My Class D amplifier gets a negative feedback / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

My Class D amplifier gets a negative feedback

Move Thread

LAN_403

capfreak

Mon May 07 2012, 05:31PM

Registered Member #4649 Joined: Sat Apr 28 2012, 04:21PM
Location:
Posts: 2

Hey guys!
First things first: IÂ´m new in this forum and IÂ´m very interessted in power electronics and different types of HV sources and tesla coils. I built a SGTC and a SSTC on my own and for a research paper for my A-levels i built a "simple" Class D amplifier without any feedback. I designed the whole thing the whole thing completly new. As one can guess as a consequence of the missing negative feedback it needs voltage regulated power supply.
I am from Germany and no native speaker, so it would be very polite if you forgave me my mistakes.

Now letÂ´s go futher to the important things: Electronics

1336409888 4649 FT0 Pw Modulator Plus Offset And Opto Coupler Schematic

One can see the current version of the PW modulator. On the left side is the "XR2206" which generates a good triangle signal with a frequency of 120khz. The signal goes into an inverting amplifier which sums up the triangle with an offset voltage so that the TLV3502 comparator, which is a very fast one and can only operate from 0 to 5V, gets a triangle with an amplitude of round about 1V which swings around a DC offset of 2.5V.
The comparator gets an audio signal which also has an offset of 2.5V from another op amp which sums a voltage of 2.5V to the audio signal. The comparator creates two 90Â° shifted square waves which go to the next device, two optocouplers, where they are seperated to prevent troubles coming from the actual amplifier.

1336410358 4649 FT0 Amplifier With Feedback Schematic 2

The current version of the amplifier WITH a NOT WORKING feedback is shown in the picture above.
The amplifier itself works pretty well! The high current drivers are capable to create a pretty high dV/dt at their gates and the filter also works well. Now there is the problem: The negativ feedback does not work.
It is a very simple P-regulator, but i donÂ´t know how to design a better one.
The present version just heats up the op amp, which generates much distortion and has a rather tiny amplitude. How can i make the device stop regulating the 120khz swing wihtout shifting the phase to much so that the amplifier still can control the loudspeaker?
Can you help me please?
If you are interested in: Here i have some measurements:
First one shows the offsetet triangle (just a qualitativ measurement)

Next one shows the PWM output signal

1336410798 4649 FT0 Pwm Output Directly Measured At Tlv3502

This one is the signal directly masearued at the MOSFETÂ´s gates

1336410911 4649 FT0 Gate Signal At A Mosfet

The next ones show the output of the half bridge with a connected loudspeaker without and with a filter:

1336411084 4649 FT0 Output Signal Amplifier 1

1336411084 4649 FT0 Class D Output Without Feedback Amplitude 15v

Thanks for the reading. BTW: The amplifier sounds pretty good with a regulated power supply! If someone wants du build this thing just say it and iÂ´ll post the original eagle datas.

Storken

Tue May 08 2012, 05:32AM

Registered Member #1960 Joined: Thu Feb 05 2009, 11:57PM
Location:
Posts: 48

How long is your deadtime? Seems like it is pulling current through the body diode of your mosfets and body diode reverse recovery charge sets off high frequency ringing. It seems you are hard switching, something that is unwanted in a Class D amplifier.

Steve Conner

Tue May 08 2012, 08:23AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Unwanted but also pretty much unavoidable. And not really an issue: the low rail voltages in a Class-D amp (compared to say an off-line SMPS) mean that the switching losses aren't too bad.

With classical PWM, it can be very difficult to get the feedback working over the whole audio band. You can take the feedback before the output filter, in which case you'll struggle to get enough feedback gain at 20kHz while still rejecting the carrier frequency. Or you can take it after the filter, in which case carrier rejection will be easier, but you'll struggle to maintain an adequate stability margin with all possible speaker loads.

Most successful DIY Class-D amps (that I'm aware of) are self-oscillating or sigma-delta. These techniques have a ju-jitsu way of using the stability problem's own power against it, and this article

gives a good flavour for how to apply them. (also see the references linked in the article)

capfreak

Tue May 08 2012, 06:35PM

Registered Member #4649 Joined: Sat Apr 28 2012, 04:21PM
Location:
Posts: 2

wrote ...
How long is your deadtime? Seems like it is pulling current through the body diode of your mosfets and body diode reverse recovery charge sets off high frequency ringing. It seems you are hard switching, something that is unwanted in a Class D amplifier.

The dead time should be about 20ns, itÂ´s generated with the circuit in front of the MOSFET drivers (of course there is a very fast diode and no 1n4001

)
What do you mean with hard switching? I built a "usual" class D amplifier how can i NOT make him hard switch?

wrote ...
Unwanted but also pretty much unavoidable. And not really an issue: the low rail voltages in a Class-D amp (compared to say an off-line SMPS) mean that the switching losses aren't too bad.

With classical PWM, it can be very difficult to get the feedback working over the whole audio band. You can take the feedback before the output filter, in which case you'll struggle to get enough feedback gain at 20kHz while still rejecting the carrier frequency. Or you can take it after the filter, in which case carrier rejection will be easier, but you'll struggle to maintain an adequate stability margin with all possible speaker loads.

Most successful DIY Class-D amps (that I'm aware of) are self-oscillating or sigma-delta. These techniques have a ju-jitsu way of using the stability problem's own power against it, and this article gives a good flavour for how to apply them. (also see the references linked in the article)

Thanks for the links! I will read them, but this could take a while.
I really want to keep the actual design and run this thing the analogue way without self oscillating!
It wouldnÂ´t be too bad to loose some dB in the in the frequency range above 10kHz, itÂ´s more important to me that there is a good working feedback which really can control the speaker, especially because this thing should work as a good subwoofer amp.
Greets

Steve Conner

Tue May 08 2012, 08:44PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

If you're not too worried about HF performance, try an I controller (integrator) instead of a P. You may need a little P for stability, but try to use as little as possible.

You might also try using a more aggressive output filter. You don't seem to be attenuating the carrier much.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob