A quiet forward converter likes only one MOSFET -- why?

Thanks for sharing this Piotr.

May I be heretical and suggest considering a bipolar? One of the Zetex low Vcesat offerings (designed as mosfet rivals) like ZXT849 or FZT853 might also perform well. When quietness is more important than efficiency they can be helpful.

piglet

Can you post the schematic? What transformer are you using?

I've blown out Cree SiC fets going just a bit over abs max gate voltage. They are serious about that spec.

The low-load increase is probably from leakage inductance ringing in the transformer. That shouldn't be much energy if you keep the frequency low. Personally, I'd put a 15 volt zener on the isolated side, rather than damping the primary.

I have found some of their parts work well at 4V below the suggested gate voltage.

In a bit of a hurry to celebrate, so here's the CAD file:

A hand-wound 20mm toroid made of F938 with one instance of the secondary rails. Wound with some random wires of different colour for easy prototyping, hence no isolation. The production unit will use a proper TIW wire.

Some more measurements: the negative rail is -3.54V when unloaded and

-3.46V when the 15V rail is loaded with 1k. No ringing at all. This is an outstanding achievment, given the simplicity of the circuit.

Sure, but even that is enough to kill your precious SiC parts.

That would work too, but since it is the secondary that is ringing, every clone of it would need a zener. That makes 20 more parts, while on the primary side it is just a single gate resistor.

Some more discoveries:

1. The inability to use any other MOSFET that IRFT825 is not related to much lower Cgd of the low voltage parts: an SQJA62 with 680pF between its drain and gate is similarly usless, but in a different setting of the trimpot.

1. Ditto for the series channel resistance. The SQJA62 with 1 Ohm in the source (good for current sensing, anyway) behaves exactly the same way.

No idea what physics is behind that and why the IRFR825 is just perfectly operational and predictable over the entire trimpot range. The resulting unit just rocks, but I would like to understand why -- this is a sci group after all. :-)

Some MOSFET god has smiled upon me, because it was the first suitable MOSFET that I found. If I started with any other part, the entire idea would get canned.

And finally, Happy New Year to all of you!

Best regards, Piotr

Just wondering if moving the catch diode to the non-dot end of the reset winding changes anything - ringing due to interwinding stray capacity etc?

Could you save the dual secondary and instead of making +15V and -3.5V just make 18.5V and syntheise zero with a resistor and 15V zener (or

Happy New Year!

piglet

The pleasure is on my side. :-)

This circuit is heretical enough to make more heresies go unnoticed. :-)

I have only ZTX851. No luck, the part is too slow, total disaster.

Best regards, Piotr

Why not a push pull 50% duty?

That would make you drive all the transformers in parallel and is dead simple

Happy new year

Post the schematic.

Provided you can do it twice!

Happy New Year

Phil Hobbs

Piotr, the three parts you mention trying in this application are dramatically different from each other. For example, the classic current/voltage short-cut name shows the differences, ranging from 6A at 600V to 60A at 60V. Rds by a factor of 300.

p/n Id-Vds Rds Ciss Coss Crss IRFR825 6n50 1.05 1346 76 15 FCP20N60 20n60 0.15 2370 1280 95 SQJA62EP 60n06 0.0037 3900 1700 50

The largest effect is on the MOSFET's capacitances, where Coss ranges over a factor of 22x. It seems your application likes low capacitances. If you've read much of my work on using power MOSFETs, you'll see that I do not think the best parts are the ones capable of the highest currents, like your 60V part. More often than not, the low capacitances are more important. Furthermore, once you select a part like the IRFR825, with its lower 76 and 15pF capacitances, you may be using it in a circuit where, say Crss, is important. Then it may be a good idea to design the circuit for a nominal capacitance several times higher than the FET, and get there by adding a cap from gate to drain. Then you can substitute different MOSFETs, provided their Crss is on the order of, or lower, than your original part, so your part dominates, and the circuit will be insensitive to the exact MOSFET.

What's wrong with the one already posted?

Best regards, Piotr

Sure, I am well aware of it. I wanted to use one with sufficient V_DS_MAX rating and with the lowest possible R_DS_ON in order to minimize the Ohmic voltage drop when loaded, but they simply don't work. At all or the useful trimpot regulation range is comparable to the Brownian motion of its slider. Only the IRFR825 has the right combination of parameters.

I'll try to find something in my stock and check this theory. I have already checked that adding more Miller capacitance doesn't help, so this opposite direction is worth checking. Let's see what I have, the Coss isn't specified on the bag.

Thanks, Win!

Best regards, Piotr

Because it is almost impossible to snub the ringing down to the acceptable level, slew rate control is required as well. But then you have the matching problem, which is not present in the case of a single switch topology.

The secondary-side rectifiers got more complicated as well, you either need to double the number of windings (just 40 secondaries, scary), use a full-wave like the BAS4002 or stay with the half-wave and introduce flux imbalance to the core. None of the problems exists in the case of a single-ended design.

I started with a push-pull. :-)

This forward power stage is so simple that you can duplicate it at almost no cost. Use two transistors and your transformer can have just 5 sets of secondaries, and so on. Even the current sense resistor can be shared. Since here I am limited by the winding area (the TIW wire just needs to be thick) and the core will always be too big magnetically, two smaller units can be a better choice than a bigger single doughnut.

Best regards, Piotr

Piotr, I maintain a giant spreadsheet of MOSFETs. It has 435 entries for 500 and 600V parts. The IRFR825 is not unusual. There are about 20 parts with Rds near 1 ohm, and most of the IRFR825's specs are similar to the others in that set, with the exception of Ciss, which at 1346pF is about 2.5x higher than the rest. Normally considered a bad thing, it does help to minimize the effect of Crss Miller capacitance. Maybe, along with your gate-trimming resistor, this helps your circuit.

There are other similar parts with an even higher Ciss/Crss ratio, such as the FDD8N50NZ.

You might consider using a hard mosfet driver and lowpass filtering before the transformer primary. That's a lot more deliberate.

I think you could adjust the drive duty cycle and use single-winding secondaries and two diodes per channel. Then maybe use commercial transformers.

I don't get the contrast between devices. Was that at the same R_G, or the best setting for each?

I made this,

The effect of limited I_G is partly voltage slew rate, but practically, mostly current slew rate. dVds/dt isn't much affected, not on high voltage types anyway (Cdg is too small at high voltages). To control dV/dt, introduce an R+C from D-G (the R prevents it from causing oscillation).

An example of that, in turn:

What are you seeing, and what is its cause? LL + Cp of the secondary, ringing with diode CJO, excited by primary side current.

Presumably, limiting dI/dt should do, but it must be done smoothly (d^2I/dt^2 and maybe ^3 as well should be limited, in effect), and you may not have obvious gate-side control over this because of acceleration and overswing. Consider if Vds falls too quickly, it basically slams into the MOSFET as it finishes saturating, or whangs into the body diode, and this in turn causes the secondary to bounce some.

In other words, it's not even just dV/dt, or dI/dt, but how quickly the impedance changes from commutation (~linear range, capacitive / high Z) to saturation (low Z) as well.

Modern MOSFETs help you out some here, with the aggressively graded Coss acting somewhat like a gas shock bumper does mechanically (though without the losses). Downside: gate drive is continuing to rise as Vds finishes settling down, which means available Id is still going up -- and the current slew rate isn't very smooth because Gm rises with Id; it's actually accelerating. You'd actually want Id to flatten out, rather than continue to rise, here during the bottom corner.

Adding a source resistor alone doesn't help this, but a source resistor /and/ a capacitor from gate to GND, can. Note how the external capacitor is able to work against Ciss with respect to Vs, whereas driving the gate with a current, it's left to its own dynamic regardless of Vs.

You could further add an RCD network, so that as Vds < Vgs, an additional loading capacitance is introduced, slowing it further. Kind of a dynamic braking Baker clamp. A resistor across the cap ensures it discharges during the off cycle.

Can also introduce inductance in the source circuit, as long as stability is maintained. Keeping it to an R||L is probably wise.

Making it all squeaky clean, I think, is going to depend upon a consistent load condition. You can tune the source circuit to degenerate a load step of, say, 1A, but it will absolutely drag at 2 or 3 or 5A. Because you're limiting slew rate absolutely, not relative.

I suppose a saturating inductance could be used, but it needs to be lossy just right, too (i.e., proportionally so). Likewise the dV/dt control stuff could be made nonlinear to get better range out of it, but it's a hard ask for ceramic capacitors that are say 100pF at zero bias but down to say 10pF at 10V, and also rated for 100!

I may just try out some of these strategies, as I'm a bit curious how to tune the edges and corners of switching stages for general EMC purposes. A somewhat more general solution would perhaps approach LT's SilentSwitcher line, and I'd be curious to have more insight into how they did things.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

This was the same forward proto board, I was just changing the transistors and looking what happens. One mysterious transistor works incredibly well, others are a total disaster. They switch well at full speed (no wonder) and go crazy when a bit asphyxiated (and this is the interesting part I wanted to share with you). I would expect the obvious gradual increase of switching losses, but it is not the case. They either start with unacceptable idle current (~60mA@12V) or trip the PSU CC limit (200mA for prototyping safety) from the beginning. The 825 part starts at ~45mA when switching hard and goes down to ~16mA when throttled. High frequency core losses of the spike snubbed by the ferrite? The common lore is that losses go down as switching time goes down, but not in this case, at least globally. That 30mA equivalent power had to go somewhere...

That was exactly the idea behind my design. Slower charge transfer => slower voltage rise at the gate => longer stay in the linear region. But the results are quite surprising, it works in a (very!) predictable way just in the case of one MOSFET type.

But R_DS_ON is high (1 Ohm here). I think it is the important part of the resulting quietness: allow the inductor voltage to rise smoothly and then drive hard. Win says it is thanks to low capacitance. I'll buy some low cap parts soon and launch an ordalium.

Surprisingly, adding 680pF to SQJA62's Cgd doesn't change anything. The instability just moves to a different point. To contradict myself: adding 400mOhm to the source of SQJA62 also doesn't change anything. Pure X files.

Buy an 825 and replicate my setup. You'll be amazed how stable and predictable this circuit is. Scope doesn't lie. For a not exactly understood reason, but this is a different issue.

But the slew rate is limited only during the switching period. Later Vgs quickly saturates (this is a capacitor, after all) and you have as much current as you would get in the case of a hard switched forward.

This is a very interesting idea, definitely worth checking!

Please do, I'm interested in your findings.

I haven't measured the EMI properties of my converter, but judging from the scope traces, it is close to what an LT3439 produces. What they have and I don't is the precise slew rate control over the entire switching edge. Yes, the IC's waveforms do look beautiful. But they have much worse V_out_loaded/V_out_idle ratio -- I have measured that with exactly the same transformer, the forward reset winding was used as the second push-pull winding. And this is pretty strange, given their output switch ON resistance: 0.5 Ohm typ., 0.95 Ohm max. Very close to that of the

Best regards, Piotr