Help with decision on Speaker-Cables for Class D amplifier: Do I need to mitigate 750KHz EMI ?

[ADanalog]

Hi
I am building a Class D amplifier, using the Orchard driver. Diving into the question of EMI:

With about 0.5v switching on the speaker cables, this should generate EMI , that will differ based on the speaker cable solution (flat cables - which I have today / twisted cables / shielded-twisted cables ).

The wavelength of this switching is crazy long (~400m), but the speaker cable are going right there next to the DAC, Streamer, this might cause coupling. So the question :
Do I need to switch to Shielded Twisted Cables ? If yes : I probably need to use a 3rd connection for the shield (as the D type are always differential drivers)

Is this something that was discussed in ASR ? Measured ?

Thanks!

 

[egellings]

Does the amp have an LC filter network built into it, between its output and the speaker connection terminals? That should mitigate radiation from the speaker cable.

 

[ADanalog]

Hi
As far as I understand every Class-D has a switching noise on it . The LC filter at the outputs is not capable to remove this (0.7v or 0.2v) of switching noise.
I am using Orchard module, that is clear in their own publish testing for wide band FFT, but the same is supposed to be true for the Purif as far as I saw on wife FFT testing.

The LC filter:
My understanding that all these base designs are trying to get the BW to go up to ~80KHz (3db my assumption) - that is barely enough to get a single decade to the switching frequency, but I think that this is misleading - as this is a closed loop (feedback) .

 

[DVDdoug]

I wouldn't anticipate any problems. People use class-D amps every day and I've never heard of any problems.

My understanding that all these base designs are trying to get the BW to go up to ~80KHz (3db my assumption)

That would be up to the amplifier designer. With an audio amp there's no need to go above the audio range.

 

[ADanalog]

I am guessing that the reason for the filter where it is, is to get a flat respond up and including the 20KHz, personally I think this effort to widen the BW / reduce phase shift, is a bit overboard. I am pretty sure that any cat or a dog would appreciate a little bit of an early cutoff

Maybe I can just tweak the caps and lower the corner frequency a bit , they they do use a post filter feedback, no idea what it will do to the stability.

 

[somebodyelse]

What does Orchard's datasheet say about it? Or ask @orchardaudio directly? Twisting the cables and minimising loop area inside the enclosure is usually recommended to reduce coupling between the high level output and low level input as they're going to be close together. I would expect the output filter to reduce EMI to the point where it shouldn't be a problem for well designed modern equipment, otherwise they'll have trouble qualifying their amps to sell in some major markets.

 

[AnalogSteph]

I am building a Class D amplifier, using the Orchard driver.

Which one in particular? Starkrimson 25 or Ultra?

With about 0.5v switching on the speaker cables, this should generate EMI , that will differ based on the speaker cable solution (flat cables - which I have today / twisted cables / shielded-twisted cables ).

This is not wrong, although differential mode tends to be reasonably well-controlled by onboard filtering, and even plain twin lead isn't that bad for stray fields - you can give it a few twists if you insist. Common-mode output can be more of a problem. If in doubt, buy or wind some big ol' common-mode chokes using toroidal cores that are nice and lossy at a few MHz. Possibly another set for higher frequencies in series for more bandwidth covered. Wire gauge will have to be chosen based on what kind of output impedance increase you can tolerate.

Before you contemplate gilding the lily with your speaking wiring, have you seen what the inside of a speaker looks like? All concerns for RF tend to go out the window very quickly there.

The wavelength of this switching is crazy long (~400m),

Switching implies quick transitions, and those contain a plethora of harmonics.

In the late 2000s my parents bought a Panasonic integrated micro stereo that had the unfortunate trait of audio output interfering with its own FM radio reception if the antenna was anywhere nearby. Yes, FM around 100 MHz, and we can assume that the switching frequency used was lower. Coax cable to the attic with a dipole up there ultimately sorted this one out. I don't think the chip used had BTL outputs though, which probably is why it was so problematic.

Side note, 400 m still is within the AM broadcast band, so some attention arguably is not unwarranted.

but the speaker cable are going right there next to the DAC, Streamer, this might cause coupling.

I would be relatively unconcerned in that regard. As an old radio nut I wish everyone was so considerate though...

 

[ADanalog]

Thanks for both the replies.

So I am building the Starkrimson Ultra (dual mono in a single box).

Orchard confirmed that the switching will be there (it is always there, on their test report too), this kind of switching is on the Purifi module too (0.2V amplitude more spread across the spectrum too)

If in doubt, buy or wind some big ol' common-mode chokes using toroidal cores that are nice and lossy at a few MHz.

You mean to run the speaker cable through a chock for filtration of the common mode high frequencies? I did not think about that at all.
Do you have a link or reference to something that can work ?


Below is from the specification published by Orchard audio (4ohm load)
There are no measurements on the higher harmonics, but those should (in theory ) really be nocked out by the LC filter I see on the module, but I will do some measurements and see once I fully assemble it .