Difficulty levels for amp building

Status
Not open for further replies.
These amplifiers use the same PSU:
Local voltage primary
18V+18V x 300VA secondary
2 x 35A bridge rectifiers or equivalent discrete diodes
CRC: 30mF + 0.1R 12W + 30mF + 2K2 bleeder per rail

They also require between 5lb and 6lb (2.3Kg to 2.7Kg) of reasonably designed heat sink per side (L + R)

border="1"
|-
| class="tcat" | Amplifier
| class="tcat" | Input stage
| class="tcat" | Output stage
|-
| Aleph J || 2SJ74 + 2SJ74 || 2 x IRFP240 (matched) + 2 x IRFP240 (matched)
|-
| F4 || 2SK170 + 2SJ74 || 3 x IRFP240 (matched) + 3 x IRFP9240 (matched)
|-
| F5 || 2SK170 + 2SJ74 || 1 x IRFP240 + 1 x IRFP9240
|-
| F6* || 2SK170 + 2SJ74 || 2 x IRFP240
|-
| M2* || 2SK170 + 2SJ74 || 1 x IRFP240 + 1 x IRFP9240
|-
| BA-3 || 2SK170 + 2SJ74 || 3 x IRFP240 (matched) + 3 x IRFP9240 (matched) or 6 x IRFP240 (matched)
|-

To save other 'beginners' for asking the same questions that I have asked, please note that the comment about the power supply is for a STEREO (2 channels) amp but the semiconductor list is for a MONO (one channel) amp. So multiply all the semiconductor requirement$ by two for stereo. And check the BOM for the full transistor list. For a stereo Aleph J, it looks like 16 transistors need to be on my list IF they are matched. If not matched, more (number of IFRP240 is a gamble?) must be bought to allow matching . How the matching works -i.e. which pairs exactly need to be matched in the output stage of each channel, and how closely the channels need to be matched (I don't want different gain in the R&L channel!)- isn't clear.
 
For a stereo Aleph J, it looks like 16 transistors need to be on my list IF they are matched.

(4) per amp channel, so a total of (8) IRF240 for a stereo Aleph J.


If not matched, more (number of IFRP240 is a gamble?) must be bought to allow matching . How the matching works -i.e. which pairs exactly need to be matched in the output stage of each channel,


The N-channel mosfet are usually fairly close if from the same production batch, so if buying from Mouser or DigiKey, they probably will be. If you buy your Fets from Ebay you are totally on your own.

I'd buy 3x the devices you need, match them up, you'll most likely have enough for another whole amp, and you can sell those as matched to the next guy and recoup most of your $. 🙂

Specifically talking about the Aleph J, there are 2 sets of (2) per amplifer PCB, one set are the output devices, the other set is the output Constant Current Source. (Looking at the schematic, the top is the CCS, the bottom is the outputs.)

Each set should be populated with a pair of Mosfet that has been matched for Vgs (Voltage gate to source), this is to make sure that the paralleled Mosfet turn on at the same time. If they are not matched, one will turn on before the other, hog all the current, and overheat, while the other one doesn't do much at all.

The output fets and the CCS fets do no need to be matched to each other, you basically need 2 patched pair per PCB, not a matched quad.

...and how closely the channels need to be matched (I don't want different gain in the R&L channel!)

Matching the fets in this case has nothing to do with gain.
 
Specifically talking about the Aleph J, there are 2 sets of (2) per amplifer PCB, one set are the output devices, the other set is the output Constant Current Source. (Looking at the schematic, the top is the CCS, the bottom is the outputs.)

Each set should be populated with a pair of Mosfet that has been matched for Vgs (Voltage gate to source), this is to make sure that the paralleled Mosfet turn on at the same time. If they are not matched, one will turn on before the other, hog all the current, and overheat, while the other one doesn't do much at all.
Thanks - the reference to the schematic makes it a lot clearer.
So this matching Vgs is different than the matching for the SJ74s which is current matching? At any rate I need to do some searching to see what is involved in the matching process (jig setup).

The output fets and the CCS fets do no need to be matched to each other, you basically need 2 patched pair per PCB, not a matched quad.
This means that: The output fets and the CCS fets in one channel do not need to be matched to the ones in the other channel. The transistor current/voltage properties don't affect the gain of the amplifier.
For each channel, the two CCS mosfets need to be a matched pair and the two output mosfets need to be a matched pair.
Correct ???
 
Thanks - the reference to the schematic makes it a lot clearer.
So this matching Vgs is different than the matching for the SJ74s which is current matching? At any rate I need to do some searching to see what is involved in the matching process (jig setup).

It's similar, but not difficult. You could use a 19v laptop brick for the PSU (in lieu of a 12v source), see diagram below -

An externally hosted image should be here but it was not working when we last tested it.


And more info on matching here - Embedded Hybrid Headphone Amlifier


This means that: The output fets and the CCS fets in one channel do not need to be matched to the ones in the other channel. The transistor current/voltage properties don't affect the gain of the amplifier.

Yes, and actually the Mosfet don't need to be matched to anything other than the one they sit directly next to. Which is to say you could build a stereo amp with (4) pairs of IRF240s.


For each channel, the two CCS mosfets need to be a matched pair and the two output mosfets need to be a matched pair.
Correct ???

Yes.
 
I didn't expect this many replies to my question...but thank you.

The F6 looks like it's the next one up in difficulty from the ACA so I think I will go with that one. Especially since the power supply can be used in other amps as well as the chassis.
 
My apology for hijacking your thread!🙂
Hopefully some of the answers to my beginner questions will be helpful to others. Thanks again (from me) to all the folks who have helped to 'point me in the right direction'.

In many ways just knowing there are other beginners is helpful. One reads so many posts by people that really understand the engineering details of these amps that it is easy to think you are the only one baffled by some step in the process.
 
Max? Well, overkill kicks in somewhere north of 88,000uF per rail. (8) 22,000uF caps total

If using normal diyAudio PCB use 15,000-22,000uF caps in each position. Nelson uses 15,000 in the factory-built ones, so anything more is just vanity... 😀 😀 😀
 
Status
Not open for further replies.