Effect of feedback in Class-D amplifiers

[kASD]

Recently , i came accross an TRANSAMP500AX amplifier which is using class-d topology and its listening tests performs very well and it surprised me a lot when i glanced the specs:
Inter Modulation Distorion=NIL
THD=0.001%
Switching frequency=500KHz
power out =500WRMS@4Ohms
This amp states that is doesnt use feedback , therefore there is no presence of IMD components in output.

I want to know that is this is possible that a Class-D amp without feedback will give such results.
CAN we also eliminate the feedback to get such results.
Then what type of effects does the feedback have on output of amp.

 

[lemondog]

lol, wot trannies does it use? and how are they biased?

 

[Cybergent]

Biased?

He said it is class-d.

 

[lemondog]

good point

thats wot you get for not sleeping :S

 

[phase_accurate]

Have you got any further info about the maunfacturer etc ? Please keep the following in mind:
1.)
Many manufacturers state that they would work without any NFB even though they definitely do.
2.)
In theory a precise PWM modulator, a stable PSU (!) and a perfect switching stage would give a perfect amp even without any type of feedback.

As we know, nothing is further from the truth in practice. But from the TI/TacT approach we know that very good switching amps are feasible without NFB.

It is generally advisible to build a switching amp in a way that it performs well, before any kind of feedback is applied. I know that it is feasible. Though I didn't measure the one I once developed without feedback, I assume that its open loop THD must have been below 3%. Now try to do this with an ordinary class AB topology !

Regards

Charles

 

[kASD]

Hi Charles,

How about this setup...

Triangle generation with LM6172 than feeding it at inverting terminal of differential comparator LM360 & audio signal directly fed to non inverting terminal and differential outputs of LM360 are connected to IR2113 gate driver thus driving the Mosfets...

Could we use LM360 differential comparator in the place of LM319+CD41070 for both Hi and Lo pulses fed to gate driver circuit...[assuming single supply operation , no level shifting]
I think the outputs of LM360 are not open collector type, so no need of pull up resistor..i am i right
Could you please post your valuable comments about this...

with sincere regards,
K a n w a r

 

[phase_accurate]

This would be a simple, elegant and fast solution. But it would be more expensive as well, especially when the speed advantage is not needed or even detrimental.


Regards

Charles

 

[classd4sure]

Have you got any further info about the maunfacturer etc ? Please keep the following in mind:
1.)
Many manufacturers state that they would work without any NFB even though they definitely do.
2.)
In theory a precise PWM modulator, a stable PSU (!) and a perfect switching stage would give a perfect amp even without any type of feedback.

As we know, nothing is further from the truth in practice. But from the TI/TacT approach we know that very good switching amps are feasible without NFB.

It is generally advisible to build a switching amp in a way that it performs well, before any kind of feedback is applied. I know that it is feasible. Though I didn't measure the one I once developed without feedback, I assume that its open loop THD must have been below 3%. Now try to do this with an ordinary class AB topology !

Regards

Charles

Hi,

Yeah tact and.... what is it.... spectron? Open loop class d.... I'm sure 90% of what their circuit consists of is correction circuitry though

W, what you've described is just not a high end open loop class d amp.

The required complication doesn't seem to warrant it?

Regards,
Chris

 

[phase_accurate]

I assume Kanwar doesn't use any open-loop topology anymore.

At least the older Spectrons use NFB as well (they even take it from the speaker posts).

Regards

Charles

 

[classd4sure]

Hi Charles,

I understand some of them are using a digital form of "room correction"? That's either got to be with a microphone or a voice coil...... lvdt?

I'm assuming anything with a delay would have a local feedback/reclock loop.

No clock at all seems the only reasonable solution, and I dont' think you can have a feedbackless self oscillating design, though ucd is said to sound like it by design.

I guess you can do it with digital open loop but the added complexity just seems pointless to me, in part count alone.

Regards,
Chris

 

[phase_accurate]

I understand some of them are using a digital form of "room correction"? That's either got to be with a microphone or a voice coil...... lvdt?

They measure the response of the room-speaker combination by the use of a computer measurement program. This measurement is used to calculate coefficients for a FIR filter that are then loaded into the system. I.e. the room correction is done in the digital domain before the amp and it is capable of correcting frequency- and phase- response. It can be regarded as a sophisticated equaliser.

The result is not a cure-all but quite impressive.

Regards

Charles

 

[classd4sure]

.... and quite complex! So it is a microphone Great explanation of it thank you.

Seems it would be worth a listening experience? I won't be building one of those anytime soon.

Regards,
Chris

 

[phase_accurate]

I won't be building one of those anytime soon.

I won't do that either but I can definitely recommend a listen if you have the opportunity.

Regards

Charles

 

[kASD]

Hi Charles,

Thanks for the answer..

LM360 cost around at 15dollars each.. i have just recieved the samples from NSC..

But i want to refine my previous designs eliminating the level-shifters....going single supplies and H-bridge
New setup;
[LM6172]->LM319->XORS->IR2110->IRF250N
Simple and cost effective...
configured in Bridged mode in BD mode..switching in phase but audio signal out of phase thus improves damping and makes the output filter small...mean while also using common-mode inductor at output....
Thus it also helps in implementing 3 - Level PWM
Now one question, Since i am going to use bridge mode[BD-modulation] ... Do i require feedback...Since even order harmonics as well as bus pumping is eliminated , but then also there are odd order harmonics out there...

I myself want go open - loop....

What would be the wise solution.....

Sincere Regards,
K a n w a r

 

[classd4sure]

Hi,

A quick look at the patent for what I think you're talking about will show you it's heavily dependant on feedback.

Maybe it's just me here but I take it for granted that open loop class d = "True Digital" with a boat load of correction circuits and tight local feedback loops. It would seem all too easy to be working backwards doing it that way, adding more circuitry, making it less efficient, reliable, etc.

Why don't you order one of those Mueta chips?

Regards,
Chris

 

[phase_accurate]

I'd prefer a solution with NFB as well.

Regards

Charles

 

[kASD]

Thanks once again, Charles.

So what are the advantages [more probably to say disadvantages in openloop] in your view about the implementation of NFB

sincere regards,
K a n w a r

 

[phase_accurate]

The advantages are the same ones as they are for conventional amps.

Regards

Charles

 

[kASD]

Ok Charles,
i would go for open-loop and should see what happens to the distortion figures, if they were not satisfying, then would go for NFB,then only..

Since i want to Drive 100V constant voltage line with frequency response from 30Hz to 15Khz only, thats why considering cost effective solution....

Secondly I also want to use complementary comparator scheme using dual comparators to obtain out of phase pulse signals...

Thanks for the help,
K a n w a r

 

[Pafi]

In theory a precise PWM modulator, a stable PSU (!) and a perfect switching stage would give a perfect amp even without any type of feedback.

I disagree. There are side-band components in basic PWM spectrum, very similar to phase modulated signal. The lower side-band reaches audio-band with significant level if switching freq/signal freq ratio is not enough.