Totem-pole gate driver VS Active pull down VS Fast Fet turn off

[dbanner]

One might consider that it's the stray wire connection inductance to the gate in series with the gate capacitance that forms a resonant tank circuit.
That is what can possibly cause the ringing. If it's in the tens of megahertz range or possibly higher.

Perhaps a mosfet with a higher gate capacitance and a series resistor on gate drive, maybe along with a snubber.
I wonder what role if any the so called Miller effect might impact the circuit. But that might be a different story altogether.

[Ferric Toes]

I have now tested 12 different Mosfets in the test jig with scope connections as per Altra's video (2). Every Mosfet type gives a different result for the ringing and switchoff time, with the SPP17N80 not being the best. Maybe the typical input capacitance of 2350pf has something to do with it. The best of the twelve Mosfets was a P5NB40 with only 400-550pF of input capacitance. The tradeoff is the higher Rds-On of 1.5 ohms. The last device tested was an IRF G4BC20W which turned out to be an IGBT. I didn't know I had one of these, but the late Reg Sniff sent be a number of devices to test a few years ago and this was one of them. It gave the cleanest switch off of them all, settling down at 1.5uS, as did the P5NB40. Most of the others only settled at 2.0 - 2.5uS

First plot is the P5NB40 and the second G4BC20W. The yellow square indicates the trigger setting on the falling edge, with the dashed line showing the pulse with the test device removed. I have now speeded up the drive fall time to 50nS but little change in overall waveforms. The cursors were set on an earlier plot to measure the ringing frequency, which is in excess of 10mHz, so it is not the coil/cable, which is about 1.0 mHz.



Eric.

[green]

I have now tested 12 different Mosfets in the test jig with scope connections as per Altra's video (2). Every Mosfet type gives a different result for the ringing and switchoff time, with the SPP17N80 not being the best. Maybe the typical input capacitance of 2350pf has something to do with it. The best of the twelve Mosfets was a P5NB40 with only 400-550pF of input capacitance. The tradeoff is the higher Rds-On of 1.5 ohms. The last device tested was an IRF G4BC20W which turned out to be an IGBT. I didn't know I had one of these, but the late Reg Sniff sent be a number of devices to test a few years ago and this was one of them. It gave the cleanest switch off of them all, settling down at 1.5uS, as did the P5NB40. Most of the others only settled at 2.0 - 2.5uS

First plot is the P5NB40 and the second G4BC20W. The yellow square indicates the trigger setting on the falling edge, with the dashed line showing the pulse with the test device removed. I have now speeded up the drive fall time to 50nS but little change in overall waveforms. The cursors were set on an earlier plot to measure the ringing frequency, which is in excess of 10mHz, so it is not the coil/cable, which is about 1.0 mHz.

[ATTACH]54527[/ATTACH] [ATTACH]54528[/ATTACH]

Eric.

Your recordings show a reverse current spike of over 1.6A at start of turn off. Is current actually reversing or what is causing the reversal?

[Ferric Toes]

Your recordings show a reverse current spike of over 1.6A at start of turn off. Is current actually reversing or what is causing the reversal?

That spike corresponds to the flyback voltage which is of reverse polarity.

[Ferric Toes]

I'm gradually cleaning up the switchoff ringing as this plot shows. It is still the IGBT in the test jig and some ringing disappeared when I switched to a battery instead of the bench power supply. It cleaned up a bit more with a 4.7uF thin film ceramic capacitor across the supply to the drive circuit. The big spike is due to the back emf from the coil at the point on the downward slope of the drive pulse where the IGBT switches off. I suspect that some remaining ringing maybe due to the plug in method of holding the device with its full lead length, rather than cut off as it would be in a through hole pcb.



Eric.

[Ferric Toes]

Attached is the cleanest plot yet. This is for a IRFB9N60 Mosfet, which I had not tried previously. . Also, I thought that I had better check my previous statement that the short positive going spike after the switchoff was caused by the flyback emf. Employing three channels, so that he flyback voltage is displayed: now I am not so sure. The peak of 600V is offset a bit later in time than the short peak. In actuality the Mosfet switches off from the 1.2A level and slightly overshooting the 0V line, then appears to switch on again peaking at 0.9A and decaying to zero in the next microsecond. The measurement of the 600V was at the junction of the HER208 diode and the Mosfet drain. It decays to a level of 250V, dropping only slowly before the next Tx pulse. This reverse biases the diode to give minimum capacitance as seen by the coil.

The supply voltage used was from a 11.1V Li-ion battery pack. Switching to my bench psu gave the same result, but, and this seems important, when I increased the voltage to 12V, a lot of ringing appeared. I then noticed that the Mosfet device was then in avalanche mode. Increasing to 15V and it was much worse. So, at the moment, the moral appears to be - avoid the avalanche situation. Also having too much drive voltage to the gate also induces ringing. I now have an adjustment to vary the gate drive amplitude so that I can go from just conducting plus 2V to make sure I have full conduction.

Finding the best Mosfet for a fast switchoff seems anything but easy. However, what I have now seems to have settled down for times >1.5uS. Next job is a fast preamp.

Eric.

[green]

Hi Eric, still trying to understand how you read non avalanche current with Rd connected across the coil. https://www.geotech1.com/forums/atta...2&d=1613682600 Then I saw the line to Rx. Maybe you are measuring the current thru Rin and avalanche current not current thru Rd?

[green]

Non avalanche current. Wasn't thinking about current flow thru series diode when coil volts was increasing at turn off, charging drain to source capacitance?

[dbanner]

I was curious to try slipping the gate leg of the mosfet through one of those tiny ferrite beads. Should suppress high frequency oscillations completely?

[Ferric Toes]

I doubt it would, as another interesting point arose today and that is, ringing only occurs when the Mosfet is in avalanche. If I reduce the supply voltage, or make the fall time slower on the gate drive, so that the flat top on the back emf becomes a smooth curve, the ringing magically disappears. I tried this with about 20 different Mosfets and all exhibited this. Some appeared worse than others because Vds differs within a range 100V - 800V. Vgs threshold also varies, with the result that if the gate drive is too high, then the gate charge and discharge time increase, which can be seen as a flat line just after the trigger point. Too high a drive voltage extends this line further to the right thereby increasing the time for the coil response to settle. I now have a trimmer adjustment to set the drive voltage just a volt or two higher than the minimum required to just turn the Mosfet on. The setting can be checked in three ways 1) the flyback volts reaches a maximum 2) the current draw on the power supply stays constant. 3) The current pulse as seen across the 0.1ohm resistor reaches a constant amplitude.

The first plot shows a 600V Mosfet with 11V supply to the device and its gate driver. The flyback rises above 600V but with a smooth peak. The second plot shows the supply increased to 13V and the Mosfet going into avalanche, indicated by the flat top. Now we have the ringing with the major excursions falling within the period of avalanche, which is only 0.248uS long, so not much in the scheme of things.

If avalanche is observed by ragged ringing then the Tx supply voltage can be reduced until it disappears, or, Tx pulse width can be reduced to bring the switchoff back to an earlier time on its current growth curve. Any signal loss can be made up by running the TX at a higher repetition rate. Depends what the final objective is.

Another point noted was that the gate resistor value has no effect on the avalanche ringing unless you go to values of 100ohms and upwards. These higher R values cause the drive pulse edge to slow up depending on the Gate /Source capacitance, thereby reducing the flyback volts.



Eric.

[waltr]

This has been some good reading.

So keeping MOSFET out of avalanche is a good thing.
Another way to do this is to reduce the peak coil current with a series resistor. This also decreases the coil Tau and allows longer TX on pulse (good for high conductive targets).

I ended up adding 10 Ohms in series with the coil on the Hammer-head I built. Without the resistor the detection was not quite stable.
Later did distance tests and found that detection distance didn't decrease with lower coil current. Possible due to MOSFET going into avalanche and first sample being cleaner.

[dbanner]

I recently had the good fortune of watching a remarkable tv series CHERNOBYL, produced by SKY, an HBO miniseries. Turns out those mk reactors were never meant to work in " avalanche".....

[kt315]

yes, those guys did the series in our sity. Ignalina's two reactors are same types. they are closed now both.
but they made the movie not on the station, but in layout for learning of staffs. it is faraway from IAES and the city.
all town plans they found/did in Vilnius. so you see in reality just Lithuania, not Ukraina.

https://www.kelioniulaikas.lt/kelion...ieska-visagine
Want to know what happened at the Chernobyl nuclear power plant?
You will find the answers to this and other relevant questions by visiting Visaginas, where the Ignalina Nuclear Power Plant, which operated until the end of 2009 and has supplied electricity to the Baltic States, is located.
Follllowing in the footsteps of the HBO series Chernobyl filmed in Lithuania, we will see an identical reactor, the explosion of which shocked the world, and by controlling the power plant simulator we will feel as if we are the owners of a functioning power plant.

[dbanner]

Hi kt, yeah. Hats off to those who gave their lives

[kt315]

yup, they were just biorobots, living biorobots... someone did want to crash USSR out and got that.