Whites Surf PI schematic

[bklein]

I recently attended a tutorial by the guy that wrote LTspice, sponsored by Nu Horizons: http://www.nuhorizons.com/xpresstrack/ltcswitcher/
They indicated that the Nu Horizons apps guy would help with any LTspice issue. Did you try contacting Linear Tech to see what they had to say?
I had not heard of the CMOS part issues. Bizarre. I share an interest in electronic music circuit design with a mail list group I'm in and they use LTspice quite often to model various module circuits that I'm pretty sure use cmos parts. I'll post a question on that list to see what they say.
Do you have a file I can forward to whatever "guru" I may find?
I am a newbie with LTspice. Just tried a simple mosfet switch circuit for switching hard drives that we use at work and used the nmos and pmos models. Don't work at all. Use real parts and it does. So what are those generic models for?

[Waikiki_Sweep]

Hi guys,
I plan on making this detector due to its simplicity etc. but ....
I recently acquired a Surfmaster PI - the real deal, with 9.5" coil.
I would like to maximize its sensitivity and really do not see anywhere the trim procedure. Can someone do a writeup of the trim procedure? I have a scope etc. and would like a description of test points and what to look for... What should the open air sensitivity be on a US cent or dime? Don't have any EU coins handy....

Hello Barry,

It is great to hear that you have original Surf Master PI.
Can you please provide a basis for compartion before you have changed anything in the schematics.

1. Distance for loud and clear signal.
2. Distance from where you can hear first quiet tones for the target.

if you have any:

1 Euro -
Euro 50 cents -
US 25 cents -
US 10 cents -
gold ring (small 3 mm.) -

What current your device consuming from batteries? (mA) It will show how device matched to the coil.

Original Surf Master PI data is very interesting for everyone here. So we can compare quality of our detectors with original to check which changes were really useful.

Thank you.

[porkluvr]

Did you try contacting Linear Tech to see what they had to say?
I had not heard of the CMOS part issues.
<snip>
Do you have a file I can forward to whatever "guru" I may find?

I don't think this is really an LTspice issue because Linear Technology Corporation does not sell CD4000 logic ICs. LTspice originally started as "SwitcherCad" (they have the most comprehensive line of SMPS controllers, imho) and it is only by accident that LTspice caught on for other uses besides modeling LTC switching regulators.

Here is the URL for the LTspice-Yahoo group. That is where I obtained the CD4000 family models. You will need to register before you can post questions or download extended libraries. http://tech.groups.yahoo.com/group/LTspice/

Helmut Senewald is the group's moderator and also the author of the CD4000 models that you can find at the website. I have thought about presenting the schmitt trigger problem to him but I don't want to bother him. I have asked him other questions and he always seems to have a ready answer but I've been willing to accept this shortcoming of the CD4000 models because they have been such a great help in other respects. I do not have an oscilloscope and don't have the Surf PI circuit breadboarded, so I can't observe the SPI waveforms first-hand. Also, nobody in these forums has posted oscillograph photos that I know of and so I can not present evidence that the models do not function correctly!! I don't have a leg to stand on so to speak, and so I hesitate to question Helmut's work. Call me chicken$hit if you will.

I'll post simulation files that I use but it will probably be tomorrow before I can get everything together. It is not a snap because support files need be collected along with the simulation file. You have an oscilloscope and can make personal observations so maybe you could ask Helmut (or anyone of your choosing) if there is an adjustment that can be made to the schmitt trigger model.

There may be people out there who wonder why even bother with LTspice when there are other simulators that come with more complete parts libraries. Well, it's free, for one thing. So unless you have a lot of money to spend or are willing to use bootleg software, then you are limited by availability. Then, there are those who believe that despite the limited libraries that come with LTspice, it is still superior to other simulators. Of course that is weighed heavily by opinion but I can say that after using PSpice Schematics and also Protel simulators, I prefer LTspice.

So what are those generic models for?

I'm not sure I understand the question... which generic models are you referring to?

[bklein]

Waikiki - sure, I'll try and get what I can for you in this regard after I finish my honeydoos today. I thought all that kind of work has already been done but not shared on the group... :-)

PorkLuvr - I did a search of the LTspice yahoo group messages and came up with:

------------------------------------------------------------
> I tried to use CD4093 (2 inputs NAND from your CD4000.lib) in my
> schematics and I found a small problem. If I connect (trough a
> resistor) negative voltage to the input of the NAND there is no
> (diode) limitation. I expected -0.6 V on the input. Try to connect -
> 15 V through a 50kOhm resitor to the NAND input. There will be
> around -14.5 V, not -0.6V. It needs "diode" protection, if we want
to
>
> simulate "real" situation. Using mixed (analog and digital) parts
in
> the circuitry is mostly a real situation, it is not rare!
>
Hello Peter,
the CD4000-library is already prepared for the input diodes.
I just left it out(commented off the lines) to increase the simulation
speed. You can enable the two lines. The input and output models are
at
the end of the CD4000 library file.
--------------------------------------------------------------
There's more after this, I just cut it short. Did you enable the input diodes?

Also there are suggestions of using other parts to kludge one or a 7400 series equivalent...

The point was made in the seminar that LTSpice has a blazillion downloads and thus is the most popular simulator out there. And he says any bugs are fixed immediately once proven. So as long as LinearTech wants to support it so well I suggest we milk it for what its worth.... He even made up a simulation of the orignal Philbrick op amp just to show it could be done... and **** off Bob Pease a bit.

[porkluvr]

Waikiki - sure, I'll try and get what I can for you in this regard after I finish my honeydoos today. I thought all that kind of work has already been done but not shared on the group... :-)

PorkLuvr - I did a search of the LTspice yahoo group messages and came up with:
<snip>
>
Hello Peter,
the CD4000-library is already prepared for the input diodes.
I just left it out(commented off the lines) to increase the simulation
speed. You can enable the two lines. The input and output models are
at
the end of the CD4000 library file.
--------------------------------------------------------------
There's more after this, I just cut it short. Did you enable the input diodes?

I have enabled the diodes you speak of.

The problem still remains. With the DISC pots set at or around the mid position the first one-shot in the chain (U7) will trigger and also U4 triggers but I
can't get the bottom leg, U3 and subsequent U2 to trigger.

Even worse, if I replace the TIP32C with IRF9640, the top leg does not trigger, only U7.

If anybody figure out why LTspice mistakenly shows this circuit to be dysfunctional I would love to know about it.
I don't particularly like posting broken simulations and this one is broken. You can see that v(trig2) does not swing low enough to trigger
the bottom leg of the sample timer circuit.

If I left out something out of the zip necessary to reconstruct this simulation please let me know and I'll try to make it right.

[bklein]

So this is fun! Learning LTspice and detecting circuits simultaneously...
I did have some startup issues getting the simulation to run as files were not located in the same place as your system. Also it seems my 555 model was different - not sure. I had to delete what came up and manually insert another to get that working. Then it seems the 4093 model is hosed like you say. The ouput seems maybe open drain as the output of U7 for example is only good if connected to the rest of the circuit. If I delete connections to the capacitors and look at the output it is then incorrect. I don't understand subckt files yet and what the parameters mean - but I guess the problem is somewhere in the text below?:
* 2-input NAND SCHMITT trigger
* tpd 180n
* tr 100n
.SUBCKT CD4093B A B Y VDD VGND vdd1={vdd} speed1={speed} tripdt1={tripdt}
.param td1=1e-9*(180-40-10)*5/{vdd1}*{speed1}
*
XIN1 A Ai VDD VGND CD40_IN_S_1 vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
XIN2 B Bi VDD VGND CD40_IN_S_1 vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
*
A1 Ai Bi 0 0 0 Yn 0 0 AND tripdt={tripdt1} td={td1}
*
XOUT Yn Y VDD VGND CD40_OUT_1X vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
.ends
My manual is at work. I'll print another copy for home tomorrow.

I tried my PI in the hills yesterday as my wife wanted me to go with her and walk the dog there. Iron crap, bottlecaps, wires... all over But depths not more than 5" or so. I still need to take air tests. I fixed a Tektronix current probe and took it home from work today so I can take current measurements along with circuit test points... I read of depths reported like 10-12" of some things which leads me to believe my unit needs adjustment.

Barry

[porkluvr]

So this is fun! Learning LTspice...
<snip>.
Barry

Does your wife know that you're deranged?

I guess the problem is somewhere in the text below?:
* 2-input NAND SCHMITT trigger
* tpd 180n
* tr 100n
.SUBCKT CD4093B A B Y VDD VGND vdd1={vdd} speed1={speed} tripdt1={tripdt}
.param td1=1e-9*(180-40-10)*5/{vdd1}*{speed1}
*
XIN1 A Ai VDD VGND CD40_IN_S_1 vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
XIN2 B Bi VDD VGND CD40_IN_S_1 vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
*
A1 Ai Bi 0 0 0 Yn 0 0 AND tripdt={tripdt1} td={td1}
*
XOUT Yn Y VDD VGND CD40_OUT_1X vdd2={vdd1} speed2={speed1} tripdt2={tripdt1}
.ends

Barry

I think maybe the problem starts with what's MISSING from those lines.
Hystereses has been left out. Compare the CD4093 model with the CD4011 and they are exactly the same. They shouldn't be.

I looked in the 74HC models file and I believe that schmitt trigger devices should have something like:

VT+ 1.18/2.38/3.14
VT- 0.63/1.67/2.26

I can't read that but I'm guessing "Voltage Threshold + and -" something or other.
Those two lines are from the 74HC132 model (but the two lines were starred out for whatever reason). Somebody needs to contact Helmut and see what gives. I may get around to it in a day or so, but I am going to be busy so if somebody else beats me to it I won't have hurt feelings.

I could maybe try and "install" hystereses in my schmitt trigger devices, but when I mess around with spice models sometimes they NEVER work again!!

Just kidding, but if I got it right it would be by accident.

[bklein]

In regards to the simulation:
1. I had to modify the .inc statement to point to the correct location on disk.
2. add .inc statement to point to the bjt library file.
3. located, download potentiometer.sub file off yahoo group.
4. delete 555 and replace with one of msc library.
5. add diodes on 4093 inputs to 5V. This will make them work. Not much
to play with in the way of pulse width though with RC we are working with...
6. I tried to attach the revised .asc file, called it a .txt - rename to .asc.

Barry

[porkluvr]

here's a book I wroted

Sorry about the omission of the potentiometer model. I almost always forget something needed to make a zip package complete.
Tonight I uploaded the potentiometer model but I had to rename it. It should be un-renamed to potentiometer.sub.

Yesterday I uploaded my version of standard.bjt because the TIP32C model was not included in the factory download.

Bklein, you used an include statement which is OK but you could also replace your LTC supplied standard.bjt with mine, or copy/paste only the line that you need (starting with ".model TIP32C...). You wouldn't need to use an include statement with either scheme.

From time-to-time LTspice will ask you if you want to update. If you update, LTspice will take whatever additions that have been made to your "standard.*" files (in the CMP folder) and add those to the updated files. (standard.mos, standard.dio, etc.) That did not happen so smoothly once upon a time and it was a real pain - but they fixed that. This means that as you add models to the various "standard.*" files you will only need to do so once. You won't have to add them again and again after each update.

I also included my personal updated "standard.mos" but it is extraneous because there aren't any mosfet models in that file needed for the Surf-PI.
These two files (standard.*) should go in the CMP folder if you choose to use them.

I supplied an IRF9640 model/symbol that work together. That IRF9640 symbol started as an "X" symbol which can be used for diodes, opamps, BJT, etcetera. I think that the "X" symbol is supplied with LTspice and there is an explanation somewhere in the help file... but I don't know where it is at this time. The "X" symbol and various standard LTspice symbols are different animals, don't get them mixed up!

Whenever I use the "X" symbol, my accompanying model file starts with ".subcircuit" instead of starting with ".model" but I don't think that is rule, it's just the way most third-party models come. If it's a ".model" I copy it and drop it into the appropriate file in the CMP folder. If it's a ".subcircuit" I modify an "X" symbol and set its attributes to match the pin assignments.
Then, I put the accompanying "subcircuit" file in the SUB folder. That makes an include state unnecessary. This is just my preference; there are other possible scenarios.

Usually the pin assignments are stated, or otherwise obvious in a subcircuit file, but not always. Most manufacturers follow the same pinout assignment standards but some manufacturers don't.

You will want to find and download amplifier model files. You can make a generic X-amplifier symbol and point the model attribute to the appropriate model file but that only works if the symbol and model file pinouts agree. Instead:

I prefer to make two symbols for each amplifier that I have a model for; one symbol with '+' on top, and one with '-' on top. AD797+.asy and AD797.asy for instance, would be such a pair. That is not such a great chore because since most opamps have the same pin assignments, you just copy a similar .asy file, give it a new name, and then CTL+A to change the attributes. Then you swap the input pins and give it another name. Bam, you're done.

The CD4000.lib.txt file I supplied has some of my attempts at modification but that should be transparent and cause no problem. The original is on Yahoo's server and I hope we can get Helmut Senewald to tell us how to edit it properly. Easy to edit, but getting it right - that's a little harder.

Bklein, you probably know more about spice than I so I hope I didn't insult your intelligence.

[bklein]

>Bklein, you probably know more about spice than I so I hope I didn't insult your intelligence.

Ha! No way. I understand what you are describing but until I actually do it it is in danger of escaping my memory cells.... In fact I don't even get this forum's reply formatter... how do you snippet/quote the previous message ?

Did the diodes I added fix your problem or should there already be diodes in the part model? I am clueless at this point how/if the part's input hysteresis thresholds are virtually tied to the assigned supply inputs. I
posted a "flailing screaming panic" message on the Yahoo list to see if this gets better described.
Why is it that your 555 does not match up with the circuit? I used your zip file contents with 3 different systems and had the same problem of having to delete it and use the one in the default LT package.
I wonder if we should split this discussion off into a separate topic of just simulation using LTspice. Then others could chime in on their attempts with other circuits too.?

[Qiaozhi]

In fact I don't even get this forum's reply formatter... how do you snippet/quote the previous message ?

Select the "quote" button at the bottom right of the message. Then you can remove any parts you don't want between the quote delimiters.

[porkluvr]

I wonder if we should split this discussion off into a separate topic of just simulation using LTspice. Then others could chime in on their attempts with other circuits too.?

Good idea but this shouldn't take much longer (uh - I hope).

Here's what happened. I edited the 555 symbol to make it more compact and rearranged the pins to make it easier to wire. I edit many of the default LTspice symbols because they are often too large and/or the value and name fields are too far from the symbol body. (BUH gag sputter.) To edit default LTspice symbols you must either click them (in windows explorer) or else drag them from explorer into an open LTspice window. Right clicking them won't work.

Sorry about that. I forgot (again). It used to be much worse because I made all my resistors and capacitors, jfet, bjt, everything smaller by one grid space. THAT was really a portability mess. I changed everything back except for placement of the text fields. Pronto.

I guess my simulation files are not really portable unless I include every symbol that's been changed (doh).

We could continue privately or else drop this until something worthwhile develops. You're right, the ins-and-outs of LTspice are not proper fodder for discussion here.

I have not tried the diodes. What I got from your LTCSurfmaster.txt is too hard to decipher. I really don't have time to milk this because I am behind on another obligation. I would like to get the schmitt trigger mess straightened out but I much prefer the idea of using CD4538 to develop sample timing. I think that gives much wider adjustment range and is worth the extra parts count. Maybe not for a beach detector, but definitely so for a land detector.

I-don't-belong-here and should go?!

Press the "quote" button and you can edit the content between the [quote]s.

Here we go again: I uploaded my NE555.asy symbol but I had to rename it for the transfer. Download it and un-rename (I love that word) to NE555.asy to use it.

I have a seperate folder in my LTspice\lib\sym folder, called 3rdParty. (That's root for my 555 symbol.). That folder is further divided into comparators, opamps, power, MOSFET, etc. After a part's symbol has been placed from a particular folder LTspice will remember where to look for it the next time it is opened.

This ain't complicated. Chemistry is complicated.

I'll go look at the yahoo group to see what's happening after I finish winding a transformer.

[porkluvr]

bklein,
You were on the right track by adding diodes on the inputs.

Installling the "starred out" input diodes makes the one-shots trigger in LTspice but I believe the resulting pulsewidths are less than 1/2 what they should be, according to the formula below. Starting with 51kΩ and 1nF I calculated the pulse width should be about 45us. The sample pulses were 16us in my simulation. (I used 5 +|-5|=Vdd)

IMPORTANT!!
When I posted the CD4000.lib.txt (included in the sPI timer sim.zip) back in post 305 I edited in some diodes that had been starred out. Well, today I found out that the CD4538 models won't work correctly with those edits.
I tried to contact the administrators tonight and ask if I could either replace or delete the zip file but my email was returned three times. A fourth attempt made it through - I suppose - and I'll wait to see what the administrator's answer is before I decide how to handle this mistake. Or he'll decide for me.

I think that instead of editing the CD4000.lib globally it is a wiser thing to extract individual devices and edit those as deemed necessary. I know that there are some differences between MC14093 and CD4093 but I think they are small so I have made a necessary substitution.

I have extracted the CD4093 model from the larger file and 'installed' input protection diodes. You can call a model file anything you want but I renamed it xxx.txt so I could upload it.

In case anyone wondered, I renamed CD4000.lib to a text file because otherwise when I click on it to view the models in notepad, Pspice opens instead.

I appologize for any wasted time somebody may have had trying to use a CD4538 model, or whatever else I may have corrupted.

[Qiaozhi]

Well, today I found out that the CD4538 models won't work correctly with those edits.
I tried to contact the administrators tonight and ask if I could either replace or delete the zip file but my email was returned three times. A fourth attempt made it through - I suppose - and I'll wait to see what the administrator's answer is before I decide how to handle this mistake. Or he'll decide for me.

I have removed the attachment as requested.

[bklein]

Well, the performance of my Surfmaster PI is not that great. Maybe a sucky coil? I get 87V on the pulse return not the >100V talked about.
open air detect measurements in a crowded full of metal garage:
quarter - 5.5" soft, 4" hard
dime - 4.5" soft, 5" hard
penny - 6" soft, 5" hard
gold ring - paralled to coil - 7.5" soft, 6" hard.
soft is where you can detect it vs background, hard is a definite detect but not the really hard bang of a very close detect.

I have taken several scope shots of various timings. I played with the damping resistor and could get a bit more detection range with a much higher resistor, like 820 ohms. But the resulting signals do not look "kosher" - large spike at beginning. I figure since the receive signal pulse is only 87V either the coil is the issue or the drive transistor should be replaced. I put back the 220 ohm damping resistor. I am not sure how to adjust R25. I can see where it works and where it doesn't but I don't get how I should best set it. ANYONE KNOW? If you are building these things you need to know!