New PassDIY Headphone Amp (now available)

Jump to Latest
#103
For your interest :
http://www.diyaudio.com/forums/head...esktop-headphone-amplifier-8.html#post4280285

"I must say I'm sold on this whole "match your transistors" thing, even if I did not, technically, match anything .. I just discarded outliers. The improvement, at any rate, over the previous build is substantial but hard to describe. Increased soundstage precision? Yes, but that's not really the main thing. Vanishing levels of grain/glare? Maybe that's closer to it. It's in the way the amp slots away each musical line so gracefully now, softly yet with absolute authority. Previous iterations of the Sapphire, while clean and powerful, never quite reached this level of refinement."


Patrick
 
#104
Hi Patrick, funny you posted that, I have a Sapphire 3 on the way. I haven't done comparisons in sound between matched and unmatched, but it makes sense to match complementary parts as P and N channel parts are often quite different. Its got to be better.
 
#105
I posted that because it was from someone who did not believe in matching.
Also he was matching BJTs and not FETs, the former having much less variations between devices than the latter.

One can achieve intrinsic linearity by careful selection of devices (matching including).
Or one can correct for non-linearities by applying heavy global negative feedback.
Using an unmatched source follower pair and applying a loop feedback round it to the frontend opamp belongs to the 2nd variant.
It is also not unlike tube circuits with output transformers placed inside the feedback loop.

Of course everyone is free to choose what he prefers. 😉


Patrick
 
#106
EUVL said:
I posted that because it was from someone who did not believe in matching.
Click to expand...

Patrick, reread what I actually posted. Nowhere did I state that I did not BELIEVE in matching transistors.

I'm well aware of the benefits of matching transistors for certain circuits and do so if required.

My point is the Pass/Colburn circuit appears to be an easy circuit to build WITHOUT the need to match devices.

I'm sure Richard's Sapphire is a nice amplifier and that he's a knowledgeable designer, but I don't buy into things like "reduced glare" just from matching a lot of BC5XX devices or taking the outer plastic sleeves off electrolytic caps for "better sound".
If you buy into such claims...fine.
 
#112
6L6 said:
IMG_3167.jpg

Next in the queue to try. 🙂
Click to expand...

I am keen on hearing your impressions on those muses OPAMPs 6L6. Have you had a chance to give them a listen?
 
#114
EUVL said:
I posted that because it was from someone who did not believe in matching.
Also he was matching BJTs and not FETs, the former having much less variations between devices than the latter.

One can achieve intrinsic linearity by careful selection of devices (matching including). Or one can correct for non-linearities by applying heavy global negative feedback.[....]Of course everyone is free to choose what he prefers. 😉
Click to expand...

1. Matching large volumes of transistors is a PITA. However, I do it so you don't have to.

2. In terms of hfe, my BJT lots as-purchased are sprinkled with outliers. Most are fine to use without further binning, but 2-5% are far, far away from the mean. If you end up with one of those in a parallel output stage running open loop, or a symmetry-reliant topology like a diamond buffer, it's going to screw things up. When I used binned transistors, the sound was smoother / less glare. YMMV.

(now back to your regularly scheduled thread)

The 4N35 is a clever way to stabilize the bias current. No downsides that I can see.** MOSFETs are voltage controlled, so its easy to adjust the bias current through the gate voltage like this, and thermal stability comes for free.

** Except in the unlikely case the optocoupler fails. The transistor turns off, gate bias doubles ... unless the source resistors R16,22 are big enough to catch up its bye bye FETs.

I wonder if this trick would work with bjt output stages?
 
Last edited:
You must log in or register to reply here.
Top Bottom