DC Blocking input to class D power amp

[Skeebopstop]

Hi all,

I've got my 2x150W class D amp up and running using the IRS2092 and now I am just entering tweaking stage.

Originally I used ceramic chip capacitors for dc blocking the input signal but did some digging to find out they are a bad source of distortion. I referenced an international rectifier design where they use a single 50V tantalum (probably 50V to remove any effect of DC bias on the capacitance?).

I tried throwing tantalums onto mine just to test(didn't notice any huge difference) and it all still works fine from 3 different devices as pre-amps (ipod, computer and blu ray player).

What I am getting confused about is that I can't imagine how 'tantalums' could possible be recommended unless it is inherently expected that pre-amps output class A type signalling with dc biases otherwise there will be a reverse bias?

Any advice on this matter? Currently I have the design slotted to use some 6.3V extra low esr 22uF tantalums (input impedance is only 3.3kOhm so need the capacitance).

Regards,

James

[wwenze]

In single-ended design, the output will always be positive w.r.t. ground.

[Skeebopstop]

What do you mean by single ended? As in not differentially transmitted?

[rascal101]

Based on below paper written by Walt Jung,

http://waltjung.org/PDFs/Picking_Capacitors_1.pdf

On page 1 - 2, tantalum caps should be connected back to back with its neg terminal connected together. This is then connected to a resistor going into a neg supply.

I have tried this configuration and so far it has worked very well.

[Pano]

Mostly tantalum caps are avoided for input decoupling. They are thought to sound pretty bad. But I've never used them in the above back to back configuration.

With your low input impedance, I can see why you might not want to use a film cap!

[AndrewT]

Originally Posted by Skeebopstop

Currently I have the design slotted to use some 6.3V extra low esr 22uF tantalums (input impedance is only 3.3kOhm so need the capacitance).

Hi,
try a pair of back to back 100uF 16V electrolytics with the common taken to a 9V bias. Bypass the series electrolytics with a 1uF 50V polypropylene.

[Skeebopstop]

Originally Posted by rascal101

Based on below paper written by Walt Jung,

http://waltjung.org/PDFs/Picking_Capacitors_1.pdf

On page 1 - 2, tantalum caps should be connected back to back with its neg terminal connected together. This is then connected to a resistor going into a neg supply.

I have tried this configuration and so far it has worked very well.

What an amazing article. Thank you very much. I think I will do a bit of an experiment.

I was able to find a Film cap (polyester) which can potentially fit into my design and is 10uF so acceptable in capacitance. I'll slot this in first and see how it sounds. I will then modify the design to use a biasing tantalum configuration and see what comes out of that also.

Cheers and many thanks to all.

[Pano]

Cool. We'll want to know how it goes for you.

[AndrewT]

10uF and 3k3 has F-3dB=4.8Hz.
A bit high.
33uF and 3k3 would pass all bass frequencies.

[Skeebopstop]

Originally Posted by panomaniac

Cool. We'll want to know how it goes for you.

So the results are in, confirmed from an independent source whom I did not tell the alterations to. Changes which were made in this final release:

1. Changed output LC filter from using a multilayer ceramic capacitor to film capacitor.
2. Better segreggation of my power, switching and signal reference polygons
3. Film capacitor used on input instead of multilayer ceramic and tantalum.

The description of improvement included a much more 'distinguishable' sound with a much 'crisper' tone and overall the sound quality is 'much better'.

So I'm not sure if all 3 things were coming together to provide this or not, but I'm sold that film caps are the way to go on anything in the audio path.

Cheers and thanks for your advice,

James