One ground plane, or two?

[eduardo1979]

I want to design the PCB for a PI detector, such as an Enhanced Goldscan 5, and I have a question that I'd like to hear the experts' opinion on. The PCB would be a two-layer PCB with all SMD components. The square wave signals are generated with a 555 and four 4538 ICs. The integrators are differential and I use a 5532. My question is whether it's better to use a single ground plane on the bottom layer or two ground planes, both on the bottom and top layers. I've done many SMD PCB designs, and the problem is the thermal noise generated by the damping and clamping resistors, the MOSFET, and the voltage regulators, which are components that get too hot and ruin the analog signals. Digital signals also generate some noise on the PCB, compromising the analog signals. I've noticed that all the PCB designs I've created in THT format perform better than the designs I created in SMD format, and I can't find an explanation for this. Conclusion: Which is better? Should I use one ground plane or two ground planes on a two-layer PCB?
Thanks in advance.

[pito]

thermal noise generated by the damping and clamping resistors, = separate tracce for that

[Detectorist#1]

Hello pito,
If damping and clamping resistors generate essential noise, why Carl recommended to use clamping resistor with value of 5K instead of value of 1K (very standard) for reducing of switching delay of 1N4148 clamping diodes?

[Carl-NC]

Flood both layers for best results, make sure there are plenty of stitch vias between them. Be careful around high voltage flyback signals, keep the flooding away from these.

[Carl-NC]

Hello pito,
If damping and clamping resistors generate essential noise, why Carl recommended to use clamping resistor with value of 5K instead of value of 1K (very standard) for reducing of switching delay of 1N4148 clamping diodes?

Running a high value clamp resistor creates more noise but allows earlier sampling. It's a trade-off.

[Detectorist#1]

Hi Carl,
It is very interesting for me - what is the difference in added delay from 1N4148 with 1K and with 5K clamping resistor? Give us some numbers please.

[eduardo1979]

Thank you very much, Carl and other forum members, for the clarification. It's now clear to me that a dual ground plane is required for better performance. I'll leave the high-voltage circuit practically without a ground plane to avoid capacitive and inductive coupling.

[Carl-NC]

Typically you can pick up 1-2 us if everything else is already optimized.

[Detectorist#1]

Hi Carl,
My measures of the discharging time of some fast diodes points for 1N4148 diode: 50ns after 0.37A peek current, 100ns after 0.63A and 150ns after 1.5A.
According these measures, the change of discharging time with change of the value of the clamping resistor from 1K to 5K have to be not more than 50ns.
Attached is the file with my measures.

[boilcoil]

Hi Carl,
It is very interesting for me - what is the difference in added delay from 1N4148 with 1K and with 5K clamping resistor? Give us some numbers please.

The larger series resistor before the clamp diodes - the smaller change in total damping resistance at the moment after switching/off the diodes, i.e. the smoother desay curve.

[Detectorist#1]

Hi boilcoil,
It is not so important to use clamp resistor with higher value than damping resistor. In my extremely sensitive to grain size nuggets pinpointer, I use 390oHm damping resistor and 200oHm clamp resistor. Parameters are: Peek TX current - 8A, peek coil voltage - 900V, first sample delay - 4us. The reason for using so lower value of clamping resistor is to achieve minimum current noise with excellent OPA1611AID first amplifier. As result, this pinpointer recognize 1/2 grain Au nugget at 1 inch distance.

[Detectorist#1]

Hi boilcoil,
Also this pinpointer works at 50Hz and the current consumption is only 16mA from standard 9V 6F22 battery.

[boilcoil]

Hi boilcoil,
Also this pinpointer works at 50Hz and the current consumption is only 16mA from standard 9V 6F22 battery.

There is nothing surprising in this,
If we assume that:
Lcoil=35uH,
Rcoil=1ohm,
Ccoil=80pf,
Rdamp=330ohm,
Timp=80us,
Usupp=9V,
Ip=8A,
Fpp=50HZ,
Icons=16mA.,
and you can starting samples 1,3 us after the pulse stops.
So, by using a small diameter coil, it is not difficult to hear a small flake of gold.

[Detectorist#1]

O.K. Your calculations are correct - this pinpointer really works. But initial discussion was about the ratio between the values of damping and clamping resistors and their influence on the parameters of PI MD.

[Carl-NC]

My measures of the discharging time of some fast diodes points for 1N4148 diode: 50ns after 0.37A peek current, 100ns after 0.63A and 150ns after 1.5A.
According these measures, the change of discharging time with change of the value of the clamping resistor from 1K to 5K have to be not more than 50ns.

Sorry for the delayed response, I had to dig out an old test circuit for this. Here are the responses for clamp resistors of 1k, 2.2k, and 4.7k, looking at the output of a 100x preamp:

In each case the damping resistor is adjusted to create about the same amount of overshoot; that is, the same amount of overall damping. From 1k to 4.7k picks up 1us. The test circuit uses an NMOS switch (IRF740) and an inverting preamp, and the preamp gain was maintained at 100.

On the TekPoint pinpointer I used a 10k clamp resistor and I think I sample at 3 or 4 us. I'm curious about your pinpointer; what coil inductance do you use, and what timing and TX voltage?