (faster delay) but I'm not too worried about that considering how I will be using it. All cable gland and O-rings will be replaced as well but don't appear to need it. I will also Plast-Dip the lower half of the coil for added insurance. These 950 Zero-Buoyancy coils are well regarded and are almost impossible to find in this condition....they are nice and heavy and require no effort to keep down.
If anyone else has any other ideas then feel free to comment. I know many of you can easily mod this detector.....having a pulse delay of around 20uS would be nice but I suspect my technical abilities is the weak point.
Thanks again.
***** I have included a Word doc that I generated from Findmall. I copy and pasted most of the relevant sections regarding the modification of the PI3000. It's quite dis-jointed but there are some ideas being thrown around. I'm not sure if there was a final outcome to all of this *****
This is the contents of the Word doc. I couldn't upload it separately. I think it is from around 2003-2004.
Hello!
I own a Whites PI3000 (the yellow one) that I'm succesfully using for beach/shore detecting.
Since I never found coins or rings deeper than a couple of inches, I'm asking myself if I can improve depth of detection by some sort of modification to its electronics.
I already replaced standard alkaline batteries (six) with a self built NiCD pack of 8 AA batteries.
I have the original Whites schematics (don't ask how I got them!) that report a possible Vin of 6.5 to 10.5 Volts at a current drain of 60 to 110 mA.
A NiCD pack of 8 AA 1,2V 800mA batteries can supply 9,6V for about 8 hours with a peak of 10.5 Volts (!) just after recharging.
I did not use NiMh batteries because I believe that power supply must be as much steady as possible expecially when a voltage regulator IS NOT employed in the electronics as is the case of my PI3000 and NiCds have the stadiest voltage supply of all.
It also appears to me that transmission power is fed to the coil directly from V+ that is derived from Vin minus 0.55 Volts circa. So my conclusion is that the higher the Voltage the higher the transmission power injected into the coil. Am I wrong?
Are there other mods to be made to the PI3000?
On the motherboard there is a variable resistor R46 100Kohm that looks to be able to trim gain of the first stage 5534 RX OpAmp. Is this a sensitivity control and what appens if I tweak it?
Happy Hunting
Luca
Being a newcomer to PI technology, it still seems as though you might be swinging the coil too quickly or perhaps, adjusting the sensitivity improperly: in Auto or too high (article below mentions retuning regularly as being necessary and Auto retuning reducing depth substantially)--which will actually reduce detection depth, since you won't be able to hear deeper signals-- Lastly, it is possible that there is something wrong with the coil or the detector itself.
Below is a link to an article about the White's PI 3000, which seems to have some good user tips.
Whell,'I'll air-test the unit and post results here. I'm using it correctly but I have no references on how deep it should go. One thing I noticed for sure is that it misses some gold, like small chains for example.
I was asking myself if someone made any modification or has someone tested it against other newer detectors.
I read tests of other brands were is claimed a detection depth of 20" on good size coins; that appears to me rocket science.
I can reach those depths only on very big targets like coke cans.
Thanbks all for the help....
Hello,
I follow my prec post with the following notes on Whites PI3000 sensitivity.
I air tested my PI3000 with some coins (euros) and a few silver,gold,copper rings and chains I found using a Whites Eagle Spectrum (very good machine) here in Italy.
The largest euro coin is the 2 Euros (worth today about 2.5 dollars) ; the PI3000 can detect it at about 10 inch while my Spectrum can go up to about 15 inch. These figures look good but gold is the real problem.
I tested a few gold rings and some (2 of them) are barely detected at coil contact!
One thin gold chain with small gold pendant and a thin silver chain are completely invisible to the PI3000.
I include a picture that shows my findings in order of detection sensitivity (right less or none detection, left maximum detection).
I'm worried I lose gold while detecting wet sand.
Did anyone test a PI3000 against other PI detectors?
Any idea?
Metal detectors see only one link of the gold chain.
This makes it hard to find gold chains. I have found only one gold chain in my 10 years of metal detecting and it was because there was a large pendant attached to the chain. It should be picking up gold rings though. You might want to send it in to Whites for a check up.
Good luck!
HH
Emmett
Luca
Lucia,
You might have to change time delay to do better on gold.Does the PI-3000, have these IC's 555 timer,4017 or 4069 IC. What part of Italy are you in.
Hello,
I'm in the center of Italy. I can reach both west and east coasts. West coasts (Rome) are heavily mineralized (black sand). Garret has recently modified its Infinium Pulse LS detector to better workout our western coasts . They call it Infinium Mediterraneo but I had no chances to try it.
So a pulse detector is a must here.
As for the PI3000 , in the schematic I see that the basic 30Khz clock is divided by 10 with a CD4017 decade divider. I don't know how long is the pulse in uS, I've no idea of how to calculate it from the schema. I'm good at digital circuits but bad in analogue.
The only thing I understood is that a pulse is transmitted every 3Khz (e.g. 30Khz div 10) but since I don't know how long the pulse is, I'm not able to determine sampling timing.
I found also the PI1000 schematic and it is basically equal to the PI3000. Same TX circuit, same RX circuit (amplifiers and everyting). Differences only in the threshold and audio so stating that the PI3000 is better than the PI1000 is simply wrong.
One idea could be to make a piggyback board to replace the CD4017 with a PIC16F84 to program timings at will (or even at the push of a button) and this is easy to do for me but first of all I have to understand all the original timings ....
Luca,
Sounds like you can repair the PI-3000, if you damage it.To lower time delay might be as simple as change-ing a Cap.My pi-1000 was modifyed by lowering a cap.white might have all ready made the change in your pi-3000.I need to see the print to know.If you have a Infinium PI, a WHITE XLT, Why bother with the PI-3000.Does it work OK in black sand.
I have a Garrett Infinium also,there is no black sand,only soft and hard packed beach sand where I live N-FLA East coast.to use the infinium here I have to adjust it back,low thershold, fast or slow,ground balance wont hold.Do you know what Garrett did to modify it.if they adjusted it back inside the detector,it might cut down on dept.
I have a Minelab SOV; nothing bothers it here on the beach.it dont go as deep,you dont dig all that trash.
Do you ever get down to Scily.My family came from there.my father's father had some kind of farm there,
they found Ceaser toom on his land a end of 1800.most likely Agustus Ceaser.The town has same name.You can research my name to locate home-site and land. The toom is in a out-cropeing of large rocks. use my E-Mail, [email protected]
I get back to you in a fue days. Frank.
Hello,
I have only a PI3000 a Whites Spectrum and a Whithes TM600 cache locator. Alla machines were made in the '90s.
Do you know where I can find the specs of the PI1000 modification ? I could check it against PI3000 schematics.
In one word I'm just worried of sensitivity to gold because I'm able to find a lot of coins but not much gold.
I'm preferring the PI3000 just because it lets me search where most of the other won't do: on wet sand and UW.
Cheers
Hello,
the PI3000 has a Clock generator made out of a CD4069 and a Clock divider (by ten) CD4017 to drive all the functions.
My opinion is that the PI3000 works really well on wet sand ; I've been able to retrieve an impressive quantity of coins, a watch and a silver ring,in a 3 days beach search last week. The only point here is to find ways to improve sensitivity to small gold. Pinpointing is good. I use always a scoop, since you can always find rusted nails to cut your fingers....
I tested pin TP2 (R12) of the clock generator.
Clock speed is 30Khz (32 in my machine) .
As obvious from the schematic, clock is fed in to Clk pin of CD4017 that acts as a Clock divider (by ten).
A TX pulse is generated by the CD4017 at Q1 pin 2. TX pulse length is 30uS.
Ater TX shutoff, sampling is started by CD4017 after a delay of 30uS (!).
Final sampling of background noise is taken 210uS after Tx shutoff.
So my conclusion is that this is a 30uS samplig detector and this well explains why it does't easily (or at all) detect small gold....
HH
Luca, 4069 is IC that will lower time delay. I modified my 1000,its not safe to tell you to make a change,with out compairing your print with my print to see if they are the same or close.Did you have any problem getting 13000 apart.make sure to use new O-rings and put water proof lub on them.
4069 IC mod; increased dept about 4".it was more sensitive to smaller non-ferrous metal and smaller gold rings.(Dont make the same mistake that I maked)I wonted a PI that would detect small nuggets.(my mistake)I lowered resastance to fast to get where I wonted to go.End results a transister got hot,detector off line now.can not get the trans;
If you lower R by putting another resister across it.Start off with a very high value resister, slowly come down to what you think it chould be.Good Luck.
Hi Luca,
Are you sure the TX pulse is only 30uS? This seems a bit short for that type of detector. Normally two or three of the 4017 outputs are OR'd together to give a TX pulse that is either 60 or 90uS wide. However, if this were done, it would not leave enough outputs to give the 210uS second delay.
If you wanted to improve the sensitivity to gold, I would double the clock frequency to give 15uS pulses out of the 4017. Then OR pins 2,4 and 7 with diodes into the transmitter drive to give a 45uS TX pulse. Take pin 1 as the signal sample pulse and pin 3 as the noise subtraction pulse. You are left with 60uS between the end of the signal sample and the start of the noise sample, which is adequate for most gold objects.
I'm not familiar with the circuitry of the PI3000 so there are bound to be other things that are not optimum, but the above is a good start on improving the sensitivity.
Eric.
Thanks for the suggestions!
I'm sure (as far as my understanding of the schematic is good)that TX pulse is 30uS.
The TX transistors are bound only to pin 2 (Q1) of CD4017 .
Signal pulse is bound only to pin 11 (Q3) of CD4017 and ground pulse to pin 7 (Q9).
So if I made it right, given a clock of 30Khz (I measured it) e.g. 30uS cycle , the cd4017 will pulse each gate for as long as 30uS.
So at Q1 we transmit (30uS), at Q3 we sample signal (after 30US of Q2) and at Q9 we measure ground signal , 7 cycles after TX pulse, so 210uS later.
My schematic of the PI3000 is very similar to the PI1000; they did't change much...
I hope I did not bother too much since this is a hold detector and , maybe, I should get a better one.
HH
Hello,
well removing the board worries me . Probably I can get it out without having to loose the front cables because it looks like there is enought free cable inside the case to let the board out.
If I undrestood your suggestion you mean I can double the clock speed working on the CD4069 and achieve a 15uS speed as Eric also suggested.
What I did not understand is how you "fried" the ZTX750 TX transistor.
My idea of a possible modification is to make a small board with a PIC16F84 processor to replace the CD4017 and program a diferent timing. This would leave the TX pulse width unchanged and not arm the TX transistor, but will the 4Mhz clock of the PIC influence the LM5534 RX first stage amplifier? Shielding required
Hi Luca
I follow your posts and have one more suggestion to adjust your timing circuit. I'm sure it works. Change/replace either R11 or R12 with a poti. One of them will work. By replacing the resistor with a poti you can adjust the clock circuit frequency.
The Clock circuit (CD4069) triggers your IC U4. With adjusting the poti you can "speed up" or "slow down" your clock circuit frequency. The clock "speed" adjustment works than like a discrimination and it gives you the opportunity to adjust your sample delay.
Chris
Hello Chris,
I know it will work, but a problem I see is that, in this way, I also shorten TX pulse lenght and this is undesiderable since , as Eric stated, the TX pulse lenght appears to be already too short (30uS); bringing it down to 15uS would limit detection depth in my opinion.
While it is a very easy modification( R12 can be replaced without disassembly of the whole detector) I still think that replacing the CD4017 would be better. I'm looking forward to see if a piggiback board for a PIC16F84 would fit.
I fried 750 by lowering R12 or R11 to fast.Not sure witch one, will check and let you know later.R11 AND R12 is same as R5-R6 on a 555 timmer.R12, start off with 470K then go lower.first thing you wont to do is lower a cap at 4069.if it has a 150 pf go to 130pf no lower.4069 is main timer.I HAD asked to see the print to compair.if the 1300 has been changed from 1000.you might frie something.I am at libary what I just said is from memery.(Large cap next to 4069 what size is it).
4017 I could be wrong,Eric siad to make a change to pin () () TO SPEAD up clock. dest-a durd-a
Frank-S
Hi Frank and Luca,
Changing the clock speed by half, up or down, should not fry anything. Whether you speed up, or slow down the clock, the transmitter current should remain within reasonable limits. That is because the pulse width and pulse rate change proportionally. In practice you will find that the TX current falls somewhat as the clock speed is raised. This is because the current is limited by the coil inductance. Assuming a coil inductance of two or three hundred uH, a 30us pulse is too short for the current to reach its maximum value, which would be limited by the circuit resistance with a much wider pulse. A 200uH coil with a total of 2 ohms series resistance, has a time constant of 100uS. To reach 95% of its final value, the TX pulse would need to be 300uS wide. A 30uS pulse will only reach about 25% of the final value current before it switches off. At this point it is still rising steeply and inducing eddy currents in the target which are in opposition to those which you want at switch off. Result; loss in detection range.
I suspect that in Frank
Hi Luca,
you can try to remove R13 from U4 PIN2 and connect the resistor to U4 PIN4 + the clock circuit mods. This mod will trigger your TX pulse later at a later stage.
Chris
Hello!
I designed, built, and tested my modification to the Whites PI3000 (veteran? toy? you name it) pulse detector.
Modification is based on the substitution of the main timing chip (CD4017) that gives the timings to the TX pulse, target sample pulse , and ground sample pulse of my PI3000.
I substitute the CD4017 with a small board that carries a PIC16F628 microcontroller at 4Mhz clock speed and some other support logic.
The board is connected inplace of the CD4017 by means of a flat cable.
The board fits nicely inside the PI3000 housing.
I have added a rotary switch so that I can select up to 10 different programmed timing combinations to eventually adapt to different ground conditions or even simulate the original timings.
Thanks to Eric Foster's suggestions I programmed a timing sequence that gives 2 more full inches of sensitivity on dimes (euro cents here).
The sequence is:
TX Pulse 60uS (originally 33uS)
RX Target delay 26uS for 15uS time (originally both 33uS)
RX Ground 50uS after RX Target and again for 15uS time (originally 165uS later ! for 33uS time).
All this to keep the original 3Khz sampling times even if I doubled the pulses and quadrupled the TX enery...
I programmed other timings as well but it's too long to list here.
I've not been able to move RX timing faster than 26uS because of the coil response in combination to the (nice) feature of automatic ground balance that the PI3000 provides.
I probably match an unmodified Whites Surf PI.
I post the schematic and I'll post a detailed "how to" only on heavy request!
The improvements reported are "on air tests" at the moment.....
Hello, it is impossible for me to make a real test here in my garage and I had no chance to try it in the open. Here I have iron everywhere.
I'll be at the beach during the weekend and I'll answer with a report next week.
The only thing I could test was that , under the same conditions ,so even inside the garage, by changing from the "original" mode to the new one I could gain about 2 inches of depth.
The improvement over the original mode is evident also with the other combinations I programmed, but only the beach will tell by how much.
There is also the possibility that the best combination for air tests won't be the best on wet sand.
I'll be back!
Greethings
Hi Luca,
Your mod; looks good. looks like you re-invented the wheel.2" increase not to good after all that work.I am sure that you will get it to go deeper.your mod is very impresive.
Reg,helped me with mine.took me about 5 mi to install a cap.with a 3" increase in dept on US nickel,Dont forget I later ruined PI-1000.dont have your skills.Now you need to find gold with it.
Been detecting over 40 yrs. might be able to help you there.Try what I call a military grid pattern on your beach. At low tide start about 70 ft above low water.walk in strate line down to water,keep coil flat about 1"from ground,keep it flat to end of your swing.at end of swing push coil forwards(still flat) half diameter of coil come back across with you swing.Do this both ways left and right.this is more than enough over lap for your PI.
The object is to locate all target.if you miss one you step on a mine and blow-up.
Now you need to walk in a strate line down to water and back up 70 ft. then back down to water until you cover the entire beach.I have watched a lot of people over the years, most of the can not walk a strate line.To over come this drag your foot in sand then look back,this will keep you strate.
Dont forget each time that you turn around on each approach, step over abut half length of your swing,and go back in strait line. good luck
Frank.
Hello Frank,
thanks for the suggestions.
Can I ask you which cap did you change on the PI1000?
Chris suggested to try to change the preamplifier opamp feedback resitor (1MOhm) to 1.2 or 1.5 MOhm (see PI1000 schematic below) to try to have a higher preamp gain.
That, in my opinion, should be changed with a potentiometer to adjust gain in accordance to ground conditions but this would mean to carry around wires that could pick up noise and get the situation worse
.
Anyway I did not try it yet, because I first want to evaluate my modification and play with it for a while.
I have to thank Eric Foster for his detailed articles about PIs that sparked my entusiasm and willing to share it with you all.
Hello,
I think that the 5.6 Ohm resistor in series with the coil has been put there by Whites to limit current flow in the Ztx750 transistor that has a maximum current rating of 2A. This is my opinion, maybe Eric has a better explanation....
<img src="/metal/html/confused.gif" border=0 width=15 height=22 alt=":?">
HH
Luca,
I am at public libary, might have told you about cap a month ago.this is from menery.Its the main cap at 4069 IC.might be 150pf dont go lower 120pf.You have all ready made a mager change.dont mess with this untill you finished your mod.
I think that Cris was talking about lowering the resistor at 4069 timer.its a 5 mi job.all you do is put one resistor on bottom(parall)of resistor.this will seedup timer and lower time delay.you can also put a pot ther for a non-ferrous ID.no need to remove resistors.
Frank.
Hi Luca,
Glad the technical info on this forum got your enthusiasm going for doing a bit of experimenting.
The 5534 in your schematic is running at a gain of 1000, which is more than I normally use for a front end. You need to look up the gain/bandwidth curves for the 5534 to see where you are sitting at the moment. Increasing the gain will reduce the bandwidth and may affect the earliest point that you can sample. The reason for this is that the amplifier takes longer to come out of saturation after the transmitter pulse. Using a 5534, I find that a front end gain of about 500 is about as high as I would want to go, to be able to sample at 15uS. If you were sampling later than that (which the White's did), then you can have more gain and a longer recovery.
Having leads and a pot to alter the front end gain is not a good idea. This will add capacitance to the feedback circuit which will result in even less bandwidth than you would expect from the curves.
I don't see the point of the 10K resistor from pin 2 to ground in the schematic.
The 5.6 ohm resistor limits the maximum coil current and improves the coil circuit time constant, so that the switch on current is not still rapidly growing at the point it is switched off.
Eric.
Hello Chris,
I have a 10 positions rotary switch to select one out of ten different timing programs.
I choose the following to start with:
N. TX Dly Target Dly Ground (all times uS)
0 33 33 33 165 33
1 33 30 33 165 33
2 33 28 33 165 33
3 33 26 33 165 33
4 66 33 33 165 33
5 66 30 33 165 33
6 66 28 33 165 33
7 66 26 33 165 33
8 33 26 15 50 15
9 66 26 15 50 15
Prog 0 is same as the unmodified PI3000.
1 to 3 just a sampling speedup.
4 to 7 double TX time
8 and 9 are special and also the most sensitive.
If you look at the numbers in prog. n. 8 and 9 I shortened the whole Tx/RX cycle by sampling for only 15uS and allowing a 50 uS only delay between Rx target sample and Rx ground sample. Apparently this should give HALF sensitivity but since this whole cycle is long HALF of the original one (e.g. prog.n. 0) this one is executed twice in the same time period.
In other words the sampling capacitor is charged always by the same time length but the signal is taken more toward the 26uS area(! <img src="/metal/html/wink.gif" border=0 width=15 height=15 alt="
I remind that I can't go faster than 26uS because I have a strange curve in the 20 to 26 uS that goes in the minus area (below ground!).
Probably this is an effect of the slow recovery speed of the 5534 Op amp as suggested by Eric. I'll try to lower it's gain to 500 times and see if this beaviour vanishes. In that case my programs will change again....
Sensitivity is really improved. I gain 2 inches on any target and I noticed a greater sensitivity to very small targets as well.
No doubts about it.
Cheers
Luca
I'll check the 5534 and half its gain to see what happens.
I admit my curve, measured at the output of 5534 is high and flat between 0 to 13us, decays to ground between 13 and 20uS, goes BELOW ground from 20 to 26-28 uS and is around ground from 28uS on.This prevents me to sample faster than 26uS.
Do you mean that I should try also to remove the 10K resistor to pin 2 and ground of 5534?
Hello,
it is not exactly what you asked, anyway I tested a solid 18K gold ring with gems in dry sand and I spot it at true 10 inches depth.
My modification does not make miracles but works!
Hi Luka,
thanks for the update. You should try to reduce the gain of your receiver like Eric suggested. This might help you to gain m
Hi Luca,
I am a bit puzzled when you say that the waveform is high and flat from 0 - 13uS. This is the period when the transmitter is on, and is usually more than 100uS for an underwater detector. At the end of the TX pulse, the waveform will swing negative due to the reverse emf generated in the coil at this time. You can see this emf if you put the scope directly on the coil. It appears as a high voltage spike at the end of the TX pulse. When you look at the coil, everything is reversed, compared to the amplifier output, as the amplifier inverts the signal.
If you try a lower gain in the 5534, you could increase the gain in the dc amplifier to compensate. Can you post a scope picture of the 5534 output waveform?
Take out the 10K from pin 2 to ground. Pin 2 is referenced to ground by the other resistors connected to it.
Eric.
Hello,
I explain...
I measure at the output of 5534, pin6.
All times are measured from TX pulse END.
After a 33uS TX pulse from 0uS to 13uS I see a flat high signal (diodes levelling?) that starts to drop in a curve between 13.8 and 20 uS. From 20 to 26uS signal goes below ground and levels at ground after 27uS.
I can't sample between 20 and 26uS because the PI3000 goes unstable, mute or almost unsensitive in that area.
I have two explanations:
1) the 5534 is unstable till 26-28uS, and this seems to be the case looking at the 5534 gain/bandwidth charts
2) I have a faulty head.
One way to check that is probably to look at the signal before and after the 5534.
I'll do it next week, as soon as I can.
As for the picture of the waveform, my problem is that I need a good digital camera. I'll get organized as soon as possible.
About the 10K resistor, could it be there to avoid overvoltage to pin 2 of 5534? I have no other ideas other than this stupid
Why not replace the 5534 with a National Semiconductor LM6365? If I'm not wrong noise figures are similar but LM6365 is as fast as 1uS at a gain of 1000 and they are pin compatible as far as I understand...
Hi Luca,
That is one amplifier I have not tried. There are a lot of other considerations when going to a faster amplifier and trying to sample earlier. I doubt if the coil on the PI1000 would be any good for delays less than about 20uS. Coil design becomes more critical and a good shield is essential. I have tried fast amplifiers such as the AD8055, but one problem is that the low frequency noise is often considerably worse. One thing about the 5534 is that it is nice and quiet.
Eric.
Hi Luca,
Pin 2 is a virtual ground and with an ideal opamp, remains at ground potential. The source drives current through the 1K resistor and the output of the opamp goes negative sufficiently to give the same current in the 1M feedback resistor. Hence the voltage gain. Just like a see-saw where pin 2 is the pivot, which is much nearer one end than the other. Connecting a 10K resistor from a virtual earth to ground does nothing. Protection diodes are necessary at this point because the coil back emf is much larger than the amplifier can handle and still look like an opamp. Some devices have the diodes built in, but I always fit two external ones in case I change to a type that doesn't.
Eric.
Hello,
I did get a picture of the waveforms on my Tek oscilloscope.
In the picture you see two waveforms:
Upper) taken at Pin2 of 5534 (Vin-)
Lower) taken at Pin6 of 5534 (Vin+)
reference is GND
Upper is 1V division, Lower is 5V division, DeltaT (12uS) is measured from X to X (see picture).
My idea is that my head is 12uS fast and that the 5534 is overloaded until 30uS (measuring from TX end), otherwise I do not understand why Lower) curve takes other 10uS and more to smooth down.
Any other explanation?
Hi Luca, I like your stubborn toilling in harness, while others wait for a copy. Throw the petrified 5534 to thrash, take some opamp with high slew rate and low noise, my favorite are OP37 and better LT1028. With the better speed you will choose new damping resistor. Also check grounding on the DPS - voltage supply, ground end of coil and the reference input of preamplifier should not be distant from each other, the best place of grounding is by the ground end of limitting diodes. Good luck. Sid
Thank you very much! I was looking for the AD767 (sounds like a jet plane!) that seems fast enough to handle 10uS signals at 1000 times gain and has a very (very!) low noise figure.
It appears to me that the real problem should be the overdrive recovery time of these op amps.
Even if they should be related to eachother, overdrive recovery time should be one thing and bandwidth another one I guess.
I'm building a small test circuit (I have a function generator) to evaluate some of them.
It will take a while.
Again I'll post my results here.
One question: How do you null OP Amp output offset? I have the example connections (a trimmer) . Should I ground Vin- and Vin+ and trim for zero output at Vout?
Hello,
The ad767 is a d/a converter. What is your opinion of the ne5539?
Hi Luca,
Good picture. I now see that it is inverted to what I am used to, as the TX is the other polarity. Both waveforms are quite normal. What looks like a sharp knee and straight line on the input, is not so when amplified to a high degree at the output. There is often a long low current tail to the TX switch off, which indicates that the coil may be over damped. This was often the case in earlier PI's where no attempt was made for ultra short sample delays. I remember when 75uS seemed short.
The longest delay between the initiation of the TX switch off and the start of sampling, is dissipating the energy stored in the coil field. This is my first line of approach, which affects both the design of the TX itself and also the coil. Having got this right, then you can use a faster amplifier if required. Beware though, that fast wide band am-lifiers, although having good noise performance at high frequencies, are often poor at the frequency components we are interested in. This will result in the detector having a rough audio threshold.
The 5534 can be used, with suitable coil and transmitter, for sample delays as short as 5uS at a gain of 500x, so I only use a faster amplifier if I need to go shorter than this. So far, this has only been for special industrial uses.
Eric.
Hi Sid,
Data on LT1028 looks good. Low noise at low frequencies too. I'll get one and try it.
Eric.
Hello,
I ordered today a couple of AD797 from RS Components Italy.
I decided to try the AD797 after some specs and literature reading.
The problem I'm trying to solve is the rather strange instability of the output of my 5534 in the 20uS-30uS area.
You suggested to check 5534 bandwidth at a factor of 1000 gain as the case is in my PI3000.
I did it. In fact,on the 5534 specs, at a gain of 1000 times, bandwidth seems to be gone for signals faster than 30uS.
Since I do want to respect original PI3000 design, I need to amplify coil signal at least 1000 times.
So my first goal is to try replace the 5534 with something faster.
You also mentioned noise, so I tried to find something that could satisfy both constraints.
The AD797 is a very low noise op amp (less than half of the 5534) and it's bandwidth curve is just above 100000 Khz (10uS) at G=1000 . This should warrantee a stable signal at 15uS delay. Is all that I am saying orribly WRONG?
As for the 500 times gain suggestion, I took it seriously and , in fact , I got a second Ad797 to eventually cascade it to the first one in a Inverting->Voltage follower chain to still obtain the required 1000 times gain (I choose gain 30 for the inverting op amp and 35 for the voltage follower).This configuration will have a badwidth in the Mhzs...
The LT1028 has a good noise figure but is a little slower than the AD797.
Look at the charts for a comparison...
I hope I'm not wrong, I'm new to this kind of electronics..
Hi Luca, you are right, but theory is grey and practise is green. Both these op amps are designed to other purpose and are tested in different conditions,eg. input resistance. The 797 gets his low noise from high biased input currentt mirror, so expect high energy consuption, high offset, low temperature stability, the more, it is about ten years old. The LTs have more progressed technology and wide spectra of properties, I could not get the 1222 and 1226, also promissing ones. Only experiments will solve these conditons of high saturation and its recovery. Good luck, Sid.
Thanks , in fact I did try to drop the 797 in the 5534 socket, having checked first to avoid connecting pins 8 and 1 . What I got is a very odd output curve.
I suspect this is due to the low input impedance requirements of the 797.... I'll try the LT1028 if I can find one.
In the meanwhile I replaced the original 5534N of my PI3000 with a 5534AN that has warranteed noise figures.
Hello,
I tried the 797 incircuit with some combinations of resistors. Lowering gain ( I tested it down to G=10) or lowering the surce resistor ( I tried 10Ohm instead of 1K) led to nothing. It looks to me that the 797 is not able to recover fast from overdrive.
It remains in ovedrive condition for about 30uS before reacting violently.
This is the only explanation my poor experience can give.
I hope I can get an LT1028 and try it...
Eric Foster has posted time and again that he has tested many OP AMPs with great specs but don't recover quickly. Looks good on paper but doesn't work out for this application. As for low noise, Eric has said many times also that the noise picked up by the coil far outweighs the small noise generated in the front end op amp and so a less noisy op amp isn't going to make a difference there. It will be interesting to see what the LT does for you.
Good luck,
FJ
Luca,
Have you tryed a LM6365 IC in place of 5534.
I am looking for a better CI to replace 5534 in my
PI detectot.one that will plug into same socket.
Hows your mod comeing on white PI-3000.
Frank Serra. in US.
Hello Frank,
I did try a AD797 opamp in place of the NE5534.
Unfortunately it is unstable.
I learned that I need to connect it correctly (compensation and decoupling) otherwise it oscillates.
My mod to the PI3000 works really well.
Recently I got water inside the coil so I'm waiting for a replacement one from Whites UK.
I had then the chance to open my old coil and count the windings : 36 windings. It looks a lot for a 11" coil.
Anyway I'm now sure that the long RX recovery time (25uS)my PI3000 takes after TX pulse is due to the NE5534 overload and NOT to coil inductance or lack of shielding.
The problem is that if you want the NE5534 to recover faster you need to reduce gain (1000 in the case of PI3000), also good power supply decoupling is advisable as I understand by studing theory.
I got all the components to build part of the PI3000 schematic and try different combinations of Opamps. When I'll have news , I'll post it!
Cheers
Luca
What do you think of LM6365 is it a waist of time.
Try winding your own coil, make it light-er.Mono coils are easy to make.wind it on a non-metal template so that you can test it and re-use wire.
36 turns of wire might not be to much.I went much higher with my PI-1000, Dont go lower 24 turns,This is from memory. Experment and be safe.
Frank.
Hello,
the LM6365 is one of the OpAmps I would have tried. It should recover very fast.
Anyway I still belive that we can obtain a very fast amplification by using two NE5534 instead of one by making amplification the square root of required one.
For example, I want to try to replace the NE5534 pre amp of the PI3000 working at a gain of 1000, with two NE5534 working at gain 33 each. Lowering amplification means grater bandwidth and faster recovery time!
Cheers
Comment