MAT02 equivalent (matched transistor)

[Taylor]

Building Harry Bissell's pitch to voltage converter. Was planning to maybe build a few of them, but these MAT02 matched transistors seem to be tough to find and quite expensive.

Anybody know of a decent equivalent that's a little more reasonably priced, or a way to assemble a workalike using separate transistors? What do I need to match?

[kupervaser]

Not on ebay as well?

[Mark Hammer]

Depending on the degree of precision needed, sometimes people will simply suggest glueing two "normal" (but reasonably matched for hfe and one or two other things) transistors together (flat-to-flat) with cyanoacrylate.  You want them to respond thermally in an identical way.

[R.G.]

+1 on Mark's suggestion.

The standard thing to do for the synth guys when they can't get matched parts is to build a rig along the conceptual lines of my JFET tester and to test a largish batch of transistors for gain and Vbe at a certain collector current, usually about 1ma. The idea is to get Vbe within 5mV of identical and gain as close as possible, since IIRC Vbe matching is more important in exponential converters.

I would set up something with a test socket for the TUTs (transistors under test; notice that's not "in test") with the collector tied to a convenient voltage, like perhaps +9V. The emitter is tied to a constant current source of 1.0ma; more on this below. The base is tied to a bias voltage through a largish resistance, perhaps 100K to 1M (I'd have to actually do math to pick a real value). 

The bias voltage holds the base in the middle of the available voltage, the current source forces the transistor to operate at 1ma. You connect your meter to first measure Vbe, then flip the meter to measure across the base feed resistor. That tells you the base current (V=I*R) and therefore the hfe (=Ic/Ib).

So, like all semiconductor testing, you label each device. I like those little 1/4"/6mm paper dots from the office supply store. You write sequential numbers on a whole batch of dots while they're on the base sheet, and stick a dot to each device as you first grab it for testing. Wear gloves or use pliers or other tool to keep your hand from warming the transistor before testing.

Now bring up a spreadsheet on your computer. Fill one column with sequential numbers matching what you used for the TUTs. Grab a device, note the number sticker, make the test, and put the tested ones in the "tested bin" you have so thoughtfully provided yourself. Type the results into rows on the spreadsheet beside the identifier number. Do this until they're all tested.

Now use the spreadsheet "sort" function to sort the rows by Vbe, then by hfe. You can pick out sets easily by eye, and the identifier number tells you which TUTs to grab out of the bin for the sets.

As an aside, I have ditched Microsoft Office products entirely. Open Office is available free, does everything I ever needed Office for, and produces files which are 100% compatible with people who then want to read them with Office. There are slight differences, I'm sure, but overall, it's great.

[fredmerts]

After finding the MAT02 or gluing two matched together, you still have a problem finding the tempco resistor that needs to be epoxied to the MAT02 case. A few synth builders have them for sale, but I have yet to find them at a big outlet like Mouser or Digikey. They are also quite expensive.

[Taylor]

Hmm, I found this by searching for 1k tempco:

http://mouser.com/Search/ProductDetail.aspx?R=RN55D1101FB14virtualkey61300000virtualkey71-RN55D-F-1

Is that not what I'm looking for?

[fredmerts]

You certainly could try it.

When these synth guys use tempcos, they usually specify 3500ppm. If you use something else, it may (or may not) cause tuning problems. The only 3500ppm tempco I could find were SMD, and soldering leads on these is near impossible, but I have seen it done. I gave up on this project after trying to solder a SMD tempco. You can find through-hole, 3500ppm tempcos at Bridechamber, Magicsmoke, and MFOS.

[Taylor]

I don't really know what this whole tempco thing is about, and though I know ppm is parts per million, I don't really know 'parts of what?'. I know it has to do with temperature stability, which I reckon is manifested here in stability of tuning.

Guess some research is in order.

Edit: ok, for future dummies like myself:

"Temperature Co-Efficient or TC:  The amount of change to the resistance in relation
to temperature.  Expressed in Parts Per Million (PPM) or Percent per Degree C.

1000ppm  =  0.1% / Degree C
200ppm   = 0.02% / Degree C
100ppm = 0.01% / Degree C
10ppm  = 0.001% / Degree C"

source: http://www.atcomhq.com/industry_info/Resistors%20101.pdf

So, I'm a little confused here. Wouldn't my 100ppm resistor be better than 3500ppm, since it would only change resistance .01% per degree, instead of .1% per degree (don't I want less change)?

[Taylor]

Not on ebay as well?

The ones on ebay are going  for between $30 and $45 before shipping. 

[R.G.]

So, I'm a little confused here. Wouldn't my 100ppm resistor be better than 3500ppm, since it would only change resistance .01% per degree, instead of .1% per degree (don't I want less change)?

The difference is that the resistor is placed in the circuit where it can compensate for the semiconductor drift with temperature by drifting in the opposite direction. It's >good< that it drifts, for this particular purpose. Low drift is good, but if you can't get low drift, cancel what you can.

[Taylor]

Hmm, I see.

Found the LM394 (seems to be a usable equivalent for the MAT02) for $5+$6 shipping on ebay. The seller seems to have a lot (based in Hong Kong). Might be the best way to go for the matched transistor.

Still unsure about this tempco business. Can't seem to find them. I knew I shouldn't have wandered into the no-man's-land of modular synth designs...

Stephen Giles will be around shortly to tell me to "stick to your fuzzboxes, kid". 

[RedHouse]

Did you do a search on the usual suspects?
(Mouser, Digikey, Allied, Jameco, etc)

Here's a couple hit's:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MAT02FHZ-ND

http://www.alliedelec.com/Search/ProductDetail.aspx?SKU=6301415&MPN=MAT02EHZ

Darn things are spendy aren't they?

Your initial post asked about equivalents, the LM394 "SuperMatched Pair" might do ya.

[Taylor]

Your initial post asked about equivalents, the LM394 "SuperMatched Pair" might do ya.

Yep, see post above yours. Thanks for the links. These guys certainly are more expensive than I'm used to for components.

[slacker]

There's some info here about using cheap SMD tempco resistors.
Also, generally circuits that use tempco resistors will still work without them, or with them replaced by normal resistors. It's just that the accuracy will drift a bit with temperature changes, or as components heat up, for example a VCO will drift out of tune.
Depending on what you're using the circuit for, the drift might not matter.