[linuxworks]

Quote:

Originally Posted by MisterX /img/forum/go_quote.gif How close is the mains transformer to the amplifier circuit board?

pretty far away. the ps is a TREAD with a 24vac xformer and about 6' of cord from the floor to where the amp is. (too long?)

its now using a 5pin din and I'm feeding 20v (give or take) on 2 pins and 5v-dc on 2 other pins (for my motors and IR logic board). no common ground - I went to extra pains to keep them separate

so, my PS sends in 20v for audio and 5v for the rest, each having their own +/ground pair.

 

[linuxworks]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif . Bottom line, I think you're haring off looking for exotic fixes, when there are a whole lot of other, more common things to look at.

to be clear, I'm not sure there *is* a problem, yet. this may very well be a case of 'bad program material' (not well recorded) and having a clean enough amp to actually/finally notice it

I may need to bring another set of ears into this. I am not going to rule out that the equipment is 100% fine and I've been hearing flaws in the recording and mixing all along

Quote:

The trimmer on the PPA is in a totally different place than R10 in the PIMETA. The PPA has an R10, too, in exactly the same place as R10 on the PIMETA. (On purpose.)

well, just a few hours ago I built a PPA (v1). it worked the first time (sort of) and now I have another amp I can play with to compare.

sort-of - meaning that being 'built' only counts when ALL the parts are installed. all means all, even 2 10ohm R's that were on 2 of the 3 channels. grin. say no more?

when I filled the last 2 R's all came to life, so it worked 'the first time' the way *I* count's 'em.

I may need some 'reference' program material. are there any downloadable (legit) test samples that might be good for testing out a new build?

 

[linuxworks]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Biasing the BUF634 into class A is a trade-off between it's current efficiency and preserving class A for low impedance headphones at high volumes. To be in the middle, set the bias current to roughly 120mA.

hang on, is it the BUFFER that we're class-A'ing or is it the op amp? I thought it was the op-amp that was being re-biased (?)

 

[tangent]

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif this may very well be a case of 'bad program material' (not well recorded) and having a clean enough amp to actually/finally notice it

True.

You test it by changing out the pieces one by one, re-testing with each change.

Quote:

I may need some 'reference' program material.

Bad software is certainly possible, but it's rare to find two pieces with exactly the same underlying flaw. If the symptom changes with the disc, the disc is probably the problem. If not, it's something else.

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif hang on, is it the BUFFER that we're class-A'ing or is it the op amp?

You've been talking about op-amp biasing parts. majkel is opining that I've done it wrong, and is giving you advice on how to build a different amp, which biases the buffers instead.

 

[rds]

Quote:

To be in the middle, set the bias current to roughly 120mA.

The middle of what? Between 0 and its max current output. That's pretty arbitrary.
The DIP and SO-8 have a max power dissipation of 1.2W and 0.9W respectively. So assuming more than 10V on the rails you will put the either version into thermal shutdown within minutes. Probably not a great suggestion.

 

[00940]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Biasing the BUF634 into class A is a trade-off between it's current efficiency and preserving class A for low impedance headphones at high volumes. To be in the middle, set the bias current to roughly 120mA.

How do you intend to bias the buf634 to 120ma ? Not by sticking a ccs at its output I hope (it's a class ab push pull diamond buffer).

Did you mean 12ma (which is possible using wide bandwith mode) ?

 

[majkel]

You can just use a single resistor, or JFET cascode (look for big current JFETs here

) wired to the negative supply voltage. Yes, I mean 120mA. For the output, 12mA would be absurd.

 

[majkel]

Quote:

Originally Posted by rds /img/forum/go_quote.gif The middle of what? Between 0 and its max current output. That's pretty arbitrary. The DIP and SO-8 have a max power dissipation of 1.2W and 0.9W respectively. So assuming more than 10V on the rails you will put the either version into thermal shutdown within minutes. Probably not a great suggestion.

So stick a heatsink onto the buffer. We're talking about things that make sense. Biasing the driver stage op-amp into class A brings no audible gain, rather loss, if only we're speaking about modern op-amps. Biasing the output into class A makes huge progress, however I tried it with other parts than the BUF634 and under lower voltages, and yes, they get hot but it's temperature is quite safe for the silicon. And again, you can think of cooling it with a heatsink or a metal bar with a silicon<copper<silver paste in between.

 

[cobaltmute]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif You can just use a single resistor, or JFET cascode (look for big current JFETs here ) wired to the negative supply voltage. Yes, I mean 120mA. For the output, 12mA would be absurd.

Just for the curious amateur, can you schematic your idea here?

 

[mono]

If someone wants to try biasing BUF634 to 120mA, I suggest the TO-220 part heatsunk, in a different amp not a Pimeta. Further, I suspect there is a point where very hot buffers degrade sound more than 100% class A operation improves it.

Edit: Oops, sorry about reviving a dead topic.

 

[NelsonVandal]

It would be interesting to challenge those tweaks considered to be an improvement in blinded A-B tests. If they don't do any good, we could build cheaper and smaller amps with better runtime and still be happy. I can see a lot of these tweaks being intellectually appealing, but do they actually make an amp sound better? The placebo effect is very strong when you've built/modded an amp yourself, or payed a lot of money to have it done.

1. Class A biasing of the input opamp
2. Isolating JFETs/resistors/other ccs on the supply rails to the input opamp
3. TLE2426 vs. resistors/capcacitors vs. resistors/opamp as input ground
4. Output ground channel amp vs. resistors/capacitors (still using an isolated input ground)
5. Multiloop vs. single loop
6. Std parts vs. "audio grade" parts
7. Cheap-o vs. decent quality copper vs. silver cabels
8. Adding bypassing caps in the power section
9. Will really large caps make a better sound in a 3 ch amp? in a 2 ch amp?
10. Class A biasing the output opamp/buffer

etc... please comment and make the list longer (or shorter if valid tests already have been made)

 

[linuxworks]

Quote:

Originally Posted by mono /img/forum/go_quote.gif If someone wants to try biasing BUF634 to 120mA, I suggest the TO-220 part heatsunk, in a different amp not a Pimeta. Further, I suspect there is a point where very hot buffers degrade sound more than 100% class A operation improves it. Edit: Oops, sorry about reviving a dead topic.

not a dead topic, its just been 'resting' for a while (lol).

I do like the idea of a dual approach towards things. there is what your ear can detect (also what you are willing to tolerate in imperfections) vs what the best test gear can measure.

its intellectually interesting and challenging to try to 'beat the specs' but that is often not going to make the sound *audibly* better.

so what do you do? build something that 'tests well' and has all the right show-off parts? or do you build something for people to listen to?

it depends

sometimes you are building for a client or to a build-spec or even as a demo. but other times you simply want to install and use the gear.

it does help to keep those 2 separate. at least in my mind, it does.

I'll soon be building a beta22. I have a 3board set all sitting here waiting for me to build it

and I'll build and test all 3 boards - BUT I'll defer my decision to install 2 or 3 after I've had some fun testing and listening. I may build 3 but I may install only 2. if that doesn't demonstrate my 2-level approach, I don't know what does

an extra board takes up a LOT of chassis room. it may not be worth it, to me (at the time) to take chassis space on a '3rd channel' if the 2 main channels sound good enough on their own.

I think it helps to balance the 2 levels and not let any single one dominate.