Class Evolution-Modulating SMPS output to track audio signal

[vectorplane]

I must admit, these guys make some pretty impressive gear.

http://www.alab-pro.com/

Although the way they specify SMPS power output is somewhat unorthodox, I admire their bold embracing of switching power technologies for audio amplification.

They describe an audio amplifier operating mode where the supply rails are modulated to always be just a few volts above the peak program waveform.
(class H -continuously variable, or H/CC)


Alab-Pro SMPS for power amps - Class "TD2" concept

This begs the question:

Is a power amplifier's gain dependent on supply voltage to any extent? PSRR data for power ampifiers is not always available, but I am willing to speculate it might be lower than the PSRR of integrated op-amps.

If you took a typical op-amp's PSRR or 100dB and assumed that figure for a power amp, that would suggest a harmonic distortion
penalty of 0.001% added by the supply rail modulation technique.
Most of us could live with that figure, however, a Power Amp's PSRR may be lower than 100db, by how much, I don't know.

Nevertheless, it is useful to contemplate such a supply rail modulation in any power amp design that uses an SMPS.

It is certainly within the transient response capabilities of any voltage-mode SMPS feedback loop to respond to changing audio signals. I would in fact suspect that audio frequency transients above 1 KHz would not even need to be followed cycle-by cycle, but followed at a lower-frequency envelope instead.

I would want a voltage-mode controlled SMPS in this case, to better track the secondary side, while latting a preregulator PFC stage take care of line regulation.

I like PFC stages, not because it makes regulatory agencies happy --I frankly couldn't care less-- but because in smoothing the line voltage, they reduce the size of electrolytic bulk caps required, with all their limited lifetime conundrums.

Comments, anyone?

Andy

[subwo1]

PSSR, I guess would generally be lower than the gain of the amp. For a class A amp, there could be some cross-cancellation, though.

Dynamically regulating the voltage from the secondary filter caps would eliminate the slow response of an optocoupler. The frequency of the tracking could be much higher, then.

I have done an LTspice simulation of a simple tracking supply which regulates from the primary side. Let me know if you are interested.

[vectorplane]

Yes subwo1, I would be interested in that primary-side tracking regulation.

Do you think optocouples would not be fast enough to track a audio signal?

There are some O/C made by Fairchild --HCPL4502--(I'm sure there may be others) that have 1MHz bandwidth.

But is it really necessary to track high frequency audio signals cycle-by-cycle, especially since they are typically much lower in amplitude than low frequencies? What if you tracked an envelope above the peaks instead? I'm betting the efficiency savings would not be much different than if you force the supply rail to dip down to zero in-between the cycle excursions. Turning one rail off while the audio program is making an excursion in the opposite rail saves no power since it happens during a time of zero current draw anyways.

So an optocoupler may not need to have a frequency response over 50 KHz or so to do well in this application.

Secondary regulation doesn't seem appealing, since I'd want all my regulation in the swich mode, and once I'm on the secondary side, I'm done switching.

Only one conundrum:
You can't use large filter caps on the supply rails any more.

They must be just large enough to filter the switching frequency, and leave alone any ripple below 2 KHz. That places the entire burden of load regulation on the switcher, and requires that the switcher not do any line regulation, since that would bring it's response below 2KHz. That kind of rules out current mode control.

All line regulation must be done in a PFC, and switching frequencies kept above 100 KHZ.

I think it's doable

Andy

[subwo1]

The HCPL4502 is like the 6N137, but with a minimum rated CTR of 17% instead of 7%. Both still have a maximum rating of 50%. They should be able to enable a frequency response for the tracking supply which encompasses the audio band.

But is it really necessary to track high frequency audio signals cycle-by-cycle, especially since they are typically much lower in amplitude than low frequencies? What if you tracked an envelope above the peaks instead? I'm betting the efficiency savings would not be much different than if you force the supply rail to dip down to zero in-between the cycle excursions. Turning one rail off while the audio program is making an excursion in the opposite rail saves no power since it happens during a time of zero current draw anyways.

Sounds right. The output transistors have to be rated for the whole rail to rail voltage this way. The tracking circuitry is much more complex, too.

Secondary regulation doesn't seem appealing, since I'd want all my regulation in the swich mode, and once I'm on the secondary side, I'm done switching.

Only one conundrum:
You can't use large filter caps on the supply rails any more.

They must be just large enough to filter the switching frequency, and leave alone any ripple below 2 KHz. That places the entire burden of load regulation on the switcher, and requires that the switcher not do any line regulation, since that would bring it's response below 2KHz. That kind of rules out current mode control.

The tracking switcher needs to be able to continuously output the peak power of the amp because the secondary filter capacitors are small. However, if the SMPS is overload protected, it also protects the amplifier.

I have not done much work with PFC since for my projects, passively using inductors for partial correction has been simpler and more efficient.

I tried out a different compression program than the one built into Windows XP and it wiped out the built-in file compression software. I will try to make the circuit available still.

[subwo1]

I can upload the diagram at least. Meanwhile, I plan to try to fix my Windows installation.

[vectorplane]

Thanks for the diagram

Sounds right. The output transistors have to be rated for the whole rail to rail voltage this way. The tracking circuitry is much more complex, too.

I'm not sure what you mean by the tracking circuit being more complex. The error signal just needs to be frequency compensated to respond well below 2 KHz and roll off above that.
And as for the output devices, they'd be rated for rail-to-rail voltage in an existing design anyways.
The intention is not to lower the output transistor ratings. It is to enable you to build a high power, low distortion class B amp, without turning it into a space heater.
Depending on how well this modulating SMPS is implemented, it could reach efficiencies very close to class D. ( possibly perturbing the relevance of the whole class D movement ).

The advantages of this is that it requires no modifications to the class B /AB amplifier, and can be re-fitted to existing designs. It suddenly removes the greatest single drawback of class B that made people look at alernatives in the first place.
It also makes possible class B amps that are slim, sleek, and light.

And yes, one more advantage :
It reduces BJT thermal runaway concerns, by lowering their temperature, and pushing their operating point deeper inside the SOAR curve, increasing reliability.

Andy

[subwo1]

You're welcome. Thanks.

I'm not sure what you mean by the tracking circuit being more complex. The error signal just needs to be frequency compensated to respond well below 2 KHz and roll off above that.

I was considering the way the simulated circuit powering a simple resistive and inductive load would work instead of the realistic, more complex one consisting of a linear amplifier driving a speaker. Using the original circuit with a 20khz signal resulted in poor tracking because of the slow response. It now seems to do better after I removed the modulated voltage source drive from the tracking feedback and replaced it with a modulated current source between the tracking power output and the load. Now the LTspice circuit should better represent the real life circuit.

Now I have zip capability again. I hope this zip file will decompress OK.

[nine76chris]

WTF???? check this:
http://www.alab-pro.com/index.php?la...172730&Ntype=6

Do you mean to tell me that I can buy a smps from these Alab people that puts out 4,200W continuous for the bargain basement price of $235USD? Is this for real? Who cares about their class-whatever, start buyin' up powersupplies and move 'em through e-bay$$$$$$$$$$$$$$$$$$$$$$$$

[Eva]

Don't get fooled, these power ratings are maximum peak, not continuous. Also, output inductors are tiny so output HF ripple is going to be 0.5V p-p or more...

[Workhorse]

Originally posted by vectorplane
I must admit, these guys make some pretty impressive gear.

http://www.alab-pro.com/

Although the way they specify SMPS power output is somewhat unorthodox, I admire their bold embracing of switching power technologies for audio amplification.

They describe an audio amplifier operating mode where the supply rails are modulated to always be just a few volts above the peak program waveform.
(class H -continuously variable, or H/CC)


Alab-Pro SMPS for power amps - Class "TD2" concept

Andy

Hi EVERYONE,

The Guys from ALAB PRO are nothing but Damn...Cheaters.....only masters in cheating big companies......products and copying them

They Copy QSC PLX & PL series amps each & every single thing...
Also These So called engineers copy also LabGruppen ... Class-TD is actually from it...
They also carry same "POWERWAVE" trademark...on their amps as well...


Compare PL series of amps from QSC with them...
http://www.qscaudio.com/products/amp...owerlight2.htm

What QSC says about this :
http://www.qscaudio.com/cgi-bin/ulti...017;p=1#000002

Duplicate product
http://www.alab-pro.com/index.php?la...204921&Ntype=1


LabGruppen Original Amp:
http://www.labgruppen.com/Default.asp?Id=6595

Duplicate product...
http://www.alab-pro.com/index.php?la...150237&Ntype=1

See the Truth with your eyes...how these R*A*S*C*A*L*S do their job so nicely....
Their products donot impress me, but hurts their orignal manufacturers alot....

Truth Investigated by K a n w a r