not to speak for Mooly - I think what his is meaning is you can remove the outputs from both rails, test them out of circuit and confirm at least one is still good. Put that one good PNP and one good NPN, replace the drivers and pre-drivers and you could test functionality to ensure you have repaired the amplifier. The amp should/will work with only one pair of outputs fitted, at very low volume so you can check for functionality.
Once amp is confirmed working, you then go and find all your output replacements, which due to supply could be expensive. I think he is giving you a way to confirm amp operation before buying a lot of expensive parts only to find out you have more things to repair.
Once amp is confirmed working, you then go and find all your output replacements, which due to supply could be expensive. I think he is giving you a way to confirm amp operation before buying a lot of expensive parts only to find out you have more things to repair.
Adding to what Mooly said ...
It will function with one pair of good outputs and drivers, just for functional test purposes only. Use modern replacement transistors, current manufacture from authorized distributors only. Newer parts match more closely out of a group, you want to match the beta of parallel transistors.
When one transistor fails in a parallel group, it is only the one that died catastrophically first. The rest were stressed beyond their ratings. Their characteristics normally shift and they may well fail later on. That goes for the NPN and PNP banks. If some work, use them in small projects below their ratings if you want.
It will function with one pair of good outputs and drivers, just for functional test purposes only. Use modern replacement transistors, current manufacture from authorized distributors only. Newer parts match more closely out of a group, you want to match the beta of parallel transistors.
When one transistor fails in a parallel group, it is only the one that died catastrophically first. The rest were stressed beyond their ratings. Their characteristics normally shift and they may well fail later on. That goes for the NPN and PNP banks. If some work, use them in small projects below their ratings if you want.
If I was in your shoes I'd replace all ten transistors on the right side of the bias spreader.Considering the advice to replace all in the stage if one has broken I would need to replace all the output/power transistors, both drivers Q625/623 and Q619/621?
Should I also replace Q613, 615 as they might have been stressed and possibly fail later?
The output devices are likely dead and those that aren't dead have been overstressed. This could have taken out (or overstressed) the drivers as well, so they get replaced too. The last two, the pre-drivers, are likely $0.50 small-signal devices. Just replace them. You've lost more money (in terms of time) following this thread than the transistors cost.
The biggest challenge will likely be to find replacements. Maybe you get lucky, but I bet that'll take some digging. And please don't buy the replacement parts on ePay/AliBlahBlah/whatnot. Go to a reputable distributor, even if it is a distributor of obsolete parts such as Rochester.
I like the idea of building up the amp with just one of the output pairs. That'll save some money should the amp decide to kill the new parts.
Tom
Just a warning that LittleDiode is a questionable supplier. I won't buy anything from them.
Understood on the output stage, thanks both.
On this:
I have a 2 pairs of the same output transistors from another Rotel amp that I am inclined to use. My other option is swapping them all to Toshiba 2SA1962/2SC5242 from a Onkyo surround amp but then one channel will be different than the other.
On this:
Is there a simple way to determine if the surviving transistors have been "too" stressed? Or what characteristics normally change when this happens?When one transistor fails in a parallel group, it is only the one that died catastrophically first. The rest were stressed beyond their ratings. Their characteristics normally shift and they may well fail later on. That goes for the NPN and PNP banks. If some work, use them in small projects below their ratings if you want.
I have a 2 pairs of the same output transistors from another Rotel amp that I am inclined to use. My other option is swapping them all to Toshiba 2SA1962/2SC5242 from a Onkyo surround amp but then one channel will be different than the other.
Sometimes, but often not until they act funny or fail. It isn't worth the wishful thinking.
Sometimes the beta changes, and you'd need a curve tracer that goe to their max breakdown to find out. At times they can misbehave once the temperature is raised. Too many variables and it only takes one to kill everything. A suckers bet.
Buy new parts in the same package with similar or better specs. I would use a modern On Semi part. Don't salvage parts, you need to match the sets of NPN and PNP by beta. Your ONLY intelligent option is to buy new, good, current parts. If you restrict yourself to "no cost to you" options, well. No comment.
Sometimes the beta changes, and you'd need a curve tracer that goe to their max breakdown to find out. At times they can misbehave once the temperature is raised. Too many variables and it only takes one to kill everything. A suckers bet.
Buy new parts in the same package with similar or better specs. I would use a modern On Semi part. Don't salvage parts, you need to match the sets of NPN and PNP by beta. Your ONLY intelligent option is to buy new, good, current parts. If you restrict yourself to "no cost to you" options, well. No comment.
👍 that's it exactly. The amp would be fully functional with just one pair fitted and still good for many watts output.not to speak for Mooly - I think what his is meaning is you can remove the outputs from both rails, test them out of circuit and confirm at least one is still good. Put that one good PNP and one good NPN, replace the drivers and pre-drivers and you could test functionality to ensure you have repaired the amplifier.
No easy way I'm afraid. In any case the device may be impaired in a way that would not even show on a normal check against the data sheet. Think of a 12 volt bulb, it may withstand 20 volts for quite a while. If you ran it at 20 for while and then used it at 12 volt would it have the same life... you just don't know...Is there a simple way to determine if the surviving transistors have been "too" stressed? Or what characteristics normally change when this happens?
Not unless you have a rig to verify all the parameters measured in the data sheet. And even then there's no way to know for sure.Is there a simple way to determine if the surviving transistors have been "too" stressed?
It's akin to ESD stress. Maybe the part survived. Maybe it now has high leakage current. The best case is that the part dies on overstress. At least then you know it's not working so you can replace it instead of having it fail later.
I cringe when I see semiconductors offered up for sale in plastic tubs. Those tubs will easily generate 10s of kV of ESD strikes in a dry environment. Anyway. I digress.
Tom
👍 that's it exactly. The amp would be fully functional with just one pair fitted and still good for many watts output.
No easy way I'm afraid. In any case the device may be impaired in a way that would not even show on a normal check against the data sheet. Think of a 12 volt bulb, it may withstand 20 volts for quite a while. If you ran it at 20 for while and then used it at 12 volt would it have the same life... you just don't know...
Running off one pair (at an impedance above 8 ohms - 20 is a good test case) is a good use for the outliers that you will get when you buy a tube of output transistors for matching. Or run at full blast with no load on a dim bulb, to ensure that nasties like cross conduction aren’t happening.
“Overstressed” parts will usually LOOK perfectly normal, unless you can do parametric tests. Super leaky parts or very low breakdown are possible, but you almost never see it. If had gotten that badly damaged it simply would have blown during the original failure. It might get leaky and die in 10 minutes of operation, but not when you put it on a “transistor tester”. Chances are it will “test just fine”. Many parameters have wide “acceptable” ranges - as long as they stay constant. If Vbe shifted by 20 mV or lost 20% of beta, how would you know what it was to start with? Even if you don’t get parameter shifts, excessive temperatures can cause multiple failure modes - some of them mechanical. And pretty much all of it cumulative.
At least these days the outliers are within the limits we used to accept as good. I used to buy 25 to 50 outputs of the same number at a time so I could get matches. Those were used in single pair amplifiers as an upgrade to what was in there.
It is so much easier to simply avoid the risk. It isn't worth the cost of checking and then having some failures.
It is so much easier to simply avoid the risk. It isn't worth the cost of checking and then having some failures.
Ok, its clear. Don't use the old ones for anything but testing 🙂
On finding new ones I have narrowed it down to Onsemi NJW0281G (NPN)NJW0302G (PNP) that I can get from Mouser. The 2nd option at Mouser is the Onsemi 2SA1962/2SC5242 that have a very similar spec but a few years older a tad more expensive.
Any general thoughts or comments on these?
https://www.mouser.se/datasheet/2/308/NJW0281_D-1813438.pdf
I
On finding new ones I have narrowed it down to Onsemi NJW0281G (NPN)NJW0302G (PNP) that I can get from Mouser. The 2nd option at Mouser is the Onsemi 2SA1962/2SC5242 that have a very similar spec but a few years older a tad more expensive.
Any general thoughts or comments on these?
https://www.mouser.se/datasheet/2/308/NJW0281_D-1813438.pdf
I
There is nothing to suggest they would not be suitable.
(Issues that arise from substituting can be changes in stability margins and in extreme cases differing DC characteristics can put things like bias setting adjustment out of range (that one is easily fixed usually by no more than a resistor tweak). All that is more likely an issue when you get to substituting the drivers and pre drivers and tbh I would not expect any issues with this one but until you try it can't be 100% guaranteed)
(Issues that arise from substituting can be changes in stability margins and in extreme cases differing DC characteristics can put things like bias setting adjustment out of range (that one is easily fixed usually by no more than a resistor tweak). All that is more likely an issue when you get to substituting the drivers and pre drivers and tbh I would not expect any issues with this one but until you try it can't be 100% guaranteed)
Wouldn’t surprise me a bit if A1962/C5242 go EOL soon, if they aren’t already. No real need for them to make both types when 03281/0302 have clearly cornered the market. The differences between all those types are small enough for stability not to need adjustment. You’ll find larger differences substituting drivers/predrivers, as quite often a direct sub isn’t possible or even a good idea. C613 and C615 are huge here and probably tolerate substitution well there too. This has “hit it with a big hammer” compensation. You’d have to do something stupid for square wave ringing to show up.
Just to report back, the output have been changed to the NJW0281G/NJW0302G and out of the 10 each I ordered I got hfe matched to between 89 and 90 for all 6. The rest were within 10% or each-other.
And a question, their hfe were however significantly higher than the original ones, 30->90 which I understand is not an issue and as long as its not "too" high it's rather a good thing as the driver stage don't need to supply as much current.
I struggle to source the driver and pre driver transistors and considering the NJW0281G/NJW0302G don't require as much current I am thinking I should be able to look at options with less current/power capabilities.
Is this generally a "good practise" approach?
And a question, their hfe were however significantly higher than the original ones, 30->90 which I understand is not an issue and as long as its not "too" high it's rather a good thing as the driver stage don't need to supply as much current.
I struggle to source the driver and pre driver transistors and considering the NJW0281G/NJW0302G don't require as much current I am thinking I should be able to look at options with less current/power capabilities.
Is this generally a "good practise" approach?
Hi A 8,
That is one reason I like the NJW0281G/NJW0302G so much. The flat beta curve on modern transistors is very helpful too.
Don't count on the higher beta. Peak current draws may be pretty similar. You've got running temperature to consider and that may depend more on standing current. Equivalent drivers will probably do fine as long as they aren't too much slower. These days our parts are pretty good. Don't sweat the Ft spec so much as long as you are still in the ballpark. As always, no matter what, check stability.
That is one reason I like the NJW0281G/NJW0302G so much. The flat beta curve on modern transistors is very helpful too.
Don't count on the higher beta. Peak current draws may be pretty similar. You've got running temperature to consider and that may depend more on standing current. Equivalent drivers will probably do fine as long as they aren't too much slower. These days our parts are pretty good. Don't sweat the Ft spec so much as long as you are still in the ballpark. As always, no matter what, check stability.
Please note that after repair the other channel also needs the same new transistors. To avoid amateurish faults with serious consequences it is recommended to:
a. measure extremely careful.
b. move test clips and/or measurement stuff only when the device is powered off.
Due to the need to replace semis in both channels one could in case of misfortune choose not to repair but to replace the complete device. Prices of parts are only going up, repair might be not economically viable.
After successful repair and replacement of all fuses please consider replacing C601 for a 4.7 µF 5 mm film cap, C607 for a 100 µF 35V MUSE ES and C609 for a 5 mm 10 µF 50V cap.
a. measure extremely careful.
b. move test clips and/or measurement stuff only when the device is powered off.
Due to the need to replace semis in both channels one could in case of misfortune choose not to repair but to replace the complete device. Prices of parts are only going up, repair might be not economically viable.
After successful repair and replacement of all fuses please consider replacing C601 for a 4.7 µF 5 mm film cap, C607 for a 100 µF 35V MUSE ES and C609 for a 5 mm 10 µF 50V cap.
Last edited:
Jean-Paul,
Why would you say that? Am I missing something here?
One channel has nothing to do with the other, they won't sound different even if the measured performance is a bit different. The average person won't hear it. That's assuming the other channel is working as designed to spec.
I would never subject a client to unnecessary work and parts. If there was a substantial improvement in performance I would give them the option.
Why would you say that? Am I missing something here?
One channel has nothing to do with the other, they won't sound different even if the measured performance is a bit different. The average person won't hear it. That's assuming the other channel is working as designed to spec.
I would never subject a client to unnecessary work and parts. If there was a substantial improvement in performance I would give them the option.
In a lot of amplifiers they will sound different (2SJ35/2SK135 were famous for it). I was trained/educated also to replace power transistors in both channels for technical reasons and aging. It is not unnecessary. Never did otherwise and since parts were cheap then it was no problem to do so either. One does hear a difference certainly when completely different types are used but it is up to the owner to decide.
Besides that one maybe just does not appreciate 2 channels existing of completely different transistor types. At least use the same brand and type as found in the working channel for the defective channel. Otherwise use identical types in both channels.
Say the same so not to use different caps or resistors in both channels and everybody will nod 🙂
Besides that one maybe just does not appreciate 2 channels existing of completely different transistor types. At least use the same brand and type as found in the working channel for the defective channel. Otherwise use identical types in both channels.
Say the same so not to use different caps or resistors in both channels and everybody will nod 🙂
Last edited:
I can see that. Most Mosfet amplifiers don't have a proper driver stage, so changing the outputs with different parts could change performance. If you replace with the same part, then no.
This isn't true of Bipolar amplifiers unless you're going from an early output type to modern, then you have stability compensation to change. My training at Marantz and others had us check the other channel carefully, including bias stability. But one channel failing was never a good reason to replace the other.
Now if both channels were badly overheated - yeah. Change them. I bought a Luxman M-02 with badly overheated outputs on both sides. They were on the same block of metal if you are familiar with Freon cooled amps (bad idea). I replaced them all without hesitation. The grease was so badly cooked it had become solid, as in use mechanical removal that included sanding blocks. The board was discoloured to a very dark brown.
This isn't true of Bipolar amplifiers unless you're going from an early output type to modern, then you have stability compensation to change. My training at Marantz and others had us check the other channel carefully, including bias stability. But one channel failing was never a good reason to replace the other.
Now if both channels were badly overheated - yeah. Change them. I bought a Luxman M-02 with badly overheated outputs on both sides. They were on the same block of metal if you are familiar with Freon cooled amps (bad idea). I replaced them all without hesitation. The grease was so badly cooked it had become solid, as in use mechanical removal that included sanding blocks. The board was discoloured to a very dark brown.
- Home
- Amplifiers
- Solid State
- Rotel RB980 accident, how much do I need to replace?