Flyback with "active" clamp is OK?

[grigson]

Hello,

I am designing a 330W flyback. (Vin = 180-373VDC, Vout = 80V, 4A)

The problem is that at light load the RCD clamp dissipates about 2.8W.

So i have come up with this switchable RCD clamp.......

http://i43.tinypic.com/2h71de9.jpg

Can you see any problems with this?

Is it OK?

Here is the LTspice simulation file in .txt formar, so converting to .asc allows it to be run in LTspice

http://www.2shared.com/document/ggMIfbbr/ACTIVE__CLAMP__TXT.html

 

[Skogsgurra]

Questions:
1. Is the dissipation lower when there's load? Or is it the same across the 0 - 330 W range?
2. Why do you think the clamp isn't needed when circuit is lightly loaded? Have you tested that?
3. A forward inverter is preferred at these power levels. Doesn't need huge ferrite cores. Have you found a core that works? At a reasonable cost?


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[grigson]

Hi,

I found ETD49 core fom epcos with 2mm gap that works.

This application is audio and average power is going to be very much lower than peak power of 330W.

The above schematic shows that the clamp circuit is different at light load.....the resistance is higher Ohmic.

So it dissipates just 1W at light load now........with a simple RCD clamp it dissipates 2.7W.

So i am saving 1.7W of dissipation at light load....and this is just enough to stop us unecesarily cooking the electrolytic caps when its left on light load for ages.

The dissipation in the "active" clamp shown is more at heavy load..........its about 9W at 330W load.

 

[grigson]

Regarding forward converter, do you mean the two switch forward converter....

...to be honest, our average power is so low (only when someone strums the guitar is load applied to Class D amp) that its not worth having any high side fet drives.

 

[benta]

Skogs, forget it. Previous posts reveal that the OP is completely immune to advice.

Your point #2 is very good "do parasitics change with load...?"

Benta.

 

[grigson]

The "RCDQ" clamp schematic above shows that the circuitry still provides an RCD clamp at light load..........just a less dissipative one than a "plain" RCD clamp.

The circuit is just a high power version of the well known RCDZ clamp.

I somehow doubt i'll be racing anyone to the patents office, but i haven't seen this particular circuit before, and wonder if i've missed something.

I appreciate this is a 330W flyback, though the world of audio is still using mains transformers for many of their power supplys.

 

[benta]

"I somehow doubt i'll be racing anyone to the patents office"

I agree.

 

[MacGyverS2000]

I always come back with a simple question... why?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[grigson]

Why a 330W flyback?
...because its cheap, and we are replacing mains transformers which we could have stuck with...so we dont really have small-size requirement, neither do we have much efficiency requirement......we have no standby power requirement, and we want cheapness and simplicity, and the peak power is 330W , but as you know, music is sound, and sound is inherently "AC" in nature....it consists of compressions and rarefactions in the air .........so there cannot be a contiuous peak "compression".........so that tells us that the compression corresponds to our 330W.......and therefore our average value will be around 0.632 of this, at the very most. In fact we will only be on around 45W average power.

 

[benta]

OK, let's look at one basic premise of your design: your output power is widely varying, right?
The fact is, that in a flyback you store energy in the transformer corresponding to the full 330 W. Every cycle.
Where are you going to put all that energy when consuming only your average 45 W?
Back on the HVDC line? Then you need a third winding.
Cycle-skipping or frequency foldback? Difficult to get stable.

The fact is, that a flyback is a good choice when your load is pretty constant. The high-voltage section in a CRT is a good example.
Flyback is a poor choice with varying loads, a single-switch forward converter would do the job much better, here you only need to recover the magnetising current, not the full energy.

Benta.

 

[grigson]

Single switch forward i did not like because at switch-off, the drain voltage shoots up to 2xvin + leakage spike......and that really is a very high voltage, with a corresponding high switch-off switching loss.

Forwards also need an output inductor, (or rather two in our case because output is +/-40V)......with current mode, these output inductors need coupling, as per page 3 , part 5 of the following......

http://www.ti.com/lit/an/slua119/slua119.pdf

Regarding the energy stored in the core when the load is suddenly removed.......that energy will just go into the output capacitors, and there will be a small amout of voltage overshoot there, but nothing unmanageable.......the overshoot will be minimal because we are doing ccm flyback in current mode with duty cycle maximum of 0.33.........the low duty cycle means our RHPZ is high and so we can get a good feedback loop bandwidth.

We like the flyback because it gives well coupled output coils without needing output inductors.

There is the Active Clamp Flyback, but since we have no size or efficiency requirement, it just doesnt seem to be worth the hassle to be fiddling with adding in extra small inductors to artifically increase our leakage to the value required to give ZVS for this topology......also, active clamp flyback is more expensive and obselesence issues are more time-consuming and expensive to handle.

Regarding the above "RCDQ" clamp schematic, i am surprised it doesnt look as weird to others as it does to me.
That Transistor sat on top of the input voltage bus looks precarious, but i cant see a problem.

 

[MacGyverS2000]

Should be fun to watch waveforms at various points in the circuit with a heavy bass beat playing through the amp.

You can install a screw with a big enough hammer, but if you're putting it into a piece of veneer, well...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[grigson]

Too be honest, its just a little better than the mains transformer that it replaces......but a lot better than that because the mains transformer only gave our required V(out) from a specific V(in).

This is an amp for a 6-string guitar, and not for a bass guitar.

We have introduced integrating current monitors which monitor the amps use and can tell when the average power draw over a period of time represents "non 6-string guitar use" or "non-musical use" of the amplifier......if this is detected, then the amp is shut down for a short period.

The integrating current monitors are just series sense resistors, whose voltage is fed to an op-amp integrator......whose output is fed to a PIC ADC....there are two monitors, one for the plus rail and one for the minus rail.

 

[LionelHutz]

OK, you posted if a 320W flyback was a good idea and yet did it anyways even with all negative feedback.

So, is there any point even commenting on this circuit since you'll likely just use it anyways no matter what feedback you get?

 

[MacGyverS2000]

Lionel, this is the modus operandi of the OP, yet for whatever reason most continue to support it. I at least hope for some amusing anecdotes to come out of it all, but so far I've been disappointed.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[grigson]

There are four sacred laws that i look for in a smps design:-

Thou shalt not overvoltage.
Thou shalt not overcurrent.
Thou shalt not overheat.
Thou shalt pick the cheapest.

and obviously there's EMC to pass, and whatever regulations apply.

The usual sacred laws of small Size, efficiency and standby power dont apply here.

The 330W flyback is just a little less entertaining than the 330W mains transformer.

As long as the four laws are not violated, then i can't see a reason way not for the 330W flyback.

 

[MacGyverS2000]

What about the fifth law?

Thou shalt not stick head in sand and ignore world outside of laws 1-4.

With a lot of time, money, and energy I can make a Yugo go 200mph... but I still have to ask, why? Drive the Ferrari, or the Lamborghini, or the Porsche... they were purposely designed for it from the beginning, they do it reasonably well, and they'll be a lot cheaper in both the short and long run.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[grigson]

In comparison to Bridge's and Forwards....i found the flyback, at the same frequency and the same V(in), has essentially the same peak primary current, and the same RMS input current as the bridge and the forward. (as you know, it depends on the Lp for each, but basically they can be the same)

Essentially, whether Flyback, Forward or Bridge, your primary current is a train of trapezoids. (considering CCM).

 

[MacGyverS2000]

As Lionel said, you continually choose to move forward with projects regardless of any advice given here, so why bother asking? Just do it, and let us know when anything patentable and/or publishable comes out of it...

Dan - Owner

http://www.Hi-TecDesigns.com