LM394 on my Minelab PCB

[ZED]

Mario on your receive fets, coil signal goes to the drain of the n fet and then from the source of the n fet to the drain of the p fet and then from the source of the p fet to the input of the preamp.The gates on both fets go to the 4093 chip.

There may or may not be a resistor connected to the gate of both fets

Whenever a fet is incorrectly wired through it in the bin.

The original 2100 uses a lm394 in a descrete preamplifier cct,the 2100v does not use a lm394 but may use a 797 instead.

The pinout for a to220 package is different than for a to92 packege.

Pull both fets of the board and check the signal going to the gates of both fets from the 4093 chip and see if your getting the correct signals...report back to me !!!

Zed

[mario]

Hi Zed!

As you suggested, I removed:
P fet
N fet
4093 IC from the board

With all 3 above chips removed, I get:
0V on 3,4,11 of IC 4093, or Z1 of N fet
0V on 2,5,10, 13 of IC 4093 or G of P fet
11V on 1,6,7, 12 of IC 4093





3 questions before I solder all new chips back to the pcb:

1)On N fet what's the proper order of SDG?

X1, Y1, Z1 == ???

2)On P fet what's the proper order of SD?

X2,Z2 == ???

3) is the pin order of IC 4093 correct?

4) any other steps you would like me to follow?


Thanks!!!

For higher resolution click here

[ZED]

Put your 4093 on first(makesure the bevelled edge along the chip is around the right way)once done connect your scope to the outputs of the 4093 and make sure your getting 11 volt pulses that are similar to the transmit pulses.

The fets dont go on till your sure your getting the right pulses to the gates

Study the pinouts in the pdf of your receive fets,it shows the pins and from what view (top)

See revised picture

The pin order for the 4093 is incorrect,put it in like that and it will blow,turn it around,all the ic chips are alighned this way.Number one pin is nearest to the large electro caps,this applies to all the chips on the board.

If you cant see the bevelled edge on the ic then dont put it in.

Zed

[Mechanic]

Well spotted Zed, that 4093 orientation is a trap!

Mario, what's with the burned spot to the right of the 4093?

Cheers Mick

[mario]

2SK216!

I knew that N fet was either 216 or 218. I just couldn't make out the initial 3 digits. I tried very hard to identify that IC.

Thank you Zed!

...all the ic chips are alighned this way.Number one pin is nearest to the large electro caps,this applies to all the chips on the board.

If you cant see the bevelled edge on the ic then dont put it in...

That quote reminds me of something:

For a long time I worked at shop rebuilding engines. It was always frustrating having to look up old diagrams to figure out cylinder order of old blocks. On a 4 cylinder engine the number 1 cylinder is easy to spot because it's at the front of the engine. But what if you have a 6 or 8 cylinder engine? The number 1 cylinder could either be on the left or right bank. It's always a good idea to label your number 1 cylinder if you're putting it back together after a rebuild.

However, what if you have an unidentified block?

One day this old mechanic taught me this great trick. He guided me to look at a block down from above. If you look at both banks, you will always see that one bank will always be closer to the crank pulley (This arrangement exists because all cylinders are in a determined order along the crank).

That's how you find your number 1 cylinder. And It applies to ALL 6/8 cyl engines.

The beveled edge trick is awesome! I know it will not apply to all IC's, but when it is indeed found, It will save me a lot of hard work when replacing chips!

I haven't gotten all ic's just yet, and now that I have identified that Fet, I will be replacing it with a newer part also. In short, I will update in a few days.

Again, thank you

-=======================
Hi,
Mick

To answer your question. I used your technique of applying heat from one side of the IC, to the next while prying the chip out. However, The last top-right most pin gave me a bit of a difficulty. I placed the edge of the hot metal plate that I was using at an angle and left an imprint at that particular location. It's just a surface burn, nothing serious (for once that white gunk was useful, since it absorbed all that heat! )

[mario]

Updated links

Sorry about the broken links, they should be working now:

High Res photos:
Here
and Here

[mario]

Question about CD4093

Hi, Mick, Zed,

I got my 4093 chips today!

I ordered two different chips:
Texas Instruments ,CD4093BM and a new part by NXP: HEX4093B

Both are pin for pin compatible. After comparing differences, the HEX4093B drives a pin output low sink current (Lol) at 1.3mA (5.2mA total) Vs the 2.6mA (10.4mA total)of the CD4093BM, so the HEX4093B not be a fit? And I only mention this because I remember Mick suggesting I needed one that could drive a 8.8mA total.

According to the data sheets, 4093 is a wave shaper, so one good thing that might come out of this is the response to the big metal objects. With this newer part, hopefully the out put from the headphones won't be one loud, head blowing garble anymore. Instead, it would be nice if the noise is a very nice/stable flutey high sound.

After my surprise discovery of the 2SK216, which seems to be the most appropriate part with no worthy replacement even though it was released back in 1996(! ), I'm a little hesitant to replace the 4093.

Which of the two 4093s IC's should I drop in my PCB?
Or do you know of an improved replacement (sorry Mick, I tried looking for MC14093B, which you suggested, but I couldn't find it in stock anywhere)?
===========

ALSO, HAS ANY ONE FOUND A BETTER REPLACEMENT FOR THE 2SK216 N FET?!?!
This part is 14 yrs old!!!


The specs could be found here.

Quick overview of what to match:
V= 200
Input capacitance Ciss 90pF
Reverse transfer capacitance Crss 2.2pF

==================================================
So far on the front end, we have found:
- a noticeable replacement for the P fet with the TP0610KL.
- a some what ok replacement for the AD797 with the LT1028 (the AD797 already appears to be a really super quiet chip)
- noticeable replacements for the NE5534 chips, also with the LT1028s (I will replace them once I get 4093,797 and P/N fets working stable)
- questionable replacements for the 4066 chips with the MAX chips (which I will also eventually replace just for the hell of it)

I'm affraid not much can be improved on the stock 2100V2 when it comes to the front end unless we find some what else to improve

[Mechanic]

Hi Mario,

But what is the on resistance for the 2sk216? The capacitance is very good.

You will have to trawl through datasheets from your suppliers and check the ma rating of the outputs. There are 6.2ma and 8.8. Try and find 8.8ma.

Also bear in mind the power supply arrangement. Batt+ is gnd, batt- is -7v then referenced to gnd(batt+) you have 5+ and 5-v. The +- 5v supply powers all the opamps and some of the logic cct too. -7v is used to drive the coil. So when ever you do measurements with your scope make sure you have the earth lead hooked up to coil gnd(batt+).

Cheers Mick

[mario]

I tried in vain finding Resistance values for the 216. They're not listed anywhere on the old Renesas (bought by Hitachi) data sheets. My guess is that it's >10 OHMs as most 9 amp parts. I measured Resistance values with my LCR just to see what I would get and compared them to the IRFU214PBF (which is a 2 OHM part) and got significantly higher figures. Still, reading forums at audio fanatic sites suggests that this is a really good chip with no worthy alternative. It's really difficult to match those capacitance numbers.

Oh well, at least DYI guys with the SD2000 have something to do if they're considering updating their detectors. As for me, now that I know the proper N/P fet "SDG" order and correct orientation of the 4093 IC, I will continue experimenting.


In regards to the 4093 IC, I found some 8.8mA parts and have them on order. However, I can't wait for them to arrive (another 5 days, since I placed the order at the end of the week!) and will experiment with what I have. I figure I could remove the chips and solder back the new 8.8mA ICs once I get them (5.2mA Vs 8.8mA, Vs. 10.4mA Vs 8.8mA, not much difference. I would like to get rid of that annoying garbled noise on the big objects).


Also, you are right about the power arrangement: " Batt+ is gnd and batt- is -7v" was really confusing me. If you notice my last update on the picture showing the 4093 inputs, I have 4093 pin 7 going to "case ground." I did so intentionally because when I used my ammeter, I wouldn't get the right polarity if I would connect pin 7 to battery +....It really drove me nuts because I couldn't figure out why that was the case. I worried there thinking I had blown a diode in addition to the 4093. In any case, I placed the negative ammeter terminal to the case ground and got the correct voltages/polarities -which turned out to be the case with pin 7. Even now that I write about it, it still doesn't make sense. All I know is that the pin 7 is correctly identified in terms of voltages and it goes to the case ground -it's the only way I can get the polarity of batt+ terminal. In addition, this is only the case when the detector is powered on!

Anyways, some of the guys on the audio forums have suggested improved performance by having dedicated power rails....which might be another project. Personally I think we will get improved sensitivity by just replacing the main processor with a faster one.... oh well, first thing first...

[mario]

Ok guys, sorry for the lack of updates. Everything is up and working now after several set backs.

I had run into problems after soldering back the original parts. I soldered back the factory N and P fets to see if the lower amperage 4093 IC was compatible. My initial impressions were not that particularly positive since I seemed to have gotten serious degradation in sensitivity. However, After reviewing my work, I noticed solder run off from my soldering area( on the other side of the pcb where the N/P fets are located) was shorting some nearby components.

Also, flux from my original soldering was shorting pin 6 and 7 of IC 4093. I went to an auto parts store and got some brake cleaner. Then I wiped all 4093 pins cleaned with a cotton ear swab.

Most Issues are now resolved with 1 exception. I'm now getting very very very faint noice from my audio amp. So faint, it's not being detected on my oscilloscope. however, I think this can be attributed to my replacing the 100ohm resistor with a 10ohm part (the resistor between the N fet and 4093 IC). I decided to replace that resistor, because the 2000SD schematics show that resistor to be 10ohms. I probably need to replace it to 40 OHMS?

I don't know just yet. I need to go out and test my unit. I haven't been able to make time. Seems like days are getting shorter on these winter days.

I'll update the that resistor and then the NE5534 chips next.

I don't think I'll update the AD797 chip since the LT1028 is very close to it in performance...


Last thing I would like to mention: It would be awesome if some one could help me identify those remaining chips on the PCB (the ones with the "??"). It's looking like we are half-way there!!!

Thanks for all the help Mick!

[mario]

Finally solved my noise problem!

It was driving me nuts.

I thought I had solved the issue by replacing the 10 ohm resistor with a 50 ohm part, since i didn't hear the faint static noise any more. Then, after assembling the unit back into its case, It came back some intermittent.

So I go back over my work, remove excess solder from every soldering point, clean all soldering points with my brake cleaner, replace all N fets, P fets, Diodes with new original parts (I have now everything in stock) and the damn noise just wouldn't go away!

After spending two days on this issue, I was pissed.

Finally, after what might have seemed the fifth time of redoing everything, I notice a pattern: the damn noise would only come back after assembling the unit.

Thinking that the lower amperage 4093 IC wasn't adequate to drive the 9amp N fet, I resolved on replacing it. However, before replacing this chip, it occurred to me to test the unit further outside its cover. After taking all precautions, with both side control panels off, I turn the unit on. I notice no noise....then, as soon as I put the control panel (the one with the coil) into it's place the noise comes back again!

It seemed surreal: the noise would only come back after dealing with that side of the control panel.

So I look into the case and notice the following: two of the four wires coming from the coil (the ones that go into the toroid) where rubbing against the IC that is literately less than 4 mm above the toroid.

I placed some silicone to cushion the wires from rubbing against the IC and no more noise!

I just came back from testing the unit, and after several hours, I can now with confidence say that the HEX4093 with only 5.5amps is a compatible IC. I know the 4093 IC shouldn't really affect performance at all, since it's sole purpose is to drive both N P fets. However, after testing, I have to say that the chip gives a sharper response. It's hard to explain, initial and cut-off time of target triggers seem to be tighter...but not by much...

Much more noticeable improvement will come later.

While I was trying to resolve my noise problem, I tested output signals from all NE5534 chips. I can verify that all 3 chips have the same function. They all process sound. However, I did study the oscilloscope wave patterns and did notice the following between the 2nd and 3rd chip: there's a slight degradation of the wave exciting out of pin 6 from those last ICs. In fact, I had to change sweep speeds on my time base to get the wave to focus. The saw tooth looking waves eminating out of the 797 IC looked sharp and distinct. Meanwhile, the ones coming out of those three chips did not resolve as nicely.

In other words, I predict solid improvement by replacing those 3 NE5534 chips.

Can some one answer me the following: why are 3 NE5534 ICs used? wouldn't it have been more realistic to just use the AD797 along with another IC to amplify the signal?

[Mechanic]

Hi Mario,

They use them because of the source impedance. AD797 has a really good noise figure for input impedance 400 ohms to 0 ohms, the closer to 0 ohms the better. The ad797 datasheet shows a table of source resistance vs noise. The ne5534 has a good noise figure for higher input impedance, these stages have between 1 to 5k source resistance so the 5534 is a better option.

Cheers Mick

[mario]

Thank you Mick,

so it sounds like the LT1028 will be a slightly better replacement, but only for the AD797.

Now that I got my machine stable I will be replacing the AD797 and search for a higher impendance chip to replace the 5534 chips later.

What impendance and/or other characteristics should I match?

[mario]

Oh, nevermind, you answered my question...

[mario]

Oh, nevermind, you answered my question...So i should not even try replacing those 5534s with the LT1028s?