I would use minimum 150V, and preferably 200V parts for both the first and second pairs, if possible. That will minimize the possibility of kablooey or avalanche-related beta degradation in the event that the amp gets unbalanced (forcing the output to the rail). This amp doesn't really require anything special in either pair - this thing came out long before the super-beta devices we use today were around.
Video outputs tend to have relatively flat hfe and a low Cob, making them decent for VAS duty. Most video outputs can be used to replace the 40412. It's 1W with no heatsink, about 5W with the heatsink. Just put a small heatsink on whatever you use. Don't worry about adding any capacitance - the stage is already heavily lag compensated.
Thanks wg_ski, I appreciate the help. I was thinking that Vceo should generally equal the + and - rail together, since there's just so many phase linear horror stories. I feel like many of them may have originated with the driver board, since the outputs *should* blow fuses if they short out. I am replacing all the electrolytics on the board with 250v parts as well.
That's why 200V is preferable for the input pairs. Normally, they have only one rail across them but if you lose DC feedback that can go higher. The tail resistors will keep the current under control even if they avalanche but it will degrade the beta if they are left that way for very long. The VAS and everything beyond it need the full 200+ volts or bad things happen.
The driver boards do originate a lot of the failures. The predrivers can and do fail short from base to emitter - causing a massive overbias condition. And if the off-board Vbe multiplier opens up (read: the leads break off fooling with the PCB) you kiss all 10 outputs goodbye before the fuse blows. And if you lose feedback, you get 100 volts (or 75 on the PL400) to the speaker. Maybe ok for a 15" woofer, but a cute little 6" will be toast before the fuse can blow.
The driver boards do originate a lot of the failures. The predrivers can and do fail short from base to emitter - causing a massive overbias condition. And if the off-board Vbe multiplier opens up (read: the leads break off fooling with the PCB) you kiss all 10 outputs goodbye before the fuse blows. And if you lose feedback, you get 100 volts (or 75 on the PL400) to the speaker. Maybe ok for a 15" woofer, but a cute little 6" will be toast before the fuse can blow.
Alright, after many an hour of careful PCB board repair, transistor replacement, etc. I made extra sure the off-board VBE multiplier was electrically and mechanically connected & sound. I fired it up. No fuse blows (it's underfused right now, 5A mains 3A per rail). Everything looks OK, except I have a DC offset of 11.3V! That seems to be more than just the input pair, or can a small HFE difference cause such a large offset? I used 2SD669A for VAS, BF421 for second pair, and the MMPS5172 off the other side for input pair.
If it is the input pair, can I install a pot somewhere to compensate? Or did I put something in backwards?
Thanks
m6tt
If it is the input pair, can I install a pot somewhere to compensate? Or did I put something in backwards?
Thanks
m6tt
Last edited:
Alright, I had the 2sd669 in wrong, it's in correctly now. Tested it to make sure it was still ok after the brush with transistor leg dyslexia. I pulled all the fuses except the line fuse, powers on. Turned it off. I removed all the output transistors except the RCA drivers. Installed + and - rail fuses for the left channel (the one I am repairing). Powers on, fuses hold. Checked dc at output -100.4v! ARG!
All board transistors I replaced still measure proper diode drops. I have triple checked for dyslexia. My wiring harness doesn't have the original colors (I think this amp has been serviced almost too much). I am thinking I may have gotten the wiring wrong reinstalling it, specifically the feedback. I did my best when connecting it to flip between schematic, board layout and harness diagram and make sure things looked right. I have three grey wires on the left channel, one is purplish. I was missing the yellow wire for the right channel entirely. Yuck.
Barring shorted transistors (none found), and the feedback issue, is there anything else that can put highest B+ across the terminals?
All board transistors I replaced still measure proper diode drops. I have triple checked for dyslexia. My wiring harness doesn't have the original colors (I think this amp has been serviced almost too much). I am thinking I may have gotten the wiring wrong reinstalling it, specifically the feedback. I did my best when connecting it to flip between schematic, board layout and harness diagram and make sure things looked right. I have three grey wires on the left channel, one is purplish. I was missing the yellow wire for the right channel entirely. Yuck.
Barring shorted transistors (none found), and the feedback issue, is there anything else that can put highest B+ across the terminals?
Schematic shows 99.3v Vceo, holds in circuit. Vbe multiplier takes the other half of the rails. Certainly a 250v part would be better, but it seems to work and survive startup etc. Indeed, it has survived longer than the part it replaced during the failure condition. If it fails I have a couple more computer monitor mainboards to pillage.
Currently the amp is doing a new strange thing, which is exploding the 10-ohm bias resistor on one rail only. It did this after I noticed the input jack ground was lifted and grounded it.
Thankfully outputs were not installed. I think I have a wiring harness problem. I have orange, yellow, multi-color, black, brown, red and input jacks certain. "Blue" doesn't exist, "Purple" is not certain, and "Gray" could be any of the three remaining. I have one darker gray, I have a gray (vaguely pinkish), and another gray (slightly purplish) wire. I think I need to retrace these across board & schematic and make sure they're correct.
There is a sticker that says "Keith" inside, I think maybe he redid the harness wiring with whatever was around. That or the insulation has changed colors over time? I have amps from the 40s with the same wire colors they had then. Certainly a puzzle.
Currently the amp is doing a new strange thing, which is exploding the 10-ohm bias resistor on one rail only. It did this after I noticed the input jack ground was lifted and grounded it.
Thankfully outputs were not installed. I think I have a wiring harness problem. I have orange, yellow, multi-color, black, brown, red and input jacks certain. "Blue" doesn't exist, "Purple" is not certain, and "Gray" could be any of the three remaining. I have one darker gray, I have a gray (vaguely pinkish), and another gray (slightly purplish) wire. I think I need to retrace these across board & schematic and make sure they're correct.
There is a sticker that says "Keith" inside, I think maybe he redid the harness wiring with whatever was around. That or the insulation has changed colors over time? I have amps from the 40s with the same wire colors they had then. Certainly a puzzle.
Alright, the bias resistor thing is gone, I think that it had drifted. I replaced it with a 1/8w "fuse" and it holds. Offset comes on as 19v, rises to 25v, drops to 19v, then creeps up toward 100. Settles around 88v now. I think my feedback must be the problem, but I'm not sure. The hard shorts etc. seem to come on a bit faster. At least all the driver transistors are holding! Sure am learning a lot about discrete SS equipment, and fire 
.
.
Without output signal that's what happens. But you just might want to be able to use the amplifier to deliver power to a speaker.
You'll get about 200V across it when approaching positive clipping.Even so, if you are getting positive output offset there has to be more problems than this but it would be possible for it to cause negative output offset.
10 ohm resistors exploding sounds like a broken predriver, driver or VBE multiplier - or a wiring error.
Last edited:
Railed on the negative, at approx -80 to -100 depending on the phase of moon, apparently. I am thinking there is a wiring error, I am creating a diagram of what is supposed to be on each end of the wire from the PCB art & schematic. I'll post it when I'm done, so anyone with replaced wires can figure out what is going on! This is the 14A pcb. No transistors test short, all read appropriate .5-.7 drops from gate. I should point out that one of the drivers (the TO-3 drivers) is getting warm, and the other is cold. I think perhaps the wires to the "cold" driver have been swapped, perhaps.
I will dig around for something to replace the 2sd669, since signal will push it over spec. I used 669 since I could heatsink it easier than some of the "tall" TO-92 parts that otherwise suited the application.
I will dig around for something to replace the 2sd669, since signal will push it over spec. I used 669 since I could heatsink it easier than some of the "tall" TO-92 parts that otherwise suited the application.
Checked the voltages again, and damned if it wasn't railed on positive this time. This must mean I'm close, right ? 2SD669A is holding with 98 on source, -102.1 on emitter, -101.6 on gate. Is .4 reasonable for Vge? Vbe multiplier voltages are: 99.8 source, 0 gate, 98v emitter. Seems wrong. Any ideas?
? 2SD669A is holding with 98 on source, -102.1 on emitter, -101.6 on gate. Is .4 reasonable for Vge? Vbe multiplier voltages are: 99.8 source, 0 gate, 98v emitter. Seems wrong. Any ideas?
Got nothing technical to add as that is well beyond me, just bear in mind that Phase Linear amps are justifiably famous for lots of power, decent sound quality and for breaking down A LOT! Some have been known to spontaneously burst into flames, hence their nickname Flame Linear. I know of one guy who until recently used a 700b in his studio but he always had two more as spares.
Some transistors made around that time actually went 'cbe' instead of 'ebc'. And the Jap's (and TO-126's) are 'ecb'. Make sure that isn't the problem before going too much further.
I've made those mistakes too many times to count.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.