2SD424 /2SB554 replacement

[afroaudio]

I am freshening up a beautiful old Fostex Lab 300 power amp. So far have replaced the relay and electrolytic capacitors and have now pulled the 8 output T-03s which are 2SD424 and 2SB554 pairs (original Toshiba)
Checking these I have one of the 424's with Hfe of 37 (min spec should be 40) the rest are close to 50 plus or minus. The PNPs are vary between 71 and 99 Hfe.

Question are these values of any real concern here? Should I forge on with the original outputs?

I have a set of OnSemi 21193/94 here ready to swap in if need be, but checking the Hfe's on these (same batch nos) I see 3 of my 21194s are in the 40s-50s and then the 4th has and Hfe of 80!

Maybe all of this is meaningless with output transistors? I am used to matching Hfe on input differential pairs.

 

[afroaudio]

Thanks inductor, I guess it also matters what current is used to measure these hfe values. I should admit I am using one of these cheap testers:

https://www.ebay.co.uk/itm/LCR-T4-M...=item2f25a53ce0:g:L9YAAOSwufpb4xmM:rk:22:pf:0

Also I should ask, do I stand to gain anything by swapping in the OnSemis? Part of me thinks peace of mind, but the other part of me wonders if I might subtly change the way the amp sounds.. and this thing does sound very good!

 

[Watthour]

Unless the existing outputs are really slow, as in < 3MHz, there should be little difference with new outputs. Most of the signal is already "colored" before it leaves the drivers, and the main outputs just increase their current to the level necessary to drive speakers. That's not to say that the new devices might not be better, but there almost certainly should be no loss if all the rest of the specs are equivalent.

 

[afroaudio]

Transition Freq ft on the 2SD424 is 3MHz so I guess these are on the slow side.. the OnSemis are 4MHz, I guess for peace of mind I may as well pop in the new parts. Seems unlikely I'll run into any trouble.. I was just surprised by the large variance in hFE on the new same batch OnSemis.

 

[Watthour]

Of course, "slow" is relative. As long as the repalcements are don't spec worse than the originals, there should be no negative change in sound, and the 4 MHZ MJEs should not present a huge problem with oscillation.

 

[afroaudio]

Thanks for feedback guys.. nobody has commented yet on the rather large spread of hFE values, does this have any bearing at all on how the outputs function in the circuit? Should I be wary of one MJ21194 having double the hFE of another in the same channel or is this entirely irrelevant? I know the production tolerances of the OnSemi parts should be a lot tighter than the old Toshiba parts this has me wondering (one measures 40 and another one 80!)

 

[Hyperion]

I think I am right in saying that wildly different hFE in OP pairs will result in slightly more distortion, but I emphasise slightly. As for speeds of OP transistors it all depends on the design of the amplifier OP stage, in extreme cases (perhaps poor designs) the use of 'slower' replacement transistors can actually result in oscillation. You should obviously try to get the specification of any replacements as close as possible to the originals, but when it comes to TO-3 case replacements the 'MJ series' OnSemi types are the only 'choice'. They are very robust and are even 'difficult to damage' in the vast majority of locations where we would use them.

 

[afroaudio]

Thanks John for chiming in, I think I am going to be fine withe OnSemis.. I have ordered a few more of the NPN's as somehow the hFE range of these seems to be broader (again noting that I am measuring at very low current which is not ideal at all for matching.

I found this great thread with regard to matching output transistors and it seems Vbe may be what we want to match.

http://audiokarma.org/forums/index.php?threads/output-transistor-matching-in-powerful-amps.730458/

I also read this from Rod Elliot

http://sound.whsites.net/transistor-matching.htm

where he states in conclusion:

"Matched transistors will rarely give you 'better sound' (whatever that is supposed to mean). In general, distortion will be almost completely unaffected, and there is no influence on frequency response or transient response. What you will get (for paralleled devices) is greater reliability, because the transistors will share the current more equally."