500w Class D

[Genomerics]

I would guess, know, that the devil is in the details. Unfortunately without enough details I am left guessing.

Being 'silly' I'll believe what you are saying especially since you mention your setup is having problems above 6KHz. That is possibly the hint....

Even with such a caveat in place I wouldn't have problems with the idea that it does a good job of 'playing music' because of the nature of the input signal.

However drzhuang has given us a 20KHz plot of his amplifiers output. Without the details it looks impressive, with the information given so far.. well I'm left having to draw conclusions.

Cheers

DNA

 

[Eva]

I wouldn't regard his plot as so impressive. I have just recorded this, it shows my circuit playing 5Khz into a speaker load (3 ohm in series with 0.3mH) just before slew rate limiting arises (20Vp-p with 30V DC single supply). I know that 5Khz is not the same as 20Khz, but my filter resonates at 3.6Khz and my switching frequency drops to 28Khz during the tops of the waveform, these are not the best conditions to generate a clean 5Khz wave. I'm showing also the output of the current amplifier going into the high/low comparator (it's average current mode like a plain SMPS). Note that the signal going to the comparator is actually nearly 180 degrees shifted with respect to the output.

EDIT: 500Hz just for comparison (and without changing the amplitude in the signal generator). Some LF content is shown because I'm using tone bursts.

 

[Genomerics]

Unfortunately you are placing me in a similar situation to the one I have with drzhuang.

However you are giving me more information to work with, I might have to throw some bits out to get an approximation but I might be able to do something with what you are telling me.

Of course you might just spill the beans and save a bit of heartache.

I'll give it a go and prove to myself that you are right, or maybe I'll just get it wrong.

Later

DNA

 

[Eva]

Do you really think that my prototype has good performance? I don't think so, at least not for audio, and I wouldn't like to put you in any difficult situation. Also, my circuit is a true joke of class D amplifier, you can see some pictures that I have just taken:
http://eva.eslamejor.com/t_sine7.jpg
http://eva.eslamejor.com/t_sine8.jpg
http://eva.eslamejor.com/t_sine9.jpg

It requires aditional +-15V signal supplies as the output of the filter is actually signal ground (globally it's an inverting amplifier). The two 8.2uF filter capacitors are connected one to each rail (they serve also as spike-decoupling) and the main 30V supply is floating. The other side of the speaker (live) is where I take output feedback and is coupled to one rail with 4700uF. I'm sensing both inductor current and capacitor current through shunts and feeding it directly to op-amps. As both switching banks float, I'm using a 6N137 50ns TTL optocoupler to transfer the High-or-Low pulses from the comparator to the lower side (that floats "slowly" following the amplified signal), which is feeding an IR2112 buffered with ZTZ851 and ZTX951 BJTs. The whole thing is actually intended to operate full bridge with 300V, not single ended with 30V, I'm even using 600V 10A IGBTs as switches.

Since you seem quite surprised, I think that I'm already disclosing more than it would be convenient for me. It seems like I have at least found some unusual approaches of doing things and this is likely to be a paid project so I should shut up now

 

[Genomerics]

Mind you, as an aside...

You recently mentioned something about slew rate limiting yourself which made me remember something else as well..

Um, oh yes. Probably not interesting because things might automagically work out but....

A Class D amplifier shunting out a 100Hz bonk is going to have problems reproducing a 20KHz squeal at the same level at the top of the bonk.

On the other hand a linear amplifier would just clip.

Sorry I'm just having a blither.

DNA

Back on the motor controller.

 

[Genomerics]

I don't understand where that came from which leaves me wondering about myself.

I do know I've got problems with word type stuff but I did not mean to upset you in the way I have just managed to...

Did I?

I mean..

Oh I sort of give up.

I feel that any apology I might give will sound hollow.

Anyway, sorry.

DNA

 

[Eva]

Don't worry, I can hardly understand what you wrote on your previous post. English is not my mother language, so you have to be very explicit if you want to be ironic or offensive with me. Figure out that I have played some 100Hz+10Khz 4:1 tones naively following what you wrote, and I increased the volume till I heard intermodulation and aliasing products, then I lowered it a little and captured the waveforms. That's the result (note the reduced scales):
http://eva.eslamejor.com/t_sine10.gif
http://eva.eslamejor.com/t_sine11.gif

I can't tell if the intermodulation products that I heard came also from the 6,5" speaker, but the aliasing of pure tones has a very characteristic sound.

 

[Genomerics]

Don't worry, I can hardly understand what you wrote on your previous post. English is not my mother language, so you have to be very explicit if you want to be ironic or offensive with me.

My problem is I believe what you say and so I am happy to spend some time looking at the information you gave me.

5KHz.gif

500Hz.gif

Unfortunately, in order to get that to 'work' I had to do this with the voltage error amplifier.

verramp.gif

My other problem is that, if I had to do that to get yours to work, what has drzhuang done with his to get 'similar' results?

DNA

 

[drzhuang]

Hello drzhuang.

I've been having a prod about on the basis of the information you have provided so far.

My first question would be about the output filter.

You suggest a 500W output power and your plots give an output of 37V RMS which implies a sort of 3 ohm load. For a 2ohm load you would need about 45V RMS.

Your filter comprises two 22uH inductors and two 470nF capacitors which are effectively in series. So the equivalent is 44uH and 235nF.

The cut off frequency would be 50KHz but the resonant impedance is about 13.5 ohms. This implies that your filter is going to be heavily overdamped.

I realise that loudspeakers are non ideal loads and there is an opportunity to 'tailor' the response for a particular load but, for a 2 ohm load your filter will assume a first order LR roll off at about 7KHz which might be too low.

There is a problem that arises from current slew rate limiting in the inductor depending upon its value and the available supply voltage. It will make the compensation networks 'interesting' as well.

You also add parallel damping to your filter with a 220nF|220R series combination which, effectively, achieves nothing. Perhaps 680nF with 15 ohms would be better.

At the moment I am slightly confused (I cannot deny I am a novice as well). I'll spend a bit of time looking at what you have given so far but obviously it will be helpful if you, or someone else, can provide further information.

Thanks

DNA

It's 360W/4 Ohms 20khz. In this amplitude, The 22 Ohms resistor Burned. Maybe there are problems as you said about output filter, but it works well. 50khz is infrequent in audio application. Why not choose half-bridge For 2 Ohms load.
The information I gave is poor. Almost consider this. did anyone search DDX or Zxcd1000 for reference firstly.

 

[Genomerics]

It's 360W/4 Ohms 20khz. In this amplitude, The 22 Ohms resistor Burned. Maybe there are problems as you said about output filter, but it works well. 50khz is infrequent in audio application. Why not choose half-bridge For 2 Ohms load.
The information I gave is poor. Almost consider this. did anyone search DDX or Zxcd1000 for reference firstly.

Excellent... my mistake, I misread the 22 ohm damping resistor as being 220R. Your figures for output power make more sense than my efforts at some sums. I understand about the burnt resistor problem and something about spectral power density of music.

In many respects it's not the modulation method that interested me. I know about the zetex IC's as I'm sure others do. I've seen some more plots from your website showing performance of your products under more 'stressful' conditions.

Interesting stuff. I'll give things another go and see what happens. Perhaps this time I will get further and prove what a fool I really am.

DNA

 

[kartino]

DZHUANG said: "Did anyone search DDX or Zxcd1000 for reference firstly?"

That's way I call your chip as "ZHUANGTEX" because your arrangement similar to ZXCDxxx reference.

But actually what is the difference between your amp compared to Zetex reference?

Thank you and best regards,
Kartino

 

[drzhuang]

DZHUANG said: "Did anyone search DDX or Zxcd1000 for reference firstly?"

That's way I call your chip as "ZHUANGTEX" because your arrangement similar to ZXCDxxx reference.

But actually what is the difference between your amp compared to Zetex reference?

Thank you and best regards,
Kartino

Most difference is no ZETEX IC on my board.

Zetex reference is for subwoofer, mine for full range.

Regards

 

[kartino]

So you produce your own chip?

OK so your amp is similar to zetex's reference, class d with single supply, full bridges N-Channel mosfets, full range.

Will the chip be sold on market?

Best Regards,
Kartino