I have several vintage AKAI amplifiers including two AM-2450's. Nice looks and good sound.
These are really old now and the power transistors are starting to degrade quite badly, some oscillations can be seen on the scope. I've already replaced the drivers with fresh transistors and also most caps.
The amp uses a pair of Toshiba 2SB688/2SD718 on the output of each channel.
I noticed it sounds much more dull compared to the other one i own and since i replaced all the other stuff the only thing left to replace are the output transistors.
Normally i would just buy the same transistors, but for this occasion i want a little upgrade, i want to put in a pair of KEC 2SB817/2SD1047. These are a bit higher specced than the 2SB688/2SD718 and have a slightly faster transition frequency.
However the 2SB817/2SD1047 also has a bit more gain and needs much less quiescent current (about half the amount). So i need to modify the bias circuit of the AM-2450, which looks like this:
Do i just increase R10? If so, what value do i need to halve the bias current?
These are really old now and the power transistors are starting to degrade quite badly, some oscillations can be seen on the scope. I've already replaced the drivers with fresh transistors and also most caps.
The amp uses a pair of Toshiba 2SB688/2SD718 on the output of each channel.
I noticed it sounds much more dull compared to the other one i own and since i replaced all the other stuff the only thing left to replace are the output transistors.
Normally i would just buy the same transistors, but for this occasion i want a little upgrade, i want to put in a pair of KEC 2SB817/2SD1047. These are a bit higher specced than the 2SB688/2SD718 and have a slightly faster transition frequency.
However the 2SB817/2SD1047 also has a bit more gain and needs much less quiescent current (about half the amount). So i need to modify the bias circuit of the AM-2450, which looks like this:
An externally hosted image should be here but it was not working when we last tested it.
Do i just increase R10? If so, what value do i need to halve the bias current?
You need to decrease R10 or increase R12
The optimum bias current is really determined by the amplifier topology and component values and would be around the 55 to 60ma region for this design.
The optimum bias current is really determined by the amplifier topology and component values and would be around the 55 to 60ma region for this design.
I always thought increasing resistance of the trimpot lowers the quiescent current. Can you explain this a little more?
Lowered R10 to 1,8Kohms and put in the new transistors, didn't even have to touch the bias adjustment, it was just right. Only had to correct the left channel a little to mirror the right channel.
Overall the sound is now much clearer again, no more oscillations. Guess that fixed it.
Overall the sound is now much clearer again, no more oscillations. Guess that fixed it.
Power transistors degrade? I have read about various issues developing in smaller transistors over time, but not outputs. I am merely asking because I did not realize that this was something to be concerned about. Are particular output transistor models more likely to develop issues?
It depends how the preset is wired. The harder the transistor conducts the lower the bias voltage available (and so the lower the current). Increasing the trim pot resistance would turn the transistor off and give high current (maybe destructively high)
That's an interesting VAS design. Looks like something of a mix between complementary differential and inductive degeneration for TR2. I wonder how it performs?
Well a transistor is something of a switch and it switches rapidly (MHz) to move energy from one lead to the other, this switching occurs even without a load (thanks to the quiescent current).
It isn't hard to imagine such a switch will degrade over time. Quality of the silicon, how it was used and temperature determine the lifetime. The amplifier i got was running every day playing radio, so no surprise there it was worn.
Oscillations and a dull sound are symptoms of such degradation.
A bit more than that. TR15 develops a stable yet temperature compensated voltage across it. It will be attached to the main heatsink to track the temperature of the output devices and that change in temperature reflects in a change in voltage across TR15. The overall result if the designer has done their job is that the bias should be relatively independent of temperature of the heatsink.
Agreed. While all transistors will eventually degrade, it usually takes either prolonged high temperatures (badly-cooled linear regulator pass transistor) and/or fabrication issues for them to do so at an accelerated pace. The output transistors would not be my first suspects in that regard - they are well-cooled, after all. Their drivers, by contrast, are not in this amplifier - while their longer case can dissipate more than a regular TO-92, it's still not that much to write home about, and presumably they're far from the heatsink. Just recently we had someone here who had to swap out such a PNP driver to fix his amplifier of similar vintage (PNPs usually are less robust).
It is quite possible that the new outputs are actually making up for deficiencies in the drivers. Even measuring hFE of those could be insightful. I'm sure someone here could recommend a good 2SC2235/2SA965 replacement in TO-126. With a little black Al heatsink they should last for quite a while longer than the originals.
It is quite possible that the new outputs are actually making up for deficiencies in the drivers. Even measuring hFE of those could be insightful. I'm sure someone here could recommend a good 2SC2235/2SA965 replacement in TO-126. With a little black Al heatsink they should last for quite a while longer than the originals.
It is probably the simple case of a new (better) OP transistors produce an overall (if marginally so) better sound. Transistors -specially silicone ones- do not degrade over time unless they are used beyond their SOA limits. I have a similar experience tough, one of my old time favourites, a 35 yrs old Aurex power amp (SC-M15) originally had the same output transistors, and replacing them with newer, triple diffused epitaxials (higher Pd, higher Ic, double Ft) the sound became definitely "faster", better defined.
I wasn't aware that this would be such a big secret, but here are two links:
www.sfsu.edu/necrc/.../thesis_report_Mandeep.p...
Transistor Aging - IEEE Spectrum
When you think about it's really not that difficult to imagine, everything is temporary in this world, especially true for anything that comes out of a factory.
Let's review here. We are talking about an amp with 37 years of daily use, the tuner was put on top of the amp, and that combination was sitting in a snug cabinet with little room to vent out the heat. Conditions were not exactly ideal...
When i got the amp it had lots of issues, one channel would not even produce any sound.
Last edited:
I have 35 year old silicon amplifiers that sound as good now as they did way back. I really don't believe output transistors degrade.
They aren't mechanical switches with contacts so how can they degrade.
In my experience a transistor is good or blown.
Now as for electrolytic capacitors that is a totally different kettle of fish as they do degrade over time. Some have lifetimes as low as 2000 hours.
Oh, it does happen - no way around the 2nd law. It is not totally unusual to see pass transistors in hot-running linear regulators that have lost most of their current gain in the last 3 decades or so. Same can happen to RF transistors (Sony's ICF-6800W is quite notorious for this, 2SC930 RF transistors affected there - possibly moisture getting in or some other fabrication problem). And we all know the issues with 2SC458 and other old NEC transistors.
Hi Nigel,
Now that sounds a lot more rational.
Hi jooch,
In truth, transistors can degrade over time. It depends on how they were made and encapsulated if they aren't hermetically sealed. What I have often seen is increased leakage, decreased beta, increased junction voltage and with smaller signal types, increased noise. Barring serious problems, the sound quality is determined mostly by the circuit design, operating points and passive components. One type of which is called a capacitor.
Capacitors most certainly do have a "best before date", and they need to be kept cool. Sounds like cheese a bit. Capacitors can lose electrolyte over time, or be damaged by peak currents that can burn away foil / film. The very fact your amplifier was oscillating tells me that it was unstable and capacitors are the most likely culprit. Amplifier designers do everything they can to keep their designs from oscillating, and you still have that problem. Check your main filter caps and voltage regulator capacitors too!
As for beefing up the drivers, that is a good suggestion. 2SD669A and 2SB649A were the parts I used often, or you can use 2SC3421 and 2SA1358. At +/- 45 VDC (90 VDC total rating required), the 2SC3421 pair is rated for 120 VDC and will do nicely. You can touch the drivers after running them for a while to see how hot they are and whether you need small radiators for them. Another worthwhile change would be to get some metal film resistors to replace R7 (47K) and R3 (3K6). Increase to 1 watt types and use the closest 1% values, do for both channels. That might reduce some background hiss by a little bit. Increase C5 to 220 uF or better and go for a higher voltage. 16 VDC is the lowest voltage you should use, and the characteristics get better with increasing voltage. If you can get a 35 V part in neatly, do so. That should reduce a lot of distortion from that feedback cap.
I have seen many Q1, 2SA979 go out of balance and get noisy. They can be replaced with a matched pair of 2SA970 or 2SA872A. This affects your DC offset and directly affects distortion. Your designed in DC offset should be on the order of +20 mV. Higher Beta will reduce that figure a little. You could always match R1 and R7 (47K 1% value), then R2 (4K7) would represent the DC offset you can expect. There are other things that create a DC offset too further along in the circuit, but this represents a non-simulated, off the cuff number to check if all is well. If that offset is negative, or past 100 mV, there is something wrong. The best way to match, or check transistor match is to put them into a "long tailed pair" or "differential" circuit with transistors in contact with each other and foam over top. If you use matched collector loads, the minimum voltage between collectors will represent the best match.
If you like sound, this is one way to detail your amplifier.
-Chris
Now that sounds a lot more rational.
Hi jooch,
In truth, transistors can degrade over time. It depends on how they were made and encapsulated if they aren't hermetically sealed. What I have often seen is increased leakage, decreased beta, increased junction voltage and with smaller signal types, increased noise. Barring serious problems, the sound quality is determined mostly by the circuit design, operating points and passive components. One type of which is called a capacitor.
Capacitors most certainly do have a "best before date", and they need to be kept cool. Sounds like cheese a bit. Capacitors can lose electrolyte over time, or be damaged by peak currents that can burn away foil / film. The very fact your amplifier was oscillating tells me that it was unstable and capacitors are the most likely culprit. Amplifier designers do everything they can to keep their designs from oscillating, and you still have that problem. Check your main filter caps and voltage regulator capacitors too!
As for beefing up the drivers, that is a good suggestion. 2SD669A and 2SB649A were the parts I used often, or you can use 2SC3421 and 2SA1358. At +/- 45 VDC (90 VDC total rating required), the 2SC3421 pair is rated for 120 VDC and will do nicely. You can touch the drivers after running them for a while to see how hot they are and whether you need small radiators for them. Another worthwhile change would be to get some metal film resistors to replace R7 (47K) and R3 (3K6). Increase to 1 watt types and use the closest 1% values, do for both channels. That might reduce some background hiss by a little bit. Increase C5 to 220 uF or better and go for a higher voltage. 16 VDC is the lowest voltage you should use, and the characteristics get better with increasing voltage. If you can get a 35 V part in neatly, do so. That should reduce a lot of distortion from that feedback cap.
I have seen many Q1, 2SA979 go out of balance and get noisy. They can be replaced with a matched pair of 2SA970 or 2SA872A. This affects your DC offset and directly affects distortion. Your designed in DC offset should be on the order of +20 mV. Higher Beta will reduce that figure a little. You could always match R1 and R7 (47K 1% value), then R2 (4K7) would represent the DC offset you can expect. There are other things that create a DC offset too further along in the circuit, but this represents a non-simulated, off the cuff number to check if all is well. If that offset is negative, or past 100 mV, there is something wrong. The best way to match, or check transistor match is to put them into a "long tailed pair" or "differential" circuit with transistors in contact with each other and foam over top. If you use matched collector loads, the minimum voltage between collectors will represent the best match.
If you like sound, this is one way to detail your amplifier.
-Chris
anatech,
At what operating temperature would you suggest adding a small heatsink to a TO-220?
Would this apply equally to drivers vs those used in regulators?
thanks...
Hi roger2,
If really high heat is coming off them, don't touch them. 🙂 Turning off might be a good idea.
Seriously, you should be able to hold your finger on the case for at least 10 sec. I would probably be tempted to install radiators even around 50 °C. This same thing applies to regulators. Ventilation doesn't normally improve when you put the covers back on.
Failure rate is exponential with temperature, so it is very important to keep things as reasonably cool as you can.
-Chris
If really high heat is coming off them, don't touch them. 🙂 Turning off might be a good idea.
Seriously, you should be able to hold your finger on the case for at least 10 sec. I would probably be tempted to install radiators even around 50 °C. This same thing applies to regulators. Ventilation doesn't normally improve when you put the covers back on.
Failure rate is exponential with temperature, so it is very important to keep things as reasonably cool as you can.
-Chris
OK thank you anatech 🙂
I have experimented with couple types of heatsinks on transistors. My experience is that even smaller ones do a surprisingly decent job. I have gotten around 10°C reduction on TO-220s. But I wouldn't want to risk damaging transistors by de-soldering and re-soldering unnecessarily.
When I do a restore I do a very thorough job. I would hope for the final result to be an honest increase in longevity of the unit.
It seems to me that TO-220s often run hot. But I do not know if that was the explicit intent of the original designer, or if it was simply not considered enough of a problem to deal with in terms of added cost. But to be fair, I mostly work on vintage stuff. And the fact that it is still here says something positive...
I have experimented with couple types of heatsinks on transistors. My experience is that even smaller ones do a surprisingly decent job. I have gotten around 10°C reduction on TO-220s. But I wouldn't want to risk damaging transistors by de-soldering and re-soldering unnecessarily.
When I do a restore I do a very thorough job. I would hope for the final result to be an honest increase in longevity of the unit.
It seems to me that TO-220s often run hot. But I do not know if that was the explicit intent of the original designer, or if it was simply not considered enough of a problem to deal with in terms of added cost. But to be fair, I mostly work on vintage stuff. And the fact that it is still here says something positive...
Last edited:
- Status
- Not open for further replies.