I posted on this Hitachi HA-4700 prev. on AAC:
https://forum.allaboutcircuits.com/...linear-ckt-design-please-help-explain.210716/
Serv. manual is here:
https://archive.org/details/manual_HA4700_HITACHI/page/n1/mode/2up

To cut to the chase, the LEFT CHANNEL seems to be acting (biasing) fine -- slowly ramping up to 7.7mV per serv manual, and holding there when warmed up-- despite , me having substituted in power transistors [Sanken 2SB1570/2SD2401 Darlingtons] that are NOT ideal for purpose. (That's all I have in my parts bin. Original replacements parts are difficult and $$ to acquire ).

Orig. Fujitsu 2SA1075/2SC1075 (both are BJT, non-Darlington )
My replacements for both L/R channels (currently installed): Sanken 2SB1570/2SD2401

I am not sure why the LEFT CHANNEL works (with decent sine and square waves), topping power at almost 80W!!

The RIGHT CHANNEL is not okay. Can't achieve bias above 0.0mV. Not sure why.
One reason for RIGHT CHANNEL issues might be the non-ideal choice of output trans. devices.

Note that the Sanken Darlingtons have DC current gain (hFE) much higher than the 2SA1075/2SC1075. I seem to have been able to control that with the orig. circuit, using traditional bias trim pots.
But I'm not sure why L channel is working.


============

ChatGPT reply to comparison query, compare "2SB1570 and 2SA1075":

The 2SB1570 and 2SA1075 are both PNP power transistors, rated for high current, but they have a crucial difference: the 2SB1570 is a Darlington transistor, while the 2SA1075 is a conventional single power transistor. That difference affects drive current, voltage drop, switching behavior, and whether they can be substituted for each other.



1. Specifications compared

Parameter	2SB1570	2SA1075
Transistor type	PNP Darlington	PNP single transistor
Collector-emitter voltage	−150 V	−120 V
Collector current (max.)	−12 A	−12 A
Power dissipation	150 W	120 W
DC current gain (hFE)	5,000–30,000	40 min. at 7 A; 60–200 at lower test current
Saturation voltage	−2.5 V max.	−1.8 V max.
Transition frequency	50 MHz typical	About 45–60 MHz, depending on manufacturer
Package	MT-200	MT-200
Typical application	Audio output stages, regulators	Audio power amplifiers, high-speed switching

Sources: Sanken 2SB1570 specifications and 2SA1075 datasheets. Specifications may vary by manufacturer and test conditions.


2. The key electrical differences


2SB1570 — high gain, low drive current

• Extremely high gain means the driver circuit needs very little base current.
• Its Darlington structure produces a higher base-emitter voltage drop, typically around two junction drops.
• Can be useful in power stages designed specifically for Darlington output transistors.

The Sanken version is listed as obsolete.




2SA1075 — conventional transistor, faster drive response

• Requires more base current than the Darlington.
• Has a lower base-emitter voltage drop, typically around one junction drop.
• Designed for high-speed power switching and amplifier applications.

The Fujitsu version is described as a high-speed ring-emitter power transistor.
fujitsu semiconductor


3. Can you replace one with the other?
Generally, no—not as a direct replacement.
Even though both are PNP devices rated at 12 A, their gain and base-drive requirements are substantially different. Replacing a 2SB1570 with a 2SA1075 could leave the transistor under-driven; replacing a 2SA1075 with a 2SB1570 could change bias, voltage headroom, and switching behavior. The pinout and safe operating area must also be checked.
For an audio amplifier, the difference in base-emitter voltage is particularly important because it can affect bias and crossover distortion.
4. Which one is better?

• For a circuit designed around a Darlington: 2SB1570.
• For a circuit requiring a conventional, fast power transistor: 2SA1075.
• For a repair or substitution: Choose based on the original circuit design, not just the maximum current and power ratings.

What are you using the transistor for? An audio amplifier, power supply, or another circuit? If you tell me the equipment model or the original transistor you are replacing, I can assess whether either part is a suitable substitute.
The 2SB1570 and 2SA1075 are both PNP power transistors, rated for high current, but they have a crucial difference: the 2SB1570 is a Darlington transistor, while the 2SA1075 is a conventional single power transistor. That difference affects drive current, voltage drop, switching behavior, and whether they can be substituted for each other.
Sanken Electric



1. Specifications compared

Parameter	2SB1570	2SA1075
Transistor type	PNP Darlington	PNP single transistor
Collector-emitter voltage	−150 V	−120 V
Collector current (max.)	−12 A	−12 A
Power dissipation	150 W	120 W
DC current gain (hFE)	5,000–30,000	40 min. at 7 A; 60–200 at lower test current
Saturation voltage	−2.5 V max.	−1.8 V max.
Transition frequency	50 MHz typical	About 45–60 MHz, depending on manufacturer
Package	MT-200	MT-200
Typical application	Audio output stages, regulators	Audio power amplifiers, high-speed switching

Sources: Sanken 2SB1570 specifications and 2SA1075 datasheets. Specifications may vary by manufacturer and test conditions.
Sanken Electric
+2

2. The key electrical differences



2SB1570 — high gain, low drive current

• Extremely high gain means the driver circuit needs very little base current.
• Its Darlington structure produces a higher base-emitter voltage drop, typically around two junction drops.
• Can be useful in power stages designed specifically for Darlington output transistors.

The Sanken version is listed as obsolete.
Sanken Electric
+1



2SA1075 — conventional transistor, faster drive response

• Requires more base current than the Darlington.
• Has a lower base-emitter voltage drop, typically around one junction drop.
• Designed for high-speed power switching and amplifier applications.

The Fujitsu version is described as a high-speed ring-emitter power transistor.
fujitsu semiconductor
+1

3. Can you replace one with the other?

Generally, no—not as a direct replacement.
Even though both are PNP devices rated at 12 A, their gain and base-drive requirements are substantially different. Replacing a 2SB1570 with a 2SA1075 could leave the transistor under-driven; replacing a 2SA1075 with a 2SB1570 could change bias, voltage headroom, and switching behavior. The pinout and safe operating area must also be checked.
For an audio amplifier, the difference in base-emitter voltage is particularly important because it can affect bias and crossover distortion.
4. Which one is better?

• For a circuit designed around a Darlington: 2SB1570.
• For a circuit requiring a conventional, fast power transistor: 2SA1075.
• For a repair or substitution: Choose based on the original circuit design, not just the maximum current and power ratings.

What are you using the transistor for? An audio amplifier, power supply, or another circuit? If you tell me the equipment model or the original transistor you are replacing, I can assess whether either part is a suitable substitute.

 

I'll post just the power-amplifier portion of the schematic below.
The focus is Q713/Q714 (a pair per channel).
So, the original 2sc2525 and 2sa1075 are not Darlington devices.
If you think Darlington's will work w/o any modification to ckt topology (other than bias current trimport adj), let me know. Or maybe drop-in Darlingtons will work, but not be stable various power levels (such as volume at low settings, etc).

 

It's hard to tell without thinking and measuring at the same time, and flowing with that.
But I'm pretty sure that your right amplifier has a problem that isn't about the transistor you're changing.
Because if one perfectly equal amplifier works with one transistor and the other doesn't do the same, you have a problem. It's not that it doesn't match very accurately, it doesn't work at all. That should be eye-opening.

So you have a good one, measure different points and compare them. Then this should give you an idea where the problem lies.

Also, if you can pass the schematic through a cleaner to remove unwanted noise, and increase the contrast, would be much easier to read.

 

using Darlington in place of normal transistor changes bias point so answer is no.
Google AI offered some alternatives to replace original complementary pair with modern versions including inexpensive one though package is not the same

 

it seems that ChatGPT is not understanding situation. Google AI gets it:

 

using Darlington in place of normal transistor changes bias point. Google AI offered some alternatives to replace original complementary pair with modern versions:
View attachment 372414

Yeah, I got pretty much the same results, two years ago, when I first got the amp and was using Chat GPT to look for substitutes. But none of those avail at Digikey or Mouser. They are all "OBSOLETE". ChatGPT now suggests:

1. Best starting point for an audio amplifier

• Replace 2SC2525 → MJE15032G (NPN)
• Replace 2SA1075 → MJE15033G (PNP)

 

I didn't see that it was a darlington in the place of a normal one. I still hold that it's weird that one works and the other no.

 

why? we don't know the details of tests...

on a large amp powering low Ohm loads (speakers), 7mV or so is practically nothing...

and since replacements are not a matched complementary pair it is normal that characteristics are off.
picking one pair of odd balls that somehow lands near target may as well be a fluke. but that is a result that is likely hard to replicate, which is just what he is experiencing.

testing amp with sine wave and square wave and looking for result on a scope is very subjective. neither waveforms are good at exposing non-linearities unless distortions are VERY obvious. one can do a better evaluation by using dual channel scope and carefully overlaying input and output signals. even better test would be triangle. that would make any nonlinearity much easier to spot - even if only using single channel scope.

 

I didn't understand before that TS wasn't using the same output pair for the left and right channel. If/When I have time I'll look again, carefully, about it. But as others mentioned, the 2 Vbe vs. 1 Vbe is what won't work here and in any other project. It must be readjusted to fit it in. So better to find another pair of transistors.

 

It's hard to not change the quiescent point and have a 2x bias as before. Instead of altering 1/4 of this circuit, why don't you buy another transistor that can handle such thing? Because of power? Put 2 in parallel, so 2 pairs of 2. Because of gain? I don't think this is the problem.
Posing the problem as to "how to fit a darlington in the place of a single one" is trying to fit 2 flies inside 1.
I'm sure that there are plenty of transistors for this power and gain.

 

[/QUOTE]

My replacements for both L/R channels (currently installed): Sanken 2SB1570/2SD2401

Well, I messed up big time with this comment. I experimented with those two years ago but forgot I removed them in favor of:









No wonder the Left channel works. But note that 2sc2922 and 2sa1494 are not Sanken datasheet compliments to each other.
The pair should be 2sc1922/2sa1216 or 2sc3858/2sa1494. 2sc1922/2sa1216 is very close to the original Fujitsu 2SA1075/2SC1075, but I've only got one 2sa1216 in my parts bin. I'll leave the L ch alone for now since it is working (but not optimal). When I have time, that sole 2sa1216 will replace the 2sa1494 in the R ch.

 

The R ch is now working. Recall the final issue --- from the long train of issues relating to original blown R ch components -- was no (0.0mV) bias current. Repeated soldering around Q702 and C704/704 had caused some fractured traces. So I had to re-join here. Now, both channels very nicely bias up together, as the unit warms up. I see good sine and sqr wave traces. Yes, I am still using slightly mismatched PNP/NPN pwr trans as noted earlier. I might match those later. But I do have some preamp issues to address first. Because the orig pwr amp section was the main issue, I had no clue about other issues upstream. This is what to expect when bidding on "for parts only" units on ebay!
One good thing to come out of the long troubleshooting process -- other than experience and building skill set -- is that many parts get replaced when one is simply, desperately, blindly, "parts cannoning". So I've got some good upgraded caps, multi-turn pots, new R's, and many new semiconductors in the circuits now .... stuff that would've been replaced, anyway, for vintage upgrades.