[jboehle]

Tangent's guide says around 1mA for biasing op-amps into Class A, but indicates that the required bias can vary quite a bit depending on the application.

My application is: PPA, OPA627BP G, OPA637BP L/R, 4x Intersil buffers per channel, driving either Senn HD497 (32 ohm) or Senn HD280 (64 ohm). I have the gain set at the default (from the schematic) right now, I think I am going to lower it a bit. I have Tangent's crossfeed board in front of the pot so the default gain isn't too bad even with the low-impedance Senn HD497s.

I am using closely matched 2N5486 (Q2) & 2N5484 (Q1) for the biasing. Right now I have 475 Ohms in R9 and 200 Ohms in R8. This gives me a bias of ~1.86mA per channel.

1. Is this too much?
2. What current do you other builders out there use for biasing to class A the same PPA config?
3. What do you use for other PPA configs (AD8610s, etc.)?
4. Any comments on using a different value for R9 (ie. 100 or 1K Ohms)? Is there any positive or negative effect on the sound with varying R9 values?

Just looking for some other experiences from you guys!

-Jason

 

[jamont]

I've wondered about this myself. On this one, I used the same JFET's and similar resistors, my bias was about 1.6-1.7 mA per channel. I built another one last year, using 3xAD8610, and the same JFET's, and wound up with 1.3-1.5 mA per channel and IIRC the resistors were the same in each channel (220R I think). I recall that 1.3 mA or so was recommended (by tangent?), it doesn't seem like the exact value should be critical. I wonder also if changing the opamps would affect the measured bias current, leaving the resistors unchanged? It's in the feedback loop, so changing the gain may affect this as well. I have not tried changing R9.

 

[IpsilonSound]

Quote:

Originally Posted by jamont I've wondered about this myself. On this one, I used the same JFET's and similar resistors, my bias was about 1.6-1.7 mA per channel. I built another one last year, using 3xAD8610, and the same JFET's, and wound up with 1.3-1.5 mA per channel and IIRC the resistors were the same in each channel (220R I think). I recall that 1.3 mA or so was recommended (by tangent?), it doesn't seem like the exact value should be critical. I wonder also if changing the opamps would affect the measured bias current, leaving the resistors unchanged? It's in the feedback loop, so changing the gain may affect this as well. I have not tried changing R9.

That's exactly what I used too in my PPA. AD8610 on all 3 channels, 2N5486 for Q2 and 2N5484 for Q1 with default resistor values so far(100 and 1K). I plan to adjust the bias later to recommended ~1.0mA value for each channel, though will have to desolder on lef of R8/R9 off the board.

I would need some more detail on that one...

All the 3 channels must be individually adjusted, right? Or maybe adjusting one channel and replicating resistor values to other 2 should be enough, since in the reality all the channels use the same JFETs and OPAMPs, I guess(except for the 637/627).

Just curious, how much did it affect the sound for you? Did you notice any improvement?

Thanks!

 

[jboehle]

Quote:

Just curious, how much did it affect the sound for you? Did you notice any improvement?

I haven't had the time to do any critical listening yet. I didn't hear any "huge" improvement in the sound right off or anything though.


Quote:

2N5486 for Q2 and 2N5484 for Q1 with default resistor values so far(100 and 1K)

100 Ohms for R8 gave me something around 2.2mA bias. Just FYI.


Quote:

All the 3 channels must be individually adjusted, right?

Yes.


Quote:

Or maybe adjusting one channel and replicating resistor values to other 2 should be enough, since in the reality all the channels use the same JFETs and OPAMPs, I guess

Not unless you've matched your JFETs. I just measured and sorted 100 of each type (2N5486, 2N5484), and the Idss values vary widely. For the 2N5486, I measured as low as 11.0 and as high as 14.5, and the 2N5484 as low as 1.8 and as high as 3.8. So, unless you match your JFETs, you should adjust the bias individually for all 3 channels.

Now, in my case, I picked JFETs that all had the same value (rounded off to the nearest 1/10th of a mA). Then when I went to try different values of R8, I found that the same resistor value for all 3 channels (1% Vishay-Dale RN55D resistors) provided (very nearly) the same bias current across R9, the 3 channels all measured within .03mA of each other.

With the 200 Ohm resistors I have in my PPA right now, my left and ground channels measure 1.86mA and the right as 1.89mA. (I trust that my meter is accurate down to those levels, it's a 5.5-digit HP 3468B that was calibrated in January.)


Quote:

though will have to desolder on lef of R8/R9 off the board

That's why you socket those positions!

-Jason

 

[tangent]

Those of you who don't have a PPA or will be building another should hold off on pursuing this question for a few weeks. When v1.1 comes out (soon), it will have a pot in place of the current R8 for adjusting the bias of the cascodes. Then you can tweak and listen, tweak and listen, until you find the value you like.

I predict we'll be seeing a lot of people giving their opinion on this subject when amps start getting built on the new version of the board. Up until then, it's just been too much hassle to try different values to make an opinion for most people.

 

[IpsilonSound]

When v1.1 comes out (soon), it will have a pot in place of the current R8 for adjusting the bias of the cascodes. Then you can tweak and listen, tweak and listen, until you find the value you like.

Tangent, that solution would be ideal and it crossed my mind many times last night when tweaking the bias by fumbling through a big pile of assorted resistors, checking, plugging in, measuring and so on. What kind of pot would you recommend, brand and range? Since it's not in the sound path, quality isn't crucial, as long it's solid, holds the value and not flimsy.

Sign me up for v1.1 asap, I'll order some soon!!! I think, even the present version of pcb can be tweaked to hold a small pot vertically, I'll think about that one. I still have another pcb/battery board left to build a second ppa.

 

[tangent]

Quote:

What kind of pot would you recommend, brand and range?

Brand isn't critical. What you want is a small multi-turn trim pot, probably in the 1K-2K range. The value will depend on the particular JFETs you have and the range of current targets you want to hit. I'll know more when I experiment more with this myself.

For PPA v1.1, the pot I've been testing with is the BC CT-94W series, from DIgiKey. I haven't tested them personally, but apparently the Murata PV36W and Bourns 3296W also fit in the same footprint. These are good general-purpose trim pots, with holes on standard 0.1" centers and a vertical layout with a top-adjust screw, so if you're going to keep at this hobby I'd get several, of different values. They're useful things to have around.

 

[IpsilonSound]

Just for clarification purposes... 1-2k range wouldn't be too much if we put 2N5486(Q2)-2N5484(Q1) jfets? Most people report 400-500 range for R8 to get ~1.2-1.3mA, right?... I think, I know what you're saying... it's for a wider range of opamps, not just 8610 or 627/637's.
What jfets would we need then? Thanks.

 

[tangent]

1K includes ~500 ohms, and in an 18 turn pot you're going to be able to hit the desired value pretty accurately.

If you wanted a very wide range of adjustment, you could use PN4392s for Q2, 2N5486 for Q1, and, oh, a 5K or so pot for R8. That should get you at least up to 5mA, maybe higher, with fine control all the way up.