Hi All,
I have an S500 that has worked perfectly for last 10 years. I replaced all output resistors and power supply about 10 years ago and now suddenly the right side has gone quiet. The 6A fuse on right side blows immediately on powerup. I had a tech look at the amp and he says he found two resistors that need replacing.
M8352
M8326
there is also M8213
1.) Where can I find suitable replacments
2.) If two are blown, perhaps all of them need replacing?
Anyone have experience sourcing and replacing these? Attached a pic.
Much obliged.
~ Lee
Bay Area Member
I have an S500 that has worked perfectly for last 10 years. I replaced all output resistors and power supply about 10 years ago and now suddenly the right side has gone quiet. The 6A fuse on right side blows immediately on powerup. I had a tech look at the amp and he says he found two resistors that need replacing.
M8352
M8326
there is also M8213
1.) Where can I find suitable replacments
2.) If two are blown, perhaps all of them need replacing?
Anyone have experience sourcing and replacing these? Attached a pic.
Much obliged.
~ Lee
Bay Area Member
Attachments
The original transistors are long obsolete, use On Semi MJ15022G/15024G or MJ15023G/15025G. Either will work, depends on what is available at the time. TO3 transistors are getting harder to source. Check Mouser or Digikey for stocks. Have used both types on Stasis rebuilds and are recommended by Jon Soderberg who is the go to guy on the old Threshold gear - also recommended by Nelson Pass.
Definitely replace all with new types.
Definitely replace all with new types.
Last edited:
I saw that too. Guess I’ll buy the 50 unit pak. Are all the transistors the same part and spec? Also, are they actually soldered to the board? I see Phillips head screws and assume they’re also soldered once I undo the screws? Thanks a ton btw!
Make a drawing of the output stage, you will see the existing original devices are marked with an N or P.
N signifies an NPN type and the P signifies a PNP type. Your OS heatsink has a total of 20 off TO3 devices - 10 will be NPN and the other 10 will be PNP. It is most important that where you have an N type - it is replaced with a new NPN part number and vice versa for the P types. On one side of the heatsink you will have one N and nine P's and the other row will have one P and nine N's.
The 2 screws on each transistor is the collector connection to the pcb secured with a nut on the pcb. Each screw should have a plastic sleeve where the screw passes through the heatsink to electrically insulate the screw from the heatsink body. Make sure you keep these to reuse. Under the TO3 transistor will be a mica insulating washer, you need these as well to reuse after cleaning them up or use new ones with new heatsink goop.
The base and emitter pins of the output devices are soldered directly to the pcb. You will need to desolder this connection to enable the old transistors to be removed. Once removed the pcb should be cleaned of all flux residue with Isopropyl Alcohol or similar.
To fit new transistors once everything is cleaned up, fit the mica insulating washer with new goop to the transistor base, slide each screw through the holes with the insulating bush in place and feed all through the heatsink and through the pcb on the rear. Most important is to check the resistance between the screw head (transistor case) and the heatsink after you tighten the nuts ( but not too tight) to read open circuit on resistance scale on a DMM. If this is OK, then you can solder the 2 transistor pins, and move on to the next TO3. Double check your work and absolutely make sure you are putting either an NPN or PNP type in the correct position as shown on the drawing you do and also checked with the pic you show in post #1.
If you have the means, the gain of the output transistors should be matched to within 20% max. That is all N's matched and all P's matched - you do not match N to P. That is you need nine matched N's and nine matched P's these are driven by a single transistor of the opposite polarity - which is the top transistor in each row in your photo.
While you have the OS apart on the bench, also measure the resistance of the emitter resistors on the pcb to make sure they are all in spec. I assume you have a schematic of the OS - if not search on the forum as I am sure it has been posted many times in the past.
Hope that all helps. Slow and steady wins the race, and double check everything you do.
You mention you had a tech look at it and he found 2 faulty resistors - I assume you mean transistors! If you don't feel confident in doing the repair perhaps you should supply the parts to the tech and get him to do it, just a thought.
Definitely replace all transistors on this OS - I hope the other channel OS is fine. Either type I mention will be fine and are much more rugged and higher spec than the original types. Once all of this is done, you will need to set the OS bias current to spec and adjust the DC offset output voltage to near zero - this is where you might also need your tech.
N signifies an NPN type and the P signifies a PNP type. Your OS heatsink has a total of 20 off TO3 devices - 10 will be NPN and the other 10 will be PNP. It is most important that where you have an N type - it is replaced with a new NPN part number and vice versa for the P types. On one side of the heatsink you will have one N and nine P's and the other row will have one P and nine N's.
The 2 screws on each transistor is the collector connection to the pcb secured with a nut on the pcb. Each screw should have a plastic sleeve where the screw passes through the heatsink to electrically insulate the screw from the heatsink body. Make sure you keep these to reuse. Under the TO3 transistor will be a mica insulating washer, you need these as well to reuse after cleaning them up or use new ones with new heatsink goop.
The base and emitter pins of the output devices are soldered directly to the pcb. You will need to desolder this connection to enable the old transistors to be removed. Once removed the pcb should be cleaned of all flux residue with Isopropyl Alcohol or similar.
To fit new transistors once everything is cleaned up, fit the mica insulating washer with new goop to the transistor base, slide each screw through the holes with the insulating bush in place and feed all through the heatsink and through the pcb on the rear. Most important is to check the resistance between the screw head (transistor case) and the heatsink after you tighten the nuts ( but not too tight) to read open circuit on resistance scale on a DMM. If this is OK, then you can solder the 2 transistor pins, and move on to the next TO3. Double check your work and absolutely make sure you are putting either an NPN or PNP type in the correct position as shown on the drawing you do and also checked with the pic you show in post #1.
If you have the means, the gain of the output transistors should be matched to within 20% max. That is all N's matched and all P's matched - you do not match N to P. That is you need nine matched N's and nine matched P's these are driven by a single transistor of the opposite polarity - which is the top transistor in each row in your photo.
While you have the OS apart on the bench, also measure the resistance of the emitter resistors on the pcb to make sure they are all in spec. I assume you have a schematic of the OS - if not search on the forum as I am sure it has been posted many times in the past.
Hope that all helps. Slow and steady wins the race, and double check everything you do.
You mention you had a tech look at it and he found 2 faulty resistors - I assume you mean transistors! If you don't feel confident in doing the repair perhaps you should supply the parts to the tech and get him to do it, just a thought.
Definitely replace all transistors on this OS - I hope the other channel OS is fine. Either type I mention will be fine and are much more rugged and higher spec than the original types. Once all of this is done, you will need to set the OS bias current to spec and adjust the DC offset output voltage to near zero - this is where you might also need your tech.
Last edited:
See this thread about idle current adjustment and zero. https://www.diyaudio.com/community/threads/threshold-s-500-power-amplifier.135737/ Pictures of the idle bias and zero pot. Also the schematic diagram.
Sometimes 40 year old mica washers fracture and short. I would replace them. Newer silicon rubber pads can have lower thermal resistance, and do not need thermal grease. If dirt on the heat sink or a stuck fan did not cause the burnout.
Generally buying from distributor in less than 50 pack, you need to buy a couple more output transistors of each type to not use the ones with unusual gain or Vbe. I made a jig with 1000 ohm base & 100 ohm collector resistors, then measured Vce for each transistor, writing the voltage on it with a sharpie. I used a heat sink. Measure each at about the same time. I had 2 odd voltages out of 12, very lucky since I needed 10.
Don't forget to check the emitter resistors (often burn out) and the driver transistors and predriver transistors. Also cross conduct resistors. Also VI limiter resistors, transistors(q19 q21), and capacitors. I had a open VI limitor capacitor.
This apparently has a bias pot that goes full value if the bias control wiper loses contact. This can cause O.T. burnout. After adjusting idle bias to 20-30 ma (on schematic), I would measure voltage across adjustment pot and clamp it with a diode stack so the voltage does not go too high if that happens. You can use silicon diodes 0.7 v and schottky diodes 0.3 v each to approximate the full voltage drop. Not much current there so 1 amp diodes like 1n4148 are okay.
PS use safety glasses desoldering, solder can splash in your eye. I use solderwick or a rubber squeeze bulb (parts-express.com). When powering up, put a 1200 watt room heater core series the AC line to limit damage if there is a short somewhere. This is too big to use a tungsten light bulb IMHO. if AC current is not too high (takes a second DVM) I used the heater core out of a heater with a bad tip over switch. It conveniently had .250 flag terminals, which can be bought in insulated male version at the auto supply.
Measure output for ultrasonic oscillation, which can happen. Requires scope, or AC voltage scale of an analog VOM (cheaper) Block negative probe of analog vom with .47 uf 200 v cap, it shows AC voltage on DC voltage without cap. If output is silent and no voltage on 2 vac scale, you are okay. I was not okay, I had .5 vac at some mhz after an op amp upgrade of a mixer. I know frequency because the voltage also passed through a 390 pf blocking cap, which audio will not. Use cheap speakers on power up until oscillation test and dc voltage out test are done. The simpson 266XLPM analog meter I own (bought 1985) has a selenium rectifier bridge in AC scale, which can measure oscillation down to millivolts. Other meters may have silicon diodes, which will miss anything under 600 mv. Fluke RMS DVM misses anything over 7 khz. Used scopes have been chock full of expired e-caps, a waste of time IMHO.
Sometimes 40 year old mica washers fracture and short. I would replace them. Newer silicon rubber pads can have lower thermal resistance, and do not need thermal grease. If dirt on the heat sink or a stuck fan did not cause the burnout.
Generally buying from distributor in less than 50 pack, you need to buy a couple more output transistors of each type to not use the ones with unusual gain or Vbe. I made a jig with 1000 ohm base & 100 ohm collector resistors, then measured Vce for each transistor, writing the voltage on it with a sharpie. I used a heat sink. Measure each at about the same time. I had 2 odd voltages out of 12, very lucky since I needed 10.
Don't forget to check the emitter resistors (often burn out) and the driver transistors and predriver transistors. Also cross conduct resistors. Also VI limiter resistors, transistors(q19 q21), and capacitors. I had a open VI limitor capacitor.
This apparently has a bias pot that goes full value if the bias control wiper loses contact. This can cause O.T. burnout. After adjusting idle bias to 20-30 ma (on schematic), I would measure voltage across adjustment pot and clamp it with a diode stack so the voltage does not go too high if that happens. You can use silicon diodes 0.7 v and schottky diodes 0.3 v each to approximate the full voltage drop. Not much current there so 1 amp diodes like 1n4148 are okay.
PS use safety glasses desoldering, solder can splash in your eye. I use solderwick or a rubber squeeze bulb (parts-express.com). When powering up, put a 1200 watt room heater core series the AC line to limit damage if there is a short somewhere. This is too big to use a tungsten light bulb IMHO. if AC current is not too high (takes a second DVM) I used the heater core out of a heater with a bad tip over switch. It conveniently had .250 flag terminals, which can be bought in insulated male version at the auto supply.
Measure output for ultrasonic oscillation, which can happen. Requires scope, or AC voltage scale of an analog VOM (cheaper) Block negative probe of analog vom with .47 uf 200 v cap, it shows AC voltage on DC voltage without cap. If output is silent and no voltage on 2 vac scale, you are okay. I was not okay, I had .5 vac at some mhz after an op amp upgrade of a mixer. I know frequency because the voltage also passed through a 390 pf blocking cap, which audio will not. Use cheap speakers on power up until oscillation test and dc voltage out test are done. The simpson 266XLPM analog meter I own (bought 1985) has a selenium rectifier bridge in AC scale, which can measure oscillation down to millivolts. Other meters may have silicon diodes, which will miss anything under 600 mv. Fluke RMS DVM misses anything over 7 khz. Used scopes have been chock full of expired e-caps, a waste of time IMHO.
Last edited:
Thank you gentleman. Really appreciate the help.
Yes Gary, I mispoke, meant transistor. But you are right. While I could do this myself, I'll likely have my technician do the work. But perhaps you guys can suggest someone better in the Bay Area? Jon Soderberg moved to Utah unfo. He was my goto.
It sounds like you are suggesting I should NOT replace transistors on the left channel? Wouldn't it be smart to just do all of them, both sides?
You mentioned I need nine matched N's and nine matched P's. Does Mouser sell them somewhat closely matched or is this something I can set? Not sure what you meant there.
Thanks IndianaJo for the link and extra deets on Biasing, etc.
Yes Gary, I mispoke, meant transistor. But you are right. While I could do this myself, I'll likely have my technician do the work. But perhaps you guys can suggest someone better in the Bay Area? Jon Soderberg moved to Utah unfo. He was my goto.
It sounds like you are suggesting I should NOT replace transistors on the left channel? Wouldn't it be smart to just do all of them, both sides?
You mentioned I need nine matched N's and nine matched P's. Does Mouser sell them somewhat closely matched or is this something I can set? Not sure what you meant there.
Thanks IndianaJo for the link and extra deets on Biasing, etc.
Transistors do not age typically. Bias pot ages and can cause failures if wiper loses contact, see post # 9 3rd paragraph. Mica washers age see post #9. Dirt on heat sink can cause failures, as can blocked fan.
Matching process discussed in second paragraph of post # 9. No mouser won't do it. Peavey will up to quantity 5 (PV-1.3k), but this is not a Peavey. You need quantity 9. 50 packs allegedly are all from same production run and might be pretty well matched, but I have not proven this theory by purchasing that many.
Note schematic in the link does not have output rf blocking coil. These is pretty universal feature of later products. Series output inductor, 14 turns around a parallel 10 ohm 5 watt resistor right next to output hot jack, can stop AM radio from coming in the speaker wires and driving the feedback transistor to amplify the unwanted input.
Last edited:
Yes, you need to gain match the transistors yourself (or your tech) If you are buying 50 of each type (NPN and PNP) my previous findings would suggest you will have no problems getting matched sets (within 20% gain is Jon's recommendation), and more than likely enough to do both channels.
If it was me and considering the amps age (over 40 years) I would be doing both channels.
I am in Australia, so outside of Jon Soderberg, I have no idea of who else you could approach. Maybe it is worth talking to Jon and see how much it would cost if you supply the transistors.
Your normal tech should be able to handle it though.
Jon Soderberg's email is: vintageamp@surewest.net I have found him to be most helpful in the past.
If it was me and considering the amps age (over 40 years) I would be doing both channels.
I am in Australia, so outside of Jon Soderberg, I have no idea of who else you could approach. Maybe it is worth talking to Jon and see how much it would cost if you supply the transistors.
Your normal tech should be able to handle it though.
Jon Soderberg's email is: vintageamp@surewest.net I have found him to be most helpful in the past.
15022/3/4/5 are probably closer in characteristic to the old 2N types. They follow the typical epitaxial-base hFE vs. Ic curves. The 2119x is a sustained-beta type. Not high fT, but flat gain out to 8 ish amps. About the same gain as the old ones at 15A, though. Below +/-80V, the old ones are more rugged. Above, the new ones are. All the new ones (especially MJL3281 and it’s derivatives) benefit from better uniformity.
Usually though, massively parallel output stages do well enough even with unmatched devices, especially in EF3.
Usually though, massively parallel output stages do well enough even with unmatched devices, especially in EF3.
Hello again gentleman,
Thank you again for the help and advice. The amp is back from a local tech who worked on it, installed all the new On Semi MJ15022G/15024G transistors as described, as I did not have the time. I supplied him with full instructions from your helpful posts and here we are, with a functioning S500. The amp actually sounds pretty good. I'm wondering if I hear more bloom and detail, amost like a more detailed painting before me than before.
That said, I am noticing the singer is off axis now, just a few degrees to the left, and the right channel heatsink definitely seems cooler to the touch. I need to re-bias the amp tomorrow, but am doubting that will solve the imaging issue. Does biasing the S500 have a sonic difference and how would you describe it?
Thank you again for the help and advice. The amp is back from a local tech who worked on it, installed all the new On Semi MJ15022G/15024G transistors as described, as I did not have the time. I supplied him with full instructions from your helpful posts and here we are, with a functioning S500. The amp actually sounds pretty good. I'm wondering if I hear more bloom and detail, amost like a more detailed painting before me than before.
That said, I am noticing the singer is off axis now, just a few degrees to the left, and the right channel heatsink definitely seems cooler to the touch. I need to re-bias the amp tomorrow, but am doubting that will solve the imaging issue. Does biasing the S500 have a sonic difference and how would you describe it?
Biasing properly just keeps crossover distortion in check. If overbiased, you just get excess heat, if underbiased it gets “grainy”. There is likely a relatively wide range where it will sound “decent”, but instruments could pick up differences. There will be a minimum for distortion.
Image being off center is due to a slight difference in gain between channels. That is set by feedback, and the feedback resistors need to be matched between channels. The only thing that’s that could happen are a feedback resistor getting cooked and having its value drift high, or the feedback’s DC blocking cap developing high ESR (often it’s value goes way down too, producing a noticeable loss of bass). Usually just regular 1% resistors are good enough, as the tracking of the volume control on the preamp won’t be that good. But a 5 or 10% mismatch would be noticeable.
Image being off center is due to a slight difference in gain between channels. That is set by feedback, and the feedback resistors need to be matched between channels. The only thing that’s that could happen are a feedback resistor getting cooked and having its value drift high, or the feedback’s DC blocking cap developing high ESR (often it’s value goes way down too, producing a noticeable loss of bass). Usually just regular 1% resistors are good enough, as the tracking of the volume control on the preamp won’t be that good. But a 5 or 10% mismatch would be noticeable.