Super TGSL With H Bridge TX Diff RX Quad Sampling

[Davor]

I have a very small update, I erased R9. Point is that it was hooked to a channel that is not a GEB any more, but a ferrite reference, so this channel will not be used for audio. We have a dedicated GEB channel now that is capable of balancing just about any terrain, including salty beaches.

I'm not 100% certain if the audio is OK, but I can't simulate it now. Most probably some level shifting is in order.

[Thomas]

Thanks much guys, now I can see what's going on.

[Silver Dollar]

So Davor what do you mean by proportional audio?

The width of the pulse changes with depth/size of object?

I had heard of a MD where the width changes but also the tone changes too so
thought I'd put that into this detector. I've restricted the change to 100 hz or
so to just give a little extra info. I still need to build it and set the range to equate
to the targets correctly. It's easy to remove if it doesn't work out...

[Davor]

By proportional I mean that envelope of the audio signal is related to a signal strength. Binary audio is a beep-beep audio, and in order to make it tolerable you must employ various modes of antichatter that effectively erase all the small targets in the ground. With well designed proportional audio chatters are present in a form of background noise, and not annoying at all.

Pulse width can be a solution to a proportional audio, however, harmonics content of such audio changes with loudness. I'm all for a completely analogue audio with the least harmonics content. Quite loud audio is achieved with up to 2V amplitude across an 8ohm loudspeaker. Near to this voltage is already present at the most of the detectors prior to audio PA, so in fact only a little voltage gain should do.

I tend to avoid feedback-based filtering in the last stages of target baseband signal amplification, and instead I like antiparallel diodes. This extends the dynamic range for a decade or so on the high side, hence you may pump the gain for the next 20dB without saturation. It also serves another very fortunate consequence, a depth-wise equalisation that is far more fit for our hearing sense. So you keep the depth perspective that works more in tune with your own intuition.

Filtering in feedback circuitry in combination with compression produces varying phase lags in discrimination channels and destroy something called "recovery speed". We don't want that, so no capacitors in parallel with diodes.

IMHO pitch is not supposed to be related with signal strength because it is confusing. Signal strength is supposed to be a representation of signal strength. Don't try to confuse your senses more than absolutely necessary. Pitch can be related to a sort of VDI. More about that one later on.

[Thomas]

I completely agree Davor, sig. strength should be indicated by vol. of tone, as you get closer it gets louder and type of target should be indicative of pitch of tone in my o pin. I hope that's what your talking about. I was thinking, if the freq. of the sig. could be stepped down and not using a digital divider, could that be used for the output tone. but I'm just a beginner and don't know if it's possible or practical. I just remembered you spoke of a chopper amp in another thread, could that work?

[Davor]

Chopper amplifier is not related to audio, but it would make for a gain stage free of 1/f noise. Trouble with 1/f noise is that it makes non-motion reception meaningless at maximum sensitivity. Servo can't help much with 1/f because a servo op amp suffers the same.

If 1/f noise is unfamiliar, it is a kind of noise that increases with falling frequency, In effect you can't expect a DC gain stage to ever hold a true zero volts well. Just as unfortunately the metal detectors use the sub 10Hz band, well under the 1/f corner of the most op amps.

Tone can be produced by VCO, and VCO can be controlled by VDI, so yes, it is a solution I'm considering for my future builds. Most of the builds by discrimination assume complete eradication of iron, and I don't think it is OK. Knowing what is there, even if it is iron, is much better than not knowing.

[golfnut]

Is it possible to include a +5v reg for the opamp positive rail. With symmetrical rails DC offsets are more likely to be less. Manufacturers of op amps always specify symmetry. S

[Davor]

True, but 1/f noise is a completely different animal. You can't cancel it with symmetry. If you are not that much into non-motion, a motion filter will kill the most 1/f noise for you.

[Silver Dollar]

So we want to let the chatter through and allow the operator to sort through the data.
I was just figuring that lower volume for deep targets might make you miss some. I guess
we need to find the smallest target's amplitude and map the volume response so it is obvious
and let the higher magnitude signals appear in the rest of the volume range.

What would be neat is to map the amplitudes across the stereo field so deep is to one
side and shallow to the other. Unfortunately that would add a bit of complexity and use more
power to boot...

Even better would be to create a 3d field so you could hear how deep the object is. Sort
of like a carver sonic hologram generator. Maybe we do not have a complex enough signal
to do that though...

[Davor]

You don't have to do all those things. Your ears do them for you.
Now, the targets respond by a 6th power law, which is too sharp transition for our ears. Sound in air gets attenuated at a much lower rate. So what you need is a compressor that would soften the 6th power law to something intuitively useful, and diode limiters do that surprisingly well.

[Silver Dollar]

I understand we can use some compression because of the wide dynamic range
involved but it always seemed counter intuitive to reduce our "working" range
to +- 0.6v. I would expect operating the opamps on +- 15v and using all that
headroom would allow for smaller target location. Of course noise might stop
you from getting any better results...

[Tinkerer]

I understand we can use some compression because of the wide dynamic range
involved but it always seemed counter intuitive to reduce our "working" range
to +- 0.6v. I would expect operating the opamps on +- 15v and using all that
headroom would allow for smaller target location. Of course noise might stop
you from getting any better results...

+- 0.6v? This is why I used LED's on the TINKERER. if I remember right, the signal amplitude went to near full voltage, but the gain was reducing above 3V.
In an other implementation, similar log like response was obtained by adding a biased transistor in the feedback loop, the gain was very high at low voltage signals and soft clipped near the amplifier saturation limit. With a PI, the amplifier saturation adds some delay time. This was avoided by this method.

[Davor]

Small 8 ohm loudspeakers are usually rated 0.1 W or up to 0.5 W, and it means you blow them with anything over a few volts, so don't overdo it. Soft compression is beneficial because it converts the 6th power law to something like a 2nd power law we are more familiar with, nothing else. You do have amplification after the diode compression, so you'll have your volume for sure.

[Aziz]

Oh yeah!, this is a nice project.

Aziz

[Davor]

Just a sidenote about a possible Rx improvement. It is regarding the bilateral switches. Recently in a private conversation regarding the choice between using FETs or bilateral switches for a PI device, I concluded that bilaterals are a better solution due to a lesser charge injection. However, it is only half way there. When something is "less" it is not completely gone, and in case of bilateral switches the clicks and pops are usually greatly reduced in case signal half way from each rail, and seeing low impedance, but not eliminated completely. Therefore, a real improvement in any future build would be a real double balanced mixer that treats charge injection as a common mode nuisance and gives it no free pass.