Simple Active Flyback Damping Circuit Experiment

[Brian Deese]

I put together a circuit idea I had for an active damping circuit, and tested it this morning. I'm going to post pictures over a few posts' so I can describe what is happening. The setup is on my test fixture 12 volts. 150uS Tx Width. 3kHz Frequency thereabouts. 400uH Coil, 2R Coil Resistance. 10R Series Resistance. Cree SiC Nch Mosfet 900V. I have banana jacks on the fixture,with a bnc adapter,most of my coils use bnc. There is an adapter with binding posts' with a 680R resistor mounted for the standard config of 680R across the coil.There is no parallel 1K ohm and inverse parallel diodes in this test. The test circuit is built on the breadboard,and wired to the bnc adapter. The test circuit is an scr MCR22-6 1.5 ampere rms rating with very high 150 ampere for 2uS rating. 1N914 diode connected to the gate,1K resistor,and 100K trim pot in series with the resistor to control gate current and thus turn on delay.680R Resistor in series with the anode,cathode connected to 0V. The circuit allows the flyback to rise to a high voltage of your choosing adjustable with the trimpot. When the scr fires it connects the coil to 0V through the 680R damping resistor and latches until...? Dont know,dont care,it will certainly turn off on the next Tx which will reverse bias it. It also blocks current in the damping resistor during Tx,so you get more bang for your buck there also. 680R not critical value I tried several and this is the highest value that will damp the coil,there are probably gains to be had by trying different values,and scr firing delays to get it down to 0V the fastest. I am seeing critically damped flyback in under 5uS.

[Brian Deese]

Here are some pics of the coil measurements,and the damping resistor.

[Brian Deese]

Undamped flyback scope is on 100V per div and if you look you can see it is -10 divisions so the center graticule is 1000 Volts, so that's 1,272 Volts the 772V on the screen plus the 500V you cant see.

[Brian Deese]

Now I have connected the 680R in the usual way across the coil. 528 Volts flyback and a nice decay curve.

[Brian Deese]

Now I have connected the active damping circuit. One pic is a high 900 Volts adjustment,another is the minimum delay 400 Volts,and the other is an arbitrary 592 volts.

[Brian Deese]

You can see the higher you let it go before you hit it with the damping resistance the faster it will come to zero. If you have a very keen eye you will have seen in my 900 volt adjustment the voltage goes below 0V just. High gain preamp will have that in the basement. I either need to fire the scr sooner,or use a lower value resistor. You might also be wondering why a 900 volt Mosfet will not avalanche at 1272 volts of flyback,I can't answer that one except for production tolerances variance,most mosfets,and igbts I have seen avalanche higher than the datasheet value.

[bbsailor]

I put together a circuit idea I had for an active damping circuit, and tested it this morning. I'm going to post pictures over a few posts' so I can describe what is happening. The setup is on my test fixture 12 volts. 150uS Tx Width. 3kHz Frequency thereabouts. 400uH Coil, 2R Coil Resistance. 10R Series Resistance. Cree SiC Nch Mosfet 900V. I have banana jacks on the fixture,with a bnc adapter,most of my coils use bnc. There is an adapter with binding posts' with a 680R resistor mounted for the standard config of 680R across the coil.There is no parallel 1K ohm and inverse parallel diodes in this test. The test circuit is built on the breadboard,and wired to the bnc adapter. The test circuit is an scr MCR22-6 1.5 ampere rms rating with very high 150 ampere for 2uS rating. 1N914 diode connected to the gate,1K resistor,and 100K trim pot in series with the resistor to control gate current and thus turn on delay.680R Resistor in series with the anode,cathode connected to 0V. The circuit allows the flyback to rise to a high voltage of your choosing adjustable with the trimpot. When the scr fires it connects the coil to 0V through the 680R damping resistor and latches until...? Dont know,dont care,it will certainly turn off on the next Tx which will reverse bias it. It also blocks current in the damping resistor during Tx,so you get more bang for your buck there also. 680R not critical value I tried several and this is the highest value that will damp the coil,there are probably gains to be had by trying different values,and scr firing delays to get it down to 0V the fastest. I am seeing critically damped flyback in under 5uS.

Brian,

When you do these tests and simulations, make sure you include all the typical capacitance that the flyback pulse sees in the following areas.
1. Coax cable length, typically about 10pf per foot and about 7pf per foot for thicker center core insulation thickness with a low dielectric constant materiel.
2. MOSFET coil driver COSS (output capacitance)
3. Coil to shield capacitance due to shield spacing from coil and shield area
4. Coil wire winding capacitance due to wire spacing, insulation thickness and insulation dielectric constant

When you see a sharp flyback spike that dies very quickly, the coil is not seeing any capacitance that would be indicated by any following ripples. More capacitance seen by the coil needs lower damping resistor (Rd) values to quench the energy in these ripples allowing you to sample the earliest with that coil, coax and circuit design. Low capacitance is the goal with higher Rd values with the consequences explained below.

Just make sure your simulations are done with a typical length coax and/or what typical capacitance in the circuits that the coil and flyback pulse would see in an operational metal detector at the end of a 30 to 32 inch coax on a pole shaft.


Theory states that the flyback pulse TC (calculated by the col inductance divided by the Rd value) should be 5X faster than the desired target TC to fully stimulate it. That is why detecting targets with a TC at or below 2uS is very iffy with a delay of 10uS (microseconds). A 2uS target stimulated fully decays in 5TCs. If your detector delay TC is 10 uS and the target decays to 0 in 10uS, guess what... no signal. If you can reduce the delay to 7 or 8 uS you might just pick up on the tail end of that targets signal. That is why eliminating or reducing the capacitance seen by the coil during flyback time will allow you to detect smaller targets. A 300 UH coil with a 750 ohm Rd has a flyback TC of .4 uS and a 2uS target needs a flyback pulse of 2/5 or .4uS.to be fully stimulated. However, the delay on the PI machine needs to be faster to detect this target. Rule of thumb: Lower delay detects smaller targets. Lower delay also means higher coil Rd values.

The input amplifier should get of of saturation very quickly and not add any time to the lowest delay time.

I hope your tests and simulations show this as it will verify this little known and less frequently discussed theory about full stimulation of targets.

Joseph J. Rogowski

[6666]

Good experiments, keep going.

[Brian Deese]

Brian,

When you do these tests and simulations, make sure you include all the typical capacitance that the flyback pulse sees in the following areas.
1. Coax cable length, typically about 10pf per foot and about 7pf per foot for thicker center core insulation thickness with a low dielectric constant materiel.
2. MOSFET coil driver COSS (output capacitance)
3. Coil to shield capacitance due to shield spacing from coil and shield area
4. Coil wire winding capacitance due to wire spacing, insulation thickness and insulation dielectric constant

When you see a sharp flyback spike that dies very quickly, the coil is not seeing any capacitance that would be indicated by any following ripples. More capacitance seen by the coil needs lower damping resistor (Rd) values to quench the energy in these ripples allowing you to sample the earliest with that coil, coax and circuit design. Low capacitance is the goal with higher Rd values with the consequences explained below.

Just make sure your simulations are done with a typical length coax and/or what typical capacitance in the circuits that the coil and flyback pulse would see in an operational metal detector at the end of a 30 to 32 inch coax on a pole shaft.


Theory states that the flyback pulse TC (calculated by the col inductance divided by the Rd value) should be 5X faster than the desired target TC to fully stimulate it. That is why detecting targets with a TC at or below 2uS is very iffy with a delay of 10uS (microseconds). A 2uS target stimulated fully decays in 5TCs. If your detector delay TC is 10 uS and the target decays to 0 in 10uS, guess what... no signal. If you can reduce the delay to 7 or 8 uS you might just pick up on the tail end of that targets signal. That is why eliminating or reducing the capacitance seen by the coil during flyback time will allow you to detect smaller targets. A 300 UH coil with a 750 ohm Rd has a flyback TC of .4 uS and a 2uS target needs a flyback pulse of 2/5 or .4uS.to be fully stimulated. However, the delay on the PI machine needs to be faster to detect this target. Rule of thumb: Lower delay detects smaller targets. Lower delay also means higher coil Rd values.

The input amplifier should get of of saturation very quickly and not add any time to the lowest delay time.

I hope your tests and simulations show this as it will verify this little known and less frequently discussed theory about full stimulation of targets.

Joseph J. Rogowski

Joe you are correct,I left out some info. The coil,and cable are as they would be on a machine with 38" of coax. Yes it is very good coax,I think I measured it at 21pF total.The coil is 24awg teflon stranded,silver plated. Wound a while ago before the old man said no silver only tin plating will do. Its bundle wound,with your pdf printed out on my lap as I worked. Pretty sure it has two layers of PE spiral wrap to space the shield. I do not remember how I wrapped the shield. I should test it for self resonant freq,maybe tomorrow. The switch is a Cree c3m0280090d 20pF output capacitance, 2pF reverse transfer capacitance,150 pF output capacitance.The test board also has a mc33153 driver for the switch that turns on through a 10k resistor,and turns off through a diode and 100R resistor. That's probably a little more stout than a 555 output. One thing about this test is that it will overload the 1k input resistor if it is in circuit,and you don't get nearly as good of results if it's in circuit. You can't heat the input resistor to the preamp,I'm pondering that dilema.

[Ferric Toes]

Joe you are correct,I left out some info. The coil,and cable are as they would be on a machine with 38" of coax. Yes it is very good coax,I think I measured it at 21pF total.The coil is 24awg teflon stranded,silver plated. Wound a while ago before the old man said no silver only tin plating will do. Its bundle wound,with your pdf printed out on my lap as I worked. Pretty sure it has two layers of PE spiral wrap to space the shield. I do not remember how I wrapped the shield. I should test it for self resonant freq,maybe tomorrow. The switch is a Cree c3m0280090d 20pF output capacitance, 2pF reverse transfer capacitance,150 pF output capacitance.The test board also has a mc33153 driver for the switch that turns on through a 10k resistor,and turns off through a diode and 100R resistor. That's probably a little more stout than a 555 output. One thing about this test is that it will overload the 1k input resistor if it is in circuit,and you don't get nearly as good of results if it's in circuit. You can't heat the input resistor to the preamp,I'm pondering that dilema.

Hi Brian, A very interesting experiment and look forward to more test results.

The Cree Mosfet looks interesting, although more expensive than other high voltage Mosfets that I have tried. If it gives greater performance then it will be worth it and I wll be getting one for sure.

A few points from the data. You mention 150pF as being an output capacitance, but this is input capacitance, Ciss, and remains the same whatever the Vds is when above 0V. The Coss of 20pF only applies for voltages above 500Vds and at 12Vds will be about 160pF. This is much less that some other high voltage Mosfets, but still significant. The reverse transfer capacitance will be about 20pF at 12V and 2pF for voltages above 500V.

The Coss is important as it appears across the coil, adding to the other capacitances. As stated in an earlier thread on Coss, this can be significantly reduced by inserting a series high voltage diode between coil and Mosfet drain. I use either HER208, HER158, or MUR460 for higher currents. See 'Caught out by the Coss' thread started on 20th Nov.2019. The series diode keeps the Mosfet Coss charged at the flyback voltage between pulses, the diode also being reverse biased has minimum capacitance of about 5pF, which is what the coil sees. You can also add a bit of external capacitance between drain and source if there is significant droop of the high voltage between TX pulses.

Try this and then add your damping circuit across the coil.

Eric.

[ivconic]

The era of your ruthless hegemony is coming to its end Eric!
Because i am just about to obtain my first analog oscilloscope, and brand new one it will be!

[dbanner]

Speaking of analog oscilloscopes, I wish I could get my hands on one of those vintage tektronix.

Unfortunately the shipping cost is astronomical. It is actually cheaper to buy an entry level digital oscilloscopes than the old behemoth analog scopes, just because of the weight.

[ivconic]

Speaking of analog oscilloscopes, I wish I could get my hands on one of those vintage tektronix.

Unfortunately the shipping cost is astronomical. It is actually cheaper to buy an entry level digital oscilloscopes than the old behemoth analog scopes, just because of the weight.

Problem is you can't do much with entry level digital oscilloscopes, especially not when dealing with PI detectors.
While any decent analog will meet the demands.
The other day i explored my options on that.
Either to invest ~200e in analog one... either ~2000e in digital one, to do the same job with it.
Tektronix is alright, providing it is in good shape.

[Qiaozhi]

The era of your ruthless hegemony is coming to its end Eric!
Because i am just about to obtain my first analog oscilloscope, and brand new one it will be!

I've never thought of Eric as a hegemon.

[ivconic]

I've never thought of Eric as a hegemon.

Neither than i. It was a joke.
But than again yes, the "hegemon" in most positive way, why not?
The "hegemon" of the finest PI knowledge.


P.S.
Really?!
George the squarepants!