Ground Balance Theory

[green]

Hi Aziz
VRM decay measurements require precise and accurate timing records.

I plot decay curves because each time I try to get them better and I learn something trying. Sometime the T.C. are different than stated in other posts. What is important in measuring ground response? I'm guessing 10 or 100 Mhz amplifier response doesn't matter and 100khz or 1 Mhz does. What is the minimum frequency requirement needed to measure ground response. I'm thinking coil resonance and damping might be important. Reading the replies above, on time is important. Is charge shape important? When does decay time start? Does a flat peak on the flyback matter? What else is important. Is there a test that be could done on the setup that would check for minimum requirements? Maybe recording a target with a 1 usec T.C. I'm thinking this is important for measuring ground response, maybe, maybe not for rejecting ground.

[Dave J.]

Green, the simple answer is that all of it matters.

--Dave J.

[Aziz]

Hi Aziz
VRM decay measurements require precise and accurate timing records.

I plot decay curves because each time I try to get them better and I learn something trying. Sometime the T.C. are different than stated in other posts. What is important in measuring ground response? I'm guessing 10 or 100 Mhz amplifier response doesn't matter and 100khz or 1 Mhz does. What is the minimum frequency requirement needed to measure ground response. I'm thinking coil resonance and damping might be important. Reading the replies above, on time is important. Is charge shape important? When does decay time start? Does a flat peak on the flyback matter? What else is important. Is there a test that be could done on the setup that would check for minimum requirements? Maybe recording a target with a 1 usec T.C. I'm thinking this is important for measuring ground response, maybe, maybe not for rejecting ground.

Hi green,

as Dave mentioned, all and more matters.

The TX coil in a PI has limitted bandwidth. You can approx. assume, that the upper corner frequency is at the self resonant frequency of the TX coil. Let's say 500 kHz - 800 kHz.

The front-end amplifier has also limitted bandwidth. The bandwidth is dependent on the circuit implementation, op-amp gain bandwidth product and gain. The higher the amplifier gain, the lower the bandwidth.
The front-end amplifier should be able to deal with at least 100 kHz to 500 kHz. If you wanna measure the VRM response only, it also makes sense to reduce the gain of the pre-amp (getting faster). A pre-amp gain should be in the region of 10 x to 50 x.

If a front-end amplifier hasn't enough bandwidth, it won't amplify the high frequency input signals (at early decay times) as much as it is specified for low frequency region. So you get a non-linearity in the early timings.

The non-linearity doesn't really matter much. Even a reduced bandwidth. As long as you can measure and model the non-linearity (transfer characteristics of the front-end) by calibrating the measurement system, you can correct the VRM measurements data easily. The non-linearity of the front-end system can be determined with a single time constant (TC) target response as it should be a straight line in the log response/linear time plot. You can use a thin single wire loop with a resistor soldered to it for this purpose. The resistor would define the TC with the wire loop diameter.

Of course, you have to measure the air only response (no targets, no ground response, simple air test) to subtract it out from the VRM measurements to eliminate the effects of the flyback damping process, DC signals, etc.

Yes, charge shape is important.

"When does decay time start?"
That is dependent on your TX circuit and TX damping. Usually at switch-off if you can damp the coil current fast enough. But it also may be prolonged.

"Does a flat peak on the flyback matter?"
Yes.

What is else important?
- Measure the air-test decay only (as reference for subtracting the distorting effects)
- Measure the VRM decay only (only ground response measurement, no target)
- Measure the target response only (with a single TC, no ground response)
You get enough information for calibration and isolating the ground response.
Ground + target response isn't important. You can look at the individual measurements (both responses superimpose).
- Temperature (VRM response is temperature dependent)
- Low noise measurement
- Stable & noise free supply voltages
- Pulse on-time (the VRM response is dependent on this)
and so on...

Cheers,
Aziz

[Aziz]

BTW,

one could also look at the VRM response with the superimposed target response and the effects of the target response to the ground response signal.
As mentioned, you would need a target only response, ground only response and target + ground response. I'm sure, there will be an interesting effect. And we would have all the infos to analyse it.

And we have the conducting magnetic susceptible/viscous minerals too, which screw our GB.
Too many variables to analyse. We would require dozens and hundreds of accurate measurements.
Anyone want to make a PhD?

Aziz

[green]

Aziz
Thanks. I work in measurement, torque, speed, displacement, temperature, etc. Metal detectors are new to me. I hope to try to measure some of the variables, mostly for fun and to learn something. Knowing what to compare should help. With all the variables it looks like it would be easy to get different decay plots.

[Aziz]

Hi guys,

here is a good and very valuable job for you to figure it out.
Investigate the relation of resistive response (R) (=target response) to reactive response (X) (=soil, ground response) in an IB coil configuration.
What is better?
On-time or off-time response relation (R/X).

Is the off-time sampling (usually in a PI configuration) really better?
Who can do the investigation (measurements, scope pics and data)?

Aziz

[green]

Hi guys,

here is a good and very valuable job for you to figure it out.
Investigate the relation of resistive response (R) (=target response) to reactive response (X) (=soil, ground response) in an IB coil configuration.
What is better?
On-time or off-time response relation (R/X).

Is the off-time sampling (usually in a PI configuration) really better?
Who can do the investigation (measurements, scope pics and data)?

Aziz

An attempt at measuring on-off time. Could measure other targets. Any suggestions for data presentation.

[Aziz]

An attempt at measuring on-off time. Could measure other targets. Any suggestions for data presentation.

Fine. Now add some magnetic susceptible samples to the measurements and compare the relation of resistive response (eddy current response of targets) to reactive response (response of magnetic susceptible samples only).
But the response is waveform dependent. Show the coil current, so we can interpret the results better.
Aziz

[green]

Attempt #2. The data was absolute valued for the log plot.

[Aziz]

Attempt #2. The data was absolute valued for the log plot.

Fine again. Thanks for your effort.

We need a resistive (R) target response, which is totally free of reactive (X) response. Nail isn't appropriate as it is made of iron/steel and is producing both responses dependent on the orientation. Use a copper or Al target for this purpose. Ferrite is ok but be careful. Not all ferrites do produce a response of a "natural ferrite" (magnetite, maghemite and other magnetic susceptible materials).
It is important to see them all together. The resistive and reactive response during the on-time and off-time. We can see, in which region we have a much better relation of R/X. And it will answer the question, why the PI technology has still merit.

Cheers,
Aziz

[Dave J.]

This is good entertainment, but I won't say why.

If anyone wants me to say why, talk to my boss first. It's his IP, he paid for it.

--Dave J.

[green]

This is good entertainment, but I won't say why.

If anyone wants me to say why, talk to my boss first. It's his IP, he paid for it.

--Dave J.

I'm playing with metal detectors to try and learn something. Some of my measurements don't match other posts. Why I don't know, probably doesn't mater. If my curves are wrong or don't make sense say so. I'm sure the curves have been plotted many times, but I haven't seen them. Entertainment might be the best reason for doing it.

[Aziz]

I'm playing with metal detectors to try and learn something. Some of my measurements don't match other posts. Why I don't know, probably doesn't mater. If my curves are wrong or don't make sense say so. I'm sure the curves have been plotted many times, but I haven't seen them. Entertainment might be the best reason for doing it.

Please Mr. green,

just go on and ignore Dave. He is trying to troll and to strangle the GB topic.

His "fish equation" is a good example to make the thread off-topic. That had no reference to our GB here and has confused more members (including myself - oh yeah!, I've learned the Poisson distribution function along with Gaussian distribution, Bernoulli process etc. etc. and I could even calculate his fish example - but that is very off-topic here).

Just troll back (like I do it: G(t) = a*(t+p)^b ).

Aziz

[Dave J.]

What makes all this talk about "the curves" so entertaining is that in the end, the curve you get depends on the apparatus you use to generate the curve. A practical ground balancing PI has to be balanced empirically.

The Fish Equation goes straight to the physics of the superparamagnetic-singledomain transition, I'm a bit surprised that Aziz is still discombobulated about the matter.

--Dave J.

[Aziz]

What makes all this talk about "the curves" so entertaining is that in the end, the curve you get depends on the apparatus you use to generate the curve. A practical ground balancing PI has to be balanced empirically.

The Fish Equation goes straight to the physics of the superparamagnetic-singledomain transition, I'm a bit surprised that Aziz is still discombobulated about the matter.

--Dave J.

Fine Dave. You're going to calculate the number of magnetic domain changes. But you still have to know the rate of change in a time interval before you can calculate the probability of the successive/other time intervals. And the rate of change must be stable (const).
But we know these processes like the radioactive decay.

Where is the logic link to the VRM decay now? Oh yeah!, the exponent b.
But we cannot say, that the exponent b stays constant. It may stay constant for one sample but for a different sample it is varying. And it is completely pulse history state dependent too. And other factors do change exponent b as well.

Well, I'm not reading always off-topic stuff. That's is probably the reason, why I didn't understand you and your fish example. So I got very confused and have tried to make the link to the related stuff. But I got more confused. You have to be more precise in your wording next time.

Cheers,
Aziz

PS: G(t) = a*(t+p)^b