[SOLVED] Active clamp voltage

Hi Guys,

I have a high side active clamp forward design with 350V in, but i am seeing 760V on the switching node (very similar to setup as pic)



Does the clamp capacitor control the amount of voltage seen at this point?

That won't work. The MOSFET parasitic diode will forward bias as soon as the HV node swings above the HV DC rail. It'll behave just like a diode and the gate drive will have no affect.
Anyway, why don't you just use a reset winding. There's not much point in an active clamp unless you are using a regenerative snubber. Are you going to operate above 50% duty cycle?

That won't work

Click to expand...

What wont work? I think in the question I am asking what controls the level of voltage that is seen on the switching node I haven't suggested changing anything.

Yes I am running this at up to 65% duty cycle. Ive had the voltage so it was lower on a previous design with similar input voltage and i think the question im asking is what effects the voltage that is seen at the switching mode. If 350V is the input voltage why is there a much larger voltage seen on the switching node in this case?

What do you mean with input voltage? Vt?

The schematic is at least equivocal. It shows the "switching node" tied to COM. What's the reference for the observed 760 V and input voltage? I think it would be helpful to show a complete schematic.

Sorry if i didnt explain in enough detail, sometimes it is hard to know how much and how little to give for each question.

Sure the schematic have is based on this


http://cds.linear.com/docs/en/demo-board-manual/dc1929afa.pdf

Input voltage is 350V (Vt referenced to ground)
And the voltage I am seeing 760V is (Com referenced to ground)

And the voltage I am seeing 760V is (Com referenced to ground)

Click to expand...

Makes no sense for the post #1 schematic which has the switching node shorted to COM.

Sure the schematic have is based on this

Click to expand...

It shows just a basic flyback converter. So we are still left in the dark about the actual "active clamping circuit".

A flyback converter with 0.66 duty cycle has a flyback voltage of at least double the input input voltage, the switching node swings to threefold input voltage referred to ground. A snubber or clamp can only reduce additional overvoltage caused by the transformer leakage inductance, not reduce the basic flyback voltage according to volt-time product.

Is the DC1929A i sent in the link an 200W active clamp forward?

http://cds.linear.com/docs/en/demo-b.../dc1929afa.pdf

Sorry, I only noticed the auxiliary flyback converter, not the forward converter in the Linear schematic. Thanks for insisting.

The volt-time product point is however still valid. The switching node must swing to a level above input voltage, the longer the primary on-time the higher the voltage.

how do linear get away with having a 650V FET for a 400V input then?

I presume the transformer winding ratio is chosen so that the converter runs at rather low duty cycle (e.g. 0.2) with 400v input.

That is what mine does, I have a turns ratio of 4.2, with the output voltage being 24V

Works out to about 0.25 if im correct?

Dmin=(Vout/Vinmax)*(Np/Ns)?

Yes, the maximum transistor voltage with ideal transformer would be 500 V. Clamp operation requires a certain overvoltage. 760V seems too high. Possible reasons:

- clamp control isn't working correctly (capacitor isn't discharged during off-time)
- capacitor too small respectively leakage inductance too high

capacitor too small respectively leakage inductance too high

Click to expand...

This was my initial thought hence post#1, i will bump up my capacitance for both clamp and snubber and see if that will help reduce it and will update.

- - - Updated - - -

I changed the value I had of 1.5nF to the same as they have at 10nF and it dropped to 720V, if i go too high of value i take it there will not be enough time to discharge the capacitor completely?

As far as I understand, there should be no problem of making C larger. You would scale both capacitors and also adjust the damping resistor, achieving a trade-off between power dissipation and possible oscillations. In other words make the series R as low as possible without getting too much ringing. Notice that the low side clamping switch reference circuit in the datasheet uses a single C instead of RCR circuit, a series resistor might not be required in all cases.

it is not uncommon for the (flyback) voltage in a forward converter to be 2x the input volts, esp at 50% duty cycle, above 50% the "flyback" or reset volts will be higher still, to (try to) preserve the volt seconds each way on the Tx primary (to reset the core flux to near zero every cycle)
So 760V not unreasonable, you cannot have a low value reset voltage without having it for a longer time (i.e. lower duty cycle)...

I changed the value I had of 1.5nF to the same as they have at 10nF and it dropped to 720V, if i go too high of value i take it there will not be enough time to discharge the capacitor completely?

Click to expand...

There are problems with the clamp cap getting too high faradic value...basso discusses these in his book "switch mode power supplies"...towards the latter part of the book.
Here is some sims in ltspice of active clamp forward.

- - - Updated - - -

If the active clamp cap is too big faradically, then it can cause problems of saturation of the primary during sudden load transients, as the voltage across the clamp cap takes time to re-adjust to the new load level, and for a while, you will not get your nice v.dt ON = v.dt OFF equation nicle satisfied.

- - - Updated - - -

But yes, you are indeed discovering that the “forward converter active clamp” is not the bed of roses that the apps engineers of the semico’s have been telling us all in their sales articles.
Yes, the FCAC does indeed have the problem of the high reset voltage that’s needed…..and to be honest, why not just use a simple forward converter.
Some say that the active clamp forward is best because it allows you to use a simple and cheap bootstrap high side fet driver to drive the clamp fet with (if its in the high side, not if it’s a pfet in the low side), but in truth, as we have discussed in these forums at length, you can indeed use a simple bootstrap high side drive chip to drive the high fet in a simple forward converter. So whats the point in using an active clamp converter?
Some say the FCAC can give you zero voltage switch on……but in truth that always needs an extra leakage inductor to be added and its more of a detriment than an improvement if you arrange for the vds to swing right down to ground before switching the fet on.

- - - Updated - - -

so what is the point if an FCAC?.....Hmmmmm.....well its interesting isnt it?.......a novel way of resetting the transformer....and theres always engineers in co's who want something novel.

- - - Updated - - -

here is a bit on how the FCAC works, so as to get you into the operation toward knowing th situation with that clamp cap.

In an Active Clamp Forward converter, it is also possible to arrange it to get ZVS of the FET, but again, rarely is it actually worth doing this, and you just use the active clamp for the fact that it resets the leakage inductance in a forward converter without necessarily needing a high side FET.
In the active clamp forward converter, you have a “reset FET”, which you switch on at certain times in each switching cycle, so as to let the leakage inductor’s current “go back the other way”, …then when you’ve got it ‘going back the other way’, you cunningly switch the ‘reset FET’ off, and then the leakage inductor current is forced to flow in a path that ends up discharging the cds capacitance of the main power fet…then when the Vds of this fet is zero, you switch it on.

So that is it in brief, in resonant converters you get an inductive current flowing, and of course, as you know, you cannot break an inductive current, so , just before you switch on the relevant fet (the one in which you wish to achieve ZVS), you re-direct the inductive current, (perhaps by switching off a fet in which it is flowing) and the inductive current then re-directs and then discharges the Cds capacitance of the fet that you are just about to switch on…..to the point where the switch-on happens when its Vds is zero….that is in brief how resonant converters often operate.

…So, you use an inductive current to discharge the Cds capacitance of the fet that you are just about to switch on….of course, since fets have a diode in them, what often happens is that the diode starts conducting and then there really is very low voltage (Doide Vf) held across the fet just before it switches on.

basso discusses these in his book "switch mode power supplies"...towards the latter part of the book.

Click to expand...

may have to invest in this book

thanks treez, probably one of the best replies ive had on the forum about a question I have asked, so I guess what you are saying is that the active clamp is more of a gimmick, rather then bringing something decent to the table? I have wondered whether this topology has been the right one, I guess this is where lack of experience comes into play.

In an Active Clamp Forward converter, it is also possible to arrange it to get ZVS of the FET, but again, rarely is it actually worth doing this, and you just use the active clamp for the fact that it resets the leakage inductance in a forward converter without necessarily needing a high side FET.
In the active clamp forward converter, you have a “reset FET”, which you switch on at certain times in each switching cycle, so as to let the leakage inductor’s current “go back the other way”, …then when you’ve got it ‘going back the other way’, you cunningly switch the ‘reset FET’ off, and then the leakage inductor current is forced to flow in a path that ends up discharging the cds capacitance of the main power fet…then when the Vds of this fet is zero, you switch it on.

So that is it in brief, in resonant converters you get an inductive current flowing, and of course, as you know, you cannot break an inductive current, so , just before you switch on the relevant fet (the one in which you wish to achieve ZVS), you re-direct the inductive current, (perhaps by switching off a fet in which it is flowing) and the inductive current then re-directs and then discharges the Cds capacitance of the fet that you are just about to switch on…..to the point where the switch-on happens when its Vds is zero….that is in brief how resonant converters often operate

Click to expand...

.

Brilliant explanation of what is going on with this circuit, something that has been lacking in most books that I have at the moment or online, most of the information I could find was over complicated and hard to interpret what is actually going on.

Also how do you use the word docs in LTspice, I cant remember how to use them?

Update: Found the issue and i think FVM called it with this, which I did think its was something to do with the clamp

- clamp control isn't working correctly (capacitor isn't discharged during off-time)

Click to expand...

For that LT part, the duty cycle on the clamp I had set to 60%, but with the main FET being at about 30% the clamp was reaching its maximum duty (at 250V in etc it was fine)

the LTspice sims can be run by changing the .txt to .asc then open them in ltspice, and hit the running man icon.
The sim should be in a folder in your c drive, and not run the sim over a network.