AMX Test Results

[Carl-NC]

Boards arrived yesterday, and I am leaving for vacation on Tuesday. So today I threw together a quick test of the TX circuit. Q1-Q6 were installed, and also the gate driver components. VTX and VTXSW were supplied by a bench supply and the TX clocks by a dual channel DDS generator.

I started out using the IRF740 for Q3,Q4. I quickly found out that if these are allowed to avalanche other parts will get fried. I had to replace Q5 and one of the gate drivers. So I added a 4.7nF "Cextra" to slow down the transitions and reduce the flybacks and everything was grand. Here is the coil current (bottom trace) using a current probe, and the flyback pulses:



The flybacks are about 200V and the current shows to be 2A-pp but I think I have the current probe scale wrong and it is actually 1A-pp. Here is a close-up of a transition:



It shows the transition to be about 6.5us or an equivalent frequency of 77kHz, The coil I'm using is 0.97mH and with the 4.7nF cap the calculated f is 74.5kHz.
I replaced the IRF740s with G3R350MT12 (1200V SiC), removed Cextra, and here are the waveforms:




Flyback is now almost 1000V and the transition is now 1.6us. Note the current transition has some funkiness to it, I don't know why but I can say that VTX did not have great bypassing. You can see a little bit of tilt in the waveform above, and it is more obvious when looking at more repetitions:



Next step is to install the TX buck regulator and the feedback current controller. That will have to wait until after vacation.

Edit: Here is a photo of the board while testing the IRF740s. Note the big brown Cextra.

[ivconic]

Am I wrong if I say that first oscillograms looks cleaner?
The ones with IRF740.
Maybe because Cextra was removed later on other oscillograms?

[Altra]

Nice looking Tx signal. What was your actual supply voltage and current? I've been trying to figure out if the Deemon TX is actually efficient? It ran about 350ma with a 10v supply. The Tx output around 2 amps pp.
Thanks

[Carl-NC]

I suspect the 1m cables and the bench supply are causing problems with the higher slew rate. When I get the buck regulator running it may get better. Or it may get worse.

[Carl-NC]

Nice looking Tx signal. What was your actual supply voltage and current? I've been trying to figure out if the Deemon TX is actually efficient? It ran about 350ma with a 10v supply. The Tx output around 2 amps pp.
Thanks

For the IRF740 VTX was 5.3V @ 430mA, for the G3R350 it was 5.1V @ 500mA. The voltage is high partly because the coil has R=3Ω, and partly for reasons I don't yet understand. I haven't had time to do anymore than a quick look at the circuit.

[green]

Interesting. What tests would show if current tilt correction is needed or even wanted? My latest spice model doesn't appear to need tilt correction.

[Tinkerer]

Interesting. What tests would show if current tilt correction is needed or even wanted? My latest spice model doesn't appear to need tilt correction.

Does your Spice include the RX damping and preamp?

[green]

Does your Spice include the RX damping and preamp?

Not looking at preamp. Rx, 300uH with 1k across it.

[moodz]

Not looking at preamp. Rx, 300uH with 1k across it.

Anything that draws energy from the primary TX coil ( targets / ground / RX coil coupling ) will cause tilt.

High energy losses will also modulate the amplitude ... this affects the RX also .....

I have been able to correct tilt and amplitude even if a target is place on the coil see pic below ... waveform recovery in around 4 milleseconds ... the green waveform is the control voltage for the tilt recovery feedback loop when a target is placed directly ( ie worst case ) on the TX coil. The current peak to peak is 1 amp. The reverse occurs when the target is removed. I am not using Carls method as I am not convinced the amplitude modulation of the current pulses is solved. The tilt "flatness" is around 300 microamps.

[ivconic]

...

[moodz]

For the IRF740 VTX was 5.3V @ 430mA, for the G3R350 it was 5.1V @ 500mA. The voltage is high partly because the coil has R=3Ω, and partly for reasons I don't yet understand. I haven't had time to do anymore than a quick look at the circuit.

I have been tooting around with a CCPI cct for the MAGPI detector.

My initial test TX board worked fine ( 1 amp peak to peak / 500 ma from the supply / 750 nanosecond current transitions etc etc ) EXCEPT i had to raise the tx supply voltage to 4.5 volts to get it to work.

DOH ... why was it not working like my previous circuit ????

What was different this time is that I used 1200 volt SIC devices for the high voltage switches and this made my circuit SICK ( pun intended )

Its the forward voltage drop of the SIC body diode !! which is about 4 volts for alot of SIC devices ... whereas the "normal" fets I was using were at 1.7 volts.
Yes the forward body diodes do conduct during the CCPI operation :-)

A high voltage SIC SHOTTKY diode placed across Drain and Source ( anode to the source ) each SIC FET recovered ( pun intended ) the situation totally

Whereas before I need 4.5 volts ... only need 1.9 volts now for the same result ( 0.5 amp draw for 1 amp peak to peak )
So the total power consumption of the TX board is 1 watt

moodz

[Riss]

Boards arrived yesterday, and I am leaving for vacation on Tuesday. So today I threw together a quick test of the TX circuit. Q1-Q6 were installed, and also the gate driver components. VTX and VTXSW were supplied by a bench supply and the TX clocks by a dual channel DDS generator.

I started out using the IRF740 for Q3,Q4. I quickly found out that if these are allowed to avalanche other parts will get fried. I had to replace Q5 and one of the gate drivers. So I added a 4.7nF "Cextra" to slow down the transitions and reduce the flybacks and everything was grand. Here is the coil current (bottom trace) using a current probe, and the flyback pulses:



The flybacks are about 200V and the current shows to be 2A-pp but I think I have the current probe scale wrong and it is actually 1A-pp. Here is a close-up of a transition:



It shows the transition to be about 6.5us or an equivalent frequency of 77kHz, The coil I'm using is 0.97mH and with the 4.7nF cap the calculated f is 74.5kHz.
I replaced the IRF740s with G3R350MT12 (1200V SiC), removed Cextra, and here are the waveforms:




Flyback is now almost 1000V and the transition is now 1.6us. Note the current transition has some funkiness to it, I don't know why but I can say that VTX did not have great bypassing. You can see a little bit of tilt in the waveform above, and it is more obvious when looking at more repetitions:



Next step is to install the TX buck regulator and the feedback current controller. That will have to wait until after vacation.

Edit: Here is a photo of the board while testing the IRF740s. Note the big brown Cextra.

The oscillogram during current reversal is of the "Pablo Picasso" type, because the output capacitance of G3R350MT12 is 10 pf , BUT ! -at drain-source voltage 800 volts . At a drain-source voltage of about 5 volts, this capacity is already about 200 picofarads …dynamic capacity change

[moodz]

The oscillogram during current reversal is of the "Pablo Picasso" type, because the output capacitance of G3R350MT12 is 10 pf , BUT ! -at drain-source voltage 800 volts . At a drain-source voltage of about 5 volts, this capacity is already about 200 picofarads …dynamic capacity change

Yes the circuit is ringing but that is because this type of circuit "damps" itself by transferring the energy from the current falling pulse to the next rising pulse.
However as I pointed out in a previous post if you use SIC mosfets most of them have very high forward drop voltages in their body diodes ( 4 to 5 volts )
... this circuit relies on body diode conduction and the diode shuts down before all the energy is transferred .... so it rings as the residual energy dissipates.
You can even see the step in current flowing in the 1000volt current results from carl above.
A SIC schottky diode across the drain source of the SIC mosfet fixes most of this problem.

Unless you use a fancy SIC mosfet with inbuilt schottky diode.

moodz

[ivconic]

Any news?

[Carl-NC]

I have had no time to work on this lately. The last thing I did was build out the current control feedback loop on the TX driver and that works just fine. The next thing to try is the tilt control. I have no idea when I will be able to work on this some more.