replacing 2SC3281/2SA1302 outputs, transistor matching

[nick78]

I have an NHT SubTwo with a history of blowing outputs.

Read more about it's history here:

needed: 2SA1302

That was two years ago, and now I have another blown output. This one occurred when I was moving the sub a few inches to reach an antenna wire behind it. The sub was in standby mode at the time, upon being moved I heard the two 10" drivers get shorted to rail voltage, followed very shortly by the fuse pop.

Damage is a failed original (non-counterfit) 2SC3281. The previous failure two years ago was one of the 2SA1302's.

I am wondering if there is a problem in the jack for the interface cable to the controller, and that moving it caused a connection failure. I'll be inspecting it carefully.

But, alas, on to the reason for this post:

Now that the 2SC3281/2sa1302 combo is no longer available, I am looking at replacing all four outputs with the newer 2SC5200/2SA1943 devices.

Question 1: Does anyone sell these in matched sets? The amp design has no emitter resistors, so I'd really like to install a matched set.

Question 2: If nobody is currently selling these as matched sets, is there a relatively benign way to match transistors? I don't mind buying a dozen or so of each and building a small circuit to match them, and my DVM has a hFE test mode as well. I don't have access to a curve tracer, though.

Question 3: are there now any better output devices on the market that would be a drop in replacement for this design? (see schematic in thread linked above)?

Question 4: any suggestions on what's causing them to fail? Poor amp design is my #1 suspect.

[hewlew1]

Nick78,
www.ampslab.com click on the component tab at the top he sells them 5 complimentary pairs for $35.00 and he accepts paypal.

[pooge]

I think what is needed more is to match the same sex devices rather that the complimentary pair. The Vbes of the same sex devices should be matched, preferably at their operating current, to force equal sharing of current. For a subwoofer, matching compliments would not be as high a priority.

[MaxS]

Hello,

You can buy some of these BJTs on eBay.com and matching them yourself. It will costs fifty bucks for ten PNPs and ten NPNs.
For the rest, don't be afraid. This kind of transistors are selling themselves. You can easily find buyers ( and I can give my adress for free postage )

Max

[jacco vermeulen]

Originally posted by nick78
Poor amp design is my #1 suspect.

Considered getting rid of the power amp section in the MA-1A ?

[jaycee]

I'd say replace them all. Consider OnSemi's MJL3281/1302... or even the 4281/4302. Consider replacing the driver transistors as well.

Looking at the schematic in the other thread, I'd say that amp is junk. Using two different types of transistors in the LTP sounds like a no-no to me. Similarly, a diode based bias network stinks, and NO EMITTER RESISTORS is just insane - no wonder this amp blows outputs!

Find out what rail voltages are used. I'd say 42V, perhaps 48V for two pairs of outputs.. If it's more than that I'd be surprised... then build a better amp in there!

I read in the other thread that this is not really what you want to do, but to be honest, the design is so poor that it's just going to keep totalling transistors, and costing you money.

[nick78]

1) rail voltages as rated on the side of the toroidal are 95.6V. There is an additional portion of the schematic (which I did not post in the other thread referenced above) with "downconverters" on each of the rails. This amp is supposedly the same as in the Sunfire VT-3. More explanation here:

http://www.diyaudio.com/forums/showthread/t-103684.html

The short short version: the downcoverters are used to keep the voltage across the output devices at ~ 6V to keep the outputs in their safe operating area. so it's a variable power supply of sorts.

the complete schematic is here:

http://home.comcast.net/~jhidley/SubTwoi_Amplifier.pdf

2) i've already ordered the 5200/1943's as replacements from digikey, ordered a dozen or so of each. Any ideas/suggestions on how I should go about matching? is there a test rig schematic floating around out there somewhere?

I realize that this may not be the ultimate end-all be-all amp, but it works fine normally so if i can get a few more years out of it, that's what I'd like to do. It has never died/blown while in use... kind of odd and yet funny.

[carlmart]

Originally posted by nick78
2) i've already ordered the 5200/1943's as replacements from digikey, ordered a dozen or so of each. Any ideas/suggestions on how I should go about matching? is there a test rig schematic floating around out there somewhere?

I realize that this may not be the ultimate end-all be-all amp, but it works fine normally so if i can get a few more years out of it, that's what I'd like to do. It has never died/blown while in use... kind of odd and yet funny.

I think you should add emitter resistors and forget about matching the transistors. Something small, like 0.1ohm wouldn't hurt that much.

MOSFETs are much better for matching and keeping that value in time.

[nick78]

Well, I guess this topic is dead. I was able to match two pairs of transistors for the outputs. I installed one of the non-matching outputs in the amp temporarily to make sure she would fire up and not blow the matching pair. unfortunately, it again blew the same Q13 or Q14 output. I traced back in the schematic and discovered full rail voltage (+130V) on the gates of Q22/Q23/Q24, so they are blown as well and who knows what else that drives them upstream. Failure at this extent is probably beyond my ability to fix.

Any suggestions on what kind of amp to build around a nice toroidal that has 95.6-0-95.6 outputs and also 19.3v outputs?

[jaycee]

This is some kind of Class H topology. Obviously if the power supply part has died, then it's going to be a complex fix.

*edit* That explains it all - Sunfire is a Bob Carver company. This guy is known for doing very clever rail-switching amplifier designs. I retract my "amplifier is junk" comment from earlier posts In this whitepaper, he describes this amplifiers topology.

The problem with a transformer rated that high, is a conventional amp is going to be massive, with a lot of outputs, and a big heat sink. So sadly, it looks like this is junk unless you also want to go to the hassle of buying a new transformer.