im repalcing transistors a1491 with 2sa1294, c3855 with 2sc3263, on a nakamichi amplifier one because the left channel has blown and took out two resistors, seems that im slightly uprating the two 15amp transistors do i upgrade the smaller ones, numbered c3421, a1358, if so what to?
and replacing all the capasitors i can with rubycon za's will this make an improvement to sound?
any help would be appreaciated thanks.
and replacing all the capasitors i can with rubycon za's will this make an improvement to sound?
any help would be appreaciated thanks.
For heaven's sake! Don't use Rubycon caps! They are by FAR the WORST electrolytics made in Japan in terms of reliability, and not very good sound performance, either. Use Nichicon!
As for the output transistors, I would either use 2SA1492/2SC3856 or 2SA1302/2SC3281. I don't think you will get good reliability with the numbers you mention. The 2SC3421/2SA1358, in this particular circuit, can be replaced with 2SD669/2SB649, but you must add insulators behind them if heatsink mounted.
As for the output transistors, I would either use 2SA1492/2SC3856 or 2SA1302/2SC3281. I don't think you will get good reliability with the numbers you mention. The 2SC3421/2SA1358, in this particular circuit, can be replaced with 2SD669/2SB649, but you must add insulators behind them if heatsink mounted.
I'll back up Pinkmouse. Rubycon actually make Blackgate (not that I advocate boutique parts).
It's also bad advice to recommend use of 2SA1302/2SC3281 as these are the worlds most faked transistors, and went obsolete years ago with the advent of their replacement 2SA1943/2SC5200.
You shouldn't need to replace/upgrade the driver transistors as the new output devices probably have higher current gain (hFE). However, if the output devices have blown it's always prudent to replace the drivers as they usually get damaged (but go undetected by simple tests) and may fail later on, taking out everything you just repaired.
It's also bad advice to recommend use of 2SA1302/2SC3281 as these are the worlds most faked transistors, and went obsolete years ago with the advent of their replacement 2SA1943/2SC5200.
You shouldn't need to replace/upgrade the driver transistors as the new output devices probably have higher current gain (hFE). However, if the output devices have blown it's always prudent to replace the drivers as they usually get damaged (but go undetected by simple tests) and may fail later on, taking out everything you just repaired.
Al,
i'm just a second rate geenie, rubbing ís no good, i broke the bottle after i emptied it.
The A1294/C3263 have almost twice the Ft as their obsolete Sanken brothers, 35MHz versus 20MHz.
Stability might be an issue, maybe that's what the term reliability referred to.
The recommended 2SA1492/2SC3856 epitaxials differ in a lower Ft number.
In power terms, A1294/C3263 are not just a lot stronger on the specs. The're the same ones as the big Sankens but in an MT100 case, their datasheet numbers are conservative.
Regular replacement for the broken Sanken numbers would be the 2SA1265/2SC3182, made by Toshiba.
(the 2SA1491/2SC3855 are an oddball couple btw, in Sanken terms they haven't been manufactured that long.)
I don't get the Rubycon/Nichicon advice either.
Rubycon has always been an elite Japanese manufacturer, long before the BG thing started.
My previous encounters with Nichicons put them on my last resort list.
i'm just a second rate geenie, rubbing ís no good, i broke the bottle after i emptied it.
The A1294/C3263 have almost twice the Ft as their obsolete Sanken brothers, 35MHz versus 20MHz.
Stability might be an issue, maybe that's what the term reliability referred to.
The recommended 2SA1492/2SC3856 epitaxials differ in a lower Ft number.
In power terms, A1294/C3263 are not just a lot stronger on the specs. The're the same ones as the big Sankens but in an MT100 case, their datasheet numbers are conservative.
Regular replacement for the broken Sanken numbers would be the 2SA1265/2SC3182, made by Toshiba.
(the 2SA1491/2SC3855 are an oddball couple btw, in Sanken terms they haven't been manufactured that long.)
I don't get the Rubycon/Nichicon advice either.
Rubycon has always been an elite Japanese manufacturer, long before the BG thing started.
My previous encounters with Nichicons put them on my last resort list.
Nexus said:Hello Guys,
just keep in mind that 2sc5200/2sa1943 have a 5MHz higher ft than the 2sc3281/2sa1302 pair.
If 2sc3281/2sa1302 pair is used, and upgraded to 2sc5200/2sa1943 than check the stability of the amp.
Best Regards:
Nexus
If the amp is compensated for a lower ft transistor, why would substitution of a higher ft transistor cause instability? It should have a higher phase margin.
Who said the statement is correct ?
Car talk,..again:
Consider the tires as the output devices.
The size of the car determines the engine horsepower, see the vehicle as the loudspeakers.
The engine torque determines the size of the tires.
The steering response should always be faster than the car, otherwise the car will run off the road.
Going through a corner, the steering will have to turn into the corner more than the direction the car is going to, difference between steering direction and car direction is the phase shift, the phase margin is what is left before the tires start to slide. The nominal contact surface of the tire remains the same, but the effective surface at higher steering versus car direction angles is reduced.
The higher the speed of the car, the greater the difference between steering and driving direction.
At the tire/car critical speed point (angle speed) the gas should be reduced or the brakes hit or the tires loose grip and the car crashes. Maximum allowable phase shift is reached, hitting the brakes reduces the speed=> freq. limiting
Replace the tires with faster ones, and the angle between steering and the driving direction of the car is reduced at the same speed because the raised grip of the tires makes the car respond faster.
=> sustained hFe value at higher frequencies.
Bigger tires allow for a bigger car, but in general bigger tires mean less cornering grip.
Putting racing tires on an Edsel doesn't make it a thoroughbred, but it generally doesn't make the car drive worse. That does not mean the tires will not have problems with the camber and so on of the wheel suspension geometry/type.
Car talk,..again:
Consider the tires as the output devices.
The size of the car determines the engine horsepower, see the vehicle as the loudspeakers.
The engine torque determines the size of the tires.
The steering response should always be faster than the car, otherwise the car will run off the road.
Going through a corner, the steering will have to turn into the corner more than the direction the car is going to, difference between steering direction and car direction is the phase shift, the phase margin is what is left before the tires start to slide. The nominal contact surface of the tire remains the same, but the effective surface at higher steering versus car direction angles is reduced.
The higher the speed of the car, the greater the difference between steering and driving direction.
At the tire/car critical speed point (angle speed) the gas should be reduced or the brakes hit or the tires loose grip and the car crashes. Maximum allowable phase shift is reached, hitting the brakes reduces the speed=> freq. limiting
Replace the tires with faster ones, and the angle between steering and the driving direction of the car is reduced at the same speed because the raised grip of the tires makes the car respond faster.
=> sustained hFe value at higher frequencies.
Bigger tires allow for a bigger car, but in general bigger tires mean less cornering grip.
Putting racing tires on an Edsel doesn't make it a thoroughbred, but it generally doesn't make the car drive worse. That does not mean the tires will not have problems with the camber and so on of the wheel suspension geometry/type.
Hi Jacco & Pooge,
does this mean we are agreed? That substituting a faster output device (within sensible limits) but retaining all the other physical attributes of a circuit should not increase the risk of instability (i.e. reduced phase margin).
To all others.
Why do so many commentators state that changing to a faster device holds the risk of instability?
does this mean we are agreed? That substituting a faster output device (within sensible limits) but retaining all the other physical attributes of a circuit should not increase the risk of instability (i.e. reduced phase margin).
To all others.
Why do so many commentators state that changing to a faster device holds the risk of instability?
In the Hitachi days the main theme was how to keep them from swinging, because they were so fast and susceptible to even a minor change of wiring. Funny stuff in the PS, and the beasts oscillated. Vibration problems are my analogy for those lateral things.
Lack of fast enough output devices create design issues, leading to various compensation measures, which then are only a compromise.(imo)
The availability of faster ones makes it easier on the amp circuitry, easier to design a good amplifier.(audio designers words, not mine) In turn, the same problems arise as in the J50/k135 days. Such as direct wired output devices: good enough for Dino 15003s but not for 60MHz Sanken devices, and stuff like the mentioned parasitic capacitances, board layout, powersupply issues, grounding, yes or no ground planes, wiring, etc. Someone who has been playing with TO3 Hitachi's in the past finds spaghetti amps pretty painfull to watch.
If all those amplifiers in need of an output stage swap were designed as high speed switching amps, with 4-layer teflon boards and the works, stepping up to higher Ft devices wouldn't be a big deal i suppose.
Recall the ever continuing debate on where exactly to place snubbers/ bypass capacitors, and which value is allowed ? Many older amplifiers, and maybe (some) recent, nowadays are seen as having a lot of flaws.
On the ksa i used some of the most fast devices around instead of the original slow ones, and had no problem getting the amps stable. Just lucky maybe, it does show that the raised finger is somewhat uncalled for.
Read Jozua's post of the Aalt-Jouk interconnect switch effect ? If it were all that easy it would swiftly be an extremely boring merry go round.
Lack of fast enough output devices create design issues, leading to various compensation measures, which then are only a compromise.(imo)
The availability of faster ones makes it easier on the amp circuitry, easier to design a good amplifier.(audio designers words, not mine) In turn, the same problems arise as in the J50/k135 days. Such as direct wired output devices: good enough for Dino 15003s but not for 60MHz Sanken devices, and stuff like the mentioned parasitic capacitances, board layout, powersupply issues, grounding, yes or no ground planes, wiring, etc. Someone who has been playing with TO3 Hitachi's in the past finds spaghetti amps pretty painfull to watch.
If all those amplifiers in need of an output stage swap were designed as high speed switching amps, with 4-layer teflon boards and the works, stepping up to higher Ft devices wouldn't be a big deal i suppose.
Recall the ever continuing debate on where exactly to place snubbers/ bypass capacitors, and which value is allowed ? Many older amplifiers, and maybe (some) recent, nowadays are seen as having a lot of flaws.
On the ksa i used some of the most fast devices around instead of the original slow ones, and had no problem getting the amps stable. Just lucky maybe, it does show that the raised finger is somewhat uncalled for.
Read Jozua's post of the Aalt-Jouk interconnect switch effect ? If it were all that easy it would swiftly be an extremely boring merry go round.
Just so people know ... I ordered C3855 and A1491 off of ebay, (chinese) thinking to myself they could be fakes because the chinese do that stuff a lot .. But I hadn't had experience with fakes yet. The amplifier came up and worked perfectly until I actually demanded significant power from it. At that point, the usual drama ensued, and now I get to look for a lot more parts. Conclusion: those transistors are probably no longer available, and anything on ebay is probably a fake .. though sadly it may be possible for also genuine parts to be on ebay.
Slow Ft transistors have gone the way of the dodo bird. The processes are abandoned, the factories knocked down and turned into shopping malls. Those factories made $11 transistors (1972 price of a 2n3055) and people won't pay more than $4.50 for "better" parts now. Fast new transistors instead of slow ones may meet spec, since Ft had a lower limit or typical, no upper limit. For example 2n3055 is still in production. But old designs act up using them.
I get genuine 2-4 mhz transistors from On semi via newark (farnell) or mouser. I install them in old obsolete amps made in the USA. Adjustments have to be made. Dynaco installed 50 pf capacitors b-c of the drivers when they went from homotaxial 40xxx (slow) to TIP3055 (faster) . (see greg dunn TIP mod) I've tried this, plus 10 ohm 1 w resistors between driver and ouput tranistor base. I got away with it even with 4" flying leads between the driver board and the output transistors.
Maybe you can try MJL4281/4302 which are ON numbers, which you can buy from authorized distributors. Also, Fairchild 2sa1943/sc5200 are now an ON division and can also be bought (sometimes) from newark.
I check for oscillation with a VOM, a capacitor on the ground lead, on 20 VAC and 2 vac scale. I'm not getting any with shorted inputs. What the circuit does to a square wave is undefined; all the scopes I've bought ******ed out in a few weeks. **** expired e-caps on glued in PCB's.
There are a couple of e-bay sellers that participate on here and are likely genuine. But they are in California and New Jersey, and Newark is closer in SC, so I didn't log the names.
I get genuine 2-4 mhz transistors from On semi via newark (farnell) or mouser. I install them in old obsolete amps made in the USA. Adjustments have to be made. Dynaco installed 50 pf capacitors b-c of the drivers when they went from homotaxial 40xxx (slow) to TIP3055 (faster) . (see greg dunn TIP mod) I've tried this, plus 10 ohm 1 w resistors between driver and ouput tranistor base. I got away with it even with 4" flying leads between the driver board and the output transistors.
Maybe you can try MJL4281/4302 which are ON numbers, which you can buy from authorized distributors. Also, Fairchild 2sa1943/sc5200 are now an ON division and can also be bought (sometimes) from newark.
I check for oscillation with a VOM, a capacitor on the ground lead, on 20 VAC and 2 vac scale. I'm not getting any with shorted inputs. What the circuit does to a square wave is undefined; all the scopes I've bought ******ed out in a few weeks. **** expired e-caps on glued in PCB's.
There are a couple of e-bay sellers that participate on here and are likely genuine. But they are in California and New Jersey, and Newark is closer in SC, so I didn't log the names.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.