M³ Project Announcement - Page 17

This looks great Morsel!

If the calculations for the cap value is similar to the PPA, you need to fit caps in the 220-330 nF range IMO. This means 15 mm lead spacing for both Wima and Rifa MKP types.

If you can make the PCB considerably smaller by going to a single pair of 18mm caps, I think it is worth it.

Sounds like it is the best decision in this case.

Once again, kudos to you and AMB (and anybody else involved in this). I'm looking forward to seeing more


/U.

Great looking project

Just a little question : would you consider to put traces for smd opamps directly on the board ? Since there's not traces on the top layer, it should be rather easy.

I'll have to take the other side on this one. If subjective or objective tests indicate that the added caps help, I'd say who cares about board size.

Looking at the PCB I see you have the left and right channels next to each other, where the PPA has the ground channel between the left and right channels. Just wondering what the reason for this is, since IMO the board would be more symmetrical if the ground channel is in the middle.

Just a simple suggestion: eliminate the hole-and-cross connections where some pads meet the ground plane. This prevents the short traces from radiating high frequency energy and just looks more professional. It's an easy change in Eagle (a property of the ground plane).

What are the trade offs in using the shared caps on the opamp power rail? I'm for fewer parts. unless there is any sonic benefit.

What are the bias parts? trim pots?

The bias parts are everything after the output of the opamps up to the MOSFETs.

I would like to hear more opinions about the bass boost pot wiring to the right channel block. As Morsel pointed out, note that the left channel has real PCB traces for this purpose but the right channel has a couple of thin red lines suggesting where such traces might go. Alternatively, we would also consider no traces, so that if you want bass boost you'd have to run a pair of wires from the right channel block to the pot. The advantage of having traces is no separate wires off-board, the disadvantage is that the traces must cross the left channel amp block and the space is tight. Let us know what you all think about this.

There are two different TLE2426 rail splitters shown on the PCB. These are suggestions of the location for a single unit. In my opinion either is fine, but we'd like you to help convince us that one might be preferable over the other.

As for Morsel's suggestion about going to one single pair of (possibly larger) C5 electrolytic caps for all three opamps rather than three separate pairs; the theory is that having three separate pairs each located close to the opamp should be better, but we are not sure there would actually be a measurable or audible difference. Going with just one pair would allow us to shorten the board size by approx. 0.4 inch vertically. This means that we could actually fit the board into a 6-inch wide (external dimension) case and have some space for clearance and case thickness. Otherwise the case will have to be at least 6.25 inches wide. This may not be a big issue unless you already have a 6" wide case, but it might be convenient to have nice round numbers for the case dimension (at the expense of a fractional board dimension).
Alternatively, the space we save could be better used to route the right channel bass boost traces. We'd like to hear your opinions about this too.

I guess my thought is use the extra space to get good routing. The board is already out of the 'mini' category at 6"x7" and there are a lot of chassis' out there with 8" dimensions. I dislike 'external' wiring, so I'd say keep as much on board as possible

I agree, I'd rather see better trace routing than a small reduction in size, and I'd definitely rather have the traces on the PCB than deal with wires. (So I'm lazy.) Also, I'm probably missing something here (I tend to miss things, heh), but wouldn't one of the big Hammond 1455s fit the board in its current size/shape? I thought the biggest one was eight point something by 6.3in, seems like it'd work for this to me..

I'd also say that if the extra caps do turn out to help, then I'd rather see them in anyway. That's a pretty big "if," though.

EDIT: Found it, sure is big compared to the rest of 'em: 1455T2201

Is there any reason you're not using the 8-pin version of the TLE2426? It looks like you have the room for it.

Perhaps a small cap from output back to CNR? Funny how the datasheet doesn't have anything