Replacement for LM394 transistor?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

user 40373

Well-known member
Joined
Nov 22, 2009
Messages
109
I'm modify a Ashly SC-50 I just picked up off eBay and wanted to replace the old LM394 resistors with MAT-02 E versions, but they are no longer available. Is there a modern replacement for the TO-78 tin can style transistors? It be great if they weren't $15 or so each like the MAT-02's  :-\
 
You can still get MAT02 from a source in Asia, there is an email I think in a recent BM thread about the MAT02. You may also try to use 2 low noise transistors (MPS-A18 or something) in their place, matched for hFE, should work OK.  Just watch the pinout.
 
that's the reason I gave up making my 9k preamp...

@mitsos, did you test the MPSA18? is it (really) as good as the lm394's??
 
Did you test the MPSA18? Is it (really) as good as the LM394s?

Forget it--the noise of the MPSA18 is way higher, and you'll never get the matching close enough for critical applications. MAT02E has been the very best NPN dual transistor--a real pitty this is a thing of the past.

Samuel
 
abbey road d enfer said:
Anyway, what's wrong with the 394? (apart from being "old")

For existing designs, not much. Besides, there's some evidence that at least for the last decade the LM394 and the MAT-02 have used the same die, in which case an 'upgrade' to a MAT-02 would be a lateral move at best.

For new designs, it's out of production and getting very hard to find, which is a great pity.

JD 'answering rhetorical questions for fun and profit' B.
 
If you're looking for MAT-02 in Europe, Reichelt still have MAT-02FH. (€10,95)

www.reichelt.de

They're the same that Digikey used to sell. (engraved text and AD symbol...)

 
These are at least 20 years old, but if I can get away with not replacing them, that would be great.
 
JohnRoberts said:
There is a nice recent transistor array from THAT corp.. (300 series) but only available in SOIC and 14 pin dip...

I have no idea of the price...

JR

Any thoughts on the viability of replacing (read "replacing" in the context of a new PCB design and some component adjustments) the LM394 or MAT02 in existing designs with the THAT 300 array?   Other than pin layout, what changes or challenges would there be?    

Since the THAT 300 has two pairs of electrically isolated transistors, could the two halves be shared between two channels, or would there be some issue with having two channels thermally coupled like that?

P.S.  Given their marketing niche, I'm surprised THAT hasn't filled the void with a plug-in replacement THAT394 or THAT02.
 
stickjam said:
Any thoughts on the viability of replacing (read "replacing" in the context of a new PCB design and some component adjustments) the LM394 or MAT02 in existing designs with the THAT 300 array?

We talked about that a month ago in this thread, but it was in the BM so not the most logical place to look for it. Some excerpts:

jdbakker said:
The hFE for the THAT300 is pretty low: 60(min)/100(typ), or 3-5 times less than the worst grade of the LM394. That leads to a sqrt(3)-sqrt(5) increase in current noise and (less importantly) higher input bias current. Not very good for a DOA input.

jdbakker said:
Any active device will have both voltage and current noise; in modern small-signal BJTs at a mA or so of collector current this is dominated by base shot noise. Shot noise is proportional to the square root of current, and for a given collector current the base current is inversely proportional to hFE. Because of this all things being equal a low-hFE device will always have more current noise than a high-hFE one.

burdij said:
While I agree that there is a substantial difference in current gain, I only said that the That part was good enough. When you are in the desert with no bottled water available, you certainly will drink from a puddle to survive. [...]
The suggestions being made here are in answer to a question about suitable substitutes in existing applications. I think this part should be given a fair trial for that reason. It may not work. Certainly, it is worth $7 to find out.

I've bought a dozen about two years ago; tested them in a few LF amps and found them reasonably well matched but noisy.

We've discussed these parts a few times in the past, but much of that was deleted when Wayne and Roger left. Searching for "that 300" gives 40+ pages of hits, and sifting through that to get to relevant posts gets very old very fast. (I've always found THAT a very unfortunate company name when searching for information).

The MAT02/LM394 offer low noise and good matching. For VCAs, logging amps, multipliers and the like you need the matching, so there the THAT parts or the LIS ones may make sense. In audio you seldom care about a mV or two of offset, but noise does matter. For that purpose a pair of low-noise TO-92s would be more appropriate, IMHO.

JDB.
[still, might be useful for the synth crowd if not for mic pres]
 
> at least 20 years old

So? If they are working, they are working perfectly. And probably as good as any part you could buy today or tomorrow.

> if I can get away with not replacing them

You can.

The Ashley has much more "interesting" problems. The 4136 chips were mundane when new, you can do far better, but IIRC the pinout is not the one which became common, so it would take some serious leg-bending to plop another chip in. The output level pot is very strange. Unless the case is spectacular, or the overload is ohh-too-phatt, I see nothing interesting there.
 
Samuel Groner said:
[quote author=pedroplanet]Did you test the MPSA18? Is it (really) as good as the LM394s?
Forget it--the noise of the MPSA18 is way higher, and you'll never get the matching close enough for critical applications. MAT02E has been the very best NPN dual transistor--a real pitty this is a thing of the past.
Samuel
[/quote]
No I did not test this (satisfactorily). I once tried a couple of BC550C but there were other probs in the circuit so I can't say 100% that it worked, and by the time I got it working, my replacement MAT02 arrived so out went the 550.  I was going from memory of the discussion JDB quoted above. The MPSA18 is a transistor that recently interests me, but I think the discussion was based on the BC550C and/or 2N5088 and was between JDB(?) and burdij. I thought the concession was that it should work.
[quote author=jdbakker link] In audio you seldom care about a mV or two of offset, but noise does matter. For that purpose a pair of low-noise TO-92s would be more appropriate, IMHO.
JDB.
[/quote]

I am uncertain how to compare the noise figure listed in most transistor datasheets to the noise voltage density in the MAT02 datasheet. Samuel?
 
So i'm guessing the big answer is there are no currently in production replacements for either MAT-02s or LM394s...

Damn, i was going to do a group buy and have cases made up.
 
mitsos said:
I am uncertain how to compare the noise figure listed in most transistor datasheets to the noise voltage density in the MAT02 datasheet.

No, that's not very easy to do; it soon devolves into apples vs oranges. What you need is more complete noise specs, preferably either voltage/current-noise vs collector current, or the noise plots like those found in some datasheets (like page four of the 2SC2240 datasheet). Sadly very few manufacturers give detailed noise specs for typical parts, let alone guarantees for maximum noise.

TheGuitarist said:
So i'm guessing the big answer is there are no currently in production replacements for either MAT-02s or LM394s...

No duals with the kind of specs we're looking for, no. Some singles, like the 2SC2240, might work if more voltage offset is acceptable.

JDB.
 
PRR said:
The Ashley has much more "interesting" problems. The 4136 chips were mundane when new, you can do far better, but IIRC the pinout is not the one which became common, so it would take some serious leg-bending to plop another chip in. The output level pot is very strange. Unless the case is spectacular, or the overload is ohh-too-phatt, I see nothing interesting there.

I put an 2 opa2604's on a Brown Dog adaptor.  Works great.
 

Latest posts

Back
Top