Not snobbery - when you start building higher voltage amplifiers you just need a higher voltage transistor. Go above 70 volt rail to rail and you have to go to the trouble of selecting 3055’s. No incentive to nowadays as the 15003 is the same price or even less if you have to buy them new anyway. Most people are stuck in that boat, and would even use 15003s or even 21194’s for a 40 volt amplifier.
I did buy about 80 pairs of 2N5879/5881 when I could get them for a song last year ($1.50 per complementary pair). Why? To use as 3055/2955 replacements so I don’t use up my stash of higher voltage parts in situations where they are not needed. It saves all my 15024’s for rebuilding old Phase Linears and CS800’s where a lesser part simply will NOT work unless you like fireworks coming from the amp rack and the sound to go out. All TO-3’s have started to go through the roof the last 3 years and at some point I’ll only be working from inventory on hand and buying new only in dire emergency. I can see $15 for a 3055, 15024, or 21196 coming within the next 5 years followed by a very rapid phase out entirely. Then it will be like trying to get Sanken LAPTs in the MT200 where the high end repair shops hoard them and charge a fortune (because once they're gone they’re gone).
I did buy about 80 pairs of 2N5879/5881 when I could get them for a song last year ($1.50 per complementary pair). Why? To use as 3055/2955 replacements so I don’t use up my stash of higher voltage parts in situations where they are not needed. It saves all my 15024’s for rebuilding old Phase Linears and CS800’s where a lesser part simply will NOT work unless you like fireworks coming from the amp rack and the sound to go out. All TO-3’s have started to go through the roof the last 3 years and at some point I’ll only be working from inventory on hand and buying new only in dire emergency. I can see $15 for a 3055, 15024, or 21196 coming within the next 5 years followed by a very rapid phase out entirely. Then it will be like trying to get Sanken LAPTs in the MT200 where the high end repair shops hoard them and charge a fortune (because once they're gone they’re gone).
and
Well the way I see it the old 3055 had a gain that dropped rapidly with increasing current (non-linear) and had very low gain at high current - compare to a modern "extended beta" device like a '3281 which holds its ~100 gain upto 4 or 5A and has a much flatter gain/current curve, and you'll see where the real beasts live!
One datasheet I find for the 2N3055 suggests gain drops from 100 to 25 as the current varies from 0.5A to 5A. The cream of 1970's performance just doesn't stack up against current devices, contributing to large-signal distortion in output stages, and pushing drivers and pre-drivers much harder.
And at last TO3 is dying, good riddance, awkward and clunky to use and always has been - a transistor with two pins and noone thought there was something missing?
Yes they look great and I even have a fond memory of them, but they simply are much harder to deal with than simple flat packages - a bad design in everything except the hermetic seal. There does need to be a non-epoxy package for devices for high reliability / extended temperature duty, but it should at least have 3 pins!
I found the TO3 packages hard to work with at one time.
I have built quite a few JLH1996 amps with MJ15003's.
I found the trick is to use the pcb as a drilling template.
Drill first hole then put M3 bolt through the hole to stop it moving around.
Drill hole at opposite end of heatsink and fit M3 bolt in that.
Drill out last two M3 holes.
Then drill out transistor pin holes about 2mm so they dont open up pcb pads too much.
Then make all holes 6mm.
Finally un-burr all the holes carefully.
The bolts through the transistors have plastic bushes inside the heat sink to stop them shorting out.
I have built quite a few JLH1996 amps with MJ15003's.
I found the trick is to use the pcb as a drilling template.
Drill first hole then put M3 bolt through the hole to stop it moving around.
Drill hole at opposite end of heatsink and fit M3 bolt in that.
Drill out last two M3 holes.
Then drill out transistor pin holes about 2mm so they dont open up pcb pads too much.
Then make all holes 6mm.
Finally un-burr all the holes carefully.
The bolts through the transistors have plastic bushes inside the heat sink to stop them shorting out.
I wouldn’t say “good riddance”, because I still like using them for smallish amps, building vintage clones, and rebuilding vintage gear. But if you’re into that start thinking about lifetime buys if you haven’t already. I probably have all the TO-3’s I will ever use now (hundreds in stock, all purchased either surplus or before 21194’s went from $4 to $7). They are a royal pain in the rear for building amps which use 20 to 40 output transistors, and some of those are on the horizon. I even started doing flatpack lifetime buys years ago, anticipating industry consolidation and price hikes. I still have original C3281/A1302 on hand, as well as hundreds of MJL21193/4 from when they used to cost a buck and a half in hundreds. Some dating back to 95 that have the”M” logo. All batches verified real by cracking some open. That will build a lot of 4 and 5kW amps.
Actually, I prefer the TO-3 cans. I always considered the higher temperature operation somehow advantageous compared with plastic devices, and the two bolts allowed for a good thermal contact.
Like Nigel I also used PCB as the template, drilling the holes in two stages.
I have a small stock of TO-3 devices for replacements, too.
4BA screws were ideal if you used RCA's 0.25-inch diameter insulating bushes. Standard washer diameter 0.3 inch. The metric near equivalent M3.5 does not quite have a large enough washer size (just about 0.25 inch) that after a while the bushes sag into the mounting hole, so I reluctantly use the smaller washers more often.
Once upon a time there was a TO-3 with three leads. That never seemed to last the test of time. I never found having to use an eyelet type connector on the collectors while soldering the emitter and base leads a problem, but mounting everything on the PCB these days (whether plastic or TO-3) has eliminated the interconnecting flying leads that used to run between the PCB and heatsink in older designs.
Like Nigel I also used PCB as the template, drilling the holes in two stages.
I have a small stock of TO-3 devices for replacements, too.
4BA screws were ideal if you used RCA's 0.25-inch diameter insulating bushes. Standard washer diameter 0.3 inch. The metric near equivalent M3.5 does not quite have a large enough washer size (just about 0.25 inch) that after a while the bushes sag into the mounting hole, so I reluctantly use the smaller washers more often.
Once upon a time there was a TO-3 with three leads. That never seemed to last the test of time. I never found having to use an eyelet type connector on the collectors while soldering the emitter and base leads a problem, but mounting everything on the PCB these days (whether plastic or TO-3) has eliminated the interconnecting flying leads that used to run between the PCB and heatsink in older designs.
Nothing actually wrong with the 2N3055 I suggest. This may partly because of a hark-back to RCA's original. I used a pair in my first amp of about 30W. Upgraded to 50 shortly after. But, comparing it with Bailey's design left something to be desired. The Bailey 30W used epitaxial transistors. The 2N3055 and MJ2955 are both epitaxial devices now, and a pair will work well in Bailey's design. That gave about 0.02% distortion, at 1kHz. I could not match Bailiey's claimed distortion at 20kHz, however, without changing the VAS for a cascode.
I now have a 50W amp which achieves about .003% at 1kHz and rises to .02% at 20kHz using a pair of epi base devices.
They are still capable of delivering good quality. I once read Douglas Self saying they weren't the best any more, just after he had discussed measuring an increase in supply current at 30kHz. That only happened in my amps with the old original RCA 2N3055. This may have caused some of the misunderstanding.
It is true that the epi base devices are not state of the art, but I think with enough base drive they can deliver a useful circuit.
Maplin's original design seemed to be very popular, but it certainly had weak spots which needed improvement.
Nigel's design is such an improvement. I also have an improvement - it was not hard given the starting point.
However, it does seem that after almost 60 years of the 2N3055 (and 70 years for the TO-3) the TO-3 is being retired. I suspect it has a few years yet.
There is always the TIP3055/TIP2955 as a plastic look-alike.
I now have a 50W amp which achieves about .003% at 1kHz and rises to .02% at 20kHz using a pair of epi base devices.
They are still capable of delivering good quality. I once read Douglas Self saying they weren't the best any more, just after he had discussed measuring an increase in supply current at 30kHz. That only happened in my amps with the old original RCA 2N3055. This may have caused some of the misunderstanding.
It is true that the epi base devices are not state of the art, but I think with enough base drive they can deliver a useful circuit.
Maplin's original design seemed to be very popular, but it certainly had weak spots which needed improvement.
Nigel's design is such an improvement. I also have an improvement - it was not hard given the starting point.
However, it does seem that after almost 60 years of the 2N3055 (and 70 years for the TO-3) the TO-3 is being retired. I suspect it has a few years yet.
There is always the TIP3055/TIP2955 as a plastic look-alike.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.