Match Pair PNP Transistor array?

[zz28zz]

Found a 5-pin IC that needs replacement. It's a pair of PNP's with common emitters. The part is no longer available so I'm looking for an alternative. Considering 2 separate transistors, but not sure if they have to be a matched set. The only data sheet I've found is mostly written in Japanese. How can you tell if the transistors need to be matched?

Thx in adv.

[floobydust]

Part number? What's it used in?

[Smith]

I would consider every "2 in 1" transistor a matched pair. It  probably has to do with keeping their temperature matched too. They cost more than two normal transistors,  so the manufacturer has at least one of these reasons in mind.

[voltsandjolts]

Hard to advice with such scant info.
If your not sure then try just replacing it with a matched pair.
https://www.digikey.co.uk/products/en/discrete-semiconductor-products/transistors-bipolar-bjt-arrays/277?k=dual+pnp&k=&pkeyword=dual+pnp&sv=0&pv74=164&sf=1&FV=ffe00115&quantity=&ColumnSort=0&page=1&pageSize=25

[T3sl4co1l]

I take it, it's one of those SIP-5, beveled slab package, common emitter types?

Using independent rather than monolithic pairs only increases the offset voltage (especially due to thermal drift), and the distortion marginally.  If you want the best replacement, hunt down a monolithic pair; if you just want it working again, use anything.

Note that "matched pair" transistors are available (especially in SMTs?), but they are independent dies on separate lead frames (well, the lead frames are just leads anyway, of course they're separate).  The thermal matching is pitiful, typically something like 500 K/W differential thermal coupling between dies.

The schematic can tell you can tell how sensitive it may be.  If the diff pair is operated at high voltage, the change in power dissipation will be large.  If the common mode voltage range is small, a cascode could be added, significantly reducing the voltage drop on the diff pair.  (Which may also bring the power dissipation within the realm of a SOT-363 device, if that's the only sort of thing you can find.)

Tim

[Martin Hodge]

Would something like this work for you? I made them to replace such a transistor in a Pioneer SX-750

[dcarr]

That's cool.  Also, what's the number for the "pin strip"---that also looks useful.

Thanks for sharing.

[Martin Hodge]

That's cool.  Also, what's the number for the "pin strip"---that also looks useful.

Thanks for sharing.

From digikey: A128053CT-ND

Be sure to order the correct size to match the thickness of your PCB.

[fzabkar]

Found a 5-pin IC that needs replacement. It's a pair of PNP's with common emitters. The part is no longer available so I'm looking for an alternative. Considering 2 separate transistors, but not sure if they have to be a matched set. The only data sheet I've found is mostly written in Japanese. How can you tell if the transistors need to be matched?

Is it a secret?

[zz28zz]

Part number? What's it used in?

2SA1928 Premier (Pioneer) car stereo amp.

I would consider every "2 in 1" transistor a matched pair. It  probably has to do with keeping their temperature matched too. They cost more than two normal transistors,  so the manufacturer has at least one of these reasons in mind.

Thx, that's what I was thinking too.

Hard to advice with such scant info.
If your not sure then try just replacing it with a matched pair.
https://www.digikey.co.uk/products/en/discrete-semiconductor-products/transistors-bipolar-bjt-arrays/277?k=dual+pnp&k=&pkeyword=dual+pnp&sv=0&pv74=164&sf=1&FV=ffe00115&quantity=&ColumnSort=0&page=1&pageSize=25

Didn't see anything on Digikey. Needs to be at least 100v VCEO and 400mW Pt.

I take it, it's one of those SIP-5, beveled slab package, common emitter types?

Using independent rather than monolithic pairs only increases the offset voltage (especially due to thermal drift), and the distortion marginally.  If you want the best replacement, hunt down a monolithic pair; if you just want it working again, use anything.

Note that "matched pair" transistors are available (especially in SMTs?), but they are independent dies on separate lead frames (well, the lead frames are just leads anyway, of course they're separate).  The thermal matching is pitiful, typically something like 500 K/W differential thermal coupling between dies.

The schematic can tell you can tell how sensitive it may be.  If the diff pair is operated at high voltage, the change in power dissipation will be large.  If the common mode voltage range is small, a cascode could be added, significantly reducing the voltage drop on the diff pair.  (Which may also bring the power dissipation within the realm of a SOT-363 device, if that's the only sort of thing you can find.)

Tim

Yes it's a SIP-5 package. It's a high end amp and so anything that increases distortion is out.. This device sees 100V across it.

Would something like this work for you? I made them to replace such a transistor in a Pioneer SX-750

Oooohh, something like that might work. Probably my best bet the way things are looking so far. How much voltage and power can it take?

Found a 5-pin IC that needs replacement. It's a pair of PNP's with common emitters. The part is no longer available so I'm looking for an alternative. Considering 2 separate transistors, but not sure if they have to be a matched set. The only data sheet I've found is mostly written in Japanese. How can you tell if the transistors need to be matched?

Is it a secret?

No secret. I had another thread going but it seems to have died. Component in question is Q552.
https://www.eevblog.com/forum/repair/premier-prs-x720-amplifier/



Here's a pic of the schematic with comments and the only spec sheet I was able to find, mostly in Japanese.
The circled components are the 2 resistors that got hot and had to be replaced. Q552 is just to the left and down a little.

[T3sl4co1l]

Actually the bases and emitter are around 0V so it's about 50V drop.  Looks to be ~1mA total, so 50mW, or 25mW each transistor, give or take.  Shouldn't be too too bad for an SMT, I might still shy away from SOT-363 but -23-6 with big pads would be more than enough.

Tim

[Martin Hodge]

Would something like this work for you? I made them to replace such a transistor in a Pioneer SX-750

Oooohh, something like that might work. Probably my best bet the way things are looking so far. How much voltage and power can it take?

Here's the datasheet. PM me your address (if in US) and I'll send you one if you're interested.

[zz28zz]

Actually the bases and emitter are around 0V so it's about 50V drop.  Looks to be ~1mA total, so 50mW, or 25mW each transistor, give or take.  Shouldn't be too too bad for an SMT, I might still shy away from SOT-363 but -23-6 with big pads would be more than enough.

Tim

I was looking at a worst case situation (1 transistor hard-on, other idle).
56k + 5.1K= 61.1K total resistance. (assume 0 ohms for transistor)
100V / 61,100 = ~1.6mA
1.6mA X 100V = ~163mW.

Your right abt base/emitter being ~ 0V, but even at that:
56k + 5.1K= 61.1K total resistance.
50V / 61,100 = ~818uA
818uA X 100V = ~81.8mW.

What am I doing wrong?

What is this 23-6 you speak of?

[zz28zz]

Would something like this work for you? I made them to replace such a transistor in a Pioneer SX-750

Oooohh, something like that might work. Probably my best bet the way things are looking so far. How much voltage and power can it take?

Here's the datasheet. PM me your address (if in US) and I'll send you one if you're interested.

Thx for the offer, that's really generous.
Looks like I'm really close to Vceo max (50V) if not slightly over. It might work. Not sure. I'll PM you.

[T3sl4co1l]

I was looking at a worst case situation (1 transistor hard-on, other idle).
56k + 5.1K= 61.1K total resistance. (assume 0 ohms for transistor)
100V / 61,100 = ~1.6mA
1.6mA X 100V = ~163mW.

Your right abt base/emitter being ~ 0V, but even at that:
56k + 5.1K= 61.1K total resistance.
50V / 61,100 = ~818uA
818uA X 100V = ~81.8mW.

What am I doing wrong?

Only 50 of the V are dissipated by the transistors (and a bit less than that even, because of the collector load resistors, but that's only a few V off), the other by the "tail" resistor.

You would see maximum power imbalance, and duration, for strong bass hits -- so strong they drive the amplifier into clipping.  The distortion or thermal recovery under that condition probably wouldn't be noticeable (for any reasonable speaker efficiency, your eardrums alone are distorting as much, nevermind the fact that, again, the amplifier itself is clipping heavily ).

High frequencies, at high levels (but not necessarily clipping), can run into the same concern (large current swing in the diff pair), because the amplifier loop gain is lower up there; but the power change at say 20kHz will be adequately filtered by the thermal time constant, so this isn't a concern.

What is this 23-6 you speak of?

SOT package type.

Tim

[zz28zz]

I was looking at a worst case situation (1 transistor hard-on, other idle).
56k + 5.1K= 61.1K total resistance. (assume 0 ohms for transistor)
100V / 61,100 = ~1.6mA
1.6mA X 100V = ~163mW.

Your right abt base/emitter being ~ 0V, but even at that:
56k + 5.1K= 61.1K total resistance.
50V / 61,100 = ~818uA
818uA X 100V = ~81.8mW.

What am I doing wrong?

Only 50 of the V are dissipated by the transistors (and a bit less than that even, because of the collector load resistors, but that's only a few V off), the other by the "tail" resistor.

You would see maximum power imbalance, and duration, for strong bass hits -- so strong they drive the amplifier into clipping.  The distortion or thermal recovery under that condition probably wouldn't be noticeable (for any reasonable speaker efficiency, your eardrums alone are distorting as much, nevermind the fact that, again, the amplifier itself is clipping heavily ).

High frequencies, at high levels (but not necessarily clipping), can run into the same concern (large current swing in the diff pair), because the amplifier loop gain is lower up there; but the power change at say 20kHz will be adequately filtered by the thermal time constant, so this isn't a concern.

What is this 23-6 you speak of?

SOT package type.

Tim

In regards to the second paragraph, you lost me. Under what conditions would all this occur?

[T3sl4co1l]

In short, the amount of "work" the diff pair itself is actually doing.  In steady state, and for small signals, it's balanced by feedback.  It only really develops any voltage drop (between the input and feedback nodes) at high frequencies, or in clipping.  And I would guess, the conditions where those occur, really aren't of interest to audio purposes, so I wouldn't worry too much about differential power dissipation.

Tim

[zz28zz]

I received a couple of transistor arrays from Martin. Unfortunately, that wasn't the fix. Audio FETs still being driven very hard even at very low inputs. Installed 1k ohm 1/2 watt resistors in series with audio FET source legs to keep from blowing them up. Getting a weak tone out of speaker but o-scope shows serious clipping. If input from signal generator is turned down to 20mV PTP and amp gain is turned almost all the way down, I get a sine wave. Everything in the amplification portion of the circuit tests OK. Both channels doing same thing. Starting to think issue may be in I/V protection circuitry.

[T3sl4co1l]

Feedback network come open?

Tim

[Martin Hodge]

Feedback network come open?

Tim

Makes the most sense. Check for cracked solder joints with a magnifier.

[zz28zz]

Would feedback network issue cause both channels to behave the same way?

[Martin Hodge]

If both channels have the same problem, yes. Example; Maybe there are two components on one heatsink that have flexed and broken traces at the same area on the PCB. 12voltvids on YouTube has many examples of this type of thing. May not be your problem, but worth it to look.

[Kleinstein]

The replacement transistor is quite a bit slower (80 vs 150 MHz Ft). This could be a problem and worst case cause oscillation that causes other parts to heat up. If one channel is oscillation this may cause the other channel to also have quite some RF background.

Another possible problem is too much supply ripple due to aged caps. This would also effect both channels.

After a repair one may have to adjust the standing current and maybe DC offset.

The circuit by it self does not look really high end, more like rather basic. A high end circuit would be more like using emitter resistors for higher slew rate and a cascode circuit of some kind to reduce the voltage seen by the input transistors (could be faster and less thermal effects). The original transistors are low noise types with high gain and still 50 V rating. So this could be hard to find a same or better performance replacement in that combination.

[shakalnokturn]

The replacement transistor is quite a bit slower (80 vs 150 MHz Ft). This could be a problem and worst case cause oscillation that causes other parts to heat up.

I've always believed that it is high gain x bandwidth that lead to unwanted oscillations? A lower Ft can cause oscillation?

[T3sl4co1l]

Think such a small difference is unlikely to cause problems here, as the loop's dominant pole will be set by the volt-amp stage, not the diff pair.

If it were down under say 10MHz, I would be concerned, as the additional phase shift may cause instability.  I'd also be concerned about the HF distortion spec being worsened, since fT is talking about gain versus frequency and low fT means less loop gain at high frequencies.

Likewise if it were much over 300MHz, I'd be concerned about oscillation with parasitics, traces and whatnot.  Ferrite beads for one or both bases might be recommended in that case.

Tim

[zz28zz]

If both channels have the same problem, yes. Example; Maybe there are two components on one heatsink that have flexed and broken traces at the same area on the PCB. 12voltvids on YouTube has many examples of this type of thing. May not be your problem, but worth it to look.

I looked around but there are no heat sinks other than the case it self. I've ohm'ed out the feedback circuits and have continuity throughout. Flexed board a little but no changes were seen.

I set up this part of the circuit in a simulator and it shows the 1k ohm resistor I have installed in series with the source leg of the audio FETs will prevent sufficient feedback. However if I connect without the resistor, the FET overheats and shorts out very quickly. Kind of a catch 22 situation it seems.

[zz28zz]

Another possible problem is too much supply ripple due to aged caps. This would also effect both channels.

After a repair one may have to adjust the standing current and maybe DC offset.

 The original transistors are low noise types with high gain and still 50 V rating. So this could be hard to find a same or better performance replacement in that combination.

All the large electrolytics have been replaced and scope on rails looks very smooth. I'm planning on adjusting idle current once amp stops blowing up FETs.  I know finding parts for these older amps isn't easy.

[T3sl4co1l]

Incidentally, a similar part is actually two matched dies:
https://zeptobars.com/en/read/2SA979-matched-pnp-bjt

It's probably a lot of fussing for nothing, about the difference between a legit part and a substitute...

Tim