Drivers need to be fastER than the outputs. With 400kHz hometaxial outputs, practically anything is faster - even TIP31/2. Now you can just throw it into a simulator, but back in 1966 things were hand calculated making many simplifying assumptions along the way. If your drivers are too slow, putting the non dominant poles too close together, those simplifying assumptions break down and calculations can be way off. If the output capacitance of the drivers is too high (even if fast enough) those assumptions also break down, which is why oversizing them should be avoided. Sometimes resulting in instability, and certainly worse frequency response than “predicted”.
But I suspect that stability really isn’t an issue here - since the amount of negative feedback available isn’t anywhere near that of a modern design. The slow drivers increase the crossover distortion. It takes longer for them to suck the charge out of the outputs so the crossover bump is larger in magnitude and happens over a longer period of time. And at higher frequencies, near fT/Beta (which can be around 20k) you’re running out of NFB so it can’t be corrected.
But I suspect that stability really isn’t an issue here - since the amount of negative feedback available isn’t anywhere near that of a modern design. The slow drivers increase the crossover distortion. It takes longer for them to suck the charge out of the outputs so the crossover bump is larger in magnitude and happens over a longer period of time. And at higher frequencies, near fT/Beta (which can be around 20k) you’re running out of NFB so it can’t be corrected.
Fairchild/ST BD139/40 are typically 50 MHz which is “fast enough”, as are MJE340/50 even though neither is specified. But the former have higher gain at 20V and below where it is sorely needed here.
243/253 is a fairly large die, not too unlike MJE1503x. 1381/3503 are not big enough, unfortunately. The C2690/A1220 are also unfortunately going the way of the dodo. Leaving us with 2 options from Toshiba and one from Sanken (get them while you can - when they go away the only choices will be the larger OnSemi units, UTC D669/A649 while THEY last, or the Central Semi repro’s of the old TO-5’s).
When ON acquired all the old Sanyo parts (what was it 15 years ago now?) I though “Great! I’ll be able to get all my favorites forever!” See how that worked out.
First assertion of that I've seen in the twenty years I've been on diyaudio. No specification of that has ever been published and I don't have any way to test Ft. The last source of toshiba or sanyo anything in USA I ever saw was multicomp, but when I bought some TO126 they were ****ese clones with a unknowable logo with the datasheet published by toshiba. When multicomp was shut down ~2015, those parts were not moved to newark SC. When I discovered KSC3503/KSA1220 with actual verbage about frequency and low Cob, I bought some from mouser. Took 8 months to get them, then 6 months after that a burglar hauled them to the steel scrapyard for me.
I can get MJE15028/29 in two days surface from 3 sources. The Ft is 30 mhz min. The cob chart shows 40 pf npn 70 pf pnp at 50 v. You will note there is NO such Cob chart on MJE15032/33 or MJE15034/35, which wg-ski talks always as if they were the same part. What I can test is recording of top octave Steinway grand versus the wood+steel original at Kentucky Center for the Arts or Louisville Highlands Methodist Church. MJE15028/29 as drivers for NTE60 (likely MJ15003 by soa spec) pass the top octave Steinway test, TIP31c/32c fail the test. Such test cannot be replicated by typical US male that is deaf above 6 khz. I'm good to 14 khz.
Per whitedragon's sim results of oscillation post #40 if the centerline of the dynaco ST120 goes off center, just goes to prove AX6 is better design with fewer parts and the same 6 transistors and rail voltage. Has DC feedback. Wonder if the mooly 1 ohm base stopper on O.T. stops the oscillation in sim? Sims do not run on my op system except a *****y browser version that crashes the op system frequently.
Last edited:
Universally not stocked in USA since I've been on diyaudio about 2011.
Except of course on e-bay. You can buy any transistor you want on ebay. Or a plastic package with the right logo and part number, anyway.
Having bought some trash from NTE, just because a part has the same part number, does NOT mean it meets the originator specifications. Even fairchild & On leave out a lot of the specifications. You would think market for BD139/140 is big enough for fairchild or ON to add one more test to their datasheet. Since 2011 the fools keep lining up to buy the trash.
MJE15028/30/32 is the same part, graded for voltage, like TIP41A/B/C. The 15034 is a similar higher voltage part. It needed some processing differences to push the voltage up to 350. It results in lower gain. Even within the 28/30/32 family, gain tends to run inverse of ACTUAL vceo breakdown. If you buy one as a 28 and it runs closer to the 120V spec than 250V, the hFE will run toward the high side of the distribution. And vice versa.
Multicomp was just MCM’s house brand - much as Dayton Audio is to Parts Express. Their transistors could come from anywhere. Last “Multicomp“ I got were Mospec 2N6609’s.
The fact that Fairchild/ON doesn’t guarantee 190 MHz isn’t a problem when 30 will do. THE problem with BD139 is the miserable 80 volts. Not a problem here at all, but there is only so much you can do with an 80 volt supply. And 100 watts at 8 ohms is not one of them.
Toshiba still makes 2 driver pairs - available from Mouser. TTC011/A006 is supposedly their “discontinued” C4973/A1837 stuck into a TO-126. But heaven knows what the lead time is. Ive seen the MJEs out of stock for long periods too. I’ve been waiting over a year on Infineon mosfets.
Multicomp was just MCM’s house brand - much as Dayton Audio is to Parts Express. Their transistors could come from anywhere. Last “Multicomp“ I got were Mospec 2N6609’s.
The fact that Fairchild/ON doesn’t guarantee 190 MHz isn’t a problem when 30 will do. THE problem with BD139 is the miserable 80 volts. Not a problem here at all, but there is only so much you can do with an 80 volt supply. And 100 watts at 8 ohms is not one of them.
Toshiba still makes 2 driver pairs - available from Mouser. TTC011/A006 is supposedly their “discontinued” C4973/A1837 stuck into a TO-126. But heaven knows what the lead time is. Ive seen the MJEs out of stock for long periods too. I’ve been waiting over a year on Infineon mosfets.
Yes, the 34 is different. But the other 3 are the same, just binned. The hFE falls off below vce=10V on the 34 a lot more than the others. It’s what you have to do to reliably raise the vceo, getting it into the 400’s consistently so that a 350 spec will yield. Not a problem with +/-150 volt supplies (and the EF3 you’d need anyway), but at +/-30 where 10V is an appreciable fraction of it you’re simply better off with the lower voltage part/family.
I have recently ended up with an ST-120 with upgraded caps but in need of the output transistors being replaced on at least 1channel. I have a set of 4 2SC1586 which are brand new. Would they work as replacements, or are they too fast and would oscillate? Any opinions would be appreciated. Thanks.
Sounds fun.
Be nice to have a few 2SC1586 laying around for a older Quasi
such as this.
I wouldn't worry about it too much.
Most the importance is the usual.
Making sure none of the other semi conductors or diodes
were damaged from the original fault.
And see if any resistors exposed to high heat show
drift or higher values from heat all their life.
Not very many caps on the board.
relatively easy to shotgun replace all the electrolytic
with new guys.
If it did oscillate would assume just basic problems
with the frontend of the amplifier/ missed damage
from the original fault.
Dont remember the topology having differential input.
So the frontend should be relatively stable regardless.
clean bill of health and good.
fun amp
Be nice to have a few 2SC1586 laying around for a older Quasi
such as this.
I wouldn't worry about it too much.
Most the importance is the usual.
Making sure none of the other semi conductors or diodes
were damaged from the original fault.
And see if any resistors exposed to high heat show
drift or higher values from heat all their life.
Not very many caps on the board.
relatively easy to shotgun replace all the electrolytic
with new guys.
If it did oscillate would assume just basic problems
with the frontend of the amplifier/ missed damage
from the original fault.
Dont remember the topology having differential input.
So the frontend should be relatively stable regardless.
clean bill of health and good.
fun amp
Thank you very much. I really appreciate such a detailed and useful response. I just got lucky on the 2SC1586. I bought a couple surplus unused Tektronix assemblies for the heatsinks. They each had a pair of 2SC1586 on them. I had no idea what they were until I looked them up later. I think I got lucky.
Thanks again for all of the great information.
Thanks again for all of the great information.
Those may be Sanken parts, but they predated the LAPT as we know it today. fT is higher than the old original parts, but not any higher than the OnSemi parts that usually get used as replacement/upgrades today. That one is more like the D424 or D555 than the 30+ MHz flatpacks they make today. I doubt you’ll have any trouble just dropping them in.
Since the output transistors are located 2" from the driver board PC-14, any epitaxial transistor is a small oscillation risk. Run the wires from PC-14 to the O.T. as far apart as possible, not closely parallel. The feedback wire that goes to the VAS, I put a 8 turn salvage coil in it. make sure the TIP mod is done. Ie a zobel on the output jack, 1000 ohm series .01 uf ceramic cap between hot speaker terminal and the speaker return. If you use 10 ohm 1 watt resistors instead of wires to drive the bases of the O.T. it can help prevent oscillation also. A tip from Mooly. I did it before internet, just to provide something to burn up instead of the driver transistors if the O.T. shorted. Hint to prevent O.T. shorting, provide more heat sink fins or a fan.