Class D amp in N&V Q&A

[CeaSaR]

My kind of build, TL074, 2N4401/03's and IRF9530/530's. Nothing fancy.

Just for clarification, "My kind of build" refers to a relatively simple or uncomplicated piece
of work. Over the years I have found that most "extras" included on a gadget are not really
needed, therefore are a waste of time to include - unless they are needed for an initial setup,
in which case they should be hidden from "normal view".

But that's just me...

CeaSaR

[CeaSaR]

Here are the scans.



That's the whole article. Have a blast with it.

CeaSaR

[Bob Scott]

This amp from PE has no power supply isolation; no power transformer. It uses directly rectified AC. If you connect this thing to stereo that actually has a grounded outlet, GFCIs will trip, smoke will be generated, connected equipment will suffer catastrophic major malfunctions, and people may get "surprised".

Is there no feedback from electrocuted readers in later issues?

[EDIT] Oh, man. There's more. The MOSFETs are 100V devices on a 102V power supply (2*51V).

[CeaSaR]

Well, the author did state that the PS was fed directly from the mains and
should NOT be touched during operation. See the 2nd to last picture (pg 34)
in the middle, near the bottom, for the warning.

For safety's sake, an isolation transformer should be used. As for the power
MOSFET's, yes, ones with the appropriate voltage ratings should be used also.

Is there no feedback from electrocuted readers in later issues?

Well, if they were electrocuted...

How about the rest of the circuit? Does it seem up to snuff otherwise? As for
feedback about that article, I don't recall any comments in later issues.

CeaSaR

[Bob Scott]

Hi Ceasar,

First off, thanks for taking the trouble to post this interesting circuit.

Well, the author did state that the PS was fed directly from the mains and
should NOT be touched during operation. See the 2nd to last picture (pg 34)
in the middle, near the bottom, for the warning.

Yes, but you still need to plug something into the input jack. It has a ground symbol on it but it is hot. If the input device is grounded with a 3 conductor power cord, sparks will fly; a breaker will trip and expensive equipment damage will probably happen.

For safety's sake, an isolation transformer should be used. As for the power
MOSFET's, yes, ones with the appropriate voltage ratings should be used also.

Also the driver transistors, the 2N4401 and 2N4403 are rated for 40V. He inserted Zener diodes in series with the collectors as a kludge in order to extend the voltage range. Why did he not simply use transistors with a higher Vcb voltage rating?

How about the rest of the circuit? Does it seem up to snuff otherwise? As for
feedback about that article, I don't recall any comments in later issues.

About the triangle wave oscillator/integrator: It could be a little simpler. I did one a few years ago. His is pretty close.

The peak voltage of a 60 watt sine wave into an 8 ohm load is 31V. He could have used +/- 31V power supplies for 60W/channel or +/-22V for 30W/channel. That makes sense. The MOSFET voltages would then be OK, and the transistors would not need those zeners added to the collectors. I think he made it higher when he decided to use directly rectified line voltage. I bet he had to add the zener kludges after he jacked the voltage to +/-51V.

One last item: With +/- 51V it is possible to drive this amp up to 163 watts RMS per channel into 8 Ohms if the volume is turned up. That will blow the MOSFETS. Too much current.

Wait! One last, last item: There should be a high pass filter on the outputs, so the tweeters don't fry. The ultrasonic modulation and harmonics are at such a high frequency that the tweeter cones cannot respond, so the energy gets dissipated as heat in the voice coils. I see he recommends using a full-range speaker. ie: no tweeter.

If I built one of these I would use an isolated power supply using a 48VCT transformer to produce +/-34VDC. I'd add +/-15V regulator ICs for the op-amps. I would add at least output chokes for filtering the audio outputs. I'd take out the zeners on the drivers, use more robust MOFETS, and slightly change the wiring of the triangle oscillator. It's set the oscillator frequency to ~100KHz.

I think electronics mags should have a competent mentor on staff to help authors polish their designs. Credibility is important.

[Joseph]

Hi Bob

I've enjoyed your input on this topic. I think the zeners are there also to prevent the gates of the MOSFETs from getting too much drive voltage. I remember reading that Popular Electronics article on the circuit that CeaSaR kindly shared. That was before class D was beginning to take off in popularity and before I knew a whole lot about the topic.

[Bob Scott]

Hi Bob

I've enjoyed your input on this topic. I think the zeners are there also to prevent the gates of the MOSFETs from getting too much drive voltage. I remember reading that Popular Electronics article on the circuit that CeaSaR kindly shared.

Ah yes. The gates (to source) are rated at just 20V . I missed that.

That was before class D was beginning to take off in popularity and before I knew a whole lot about the topic.

I think that the availability of power MOSFETs was a big boon to Class D designers. Before MOSFETs, there was a California company by the name of "Infinity" if I recall correctly, had a high end audio, high power, class D stereo amp circa 1976 but it boasted of incredibly fast switching bipolar output transistors. That amp had reliability issues with the output section. Every time our retail store owner tried to phone the company, they were all "out at the beach".

[Joseph]

I think that the availability of power MOSFETs was a big boon to Class D designers. There was a California company by the name of "Infinity" if I recall correctly, had a high end audio, high power, class D stereo amp circa 1976 but it boasted of incredibly fast switching bipolar output transistors. It had reliability issues. Every time our retail store owner tried to contact the company, they were all "out at the beach".

I first tried to do it with bipolar transistors about twenty years ago. I still have some of those unsuccessful breadboards, I think. LOL--at the beach.

[CeaSaR]

What would be the limiting factor for using Bipolar devices instead of MOSFET devices if they both
have the same frequency limit?

For example, if both types have a fT of 300 MHz, what prevents the Bipolar device from working
correctly in a circuit where the operating frequency is ~100 KHz? Is it turn on time?

I would greatly appreciate any information on this subject.

CeaSaR

[CeaSaR]

Poking around (again) and found an open source site for Class D.
Haven't had much time to look at it, but thought I'd pass it along.

The Sorenson Audio Experiment.

CeaSaR

[Bob Scott]

What would be the limiting factor for using Bipolar devices instead of MOSFET devices if they both
have the same frequency limit?

For example, if both types have a fT of 300 MHz, what prevents the Bipolar device from working
correctly in a circuit where the operating frequency is ~100 KHz? Is it turn on time?

I would greatly appreciate any information on this subject.

CeaSaR

I thought about this but forgot to post.

And before I forget again, class D is also called PWM for Pulse Width Modulation.

One big difference between power MOSFETs and bipolar transistors is that bipolar transistors are subject to secondary breakdown and SOA (Safe Operating Area). Secondary breakdown was hard for me to understand because explanations of how it works said that this breakdown is not due to thermal considerations. Seeing as how I had sawed the tops off of several APEX power op-amps and seen melted circuits under the microscope, I found this hard to believe. It is not due to the general temperature, but it is due to a sudden increase in power causing localized heating, and because BJT junctions are current hogs. They pass more current when warm. FETs are current friendly. They pass a little less current as they get warm.

Don't miss out on Apex Microtech's excellent technical documets. They have been bought out by Cirrus. Find them here:

http://www.cirrus.com/en/products/apex/documents.html
http://www.cirrus.com/en/products/apex/ ... minar.html

[Bob Scott]

I also suspect that bipolar transistors may fail in class D amps because the designers imagine them driving resistive loads. Speakers can be inductive, so when a transistor in ON to drive voltage through the load, it will be required that the circuit actually sinks current instead. So you have to have diodes wired from output to the supply rails to limit the voltage generated by reactive current from the load.

It's either that, or the transistor are working outside the SOA. The SOA is limited at high power by the secondary breakdown curve, something the FET is not subject to.

Visit that Apex Microtech site at Cirrus. I found their technical documents to be invaluable. I have them on CD. They have a nice explanation on Howland Current Pumps too; - wiring op-amps as current sources.