Maker Pro
Maker Pro

What specs are the most important when swapping transistors with another type?

LearningSomethingToday

Mar 4, 2018
23
Joined
Mar 4, 2018
Messages
23
I am have an amp that has a complimentary pair of Toshiba 2SD845 and 2SB755 which have given up the ghost.
It is laughable at how many counterfeit option that are out there. I think even smaller reputable sources may have some fraud inventory :)

So I have been looking for a more modern alternative to replace all of the output transistors on this amp.
I want something that will not require modifications to the heat sink. (MT200)
The Sanken 2SC2921 seems like a good fit. It has a little more power handling capacity.
However I know there must be some trade-off or caveat that I am missing.

What are the specs that make the most impact on finding an adequate replacement?

I don't expect anybody to do the leg-work for me so I put this data together. I think the Sanken devices seem like a good fit. What do you think?
Is this a viable alternative? Any better suggestions? Am I better off waiting until I can find some very old used Toshibas (even then they could be counterfeit)?
I do like the idea of working with newer devices that I can get easier through more reputable distributors.

Thanks for any advice!

s-sheet.jpg
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I would check that the Vbe is also similar (although I expect it would be).

Assuming the amplifier has some sort of adjustment for setting the biasing, you should be able to use a different, albeit similar pair of transistors.

Whatever you choose for one, I would check what the manufacturer recommends as the complement and double check that the specs are similar.

Beware of different lead configurations, and try one channel first (presumably the dead one).

Also take a look at the safe operating area graphs (SOA graph) and make sure the replacement is the same or greater (especially in the top right secondary breakdown region).
 

LearningSomethingToday

Mar 4, 2018
23
Joined
Mar 4, 2018
Messages
23
Assuming the amplifier has some sort of adjustment for setting the biasing, you should be able to use a different, albeit similar pair of transistors.

Thank you for your input. The amp does have adjustable bias.
Sanken publishes the 2SA1215 as the complimentary device.
Both are available through Digikey or B&D Enterprises (a little cheaper at B&D)
I have checked the pinouts of the devices and they are the same.
Pin1.jpg

With my limited knowledge of these devices, the Toshiba (left) and Sanken (right) seem to have similar curves.
The Tosh seems to break left with collector current lower, @ 0.2 amps (DC)
The same occurs with 10ms pulse breaking left at a lower @ 0.4 amps
I am not sure what those small changes would mean to me.

Pin2.jpg Also would I have to worry about any oscillations using a different transistor?

Thanks
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I can't think of any reason why they may be more likely to oscillate. They appear to be quite similar.

I also agree that the SOA graphs are substantially similar.

If certainly give than a go.
 

dave9

Mar 5, 2017
1,188
Joined
Mar 5, 2017
Messages
1,188
What amp? Consumer or (boutique or really old) Is this amp entirely discreet or with a precision feedback loop? If the latter, you can stuff just about *anything* in so long as you get NPN/PNP right.

Overgeneralizing, fewer, higher power handling transistors have reduced treble definition, soundstage. At the same time if the design was too bright they can correct that... depends on how it was developed, what it does or doesn't need for balanced sound.
 

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
The higher ft (transition frequency which defines the point at which the transistor current gain becomes unity) may result in a wider open loop bandwidth. It depends on the frequency compensation applied to the circuit to keep it stable in closed loop which may be marginal or inadequate.
When you replace the transistors you will need to monitor the current demand of the unit on test from its power supply (you will need to do this anyway to set the quiescent current of the output stage). If it is very high, then the amp is probably oscillating. If you can prove it is not then (scope), you have some other issue. These test should be carried out with the input shorted.
If the O/P transistors have failed, then it would be good practice to replace all the other transistors on the board. The reason for this is that if the circuit is all DC coupled (most likely) then there is a very good chance that the other transistors have been damaged or at best, stressed, which would most likely cause them to fail at some later date and you find yourself back in the same situation.
 

LearningSomethingToday

Mar 4, 2018
23
Joined
Mar 4, 2018
Messages
23
What amp? Consumer or (boutique or really old) Is this amp entirely discreet or with a precision feedback loop?

Just as full disclosure, I have limited knowledge on most of this. I am learning a little about this daily and I know enough to be only dangerous. Small baby steps is what I am taking.

These transistors are for an SAE A202. The 202 does not have known schematics but has some very similar design attributes that is shares with the A502. The 502 does have known schematics, which I have. The A502 has a sort of auto-pilot bias system. The A202 has a manually adjusted bias circuit with temp compensation. Oddly enough I think the 502 may use the same Tosh 2SB755 /SD845 as drivers to ultimately drive 2SA1494/3538 Sanken output devices. I could probably use those but I think the 2SA1494's may be overkill in the 202. The 2SA2921 seems closest to the original Toshibas with slightly higher power handling capability. Here is a screenshot of the 502:

saecap.jpg
 

LearningSomethingToday

Mar 4, 2018
23
Joined
Mar 4, 2018
Messages
23
It depends on the frequency compensation applied to the circuit to keep it stable in closed loop which may be marginal or inadequate.

This is when I know I may be out of my league. When you started talking about freq compensation, my train just falls off the tracks. I am assuming the upper limits of the of fT would be constrained by some type of RC circuit, correct?

I may need to call back on you for some further understanding of how to determine if oscillating is occurring. I have read articles that speak of the dangers of oscillation but I have never seen it in person. It is just one of those things I tucked away in my memory hoping to learn more about later. I understand the part about shorting the RCA input. Is oscillation something I can see with a scope at the base of the output devices or at the speaker outputs? You mentioned "power supply". Does oscillation show up on the rails? Mind you, I am always leary about where I decide to poke my oscilloscope. I don't have any specialty probes (ie differential, isolated etc) so I am always worried about harming the equipment or myself.
 

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
To see if the amp is oscillating, the best place to monitor is at the amplifier output terminal with you speaker connected.
If there are fundamental differences between the two designs as you seem to indicate then, the circuit provided is of limited use.
The compensation in modern amps is usually done using a dominant pole. The circuit supplied looks overly complicated. I tend to err on the side of simplicity using the smallest number of transistors that I can get away with.
 

LearningSomethingToday

Mar 4, 2018
23
Joined
Mar 4, 2018
Messages
23
The circuit supplied looks overly complicated. I tend to err on the side of simplicity using the smallest number of transistors that I can get away with.

The input stage up to the driver transistors are almost identical in the 202 and 502, just laid out on the boards differently. The 502 uses larger Sanken devices because it was a much more powerful amp.

SAE was not exactly known for being tech friendly. They were known for being a little complex. I am of the opinion that the design, for its time in the mid 80's, was forward thinking. I guess they offered a lot of engineering for a small amount of money. I can't say it was the smartest engineering but it was great sounding gear for the price. Take a look at the schematics of a Mark Levinson 33(x) of the same era and be prepared to fall out of your chair with the level of complexity. The 80's was a great time to make overly complicated stuff :)

I have ordered the set of Sanken transistors. I will post some results once they come in and I get them mounted.

Thanks
 

Sunnysky

Jul 15, 2016
541
Joined
Jul 15, 2016
Messages
541
The devices you chose look like good replacements. I believe they include a small emitter resistor to share currents in multiple emitters inside and also give more consistent current sharing with parallel devices. Normally this is also shared externally by adding emitter R's for each device to perform the same impedance equalization so one device does not hog all the current. Normally hFE linear and ESR near saturation have a tolerance of 50% which is best measured near saturation or Vce=2~4V at max current. They say "Wide ASO because of on-chip ballast resistance." In theory with these devices the power lost in added emitter R's could be reduced because of this to gain less voltage drop, lower output impedance or raise damping factor say from 300 to 1000 in high power amps which results in less bass distortion from back EMF.

The last schematic is excellent for power sharing and linearity because of this to that device parameters are less critical. It has excellent DC tracking and the complexity is needed for linear power sharing with impressive THD specs.

The package size is important for thermal resistance and the max power and current, linearity and other parameters
 
Last edited:
Top