Driving 'Fiddy's Flybacks' in flyback mode. / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Driving 'Fiddy's Flybacks' in flyback mode.

first previous 5 6 7 8 next last

Move Thread

LAN_403

Ash Small

Fri Nov 14 2014, 02:58PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Wolfram wrote ...

In a regular flyback topology circuit, the MOSFET reverse diode will never be forward-biased as far as I understand, so bypassing it with a schottky diode will not make any difference, aside from increasing switching losses slightly due to the increased drain-source capacitance.

I'm glad you raised this point, Wolfram. I'm aware that this has been discussed here before.

My present opinion goes something like this: When the MOSFET switches off, current still tries to flow, resulting in a forward voltage spike, which, as you point out, will not forward bias the reverse diode.

However, due to the 'transformer effect' during switch off (or is it due to leakage inductance), there is a 'back EMF' which results in a spike in the opposite direction (I'm not completely sure about this, I'm currently reading up on all of this stuff), which WILL forward bias it. (Im not certain about these points)

I think my 'scope may be too slow for me to see exactly what happens when I've played with flybacks in the past, but due to the fact that there is definitely 'ringing', which I have seen and not been able to completely damp in the past, this suggests that the voltage across the MOSFET and reverse diode goes positive and negative.

The bottom line is that it can't do much harm, and 'may' do some good, although, as I said, I'm not certain about this either.

I'm working on snubbers and protection at the moment, and will update the schematic later with the ideas I've arrived at, although I do welcome input regarding this subject.

hen918

Fri Nov 14 2014, 04:47PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

What you are referring to is the flyback topology: what gives flyback transformers their name (even though they are not actually transformers, but coupled inductors) and this is what gives them their unusual characteristics.
This is where the ringing comes from: when the MOSFET switches off, collapsing the magnetic field in the flyback transformer, the voltage reverses and causes a high voltage spike which, in a boost converter (which is very similar but with the "primary" and "secondary" as one winding), would be tapped off with a diode. Stray capacitance creates a little LC circuit that continues the high frequency sine wave until it dies away due to stray resistance.

Ash Small

Fri Nov 14 2014, 05:15PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

hen918 wrote ...

when the MOSFET switches off, collapsing the magnetic field in the flyback transformer, the voltage reverses and causes a high voltage spike

Yep, I'm familiar with the 'coupled inductor' bit, and with the 'stray capacitance' and 'leakage inductance' bit, but it also acts as a 'conventional transformer' while the MOSFET' is 'switching off', as both primary and secondary are conducting during this period.

During this, the 'turns ratio' comes into play, inducing a voltage in the primary that is equal to the secondary voltage/turns ratio. Once current starts flowing in the secondary, the voltage drops, but current can't start flowing instantaneously, due to the inductance of the secondary.

I think this is the 'back EMF' bit, and that this bit would 'forward bias' the reverse diode, but I'm not sure about this.

EDIT: I think there are three different factors involved here, firstly, the primary wants to continue conducting, creating a forward voltage spike, then there in the 'stray capacitance-leakage inductance' bit, then there is the 'transformer action' creating a 'back EMF', but I'm not certain.

I do know you get ringing and voltage spikes, which, I think, means you get both positive and negative spikes, but I don't think my 'scope is fast enough to ascertain exactly what's going on.

Sigurthr

Sat Nov 15 2014, 12:54AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

My scope is too slow as well to really SEE the reflected spikes, but I have run tests with and without body-diode-isolation, and the fet always runs cooler with isolation. Usually I can spare the wasted power from a schottky, so I just add it in anyway now as default. Remember it takes TWO diodes to blockade a body-diode. One forward biased in series with the DS junction, and one antiparallel to the combined DS-Schottky series string.

I will say that the reflected spikes can be surprisingly bad in some configurations. I had been working on a flyback driver circuit for a friend a few months back and spent hours blowing fets (not the FDLs though, cheap 100V ones) as I expected about 65V reflected. Instead I was seeing (on my slow ass 40MHz CRO scope!) spikes over 390V which were causing DS avalanche and massive heating. The culprit turned out to be that the core had a chip missing from it where the air gap spacer should have been, resulting in a HUGE gap (compared to as normal). I sanded the core down to be flush on both halves and reassembled with a new gap spacer and the reflected spikes settled back down to the <75V I was expecting, and the heating went away. It ran happy for hundreds of hours after that.

For Schottky diodes I typically use MUR1560s, because they're inexpensive and I have tons. I haven't had any projects where >15A continuous is exceeded, and I've used them for ~22A pulsed, so if you want more current capacity you'll have to worry about balancing parallel diodes and the inherent flaws, or find a beefier schottky.

Ash Small

Sat Nov 15 2014, 01:44AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sig, can you elaborate on this bit please:

Sigurthr wrote ...

Remember it takes TWO diodes to blockade a body-diode. One forward biased in series with the DS junction, and one antiparallel to the combined DS-Schottky series string.

I pretty much follow/agree with the rest, but I'm not clear about the 'antiparallel' thing.

Thanks for the input.

Sigurthr

Sat Nov 15 2014, 01:49AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Ash Small wrote ...

Sig, can you elaborate on this bit please:

Sigurthr wrote ...

Remember it takes TWO diodes to blockade a body-diode. One forward biased in series with the DS junction, and one antiparallel to the combined DS-Schottky series string.

I pretty much follow/agree with the rest, but I'm not clear about the 'antiparallel' thing.

Thanks for the input.

Not sure where it was coined, but I picked the term up. It means "reverse biased, and in parallel to". I.e. Anode to Source, Cathode to Anode of the series diode.

Ash Small

Sat Nov 15 2014, 02:24AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

It was the 'it takes two diodes' bit that I don't follow.

I can put a Shottky in parallel with the 'freewheeling diode' that forms part of the MOSFET, but where does the second diode go?

Or have I misunderstood something?

EDIT: I agree with adding the Shottky, and I agree that it can make the MOSFET run cooler, but I don't understand the bit about a second diode, unless I've misunderstood something.

Sigurthr

Sat Nov 15 2014, 04:02AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

The losses in the body diode aren't mostly from conduction, but from charge accumulation (and charge neutralization), so simply putting a schottky in parallel doesn't really prevent the body diode junction from charging as there will be a voltage developed across the schottky which will also be seen by the body diode. You have to totally isolate the body diode by preventing it from seeing that reverse voltage at all. See snipped schematic:

DFEe4fr

The image is from my Half Bridge schematic, but you can see the two-diode-to-one-fet arrangement.

Andy

Sat Nov 15 2014, 06:27AM

Registered Member #4266 Joined: Fri Dec 16 2011, 03:15AM
Location:
Posts: 874

What about 1 analog to digital and 1 digital to analog resistor network, with a resistor/ cap time delay. The first ADC reads the input voltage, and triggers a mosfet to open, when more voltage in, more voltage is sent to the R/C, the R/C has a fixed 3.3 hex inverter(10-15v) that powers the DAC, which adjust the the ADC by mosfet/resistor voltage divider.

To get it to work, you increase the voltage with something, which adjust the on time, you send a control signal to the DAC to control the off time.

Hope it makes sense

Ash Small

Sat Nov 15 2014, 12:07PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sigurthr wrote ...

The losses in the body diode aren't mostly from conduction, but from charge accumulation (and charge neutralization), so simply putting a schottky in parallel doesn't really prevent the body diode junction from charging as there will be a voltage developed across the schottky which will also be seen by the body diode. You have to totally isolate the body diode by preventing it from seeing that reverse voltage at all. See snipped schematic:

Yep, that makes sense now, the second Shottky goes in series with the MOSFET. Thanks for posting.

@Andy, I think you'll need to post a schematic for me to follow.

EDIT: I've had a look at the MUR1560 datasheet

and, while they are pretty fast, instict suggests that Shottkys may be better here (lower forward voltage drop and instantaneous reverse recovery), although the MUR1560 is a lot cheaper than the Shottkys I was previously looking at.

first previous 5 6 7 8 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob