The 992/1845's I used were on the borderline between B & C for the BC550/560. But I think the more gain the better, so the Cs should work great.
Not this amp but similar.
For EUVL's UTHAiM I am matching quads (for mod Wilson current mirrors) BC327-40,BC337-40. Suffix 40 means 250-630 hFE. This is in line with hFE reported by Jeff a few posts above.
For EUVL's UTHAiM I am matching quads (for mod Wilson current mirrors) BC327-40,BC337-40. Suffix 40 means 250-630 hFE. This is in line with hFE reported by Jeff a few posts above.
They are not terribly expensive. It is not a problem to find six identical ones for one channel and six identical ones for the other out of 400 pieces (200 each). "Identical" within the scope of our testing equipment and testing abilities, of course. I'd say it's worth doing so anyway, because this amp deserves respect 🙂
BTW, 2SD2081 and 2SB1259 are in stock in Digikey.
Sure, I was kind of discouraged by the gap between npn and pnp in my samples (30 each). I could find lots of matches within one type. But it looked like pnp ware around mean for that type and npn at the high end. Too big gap between npn and pnp. Ordered more...
Get 'em while you can. They're parts that don't have a lot of good replacement options....
Someone PM'd me for the button board Gerbers. I thought I had posted them, but maybe my memory really is as bad as my wife says it is.
Anyhoo, here they are.
Anyhoo, here they are.
I need some help with a HPA1 build. I stumbled on this forum a couple of month ago. 1 WHAMMY and a month or so later I'm trying to finish up a HPA 1 based on Jeff's design. A few questions.
1, Where should I be measuring the bias current/what value am I looking for?
2, R11(332ohm) and R41(475ohm) are they supposed to be different, will the offset pot make the difference null? Or should they be the same.
Thanks in advance!
1, Where should I be measuring the bias current/what value am I looking for?
2, R11(332ohm) and R41(475ohm) are they supposed to be different, will the offset pot make the difference null? Or should they be the same.
Thanks in advance!
Don't have a schematic in front of me right now, but if you read the thread carefully I think you may find some gain is picked up in the mirror, hence the different value resistors. Best as I recall anyway.
The R11/R41 difference is a typo. They should both be the same although which value doesn't matter much. (But yes, even the mismatch will work as the pot is used to balance it with R13/R43 anyway.)
Bias current is measured across R21 or R22. They will be slightly different so you can either average them or just pick the same one to measure for both channels. I did the latter and went for 100mA.
Cheers,
Jeff.
[Edit: I can't believe no one caught that before now....😎 ]
Bias current is measured across R21 or R22. They will be slightly different so you can either average them or just pick the same one to measure for both channels. I did the latter and went for 100mA.
Cheers,
Jeff.
[Edit: I can't believe no one caught that before now....😎 ]
Last edited:
I guess that most people are looking just into one channel 🙂 One other thing bothers me though. Why C3/C23 don't have their equivalents on negative MOSFETs?
Jeff,
I see you used a 120VA transformer. I assume that is pretty conservative. I'm trying to fit into a smaller chassis. What do you think of something like 50VA ?
Terry
I see you used a 120VA transformer. I assume that is pretty conservative. I'm trying to fit into a smaller chassis. What do you think of something like 50VA ?
Terry
I agree that conservative is good if you have the luxury. Sometimes reality (physics) steps in.
It looks like the bias current should be roughly 300mA. So 600mA total per winding. A 50VA transformer will give me a little over one amp per winding (at 24 VAC), so that seems reasonable.
The output voltage swing (based on the specs for the HP-1) should be about 11V pk. So the 24 volt rails might be a bit higher than needed also. I might play with lowering them a bit to minimize the heat rise in a smaller enclosure.
Terry
It looks like the bias current should be roughly 300mA. So 600mA total per winding. A 50VA transformer will give me a little over one amp per winding (at 24 VAC), so that seems reasonable.
The output voltage swing (based on the specs for the HP-1) should be about 11V pk. So the 24 volt rails might be a bit higher than needed also. I might play with lowering them a bit to minimize the heat rise in a smaller enclosure.
Terry
The heat inside the case is the real issue. I'm sharing my experience with it in post 162. Larger heatsinks really did not make much of a difference for me. It seems that the capability of the case to accumulate the heat inside kept MOSFETs hot despite larger heatsinks. This amp really needs a large or well-ventilated case to manage the heat, mine is too small, yours maybe even smaller. What I ended up with eventually is that I have removed MOSFETs from the PCB and mounted them on the side of the case, effectively turning it into a heatsink. And now, transistors are just nicely warm instead of hot.
For small cases and less heat, I can recommend the WHAMMY amplifier by Wayne Colburn. It is surprisingly good. It may be too good actually because it's so inexpensive, so easy and safe to build and well-sounding. It's a great value indeed. It also allows to roll op-amps and power devices if one wishes to experiment.
As for the 50W transformer, each rail takes about 110mA per channel @idle, that's 4*24*0.11==10.5W. On top of that, spec says that the amp gives 3.5W @20ohm, but you won't be listening that loud, so 50W more than doubles the peak demand from the amp, even at the highest volume levels.
Last edited:
Piotr,
Thanks for the suggestions.
I have considered using the case as the heat sink. That's a good way to dump the heat to the outside ambient air.
I've built a WHAMMY and I thought I'd see how this compares.
How did you decide on 110mA per channel? The HPA-1 says it can do 3500mW into 20 Ohms, so that means about 300mA bias per channel. 600mA per winding.
A place I might save some heat is the rail voltage. The HPA-1 says 200mW into 600 Ohms or about 11V peak output swing. I haven't looked closely at it yet, but I'm hoping that means the rails might be good as low as +/- 18V. That would help a lot with the power dissipation.
The voltage swing is important for high impedance (600 Ohm) phones which I'm least likely to care about, so if I compromise anywhere that's probably a good place to start. Lowering the bias current a tad would be the next thing.
Terry
Thanks for the suggestions.
I have considered using the case as the heat sink. That's a good way to dump the heat to the outside ambient air.
I've built a WHAMMY and I thought I'd see how this compares.
How did you decide on 110mA per channel? The HPA-1 says it can do 3500mW into 20 Ohms, so that means about 300mA bias per channel. 600mA per winding.
A place I might save some heat is the rail voltage. The HPA-1 says 200mW into 600 Ohms or about 11V peak output swing. I haven't looked closely at it yet, but I'm hoping that means the rails might be good as low as +/- 18V. That would help a lot with the power dissipation.
The voltage swing is important for high impedance (600 Ohm) phones which I'm least likely to care about, so if I compromise anywhere that's probably a good place to start. Lowering the bias current a tad would be the next thing.
Terry
I understand where are you coming from. I was thinking about the same thing, but lowering the rails voltage is not as simple as it looks. This will change the Vds of the MOSFETs effectively pushing them into a different place in their characteristic. It may potentially require changing the value of source resistors. You are basically redesigning the circuit and it may be questionable to claim that it sounds like HPA-1.
I'd say that the safest option will be to use the chassis as a heatsink. It has the additional benefit: you may experiment with higher bias if you wish.
As for the 110mA thing, I was saying that each channel takes roughly 110mA per rail at idle (with input signal ==0V). That's 100mA per MOSFET + some 10mA or so for FJETs / current sources. Two 24V rails and two channels make it 10.5W at idle.
I'd say that the safest option will be to use the chassis as a heatsink. It has the additional benefit: you may experiment with higher bias if you wish.
As for the 110mA thing, I was saying that each channel takes roughly 110mA per rail at idle (with input signal ==0V). That's 100mA per MOSFET + some 10mA or so for FJETs / current sources. Two 24V rails and two channels make it 10.5W at idle.
Last edited: