TGSL Experiments

[simonbaker]

LM308 offset

Here is something I'm looking into that could greatly affect the sensitivity and depth of our TGSLs.

The quiescent output voltage of the LM308 (pin 6) is supposed to be 0 volts.

If it is not, then it will change the detection to varying degrees.

The threshold range of the U106 comparators is about 3.6 mV to 36 mV (controlled by sensitivity pot R36). If the LM308 quiescent output voltage is not zero, it will directly affect the the detection.

It is also important the both LM308 outputs have the same quiescent output voltage (if not zero), otherwise the GB and DISC channels will detect at different thresholds.

I just measured mine by removing the RX coil.

My disc LM308 output voltage is about 7 mV, and my GB LM308 output voltage is about -2mV. (I see some high freq noise there, but not sure if significant or just something my scope sees.)

So my case is not ideal, but no too bad, but if sensitivity threshold is at 3.6 mV, you can see there is a big relative difference. A negative offset would definitely reduce your detector sensitivity. In my case, I don't know if the amount I have is a problem or not. It could mean my LM308 precision ICs are not so precision as they should be. I got them for about 50 cents each. Or maybe a voltage gradiant along the ground buss is responsible, although it seems to check out fairly well. Could it be some high freq noise that get's rectified at the inverting inputs creates a bias??? Anyway, an area to check out.

QUESTION FOR OTHERS

I am interested if others could try disconnecting their RX coil and measure DC voltage at LM308 pin 6 for both LM308s (or whatever IC you have there).

-SB

[simonbaker]

Sir, step away from the circuit board...

Some people should not be permitted to use sharp objects, power tools, or motor vehicles...

I decided to try swapping my LM308s to see if the different output offsets were due to the chips or something about the circuit.

When the speaker went into overdrive, I cut the power and removed the chips. Finally noticed I swapped an LM308 with the LM393 by mistake. Hey, a chip's a chip isn't it?

Happily restored the chips to their correct location and turned it on. Voltage didn't look quite right.

Pretty soon noticed a burning smell. Glanced at ammeter, .8 amps -- bit on the high side I thought. Cut the power and the LM7808 was getting ready to roast a chicken. LM393 was just right to serve with nice Chianti.

LM393 was in backwards.

[Dennis the Mennis]

I do not use back-to-back 10 uF caps for C14 and C17, rather I use single 4.7 uF nonpolar caps.

I built 2 TGSLs, one with back to back 10uF caps, one with single 4.7uF cap for each channel.. I can't tell any difference in audio or detection range. Anyone else know if there is a difference?

One thing I've notice is Ivconic and dfbowers and others have coil start and end in the middle of the straight section, whereas I always put it at the corner. However I don't see yet that this should matter much.

I pondered the differences too. Most commercially made coils have the cable exit somewhere away from the center. There must be a reason. Also, I don't know if it makes any difference on where the gaps in the shields go.. it just seemed to be a convenient place for the wires to exit.
Anyone have any ideas?

Don

As far as I can remember Ivconic wrote somewhere that he had experimented with both type caps ( back to back and the bi-polair ) and in that case in didn't seem to bother , no differences in air depth between the both types .

The same story for the gaps ...

I'm still experimenting with the TGS with the DD coils and my problem seems to be the same as some of the TGSL : 'only' 20 -22 cm detection on an 1 euro coin .

Also this (TGS) pcb design has proved its self only the Tx output is lower as the TGSL , but as Don mentioned sometimes the Tx output isn't always crucial.

So is there some mysterious relationship between the parts chosen for the pcb's ?

And what if -SB sends a set of his coils the DFBowers for comparisment ?


regards

Dennis the Mennis

[simonbaker]

For now I test with the GB wide open (no rejection) just to figure out this depth problem.

Probably a safe thing to do.. but may want to do some test with GB set like its used so as to get a like comparison. How does your circuit respond in the AM mode?


My RX cap C6 soldered in is low, about 13 pF

You meant 13 nF for C6, right?


I do not use back-to-back 10 uF caps for C14 and C17, rather I use single 4.7 uF nonpolar caps.

I built 2 TGSLs, one with back to back 10uF caps, one with single 4.7uF cap for each channel.. I can't tell any difference in audio or detection range. Anyone else know if there is a difference?

Now I did add the custom voltage divider circuit between pin J1 and the GB and Disc circuits as mentioned previously. It creates a small phase shift, like 3 to 5 degrees, but it seems negligible to me, and I don't think it causes a loss of sensitivity somehow -- but I'll probably try bypassing it as a test also.

I didn't see the thread on using a voltage divider. Does it provide some benefit?


One thing I've notice is Ivconic and dfbowers and others have coil start and end in the middle of the straight section, whereas I always put it at the corner. However I don't see yet that this should matter much.

I pondered the differences too. Most commercially made coils have the cable exit somewhere away from the center. There must be a reason. Also, I don't know if it makes any difference on where the gaps in the shields go.. it just seemed to be a convenient place for the wires to exit.
Anyone have any ideas?

Don

Probably a safe thing to do.. but may want to do some test with GB set like its used so as to get a like comparison. How does your circuit respond in the AM mode?

I'm not sure, but I think it is as expected. I find both modes react a little funny to ferrous metals, depends a lot on the shape. In AM mode it seems moving the GB up sometimes gets a cleaner signal with ferrous, not sure.

When I look at the waveforms for the AM mode, I don't really understand what principle they were shooting for. I would actually expect a little less depth in AM mode for coins and such, but I can't seem to demonstrate that.

You meant 13 nF for C6, right?

Thanks, yes, 13 nF -- too late to change it darn.

I didn't see the thread on using a voltage divider. Does it provide some benefit?

Probably earlier in this thread. I designed it to clean up the jaggies that come out of the LF353 pin 7 for TGSL circuits that don't have the feedback JFet in the oscillator and therefore have higher voltage. The high TX voltage seems to slam the LF353 input and cause some oscillations. There might be other ways to reduce them too.

But -- people make good working TGSL even with the jaggies, so I'm carrying a lonely torch!

If you look at your (dfbowers) photos you took of U102 pin 1 with disc pot set to minimum, its possible those double humps are because of the jaggies from the LF353 -- it really should be a single square wave. But the way I wired my pot, I would expect it on the maximum (CW) setting. Maybe I have mine backwards. Its the setting where you select all the signal from the LF353 pin 1.

But darned if it seems to matter. It bugged me though, I wanted a good square wave and my voltage-divider mod seems to work pretty well. Just not 100% sure if it is free of other side effects.

-SB

[simonbaker]

As far as I can remember Ivconic wrote somewhere that he had experimented with both type caps ( back to back and the bi-polair ) and in that case in didn't seem to bother , no differences in air depth between the both types .

The same story for the gaps ...

I'm still experimenting with the TGS with the DD coils and my problem seems to be the same as some of the TGSL : 'only' 20 -22 cm detection on an 1 euro coin .

Also this (TGS) pcb design has proved its self only the Tx output is lower as the TGSL , but as Don mentioned sometimes the Tx output isn't always crucial.

So is there some mysterious relationship between the parts chosen for the pcb's ?

And what if -SB sends a set of his coils the DFBowers for comparisment ?


regards

Dennis the Mennis

What is your TX oscillator voltage? It does have some bearing, but deceptive because what may matter more is the current in the TX coil and that depends on Q of coil as well as the voltage.

Can you disconnect just your RX coil and measure the DC voltage (in millivolts) at output (pin 6) of both LM308?

Then reconnect RX coil and measure typical noise level (with scope) at output of LM308?

I wondered about sending my coils too! Good idea. Also, sending back dfbowers coils to see if changed somehow -- that's really desperate !

-SB

[simonbaker]

LM308 quality

Back to the LM308 quiescent output voltages.

I learned that the LM308 quiescent output voltage seems to vary with the chip. So the offset is due to each chip, not the circuit.

I swapped the chips and now my TGSL works a little differently. Actually seems a little less sensitive with -3 mV offset on the DISC LM308 and about 7 mV on the GB LM308. In other words, the positive offset on the DISC channel may actually have been helping a little to boost weak signals when combined with noise and for certain sensitivity settings. But this is sort of a comedy of errors.

However, these little offsets are dwarfed by the typical noise signal I see there in my workshop. So you would think the would not be noticeable.

But because a lot of the noise pulses are filtered out by the low pass filter that follows, I think even the small offsets do have a noticeable effect.

What I would like to see is a lower noise level and LM308 quiescent output very close to zero.

I think I'm going to need to get to a noise free environment to really start to analyze this.

-SB

[dfbowers]

TGSL airtest 40+ cm..

I had some fun experimenting today with the TGSL.. a rediculous 40cm + for on a 1 euro coin. Simon said something and it made me think.. So I made a mod (Just to see what would happen) and Voila!

I had to do this test far away from the house because of all the noise. Also the video is noisy because I used a metal ruler and the table was just a sheet of cardboard and everything was in close proximity. Otherwise, the circuit is stable.

Also, as Ivconic once posted, probably not much use in the ground.
I will post the mod shortly when I finish a few measurements.

http://www.youtube.com/watch?v=l_fXBwojxVg

Don

[dfbowers]

TGSL airtest 40+ cm..

Another view..
http://www.youtube.com/watch?v=KytiLKZ-ybc

[ODM]

dfbowers: What type of coil are you using in these videos? Manufacturer/product name, or notes about construction if it's a coil made by yourself, would be welcome.

[dfbowers]

dfbowers: What type of coil are you using in these videos? Manufacturer/product name, or notes about construction if it's a coil made by yourself, would be welcome.

ODM,

Just handmade coils. 25mm wire (with enamel). I vacuum form and fill my own shells.. Nothing special.

Don

[lunamay]

dfbowers your MD looks great and works very very good..i can't wait for your mod to be posted

best regards

[simonbaker]

Another view..
http://www.youtube.com/watch?v=KytiLKZ-ybc

That's incredibly impressive. It would be fantastic if there was a way you could scope the background noise level at LM308 output pin 6 (and DC offsets) for that test. And describe your mod. Nice job.

-SB

Edit: I see you posted results in other thread. Do you think you could measure noise level out there in the quiet field? Need a good portable scope...

[dfbowers]

That's incredibly impressive. It would be fantastic if there was a way you could scope the background noise level at LM308 output pin 6 (and DC offsets) for that test. And describe your mod. Nice job.

-SB

Here is all I did! (plus a noise free environment) It just allows pins 2 & 6 on U106b to use a minimum of -.03V as a reference. The "threshold" of my LM393s is -.25V. This mod is impractical though, as in actual use my circuit functions just fine.. and with R36 set to a minimum tends to be a little noisy as it is. But, I wanted to spark some ideas on what might be going on with other circuits.

One thing I might actually use this for is use with my 7" coil. It air tests at around 33cm.. but it might give the detector a little more edge in the soil. The 7" coil is very quiet.

Don

[dfbowers]

That's incredibly impressive. It would be fantastic if there was a way you could scope the background noise level at LM308 output pin 6 (and DC offsets) for that test. And describe your mod. Nice job.

-SB

Edit: I see you posted results in other thread. Do you think you could measure noise level out there in the quiet field? Need a good portable scope...

Yes, I can do that.. My tool shed is probably 150ft from the house and I do have an outlet there. I will take a scope out later on.

Don

[simonbaker]

Yes, I can do that.. My tool shed is probably 150ft from the house and I do have an outlet there. I will take a scope out later on.

Don

Really appreciate your efforts for the rest of us!

Maybe we'll actually get to the bottom of what's going on here.

-SB