[amb]

Quote:

Originally Posted by 00940 what about M4 and balanced operation ?

Since that would require that you re-wire your headphones, I question how popular might such an amp be. I wonder how many people have built the Gilmore balanced, for example, compared to the standard Gilmore dynamic?

One benefit of a balanced 4-channel setup is double the maximum output voltage swing, which might be useful for very high impedance and inefficient headphones, but I think that's of limited interest to most people.

The other benefit of a balanced 4-ch design is that the output signal return is not dumped into signal ground, but our ground channel also gives the same benefit.

Four channels will also significantly add to the size and cost (not only for the extra amplifier channel, but also increased power supply demand as well as increased PCB real estate). This would be contrary to our goal, which is KISS yet strive for excellent sound quality and performance.

-Ti

 

[Sovkiller]

Well who's gonna work to try to organize the ideas and draw the first layout for the board, and maybe who knows later on organize a group buy through PCB or so....????

 

[Budgie]

I like the looks of this! I was thinking of adding a front end to my current class A amp, so the timing is perfect. Can't wait to see how it all turns out!

Thanks you, Morsel and amb!

Do you think an active current source for the output stage might be an (optional) upgrade? Something easy like an LM317/resistor combo?

 

[Earwax]

Quote:

Originally Posted by Sovkiller morsel: Any way we could use another OPamps, like the BBs OPA627/637 or the LT1122, LT1028, etc....??? Any way of getting included any onboard processor like crossfeed, bassboost, etc...even if optional or defeatable???

Is a crossfeed a good use of board space? Which circuit would be used? I like the Meier crossfeed best, but maybe someone else wants a Linkwitz or something else. Overall, wouldn't a daughteboard approach for crossfeed be better?

Any opamp at all, including Bipolar types (LM6171)?

 

[aos]

Wimpy output?? You can bias them into class A from 50 to 150mA, as I did
in my SDS labs amp (which has since fallen apart as this was 4 years ago). If 150mA isn't enough for a headphone amp, what kind of mutant
headphones are you trying to drive with it? Well I don't know how
much AKG1000 needs but what you are making then is a small speaker
amp. I found these Toshibas to sound much nicer than the IRF.
I do understand the monetary concern though - and the resourcing one
even moreso. I guess a board can always be hacked to use the
other pinout. I would push for wire bridges in order to allow different
pinouts but that would mean that no matter what layout you use
you'd have to make some bridges and that wouldn't fly.

Oh by the way, I personally like the idea of individual heatsinks as you mentioned. That works out well, as long as you don't need insane
power. Single heatsink is by definition going to be custom, and I'd rather
etch my own board than drill or God forbit cut a heatsink (I may exaggerate
a bit since I do have a drill press). The same money/resource issue applies here - single heatsinks are easy to buy and relatively cheap. Consider custom heatsink ONLY if you're going to offer
it alongside boards as well.

 

[tangent]

Quote:

We found someone besides Kurt who wants so8 pads.

Find ten and you enter the realm of statistical significance. >

Regarding the amp, I assume from the 2N3904/6 that you're using in the power supply CCS that each channel consumes under 100mA?

 

[aos]

tangent, the output mosfet's are not driven by the voltage mutliplier but by the rails directly. So just the opamp will certainly not go over 100mA. There's no way they'd be talking big power if that poor thing was to have to pass that current through anyway (or they'd have made a major design mistake).

Oh yeah, morsel, why the non-standard FETs in the voltage multiplier? Just about anyone who built PPA has probably some 5458 left over. Or is it Digikey availability? I know they have 5457.

(tangent can probably sit back proudly, because whatever he uses becomes a "standard" part

).

 

[tangent]

Quote:

the output mosfet's are not driven by the voltage mutliplier

Yes, I thought of that and came back to obliterate my post, only to find your answer already.

Quote:

why the non-standard FETs

My guess is that Digi-Key carries them, along with all the other parts they spec. You have to go to Mouser for 5484s.

 

[tangent]

That brings up a new question, then. How much does each channel draw, quiescent?

 

[tangent]

Quote:

Is a crossfeed a good use of board space? Which circuit would be used?

Morsel said above, resistive crossfeed. Just a resistor between the channels to leak some sound through. Think "fixed pan slider".

You won't see anything else on a Morsel amp, as she dislikes the nonlinear crossfeed circuits.

Quote:

I like the Meier crossfeed best

It won't happen on that circuit, as designed. The Meier circuit depends on being isolated between two active stages, due to its impedance requirements. If you put it ahead of the amp circuit like we do with the Linkwitz, its behavior varies wildly with the source's output impedance variations.

Quote:

wouldn't a daughteboard approach for crossfeed be better?

If it just means putting in some PPA-style mounting holes, I'm sure they could be talked into it....

 

[aos]

Also, you can't use Meier's circuit without permission. You can add it on your own amp if you wish but I don't think it could be included on board as default.
That's why a daughterboard (or another kind of allowance for crossfeed)
is a good approach to adding crossfeed.

tangent, I guess the current draw is whatever you set the bias of the MOSFETs to. The opamp draw can pretty much be ignored (it will be
order of magnitude less). I think they might be talking several hundred
mA per channel, perhaps even 0.5A per channel.

 

[jcx]

"AC only, no compromises for battery compatibility
Unimpeachable discrete output stage
Better price/performance ratio than the PPA
KISS philosophy (Keep It Simple, Stupid)"

?? supply splitter is "no compromise" ??

i'm assuming you do this to allow cheap single polarity wall warts - but it is definitely a compromise

accepting the splitters, it looks to me as if you are proposing shorting at least 2 of 3 tle2426 directly to each other - care to clarify these ground connections and show how to prevent this when the common ground of the external signal source is connected?

 

[morsel]

00940: Balanced operation, as AMB points out, is not very popular. We discussed this during the PPA design phase and concluded that if you want true balanced operation you can use only the left and right channels of 2 boards.

With regards to the layout, I will try to align all the MOSFETs if possible, with the input circuit on one side and the capacitors on the other side. This would allow for one big bar bolted to all of the MOSFETs, but would not be conduicive to bolting them to the case wall. The Aavid extrusions require about 1.5 inches of linear space, that's 9 inches for all 6 if they are lined up. It may be that we will either have to sacrifice the Aavid extrusions, or rotate them 90° to reduce the length of the footprint.

Sovkiller: Team M³ is designing the M³ with DIYer input and arranging for board manufacture when the time comes, as a community service.

Earwax: As you say, crossfeed is best left to daughterboards. As Tangent hinted, we will probably include PPA compatible daughterboard holes.

Aos: OK, relatively wimpy by comparison, then.

I don't want to consider any output device less than 1 Amp. May I ask if you have tried the IRFZ24N and IRF9Z34N MOSFETs? We tried the Renesas MOSFETs (if I remember correctly) and didn't find them superior.

Regarding the non-standard FETs, I knew that would come up.

That capacitance multiplier might not be used. It has yet to face close scrutiny by Team M³. To answer more directly, the 2N5457 and 2N5460 have a low, consistent Idss of about 1mA which makes them convenient for this purpose. I was just suggesting to AMB recently that people would not be happy about this and that we could use other FETs or at least use the N channel on both rails by turning it 180° in place of the P channel FET.

MOSFET bias is adjustable, with 80mA being the minimum we are considering for now. Once prototyping and testing has proceeded to a more advanced stage, we will study distortion .vs. bias to determine the optimum value.

JCX: We said no compromises for battery compatibility, which to me means compromises like low quiescent current, class AB output, keeping the amp to a portable size, etc. The power supply is still very open ended. That is one of the issues we intend to discuss. I would get into more detail right now, but I am late for a movie. I'll be back in a few hours.

Oh, and Jamont, going back to the eurocard format issue, perhaps that is best left to PPA v2. No, there is no info on PPA v2 at this time.

 

[jamont]

Quote:

Originally Posted by morsel Oh, and Jamont, going back to the eurocard format issue, perhaps that is best left to PPA v2. No, there is no info on PPA v2 at this time.

Oh, dear. I gots to know...

 

[Nisbeth]

Quote:

Originally Posted by morsel PPA v2.

Ugh, did you have to say that

Now we'll all be waiting for that as well

/U.