Source of propagation delay?

[VinceC]

I have been examining the circuit operation of my Surf Pi 1.2 to better understand some of the issues associated with the design. The attached are screenshots from my scope that shows the trace of the voltage at the coil simultaneously with the output of the preamp. There are a couple of items that I noted.

1) It appears that using the output of the preamp (red trace) to select the value of the damping resistor (which is what I did) may not be the best approach. The coil voltage measurement (yellow trace) clearly shows significant ringing.

2) The output of the preamp (red trace) does not settle until ~25us but the coil (yellow trace) would appear to settle in <10us (assuming it was critically damped). Why the huge differences in time? Is it the opamp which is taking that long to recover? That does not make sense. Thoughts?

Vince

[Davor]

With exponential decay (and linear time) you may start observing signal at any point of time until it sinks under noise, and observe the same decay shape. With enormous gain of a preamp this is precisely what you get. Same mechanism applies to sinusoidal exponentially decaying signal, and again, with sufficient gain you'll be able to see exponential gain further on in time, just as you did.

[Qiaozhi]

1) It appears that using the output of the preamp (red trace) to select the value of the damping resistor (which is what I did) may not be the best approach. The coil voltage measurement (yellow trace) clearly shows significant ringing.

2) The output of the preamp (red trace) does not settle until ~25us but the coil (yellow trace) would appear to settle in <10us (assuming it was critically damped). Why the huge differences in time? Is it the opamp which is taking that long to recover? That does not make sense. Thoughts?

Vince

The output of the preamp is the correct place to monitor while adjusting the damping resistor.
If you look at the preamp output in your scope images, you will see the coil is not critically damped, and it appears to be following the coil signal correctly. However, there is a glitch at around 5us caused by the monotonicity of the input signal (probably external noise) where the opamp briefly comes out of saturation. The huge gain of the preamp (60dB) drives the output into saturation and the settling time is affected by how quick the opamp can recover. This is why the MiniPulse Plus uses a 2-stage preamp with separate gains of 30.4dB. So even though the total gain is higher (60.7dB) the settling time is shorter.

[green]

A LTspice circuit that helped me understand what should happen. L1 300uh, the rest blank. C1 84pf for 1 Mhz coil resonance. R2 10 meg for no R damping. Circuit resonance 945khz (17 cycles/18 usec). For critical damping, R2 = pi x circuit resonance x L (3.14 x .945 x 300 = 890 ohms). Try shorting D5, circuit resonance should be a lot lower. Try 336pf for for C1 (coil resonance = 500 khz). A couple things I expect to see with LTspice and on the bench. R damping should be close to (pi x circuit resonance x L) and if critical damped the coil should decay to less than 1 mv in (3/circuit resonance + fet clamp time). 3/945khz = 3.2 usec. With spice I can expand the amplitude scale. My scope over loads if I try to look at the coil mv. Amplifier circuit effects what I see on the bench. Just a few things to try. Can add amplifiers and target signals after playing with the coil driver.

[VinceC]

The output of the preamp is the correct place to monitor while adjusting the damping resistor.
If you look at the preamp output in your scope images, you will see the coil is not critically damped, and it appears to be following the coil signal correctly. However, there is a glitch at around 5us caused by the monotonicity of the input signal (probably external noise) where the opamp briefly comes out of saturation. The huge gain of the preamp (60dB) drives the output into saturation and the settling time is affected by how quick the opamp can recover. This is why the MiniPulse Plus uses a 2-stage preamp with separate gains of 30.4dB. So even though the total gain is higher (60.7dB) the settling time is shorter.

Why is the preamp the correct location?

If I understand correctly, the use of a cascaded multi-stage preamp using high GBW opamps will greatly reduce this effect and reduce the settling time significantly. If I prevent saturation I suspect that will help even more, correct?

Vince

[VinceC]

A LTspice circuit that helped me understand what should happen. L1 300uh, the rest blank. C1 84pf for 1 Mhz coil resonance. R2 10 meg for no R damping. Circuit resonance 945khz (17 cycles/18 usec). For critical damping, R2 = pi x circuit resonance x L (3.14 x .945 x 300 = 890 ohms). Try shorting D5, circuit resonance should be a lot lower. Try 336pf for for C1 (coil resonance = 500 khz). A couple things I expect to see with LTspice and on the bench. R damping should be close to (pi x circuit resonance x L) and if critical damped the coil should decay to less than 1 mv in (3/circuit resonance + fet clamp time). 3/945khz = 3.2 usec. With spice I can expand the amplitude scale. My scope over loads if I try to look at the coil mv. Amplifier circuit effects what I see on the bench. Just a few things to try. Can add amplifiers and target signals after playing with the coil driver.

Building this model is presently beyond my skill level, though in a couple of weeks I might be able to do so.

Would you be willing to share the LTSpice model to allow me to play with it?

[VinceC]

With exponential decay (and linear time) you may start observing signal at any point of time until it sinks under noise, and observe the same decay shape. With enormous gain of a preamp this is precisely what you get. Same mechanism applies to sinusoidal exponentially decaying signal, and again, with sufficient gain you'll be able to see exponential gain further on in time, just as you did.

But wouldn't a preamp design that reduces the time to settle be preferable?

[green]

Building this model is presently beyond my skill level, though in a couple of weeks I might be able to do so.

Would you be willing to share the LTSpice model to allow me to play with it?

Don't know how to do share it. I can tell you how I do it. Draw schematic. Right click V1 and enter data for voltage source. Click simulate at top, click simulation command and enter data. Right click and hit RUN, click on schematic where you want to see voltage or current. I've got a lot to learn, there are probably better ways, but this works. The SPA11N60C3 and MUR460 are in the parts list, you just have to look for them. If your like me you need practice doing it.

[Qiaozhi]

Why is the preamp the correct location?

Quite simply because any ringing is more evident at the output of the preamp, and the sampling integrators get their input from the preamp.

[VinceC]

Don't know how to do share it. I can tell you how I do it. Draw schematic. Right click V1 and enter data for voltage source. Click simulate at top, click simulation command and enter data. Right click and hit RUN, click on schematic where you want to see voltage or current. I've got a lot to learn, there are probably better ways, but this works. The SPA11N60C3 and MUR460 are in the parts list, you just have to look for them. If your like me you need practice doing it.

I will give it a try! Thanks!!

Yes, but it appears that I have been doing a lot "more practice" in the last few days!

Vince

[VinceC]

Quite simply because any ringing is more evident at the output of the preamp, and the sampling integrators get their input from the preamp.

But wouldn't the decreased frequency response of the high gain opamp mask much of the ringing?

[VinceC]

Don't know how to do share it. I can tell you how I do it. Draw schematic. Right click V1 and enter data for voltage source. Click simulate at top, click simulation command and enter data. Right click and hit RUN, click on schematic where you want to see voltage or current. I've got a lot to learn, there are probably better ways, but this works. The SPA11N60C3 and MUR460 are in the parts list, you just have to look for them. If your like me you need practice doing it.

I appreciate your help. With the screen shots that you provided I was able to get your circuit working and duplicated your response. I have also been able to observe the significant improvement that the diode provides for the coil settling time. Good work!

I now have a couple of questions.

1) What is the purpose of R1?
2) Why isn't the resistance of coil (1-2 ohms) included in parallel with the coil? (I tried adding it and the model gave strange results.)

Thanks!
Vince

[green]

[1) What is the purpose of R1?
2) Why isn't the resistance of coil (1-2 ohms) included in parallel with the coil? (I tried adding it and the model gave strange results.)]

I had seen a lower resistance for R1 in another thread. It worked better with the 10meg, so I left it there. Probably don't need it.

The coil resistance should be in series, not parallel. Try adding resistance. I think all it will do is lower peak coil current

[VinceC]

[1) What is the purpose of R1?
2) Why isn't the resistance of coil (1-2 ohms) included in parallel with the coil? (I tried adding it and the model gave strange results.)]

I had seen a lower resistance for R1 in another thread. It worked better with the 10meg, so I left it there. Probably don't need it.

The coil resistance should be in series, not parallel. Try adding resistance. I think all it will do is lower peak coil current

The reason that I asked about R1 is that when I removed it the simulation performance was actually better. I am nowhere close to being able to run a .NOISE analysis for any circuit yet, but I do know that high resistances are a significant source of Johnson Noise, and I am looking to be able to limit the amount of intrinsic noise in the circuitry ahead of the preamp. It was also highly instructive for me to observe the impact of the various coil parameters and the effects of the damping resistor.

Oops! Stupid mistake!! Thanks for the catch.

One additional question - what is the purpose of the cap in your circuit? Is it to account for the coil winding capacitance? If so, why not include it in the dialog box for the coil characteristics directly?

Vince

[green]

The reason that I asked about R1 is that when I removed it the simulation performance was actually better. I am nowhere close to being able to run a .NOISE analysis for any circuit yet, but I do know that high resistances are a significant source of Johnson Noise, and I am looking to be able to limit the amount of intrinsic noise in the circuitry ahead of the preamp. It was also highly instructive for me to observe the impact of the various coil parameters and the effects of the damping resistor.

Oops! Stupid mistake!! Thanks for the catch.



One additional question - what is the purpose of the cap in your circuit? Is it to account for the coil winding capacitance? If so, why not include it in the dialog box for the coil characteristics directly?

Vince

I don't think R1 would contribute to noise. The voltage on R1 is over 200 volts when the coil volts is at zero Isolated from the amplifier. I'm not a EE so could be wrong. The cap could be either place for the model. Putting it external allows someone to see the value looking at the schematic.