PPA v2 construction discussion

[morsel]

PPL, those references are fine and dandy, thanks for posting them, but they do not answer my basic point, which is whether the more complex odd/even bypassing arrangement makes a measurable real world difference in the PPA v2, as opposed to being a good idea on a purely theoretical basis.

 

[tangent]

Quote:

Morsel's bass boost calculator was the only one I could get working

Sorry, that's fixed now. I tested in Firefox and Safari on OS X.

Quote:

I have no doubt Tangent measured something

No, ppl mentioned the idea, and I agreed with it on general principle. As Morsel says, it can't hurt and it may help, so I recommend it.

Quote:

Tangent still uses my 3 year old numerical approximation

Not any more.

Quote:

as opposed to being a good idea on a purely theoretical basis.

That argument would hold more water, Morsel, if it hurt to implement the idea. It doesn't, so why fight it?

 

[morsel]

It hurts in the sense that some people obsess, perhaps unnecessarily, about caps and buy more types of caps in order to comply with theories that may not have been clearly demonstrated to produce real world improvements in PPA v2. It is in my nature to try to keep things as simple as possible. I'm not saying the odd/even idea is bad or wrong, I'm asking for proof that it makes a difference.

PS - your bass boost calculator does not handle bass boost gain < 6dB properly.

 

[ppl]

Wow just let Morsel define what Good theory is, as opposed to the many experts that wrote those articles I linked. Well moving along it is quite common knowledge even around hear that multiple values of bypassing operate most effectively at different frequencies the Big electrolytic start to poop out around 100 KHz and a noticeable rise in impedance is noted as low as a 500 HZ on some units. This is why we use smaller film caps in parallel to them. This same concept is carried to three levels as opposed to two levels if the mixed 0.1uF and something 10 times lower in value so this would be 0.01uf. Myself i tried 0.22uF and 0.022uF in addition to 0.56uF Teflon populating some C1 spots and so i have 4 levels of absolutely the best bypass arrangement both sonically as well as technically. This puts me at 4 that’s FOUR Levels of supply bypassing with the dielectric chosen so as to be most effective in its ideal frequency range. I will post a photo later of the innards of the Prototype showing the verity of the possible capacitor options that fit in the standard Hammond case on Version 1.916 anyway.

 

[Glassman]

I'm dealing now with board layout and supply decoupling for Spartan3 FPGA, so I've been reading some Xilinx application notes and this is exactly what I have found there.. it is important though not just to vary the value of the cap, but each cap should have different inductance, the smaller the cap is, the lower inductance you want, so picking 0.1 and 0.01 from the same line of caps does nothing in fact, each of the step has to have different parasitic inductance.. and yes it doesn't hurt using it this way of course.. on my buffers I rely upon 1uF polyester C2 caps on PPA board and use 0.1uF polypropylene on the buffer boards themselves..

 

[tangent]

Quote:

Originally Posted by Glassman picking 0.1 and 0.01 from the same line of caps does nothing in fact

Not completely true. Yes, inductance also matters, but simply paralleling a lower-value cap will add another 'V' with a higher resonant point to the system.

(If you've seen the impedance vs. capacity graphs in some datasheets, particularly film and ceramic types, you know what I mean about a 'V'.)

 

[morsel]

Relax PPL, I'm not trying to contradict established theory. I am quite familar with the concept of multiple values of capacitors for bypassing. The question is whether it will actually result in a real world improvement in PPA v2.

Those of you with scopes and RMAA could try removing just the odd caps, then just the even caps, using all the same value, etc., and see if you can measure or hear a difference.

Sorting the wheat from the chaff is an honorable and useful goal.

 

[individual6891]

Quote:

Originally Posted by ppl Matching transistors are not mandatory however lower DC offset and perhaps better sound with slight measurable Right mark performance will be obtained with matched devices. The output transistors should all be matched in all three channels because of the symmetry created by the ground channel with respect to the right and left channel.

So I measured the hFE using a standard DMM, but the PNP transistors have a lower gain than the NPN devices, I suppose this is due to the way they are designed?

Here are my tables (2N5088 on left, 2N5087 on right)... just trying to decide which ones to use:

And for the PN4392...Only a small amount of the them put out over 30mA - is this a problem? They seem to hang around the 30mA mark.

This is my 2N5486 list:

Is around 17mA enough for the OPA637 for Q3? Looking at the datasheet, the specs say the chip uses 15mA max, but is there enough in my set of 2N5486s to provide enough? (with about 1mA bias).. Seems a bit borderline to me?!

 

[morsel]

The FET current sources only need to supply quiescent current + 1mA bias. Iq for OPA637 is 7mA, way under the 17mA Idss you measured. It is normal for PNP hfe to be lower than NPN hfe. The PN4392 Idss of 30mA is fine.

 

[individual6891]

Quote:

Originally Posted by morsel The FET current sources only need to supply quiescent current + 1mA bias. Iq for OPA637 is 7mA, way under the 17mA Idss you measured. It is normal for PNP hfe to be lower than NPN hfe. The PN4392 Idss of 30mA is fine.

Ahh, it said -/+7 typical and -/+7.5 max, so I presumed that it meant 15.. my bad

 

[Glassman]

Quote:

Originally Posted by tangent Not completely true. Yes, inductance also matters, but simply paralleling a lower-value cap will add another 'V' with a higher resonant point to the system. (If you've seen the impedance vs. capacity graphs in some datasheets, particularly film and ceramic types, you know what I mean about a 'V'.)

the right part of the V won't change, just the left one, that sure shift the resonant frequency higher, but the overal impedance at that frequency will be about the same.. the point is to move the right part of V further to the right, in other words reduce the inductance.. using many smaller caps in parallel effectively reduce inductance..

 

[bg4533]

Quote:

Originally Posted by individual6891 Ahh, it said -/+7 typical and -/+7.5 max, so I presumed that it meant 15.. my bad

I think that means 7mA from the + rail and 7mA from the - rail. Each rail has its own jfet.

 

[Clutz]

So, I've read this thread a few times and have now become a little bit intimidated of assembling a PPA- it seems that a lot of rather more experienced DIYers than I are getting confused about assembling the PPA which makes me question whether or not I should attempt it (especially given the relative costs of building a PPA)

I have ordered a Dynalo board (actually, two) and now I am afraid of trying to assemble that.

So from some of the more experienced builders - after building three Cmoys and having a 4th laid out using a TLE2426 instead of the volage divider resistors - waiting to find somewhere I can buy TLE2426's in Vancouver (I don't want to order them online.. I don't know why..) - is doing a PPAv2 too advanced for me- esp. given the experienced DIYers haven't worked out the details yet?

Thanks for your support.
Cheers,
Clutz

 

[amb]

Quote:

Originally Posted by Clutz is doing a PPAv2 too advanced for me- esp. given the experienced DIYers haven't worked out the details yet?

LOL. What you read were experts chewing on some very fine points that "most people" could safely ignore.

Don't let that discourage you. The PPA is quite a step up from a cmoy but if you simply read through all of tangent's docs, get all the recommended parts and build the amp as-is, you should do just fine.

 

[ppl]

Quote:

Originally Posted by Glassman the right part of the V won't change, just the left one, that sure shift the resonant frequency higher, but the overal impedance at that frequency will be about the same.. the point is to move the right part of V further to the right, in other words reduce the inductance.. using many smaller caps in parallel effectively reduce inductance..

And we have a winner finaly someone that understands the finner points of the Art and yes folks its an art of decoupling