Simpler PI...

[Waikiki_Sweep]

Oooo... OPA37 from Burr-Brown is celebrity in opamp world. http://www.pci-card.com/opa27.pdf

Probably OP37 from Analog Devices is the same but $1.30 cheaper http://www.analog.com/static/importe...heets/OP37.pdf

Just got OP37 to build Chance MD.

Also found interesting solution in OPA37 data sheet.
If you have Power and Money you can paralleling several OPA's to significantly reduce noise. Sure it is for High End constructions not for this Simple PI.



Good marketing trick from Burr-Brown. Why to buy one? Buy 10!
Why not to put 10 opamps in one plastic case and sell it as low-low-low noise opamp?

[green]

[Allowing 20 kHz for safety, any opamp configured to linearly reproduce the pulse properly should have a max gain = its GBW product / 20 kHz.]

Playing with spice to get minimum delay it looks to me that max gain = GBW/ coil circuit resonance. A lot different, don't know which one is closer to being correct.

[Beenthereuk]

Had some time to look at the signal closely. After diode clipping a signal with a 20 uS exponential decay, an FFT reveals a bandwidth of around 12 kHz. Allowing 20 kHz for safety, any opamp configured to linearly reproduce the pulse properly should have a max gain = its GBW product / 20 kHz.

With this in mind I simulated a number of opamps, each with its gain adjusted to suit its GBW product and the OPA37 was only beaten by the ideal opamp macromodel. I then built an inverting amplifier with a OPA37, a 1 M ohm resistor, setting the gain to 1e6 / 300 = 3300x (source resistor used was 300 ohms) and biased it to within 2 volts of the positive supply rail and it worked beautifully.
I should mention that the OPA37 GBW=63 MHz so gain should = 63e6 / 20e3 = 3150x

The TL082 doesnt even come close. The LM358 is way too slow and the LM318 was second best but could not drive its output below +2 volts.

[Beenthereuk]

Green: its OK, I uploaded the circuit already.

Davor: funnily enough, I reached the same conclusion and simulated a decent response with an LM311 but am not sure about the noise etc. It seems funny to ask a comparator to amplify a signal linearly until we remember that they are incredibly fast. I have to try some chips out and maybe different amplifier configurations too. The circuit as it stands is just a very low input impedance non-inverting amplifier with high gain that pumps a passive RC integrator. The output FETs threshold is fixed at around 2V and the slope from the signal on the integrator capacitor can be varied by adjusting the integrator resistance to vary the duty of the pulse produced. There must be many different ways of reproducing this..

[green]

[I will attach the circuit for LTSPICE if you can explain how I to, as it only seems to accepts pictures or videos ?]

Not sure if this what your asking. I do a (alt,prt scr) and paste to paint.

[Davor]

LM358 is a much better choice than TL082 as it handles comparator-like operation much better. Simulations are not giving the correct results for many op amps when pushed into comparator-like operation. Actually a real comparator would be the best choice.

[Beenthereuk]

Yes, it may be better that way if you prefer. I think the output transistor would also have to be pFET though because the signal will be inverted coming out of the opamp but have not simulated it. Either way is easily possible with small changes.

[Waikiki_Sweep]

Looks great! Usually if P-channel is used schematic allows to connect coil to ground. It helps somehow to keep coil shield and coaxial shield connected to ground. Probably it is good to connect shields to Vcc too (up side down) but anyway grounded shields looks more natural. That is why I like those rare P-channel mosfets for their straight orientation.

[Beenthereuk]

OK, here is the circuit with a choice of clocks for either a pMOS or nMOS FET coil drive stage. You only need to make one clock using the second TL082. Both circuits produce pulses that are 50 uS wide with a frequency of 333 Hz using the values shown. Make R15 a 200 k pot to tweak frequency.






And the LTSPICE file for the whole circuit:

[Waikiki_Sweep]

... and low noise. (at 10 Hz) Looking forward for your opamp choice.

Making coil always required too much labor. If that project declares to be simple making monocoil is simplest task ever.
So for that project making separated RX coil looks like installing diesel engine on moped. I believe we don need it yet.

Probably anything you can attach to attachment section of post if click on "Go Advanced" button.
I am not sure is there any limitation on file extensions, if any, you can compress it and attach .ZIP file.

[Beenthereuk]

Yes, some opamps will work as is but others require bias changes.
Ideally the opamp should be JFET input, single supply, high-slew rate (10us/v +) and common mode in rail to rail capable but just about any can be accommodated with a little tweaking. I will look at generating the pulses with the second half of a TL082 too. Might as well make it small.

Do you think there would be much benefit in using a second, higher inductance coil as a receiver here?

I will attach the circuit for LTSPICE if you can explain how I to, as it only seems to accepts pictures or videos ?!

[6666]

Beenthereuk would you mind attaching the .asc file here so we can have an experiment with your circuit ? thanks

[Waikiki_Sweep]

I see it improved and simplified.
Cool solution with C5.
Because TL082 is dual can you use second amplifier to make a pulse generator?
If so, it goes to be a really small PI.

Probably it can use lower noise opamps too.
Or simulation is required to change opamp?

[Beenthereuk]

Just tweaked the gain on a TL082 (for those of you who prefer op-amp simplicity) and changed to a single stage piezo driver which should be loud enough given the voltage dropped across the transducer.

The feedback resistor split and capacitor decoupling boost the gain but can lead to instability if large values of C5 are used, 100 pf to 200 pf should be more than enough. Drive it with any 50 uS pulse generator at any frequency < 2 KHz.
Uses a 12 V supply. Would be interesting to note the boost in sensitivity if 18 V is used.

R7 is a 10K pot.

Any coil with 600 uH > L > 300 uH should work.