Serbian "Most Wanted"

[eduardo1979]

Which version works well without changes in PCB

The version of the post 559.

[Davor]

I checked the preamp with common base and, well, I'm not too happy. True, it provides somewhat lower noise than the economy range low noise op amps, but barely so. And with very small current draw too. With about 2.3nV/sqrt(Hz) it is over 6dB worse than the original SMW, or my redox version. Point is that common emitter preamps with transistors chosen for low Rbb (say, BC337) are much quieter, and noise below 1nV/sqrt(Hz) is easy to achieve - just like with the original SMW.



Regarding the Fe channel, it is easy, just copy the existing Disc channel and use the opposite polarity input at phase shifter. An additional audio oscillator will be needed.

I'll revisit the gain blocks and give my suggestion.

[Davor]

I also checked what would happen if a good FET is placed instead of a BJT, as FETs are quite happy in common gate configuration. Well, not much happen. Practically same thing. At the same time a 2SK170 is not the cheapest component around. I tried it at currents this component is used to, but I get coil loading, a very spiky transfer function, and only slightly better noise than with BJT, as I predicted earlier. Due to the spikiness it would not be a practical one to build.
See for yourself...

[eduardo1979]

I saw that in your diagrams, you are using a coil 6mH. Please try one of 13mH, and see what happens.

[eduardo1979]

RX

Here is another example. To see if this is better.

[Davor]

The whole point is that given the same ... everything else, this common base configuration is still over 6dB noisier than any common emitter or differential amp solution. Of all the features in SMW I respect the preamp the most. It can be simplified, But it simply can't get any better.

13mH coil would give some 3dB more input voltage than the 6mH one, but regarding the noise - original SMW is still better.

The original SMW would work perfectly well with differential mode coils also, provided the Q2 base is allowed to receive the input signal. It is grounded It should work miracles in wet grass if the coil is hooked in differential mode, with or without coil centre tap.

I also think a single 9V battery should be a perfect solution for powering this device, without any stabilisers. Point is - when 12V source drops below 10V, the stabiliser performs miserably. By that logic, a simple 9V battery (with some reverse polarity protection) should perform well enough on it's own. For increased autonomy you simply plug in two batteries and switch to the other one when the first one vanes. That should also give you a pretty good idea of how much more you can go. 9V rechargeable cubes come very cheap in Lidl stores occasionally.

[eduardo1979]

The whole point is that given the same ... everything else, this common base configuration is still over 6dB noisier than any common emitter or differential amp solution. Of all the features in SMW I respect the preamp the most. It can be simplified, But it simply can't get any better.

13mH coil would give some 3dB more input voltage than the 6mH one, but regarding the noise - original SMW is still better.

The original SMW would work perfectly well with differential mode coils also, provided the Q2 base is allowed to receive the input signal. It is grounded It should work miracles in wet grass if the coil is hooked in differential mode, with or without coil centre tap.

I also think a single 9V battery should be a perfect solution for powering this device, without any stabilisers. Point is - when 12V source drops below 10V, the stabiliser performs miserably. By that logic, a simple 9V battery (with some reverse polarity protection) should perform well enough on it's own. For increased autonomy you simply plug in two batteries and switch to the other one when the first one vanes. That should also give you a pretty good idea of how much more you can go. 9V rechargeable cubes come very cheap in Lidl stores occasionally.

Thanks Davor.

[Davor]

I think I just discovered the root cause of SMW's poor discrimination: the phase shifters.
The original setup responds like this (I replaced the comparators with switches to speed up the simulation):



It is obvious that GEB and DISC phase shifts are too close by, and there is not much control over their response. Unfortunately that can't be remedied right the way. The GEB side control can be improved by beefing up the 220p capacitor to 470p, but this approach is nort enough for Disc. With additional RC and a bit tweaked component values it happens beautifully:



In light of this finding, it may be advisable to simply remove the existing Discrimination control (laid out in Fisher style), and instead use the control on the phase shifters.

[Davor]

I found a better solution to a phase shifter. The previous beefed up solution is somewhat frequency dependent, but this one is much better.

[eduardo1979]

This convinces me more than the last. Whenever I get free time, I will put it into practice to see results. Thanks for taking your time.

[smity]

I gathered together this device. Everything worked fine until I tried to install the sound of the "bell". After that, there was a terrible sound background. I removed the capacitors, but remained silent background. Replaced all the transistors in the audio part, replaced the chip ... nothing helped. In what could be the problem? What to look for? Thank you.

[Davor]

I think I have the audio fixed. There are some odd things going on with the SMW audio with 555. For a start, the resistors around 555 seem to be swapped because they are configured to produce needle like output. The audio is configured for binary response.
What follows is a simulation of the original circuit with input span from -0.5 to 0.5V, mimicking the signal at the output of discrimination comparators:

[smity]

Dear Davor. Possible to model the sound of the 555 with these capacitors? Ivconic this modification suggested in the message #55.

[Davor]

Sure it is possible. But the original circuit offers only binary response and all gains are waaaaay over the top.

And now something completely different

Actually there are only some minor differences that would skip the untrained eye.
555 is fixed to produce meaningful signal.
LM358 circuitry is convinced to produce proportional response (gain is reduced), and filtering capacitor is reduced to improve response time.
"Volume" transistor is removed and in it's place there is a 555 signal injection, and some smart level shifting. U2&R7 are at about 68k, but I placed a trimmer there simply to accommodate offset for everyone's taste. It dictates small signal response. Perhaps it could be replaced with 56k or 68k resistor.
Volume potentiometer is placed at a bit different place, instead of the existing diode. See - only minor differences.
The other diode is also removed to improve noise immunity.
All in all - there are only a few small differences against the original circuit.

Enjoy:

Now I must rush, so I'll post the LTspice files later on. Please give some comments so that I can implement any changes if necessary.

[eduardo1979]

The solution is good, but there is one problem. I've tried what you just posted, and it is accompanied by the sound of a blow to the speaker. If the coin passes very close to the coil, the knock on the speaker is not seen, but as the currency is moving away from the coil, the sound is weaker and blow the speaker is appreciated more. I have tried over 4 different types of audio oscillators, and now I have not found any that I like. There is a knock on effect of the speaker, after the sound emitted. So far I have not found the solution to that. Thanks for your efforts and for your kindness to share with us here.