Caught out by the Coss - Effect of the output capacitance of a mosfet TX in PI.

[green]

HI Eric
A chart I posted few years back. I have been using a diode for a long time with my testing and the detector I'm playing with. Can sample sooner. Looking at the chart, don't see why there would be a difference in detection distance with the targets charted. Think higher amplitude with no diode is due to a higher charge volts(no diode drop)? Would expect a difference with a target time constant around 1us(1 or 2 grain nugget). Maybe I'm missed something?
Interesting thread

[KingJL]

Being in an experimentive mood, I thought I would see what effect adding more capacitance to the Coss would have. I added 1000pF in the form of a 350V polystyrene capacitor between source and drain of a 24N60 mosfet, and then a humble IRF740. The result was surprising. Below are scope shots for the IRF740. First, without the 1000pF and the second, with. With the capacitor there is no measurable drop between pulses of the 400V, and we are getting the full 400V instead of 275V as in the first shot. The 24N60 had similar results but a slightly lower voltage. The receiver waveform showed no change in either situation. Counter-intuitive but it appears to work.

[ATTACH]48311[/ATTACH] [ATTACH]48312[/ATTACH]

Eric.

Surprising indeed! It would appear by the fact that the receiver waveform showed no change, that the added 1000p added across drain-source had no effect on the capacitance seen by the coil/damp circuit... interesting. Also, it appears that the bleed-off (without the added 1000p) is probably occurring across Cdg as it is the primary contributor of Coss (according to documentation I have read on MOSFET theory, Coss is made up of primarily Cdg plus body diode capacitance). It would be interesting to know the in circuit dynamic resonant freq of the coil and then calculating the capacitance seen at the coil. I think you could increase Rd untill you got significant ringing and then adjust scope trace to only see the ringing and use the FFT view to see the frequency. All of this is extremely interesting and like you say... "counter-intuitive". And I really like the last observation... "but it appears to work"!

[bbsailor]

Being in an experimentive mood, I thought I would see what effect adding more capacitance to the Coss would have. I added 1000pF in the form of a 350V polystyrene capacitor between source and drain of a 24N60 mosfet, and then a humble IRF740. The result was surprising. Below are scope shots for the IRF740. First, without the 1000pF and the second, with. With the capacitor there is no measurable drop between pulses of the 400V, and we are getting the full 400V instead of 275V as in the first shot. The 24N60 had similar results but a slightly lower voltage. The receiver waveform showed no change in either situation. Counter-intuitive but it appears to work.

[ATTACH]48311[/ATTACH] [ATTACH]48312[/ATTACH]

Eric.

Eric and all interested members

Series capacitance such as putting a diode (with it's capacitance) in series with the coil and MOSFET will limit the capacitance as seen by the coil due to the fact that the total capacitance will always be slightly less than the smaller of the series capacitances. Example: a 1000 pf in series with a 100 pf will be calculated to be 90.909 pf. From my experiments, anything you can do to reduce the total capacitance as seen by the coil design itself, coax cable capacitance and TX circuit will allow a higher Rd value which in turn will allow earlier sampling. Once more point that I saw on Eric's forum was a research article that said that to fully stimulate a target the discharge coil TC (coil inductance divided by Rd) should be 5 times faster than the target TC. Herein lies the reason why detecting small 1us TC targets requires full stimulation to assure maximum sensitivity to these targets. If a 300 uH coil has a 1000 ohm Rd the coil discharge TC would be 0.3uS. This would be good for a 1.5uS target. The best Rd for this 1uS target would be 1500 ohms to fully stimulate it.

The next issue is the coil diameter relative to the target size and TC but requires understanding how to do this critical balancing act to maximize the total performance on low TC targets. Eric was one of the first PI designers to use the Lock-in Amplifier effect to integrate many RX samples to improve the signal to noise ratio and detect very weak signals normally lost in noise.

Eric, do you still have that article that describes full target stimulation being five times faster than the target TC?

Joseph J. Rogowski

[Ferric Toes]

Surprising indeed! It would appear by the fact that the receiver waveform showed no change, that the added 1000p added across drain-source had no effect on the capacitance seen by the coil/damp circuit... interesting. Also, it appears that the bleed-off (without the added 1000p) is probably occurring across Cdg as it is the primary contributor of Coss (according to documentation I have read on MOSFET theory, Coss is made up of primarily Cdg plus body diode capacitance). It would be interesting to know the in circuit dynamic resonant freq of the coil and then calculating the capacitance seen at the coil. I think you could increase Rd untill you got significant ringing and then adjust scope trace to only see the ringing and use the FFT view to see the frequency. All of this is extremely interesting and like you say... "counter-intuitive". And I really like the last observation... "but it appears to work"!

A few more quick tests today. I have a few 2000pF polystyrene caps in my component store so I had 2 in series for 1000pF; 1 for 2000pF and 2 parallel for 4000pF, and 3 parallel for 6000pF Using a IRF740, I get 400V for 1000pF, 350V for 2000pF, 300V for 2//, and 260V for 3//. Since the back emf charge is the same, it make sense for the voltage to go down as capacitance is increased. For the internal Coss, as it is a voltage dependent capacitance, could it be that as the charge leaks away, the capacitance rises, giving of itself a drop in voltage, hence contributing to the slope when no additional fixed capacitance is used? None of the above capacitor values has any effect on the RX recovery for which sampling could occur at 10uS with the coil and cable used
at 2A peak coil current.

Eric.

---

[Ferric Toes]

Eric, do you still have that article that describes full target stimulation being five times faster than the target TC?

Joseph J. Rogowski

Hi Joe, I believe it was a paper entitled 'A pulsed bomb locator' from 1956. Maybe it was recognised earlier in geophysical papers by J. R. Wait. I still have the papers but recently having moved, they are still safely packed away, somewhere.

Eric.

[Ferric Toes]

I now have a 1kV high voltage probe and the scope pictures are quite different, as I expected they would be. Gone are the curves and the true transient conditions are revealed.

The 1st picture shows the voltage steps at the junction of the diode and the mosfet. Vertical scale is 200V per large division for top trace. The slope toward the right is the charge leaking away in the time period between pulse. The bottom trace is the coil current waveform taken across a 0.1 ohm thick film resistor. About 2.2A for the 24N60 transistor. Next is the addition of a 2200pF 600V polyester capacitor across drain and source. Here we see that 450V is maintained throughout the off period., resulting in a very low capacitance across the coil. Measurement of the coil plus cable resonance is 769kHz by signal generator only, and 714kHz when connected to TX minus the damping resistor.

This is an expanded view of the switch off of the TX pulse showing the rise time of the flyback (top trace) and the voltage across the 0.1 ohms in the TX return. The time scale is 0.5uS per large division a 100V/div. for the top trace. It is interesting to see the initial rise goes to 650V and the flat top indicates that the avalanche voltage has been reached. Notice the ringing at the end of the switch-off period. This is not due to the coil, as lowering the value of the damping resistor has no effect. There must be some small spurious inductance somewhere.

The last picture shows the effect of the addition of the 2200pF capacitor. The ringing on the voltage trace has cleared up but there appears to be some on the fast edge of the TX current decay. this does not appear on the RX output which was not connected for these tests.

Eric.

[KingJL]

I now have a 1kV high voltage probe and the scope pictures are quite different, as I expected they would be. Gone are the curves and the true transient conditions are revealed...

Great information... your efforts/experiments confirm some of my assumptions.

... resulting in a very low capacitance across the coil. Measurement of the coil plus cable resonance is 769kHz by signal generator only, and 714kHz when connected to TX minus the damping resistor...

By my calculations 769 kHz for a 300 uH coill calculates to ~142 pF... 714 kHz calculates out to ~165 pf... So, the coil sees only ~23pF added by the TX circuit (MOSFET, diode, stray capacitances, etc.).

Great work!!

[dbanner]

I wonder if there is perhaps a suitable use for varactor diodes somewhere?

[ivconic]

I now have a 1kV high voltage probe and the scope pictures are quite different, as I expected they would be. Gone are the curves and the true transient conditions are revealed.

[ATTACH]48367[/ATTACH] The 1st picture shows the voltage steps at the junction of the diode and the mosfet. Vertical scale is 200V per large division for top trace. The slope toward the right is the charge leaking away in the time period between pulse. The bottom trace is the coil current waveform taken across a 0.1 ohm thick film resistor. About 2.2A for the 24N60 transistor. Next is the addition of a 2200pF 600V polyester capacitor across drain and source. [ATTACH]48372[/ATTACH] Here we see that 450V is maintained throughout the off period., resulting in a very low capacitance across the coil. Measurement of the coil plus cable resonance is 769kHz by signal generator only, and 714kHz when connected to TX minus the damping resistor.

This is an expanded view of the switch off of the TX pulse showing the rise time of the flyback (top trace) and the voltage across the 0.1 ohms in the TX return. [ATTACH]48366[/ATTACH] The time scale is 0.5uS per large division a 100V/div. for the top trace. It is interesting to see the initial rise goes to 650V and the flat top indicates that the avalanche voltage has been reached. Notice the ringing at the end of the switch-off period. This is not due to the coil, as lowering the value of the damping resistor has no effect. There must be some small spurious inductance somewhere.

The last picture shows the effect of the addition of the 2200pF capacitor. [ATTACH]48373[/ATTACH] The ringing on the voltage trace has cleared up but there appears to be some on the fast edge of the TX current decay. this does not appear on the RX output which was not connected for these tests.

Eric.

I am having diy Pulse Star 2 on the bench these days.
It shows up great potential in course of the "depths" , considering what i can see now in experimental conditions.
And when i say "deep"; i mean literally "deeeeep".
Also very agile and fast recovery on targets (LF411 in front end).
Fet is IRF630.
However... there is a 150v/2W (i put 180) zener diode across the fet.

How about that? Any free thoughts?

[eclipse]

Zeners are quite noisy.
However Pulse Star 2 doesn't seems to me like a device doing early sampling (probably more like >=20uS at the lowest setting) so the zener effect won't be important.

[Ferric Toes]

Great information... your efforts/experiments confirm some of my assumptions.By my calculations 769 kHz for a 300 uH coill calculates to ~142 pF... 714 kHz calculates out to ~165 pf... So, the coil sees only ~23pF added by the TX circuit (MOSFET, diode, stray capacitances, etc.).

Great work!!

I will have to repeat the resonant frequency tests as I believe the probe capacitance across the coil may have given an erroneous measurement when testing the coil on its own. The junction capacitance of the MUR460 should be 10pF or less at 400V reverse voltage. As bbsailor pointed out, the extra capacitance seen by the coil from the TX circuit should be <10pF.

Eric.

[bbsailor]

I will have to repeat the resonant frequency tests as I believe the probe capacitance across the coil may have given an erroneous measurement when testing the coil on its own. The junction capacitance of the MUR460 should be 10pF or less at 400V reverse voltage. As bbsailor pointed out, the extra capacitance seen by the coil from the TX circuit should be <10pF.

Eric.

Here is another unusual capacitance effect on PI coils. Only about 20% of the coil to shield capacitance affects the resonant frequency lowering. If you measure the coil to shield capacitance with an LCR meter by putting one probe on either coil wire and the other probe on the shield you will find out that not the full coil to shield capacitance lowers the self resonant frequency from measuring it without any shield. This is called distributed capacitance and is much easier to measure than attempt to calculate. I discovered this when writing my Fast Coil article which is in the projects section of reference articles on this forum.

Joseph J. Rogowski

[KingJL]

I will have to repeat the resonant frequency tests as I believe the probe capacitance across the coil may have given an erroneous measurement when testing the coil on its own. The junction capacitance of the MUR460 should be 10pF or less at 400V reverse voltage. As bbsailor pointed out, the extra capacitance seen by the coil from the TX circuit should be <10pF.

I am used to working with the government... plus or minus 10%... and that depends on whether the government is the provider or the procurer of service... if its the government as the procurer then you need to be within 0.01%!!!

[Ferric Toes]

Well, I am happy with 1% accuracy. I did some tests today and found that the scope leads did have a small effect, particularly on the x1 setting. x10 is better and best is the x100 high voltage probe. My initial method was to insert a 47K resistor in series with the coil and drive from a signal generator with the scope connected to the live coil end. The signal generator was then tuned to give a peak waveform on the scope, and then the period was measured with the cursors. A better way is to have a small PI coil driven by the PI Tx and Rx and have it near to the unconnected coil under test. The ringing of the unconnected coil will appear on the output of the Rx amplifier where it can be measured without the effect of scope probe capacitances.

The shielded coil on its own with the shield connected to one end of the coil measured a period of 0.96uS (1.41MHz). With 1.6m of Vandamme coax video cable, the period was 1.46uS (685kHz). The cable on its own contributed 120pF. Coil by the way is 298.8uH, 1/0.4 PTFE insulated wire, and the coil/shield capacitance is 122pF.

This coil and cable when connected to the Tx consisting of 24N60 mosfet; MUR460 diode and 1000pF capacitor across source and drain, then gave a ringing frequency of 1.6uS (625kHz) with the damping resistor disconnected.
Increasing the added capacitance to 3000pF had the unusual effect of reducing the period to 1.576uS (634Khz) and still further with 4000pF to 1.53uS (653kHz). There are other variables of course as the back emf reduces and also the switch off rate with the higher capacitances.

This certainly shows that a suitable diode is very effective in isolating the coil from any capacitances in the mosfet circuit and allows high voltage, low Rds, mosfets with their high Coss at voltages below 100V to be used effectively.

Eric.

[KingJL]

Well, I am happy with 1% accuracy...

It reminds me... Many (many) years ago I worked in an outfit called GEEIA (Ground Electronics Enginneering and Installation Agency)... our motto was "Measure with a micrometer, mark with a chalk line, cut it with an axe".

This certainly shows that a suitable diode is very effective in isolating the coil from any capacitances in the mosfet circuit and allows high voltage, low Rds, mosfets with their high Coss at voltages below 100V to be used effectively.

All this is great information... when I have a bit more time, I need to capture all this in my notes!!!