View Full Version : 9090 driver transistor substitutions - noisy channel


Macdude
04-03-2011, 06:03 PM
I've done a fair amount of searching but can not find much related to the topic of driver transistor substitutes for the 9090. I've seen lists for the DB version, but as we know, this boards a bit different. It's an F-2436-1. Transistor controlled bias. About a year ago, when I replaced the fusible resistors, I accidentally bridged two components and saw some smoke shortly after powering them up. For the life of me I can't remember what smoked, but it was accidentally connected to a leg of one of the heatsink mounted B536 or D381s. I undid the bridge and from then on, for the past year of regular use, it's worked just fine aside from a sort of low level static/whining noise in the right channel. The sound comes and gos. It isn't constant and sometimes isn't present at all. The volume control has no effect and the music usually easily overpowers it. I figure it's probably one of the transistors mentioned above. That it was damaged and became noisy.

What I would like to do is replace all the transistors on the board. This 9090 has been mostly re-capped (all but the mains). I have also already added bourns multiturn trimpots for bias/offset. They work great. Anyway, here is a list of the substitutes I have been able to find:

B536 -> MJE15031
D381 -> MJE15030
2SC1400 -> Zetex ZTX694B (replaced in matched pairs)
2SC1720 -> ??? (black, cylindrical heatsink)
2SA849 -> ??? (black, cylindrical heatsink)

There are also two transistors mounted on heatsinks, TR07 and TR08. The service manual lists them as being 2SC984.

I have a ton of capacitors coming from Mouser for a Pioneer SX-1010, so if Mouser has the parts I need, I would prefer add them to the list and save a few bucks on shipping. Also, the Zetex 694's are out of stock on Mouser :thumbsdn:.

I really need to figure out the magic behind transistor specifications so I can stop bothering the wise mentors of these forums.

Thanks!

hatman
04-04-2011, 12:01 AM
I am not an expert, but have collected subs recommendations from AK members for my future 9090 rebuild. These subs should show up on an AK search. Maybe the AK experts will help to narrow the field.

B536 & D381 - the MJE looks good
2SC1400 - ZTX694B looks good (EBC pin-out).

2SC1720 - ZTX696B (EBC) or 2SC3502, use flag heatsinks on these TO92): or 2SC2911(TO126), 2SC1953(TO126) , KSC3503(TO126)

2SA849 - KSA1381ES

2SC984 - at MCM as Hitachi $1.39(TO92 50V) or MPS2222AGOS (40V TO92) at Digikey about $0.26

CDFixer
04-04-2011, 02:33 AM
Careful with subs for the 984s the 984s have the case isolated from the transistor substrate, most subs I have found have the case tied to the collector which will cause massive havoc if installed.

Macdude
04-06-2011, 09:50 PM
Do you happen to remember which 984 subs do not have the collector tied to the enclosure?

Also, are there any other subs for the 2SC1400 besides the Zetex ZTX694B?

EchoWars
04-06-2011, 10:03 PM
Also, are there any other subs for the 2SC1400 besides the Zetex ZTX694B?The C1400 was a very high-gain device. Hard to find something that is capable of replacing it, which is why I like the ZTX694B for a differential pair transistor...high gain, good bandwidth, and good current & power handling. The only drawback is the pinout, but it's far from a deal killer.

avionic
04-06-2011, 10:18 PM
The C1400 was a very high-gain device. Hard to find something that is capable of replacing it, which is why I like the ZTX694B for a differential pair transistor...high gain, good bandwidth, and good current & power handling. The only drawback is the pinout, but it's far from a deal killer.

How many do you usually get to obtain a matched pair?

CDFixer
04-07-2011, 01:45 AM
Do you happen to remember which 984 subs do not have the collector tied to the enclosure?

Also, are there any other subs for the 2SC1400 besides the Zetex ZTX694B?

No I don't I found some 984's a couple of years back still have some pm me

EchoWars
04-07-2011, 11:57 AM
How many do you usually get to obtain a matched pair?Yeah, I guess that's another drawback...the gains are all over the map (lows of around 475, highs of 950+, average in the low 700's). I buy 100 at a time, so finding a match isn't too tough. But at $0.75 each, buying 100 transistors is a tough sell. Out of two dozen or so, though, you ought to be able to get a couple of pairs within 5%.

Macdude
04-07-2011, 01:14 PM
Yeah, I guess that's another drawback...the gains are all over the map (lows of around 475, highs of 950+, average in the low 700's). I buy 100 at a time, so finding a match isn't too tough. But at $0.75 each, buying 100 transistors is a tough sell. Out of two dozen or so, though, you ought to be able to get a couple of pairs within 5%.

Any chance you've got two matched pairs you'd want to sell me + some money for your matching time?

EchoWars
04-07-2011, 01:28 PM
Please PM me, or send an email through my profile.

EchoWars
04-08-2011, 08:27 PM
Also, are there any other subs for the 2SC1400 besides the Zetex ZTX694B?After looking at the board itself, it seems that the input pair transistors have their pins all lined up in a neat row. For that reason, I'd change my recommendation to the KSC1845U, which would make mounting a bit easier.

Macdude
04-09-2011, 09:17 AM
Seems the only KSC1845 still available through Mouser/Digikey is the KSC1845FTA. Fairchild shows this to be a "direct" replacement for the UBU. Thoughts?

I believe I even have some of these 1845FTA's around somewhere.

KSA1845FTA Info (http://www.mouser.com/ProductDetail/Fairchild-Semiconductor/KSC1845FTA/?qs=UMEuL5FsraDCmxWozTypUA%3d%3d)

EDIT:

Actually, here's whats available from Mouser as far as a KSC1845 goes:
Mouser KSC1845's (http://www.mouser.com/Semiconductors/Transistors/Transistors-General-Purpose/Bipolar-Small-Signal/_/N-7h65w?Keyword=KSC1845&FS=True)

I understand that to find the best substitutes, one must compare data sheets and find similar (or better) specs. My question is, which specs must be kept the same or higher? Is there a rating which I cannot go lower on, like microfarads/voltage on an electrolytic? I'd assume it also depends on the circuit being serviced?

EchoWars
04-09-2011, 02:16 PM
High gain and decent bandwidth are good for the input stage. And I knew that the high-gain 'U' version was being discontinued, but I didn't know that Mouser had run out of them.

Gain groups for the Fairchild KSC1845:
U=600-1200
E=400-800
F=300-600
P=200-400

The 'U' was the most desirable. Yes, all the different gain groups will function, but we have a specific task to accomplish here, and the 'U' device was the one we wanted. You could buy a pile of the 'E' devices (still some left at "M") and hope to find a couple of matched pairs with gains of over 700, but I suspect that the average 'E' device will have gains in the upper 500's (the average 'U' device had gains in the low 700's).

I guess I'd revert back to the Zetex parts here, but it means buying a pretty fair pile of expensive devices to get a couple of reasonably matched pairs.

rBuckner
04-27-2011, 01:31 PM
I stumbled upon this thread, having the intermittant noise problem in a 9090 I'm redoing. I narrowed it down to TR06 and TR10, the TO-5 sized ones with the black tube heatsinks. Mine are 2SA818 and 2SC1628. I subbed in KSA1220 and KSC2690, which are the complements of each other per the datasheet. Both parts had a similar HFE reading on my meter to the existing transistors, which were removed at the time.

Brought it up slow with a variac and DBT with a current meter in line. It came on and I adjusted the current to 45mA (final value to be 50mA). Listened for 30 seconds, it sounded fine. The current was still at 45mA or so and hit the DBT bypass switch to give it full AC line voltage. It played another 15 seconds or so and after a brief buzzing sound (maybe from the variac), there was no audio, the red protection LED was blinking away and no current was flowing as it blew the meter's fuse.

After finding no blown semis, I retried coming up on the variac, DBT inline again and I got the bright bulb/overcurrent action. I put the original transistors back in and now it works again! I would love to hear some theories from you all!

One guess of mine is the new transistors, which still test fine still using the diode check method, broke into high frequency oscillation and upset the biasing and caused the current drive runaway. The excess current changed some characteristic of the two substitute parts so that they no longer work.

Can anyone suggest better transistor substitutes?

Thanks all...

P.S. The driver board was recapped, the 8.2K 1 watts replaced, diff pair replaced with matched 1845's and new drivers (15030/1) installed. All that working fine for many hours.

EchoWars
04-27-2011, 07:39 PM
Nothing wrong with your choices for replacements, but once you determine that the bias current is adjusting as it should while on the dim-bulb tester, you MUST readjust the bias current to minimum before powering off and powering up again directly from the AC line! Setting the bias current with the DBT in place is a no-no.

The original devices have a piece of tubing over them (which acts as the opposite of a heatsink, BTW) to ensure than the metal tab (which is at collector potential) does not short against the driver transistor heatsinks. Replacing these transistors with a fully plastic-encapsulated TO-126 device, like the Fairchild transistors you had, is always a good idea. And of course, you certainly don't re-use the plastic tubing. ;)

rBuckner
04-28-2011, 12:23 AM
Thanks for the advice!

Actually I thought I did it that way even though I failed to mention it. I did reduce the pot to where the current would be minimum and then came up on the variac. At least I thought I did, it was late... I'll try again and report back.

rBuckner
04-30-2011, 01:01 AM
I have found out that what is happening is it is breaking into oscillation at about 180kHz. I have the bias pot set for minimum current and it is about 17mA with the adjusted value being 50mA. I did try several values of bias current with the same results so brought it back to minimum while I fool around. It will not break into oscillation until the volume pot comes up some. If the volume is not brought up and the circuit is probed with a DMM, that it enough to get it started too.

I double checked all of the resistors and ceramic caps in that channel and found all to measure correctly. I also tried adding a 100pF cap to all of the transistors, one at a time, from B to C to no avail. Also the snubber resistor was lifted with no effect. I tried switching the two transistors to MJE15030 and 15031 to no effect. The only thing I did that made a difference was to remove the 15pF B to C cap on the transistor right after the diff pair and it just went into oscillation every time it was powered up. The same complement of transistors is in the left channel and does not oscillate.

Any ideas anyone?

ManyMoonsAudio
04-30-2011, 02:14 PM
If you pull out the pre-amp jumpers, can you get it to act up then?

rBuckner
05-01-2011, 11:23 AM
Yes it does. But I found out something interesting. I have scope probes on the amp outputs and hooked an audio gen. to the main amp input jacks. I brought up the level until it went into oscillation and I killed the power. I noticed that the oscillation left as soon as the power was off and the amp worked normally until the power supplies drained too much. When the power is switched back on and the protection relay has not kicked in, the signal looks normal. The moment it kicks in the oscillation takes off! I had never driven the receiver with a signal upon power up after changing the two driver board transistors so I'm glad you asked that question!! Thnaks!!

Makes me think something is wrong with the protection board. It has been recapped and has a new relay but here's the deal... the driver board was missing from this 9090 when I got it and I sourced one elsewhere. I had to jumper both channels on the driver board to bridge the outputs to the connection pins wired to the protection board. So there's a possibility some other change is required connection-wise. I'll be digging into it more and will report back.

ManyMoonsAudio
05-01-2011, 11:42 AM
All that's there is the relay and perhaps the Zobel network. What's it do with a load present?

rBuckner
05-01-2011, 01:45 PM
Same thing with or without speakers. I've been out doing yardwork but will be back out to play in a bit. Thanks!

EchoWars
05-01-2011, 10:02 PM
F-2547. Change out R13, R14, C08, and C09. Use high-quality parts for the replacements. Panasonic 0.1f 250V polypropylene (#PF2104-ND), and a Vishay 10 ohm 2W resistor, about the same size as the original 1W resistor (#PPC10W-2CT-ND).

ManyMoonsAudio
05-01-2011, 10:32 PM
F-2547. Change out R13, R14, C08, and C09. Use high-quality parts for the replacements. Panasonic 0.1f 250V polypropylene (#PF2104-ND), and a Vishay 10 ohm 2W resistor, about the same size as the original 1W resistor (#PPC10W-2CT-ND).

Bet that does it.

rBuckner
05-01-2011, 11:03 PM
I'll look into that. I was playing around some more and don't believe it to be anything on the protection board after all. The oscillation is present on the collector of the first stage after the diff pair. I changed the 15pF cap up to 100pF and it became stable. The freq. response went to pot as expected. I went back to two 15pF's in parallel which was pretty stable, I could still get it to oscillate at a high drive level. Then poof, no sound but the two 10A fuses for the + and - supplies blew. A quick check of the driver board transistors showed no blown ones so I bet one or more outputs blew.

I was pretty frustrated then so I quit for the day and will get back at it tomorrow night. Once it's running again I'm going to try inserting a small resistor in the base, like 47 ohms to try and "cool" off the gain a bit. We'll see what comes of that. Thanks fellas!

EchoWars
05-01-2011, 11:21 PM
Of course you'll see the oscillation back at the input stage...that's what feedback is all about.

The main causes are generally:
#1 - Bad film caps or resistors in the Boucherot cell (some call it a Zobel). The 9090 has one on the driver board, and another on the protection board.
#2 - Bad ceramic Miller caps on the driver board (C05, C06, C17, C18). The 9090 also has a odd decoupling film cap on the collector of the NPN driver. Unlikely to be a culprit, but stranger things have happened.
#3 - A damaged P-N junction on one or more of the transistors in the amplifier. Can be anywhere from the diff-pair to the output power transistors.

My money is on the most likely, which is #1, since I've seen this on other Sansui units using those cheap little green polyester caps.

rBuckner
05-02-2011, 11:55 AM
Good point. I'll get the parts on order. All transistors were replaced on the driver board prior to narrowing it down to the last two (of course!) as well as all new outputs. I will report back so as to have this thread as a reference on this problem.

Thanks EW!

rBuckner
05-09-2011, 12:40 PM
Got the parts in and it still has the same issue. I experimented with putting in resistance in the emitter of the first stage off the diff pair. By going up to 82 ohms I got both channels stable up to clipping. Then I hooked up some music and it went off on me at around 10 watts on the meters! Hooking back up an audio generator I found that the higher the sine wave frequency, the easier it oscillated. Makes sense. With the ferquency at 10kHz and higher and the levels low enough at the output, I noticed you could see the ringing on the waveform.

Unfortunately I had to set this aside for a bit. I will start a new thread when I'm back on it and have discovered any new info.