A new (Stupid?) class d idea

[Tekko]

I got this idea last night of using a GDT instead of lets say a IR2010 which would spare the gate driver IC´s if the fets blow.

Now i only have IRFP460 but heres the schematic of my idea:
Suggestions are welcome.

[djQUAN]

but how would you introduce dead time?

and also, remember the switching frequencies are also modulated by low freq audio so the driver transfo might saturate......

[Tekko]

I have ran this output stage setup with other stuff and havent experienced any problems with cross conduction, but that was a SSTC and induction heater so i dunno how it will behave here.

Only one way to find out.

But i am not sure on which input of the comparator the feedback should go, actually im not really a fan of self oscillating class d circuits as they tent to lock onto the rails when you push them hard.

Another idea i have is no GDT but putting an optocoupler on the highside gate driver. Also that would require a deadtime circuit, but i think i have one of those laying around on the drive, that one with XOR gates and RCD circuits.

[djQUAN]

I also thought about optocouplers before but it is very hard to find ones that can switch at high enough frequencies.

[Tekko]

Yah and they are also expensive.

[djQUAN]

so the best options are mosfet drivers specifically designed for the purpose or the UCD concept but we already know that.

[Pafi]

Tekko!

If you limit output voltage to 40...60% of supply voltage, and if you modify the feedback, then theoretically it can work. Somehow...

"i am not sure on which input of the comparator the feedback should go"

It depends on the polarity of the transformer, but this fedback method is basically wrong. 2 correct way:
- 2nd order LPF in feedback loop with 200-300 kHz cutoff freq, or
- feedback from after LC, and phase compensation in feedback network. Check UcD!

[ChocoHolic]

I was also thinking about a transformer drive for the power stage. But djQUAN's note about the low frequency modulation was exactly the reason why I skipped this idea. Already from theory the resulting number turns were unpractical.

Please note the transformer has to transfer the low frequency signal of music. Especially bass is causing issues, when driving at strong modulation, means high music signal levels.
Consider up to 95% modulation, close to clipping.
Consider about +/-22V peak output of your drivers... means your transformer has to handle 0.95x22V/1.41=14.8Vrms at may be 20Hz. Unsaint conditions for small signal transformers. This massive low frequency signal is existing in the gate drive signal in the same way as in the half bridge output....
If you ignore this, then at high levels you might run into trouble with poor gate signals and in worst case with killed MosFets. At low levels probably nothing bad would happen.

[Pafi]

Maybe this way, or something like this.

Though I'm not a big friend of gate transformers. I tried everything, but I couldn't reach low enough leakage inductance. I ordered a sample from a factory, but it was much worse then mine.

[Tekko]

I´ll just sample a few HIP4080´s

There problem solved.

[Tekko]

Is it possible to use the internal HIP4080 comparator for the PWM generation ?

That way only an external triangular wave generator would be needed, or a feedback cir if self oscillating.

[Tekko]

Heres an idea:


What do you think ?

Edit: Oops those connections to the left should be before the filter, not after.

[Eva]

Transformers require a waveform with the same volt*seconds product in both directions. Also, leakage inductance prevents a fast enough turn off.

[Pafi]

What do you think ?

Sorry, but I think first you should learn about basic operation principles of PWM modulators.

[Tekko]

So scrap this idea too then.

Problem is i already ordered the gate drivers.

[Pafi]

There is no problem with the gate driver, but you should design a usable feedback.

[Tekko]

Then trash the feedback, i hate feedback anyway.



This is just a slight modification of my previous class d build.

[Tekko]

I take your silence as this will work perfectly without hitch.

Actually ids a modified version of my previous class d build, but instead of two comparators in the pwm, heres only one.

If there are other ways of doing the HIP4080 input without adding more IC´s, tell me about it.

I am thinking about skipping the external comparator and use the HIP4080´s internal one.

[Pafi]

I take your silence as this will work perfectly without hitch.

I didn't find serious error in it, but I didn't check every details. Just try it!

[Tekko]

Btw checking yer site pafi, 2kW pwm, yer crazy!!!

How will IRFP460 (all i have for the time beeing) stand up with a 2ohm load with lets say 50-60 volts in from the power supply?

This thing is for driving a sub, so....

[fokker]

Originally posted by Tekko
Btw checking yer site pafi, 2kW pwm, yer crazy!!!

but I don't think Pafi's amp can be loaded with a lower impedance at all. that just isn't simply how physics works.

[Tekko]

I trust that he knows what hes doin, fokker.

That thing can do 10kW if the fets have low enuf Rds(on)

[Pafi]

Tekko!

It can work with IRFP460 (somehow), but surely this is not the best choice. IRF540N is much cheaper and better (faster, easier to drive, have higher efficiency) if you have only 50-60V supply.

Thanks! You are right: in this amp I used very fast FETs with Rds=0,036 ohm, Vdsmax=150 V, 2 paralelled, in full bridge. Protection takes account chip temperature, so it allows higher current for short time, this results a ~120 A peak current (I measured it with 2 ms impulses on 0,1 ohm load). In real overload-test I attached 7 pcs of 8 ohm speakers parallel (Eminence Kappa 15 Pro), and overdrived by 2-3 dB with bass signal, and protection tripped 3-4 times in 15 minutes. Temperature of heat sink stayed below 50 degrees. The reason why I call it "2kW amp" is at the time of specification I didn't have big enough stabilized power supply, and also I wanted to have significant headroom for music signal.

[Tekko]

My problem is that i cannot buy any fets (granny says no) so i have to use what i have.

[Pafi]

This is incredible!

Should I send some?