Noob questions regarding matching trannies

Hi, get out the newbie repellent...

I've purchased a B&K 162 transistor tester as some auctions mentioned that it is a great way to match trannies...I'm directly cloning the Moog 901a/901b oscillator controller/oscillator, and each of these has a "selected" transistor. There is no info on what it is selected for, and the 162 manual does not describe any process for matching. I'm now not sure if it's capable of doing so, due to having zero knowledge of such things.

The transistor tests it performs are:
Beta
ICBO
ICES
ICEO
with ranges of:
100uA
1mA
10mA
100mA
1A

If I should take the time and effort to search out all of this in order to gain a working knowledge, I will, but it's not something that will be of regular use to me. I know that might sound blasphemy here, but I'm just looking to see if anyone has a simple procedure that would allow my project to move forward.

Should I just breadboard a DUT VBE measuring bed instead?

The B&K also measures IDSS for FETs, and I need to measure some there as well (2N4339 for 1.0), and the manual is too oblique... Thank you in advance for any hand-holding.

http://www.trannyweb.com/trib-2005-06-Jun.shtml

sorry I couldn't help myself.
I thought you were talking about transformers...

ok I'll read your post now.
Sleeper

I would bet that the transistors are matched for hfe or gain, my DVM tests for that, and it's been my experience that most circuits that ask for transistor matching ask for that. These other things are part of the gain equation, but that's where my knowledge starts to fizzle, I suspect if you look at a few transistor data sheets you'll start to see the correlation. search for transistor matched pairs matching, something like that, also jfet matching (because there's been a lot of discussion here in relation to the hamptone jfp amplifier and also 1176 clones.
Good luck
Sleeper

FYI - Trannie or Tranny means transformer(s) .
At least in electronics it does.

For transistors we just say transistor.

There, now you know.
:wink:

> Moog 901a/901b oscillator controller/oscillator, and each of these has a "selected" transistor. There is no info on what it is selected for

You need to find out. Or back-think what the circuit does. Since I never got paid to work on Moog, and forgot what little I understood about ARP, I'm no help.

Is this the plan?
http://www.freeinfosociety.com/electronics/schematics/audio/pictures/moog901b.jpg

That looks like PAiA design.....

I looked at the schematic and I don't see anything on the 901a except that the mirrored pairs of transistors in that configuration ought to be matched for as many qualities as possible, but hfe ought to do. the 901b looks like you want to test for a voltage-in circuit.
The way I'm reading that schemo is that you had better put sockets on your board for Q7,Q8, and Q9 and start swapping in and out until you are reading 2.45 volts at point A and you have a nice smooth undistorted sine wave.
Transistors are cheap.
Buy a handful. I guess if you test them as you go you will be able to start to figure out what you are looking for in terms of specs on the transistors.
i.e. if you pull one out and the next one brings you closer to target then see what the difference is and what look for.
sleeper

Thank you to everyone for the speedy replies. I don't know anything about this so thanks for not leading me around by the nose as I pursue what I know to be the finest sawtooth core oscillator ever produced.

There are a few versions of the 901 pair, and the modules I have here match what's in the Moog Modular Service Manual.

Thank you for the clarification regarding "trannie" versus "transistor". See what you get for being cute when you don't know what you're talking about? (I'd thought the lingo around here for transformer to be "trafo"..)

I do have an HFE tester, and the three 2N4058 are all 325-ish. 4058s from Micro Electronics max out around 200, if they get above 160 or so... I've found some which match that HFE, and will try them in-circuit. They obviously have to have similar response at various voltages...

http://www.freeinfosociety.com/electronics/schemview.php?id=905

Q7 is the 2N4058 in question (mine are slightly different than the schemo), and there are two selected resistors.

PAIA :wink: Heh..I'll be posting the PCB artwork in The Lab at some point if anyone is interested in such wild, untamed items.

I wouldn't worry about that PNP. Get some Toshiba 2SA1015GR parts and I'll bet they work just fine. Bob Moog did do some tweaky things but I'll bet a too-high beta PNP wouldn't hurt that circuit at all.

The bigger problem, I suspect, is a reliable source for the 1N34 germanium diode.

> Q7 is the 2N4058 in question

The plan will 'work' with ANY silicon PNP there.

(I don't see how it can work with C4-C7 connected as shown.... I' probably missing something, but you must not deny the possibility that the factory schematic has errors, accidental OR deliberate.)

"Selection" would be about making all modules in a single synth track. Or track as good as early Moogs ever could. Carlos, Hyman, and others have remarked how unstable pitch-tracking was on these machines.

P4 gives adjustment for PUT ratio variation, and I know such a trim can hold 1% accuracy for years. I would bet that R8 and R9 are "selected" -only- when a way-off PUT is found. It is clear from the way the ramp-tri converter is biased that the ramp is centered above ground, perhaps this "2.54V TPA" test. P4 ought to cover that.

The 270K and 43K provide first-order approximate corrections for Ico, Hfe and Vbe.

Considering the state of silicon production then and now, I think: you build these with random 2N3702 from 1970, 90% of them can be trimmed-in with P4 and P2, 10% just won't trim at low and high pitch....

Ah. Q7 and Q9 could match for Vbe. Not necessarily exact match, more like a constant offset at any current. Test at 0.5mA and also at 5mA. However using today's silicon, I bet trannies (yes, you can say that) from a tape will match far better than Bob ever had in his dreams, and 100% will turn out as good as it gets.

Do you have the factory Service Manual? The one for the ARP right after this (full modular) went into considerable technical detail (except about what was inside the black epoxy modules). All post-repair re-trims were documented, and since in those days we still expected transistors to fail occasionally, that means all parts (except the top-secret black epoxy bits).

Thanks, I believe that I have a few 1015GRs lying around...and I have a batch of 1N34A already. eBay sometimes delivers...these vary wildly in price. I also have the other transistors mentioned on the schematic, in case I wish to build the earlier versions. 2N2926, etc. These things were easy to find, compared to the near unobtanium components for the Buchla 258b oscillators that I'm also working on- that's where I need to run the IDSS test to find FETs which will produce an optimum sine wave.

The only part which eludes for the Moog project is the CL1 100V rectifier diode, which is also in The Filter. Search engines and my favorite sources disappoint.

I know that matching diodes would be as simple as setting the voltage range/beta for the particular transistor, and then swapping out others until something shows up as closely as possible- I just don't know which range(s) are important here...

[quote author="PRR"]> Q7 is the 2N4058 in question

The plan will 'work' with ANY silicon PNP there.

(I don't see how it can work with C4-C7 connected as shown.... I' probably missing something, but you must not deny the possibility that the factory schematic has errors, accidental OR deliberate.) [/quote]

I'm told that the later oscillators, the 921b, are purposely incorrect in the manual. For the 901s I'm duplicating, I've been working from actual modules. I've confirmed parts values and most of the circuit paths but am going by the module for the capacitor/switch wiring.

"Selection" would be about making all modules in a single synth track. Or track as good as early Moogs ever could. Carlos, Hyman, and others have remarked how unstable pitch-tracking was on these machines.

Click to expand...

I once had a set of four 901a/b and they were pretty good together over around 4 octaves. Drifted like mad, but tracked fairly well together. I've also owned and used others which were nearly unusably inaccurate, so there must be a happy medium.

I'm after this design from having fallen in love with the very interesting anomalies (mistracking, drift), after having programmed synths for a living since 1987 (and having played with them since 1980). I can't take digital accuracy, not another day. I've helped voice digital synths for a few of the
big manufacturers (even if just a few sounds instead of an entire factory bank), and spent most of my time futily attempting to make ROMplers and drum brains more "alive" through extensive modulation. Analog synths which pursue "accuracy" and "stability" =produce= the problems I've spent years attempting to solve, and are thus tragic errors, IMO. I have a place for such things such as Chowning FM, but for what they are, not to imitate analog.

I have zero interest in this recent virtual digital, as I call it...but I'm an anomaly in my field, just like those here who hand-wire tube limiters, compared to most...

Considering the state of silicon production then and now, I think: you build these with random 2N3702 from 1970, 90% of them can be trimmed-in with P4 and P2, 10% just won't trim at low and high pitch....

Ah. Q7 and Q9 could match for Vbe. Not necessarily exact match, more like a constant offset at any current. Test at 0.5mA and also at 5mA. However using today's silicon, I bet trannies (yes, you can say that) from a tape will match far better than Bob ever had in his dreams, and 100% will turn out as good as it gets.

Click to expand...

I have NOS 2N3392 as used in the modules I have here...rounded base, extended cylinder case. I do have some modern 3702s, and I'll try both and see what happens. When I learn how to test for Vbe, that is.

Do you have the factory Service Manual?

Click to expand...

It doesn't go into theory the way that the Minimoog service manual does..

I appreciate the time and information. Thank you for all of the assistance.

me> Do you have the factory Service Manual?

pages 49-52, http://www.freeinfosociety.com/electronics/schematics/audio/moogmodular.pdf

This is much less than ARP gave us, and there's probably more MOOG matter but you will have to pry it out from under the old techs who hoard this stuff.

> for a living since 1987

I must bow to your superior.... wisdom? experience? persistence? stubborness?

Technically I "made a living" maintaining and teching the ARP and Other Duties As Needed (hanging rugs, running PA system, pouring musicians into their cars). But we put the ARP aside about the time you were getting into it as a living.

I know what you are saying about too-too-perfect synths. But I sat through a decade, 20 semesters, of classes teaching audtitioned students both how to hear and how to use the Lab. Over 200 students, and less than 10 would hear the things you hear, maybe 5 would actually use it. Just as 99% of the general public won't know the sound of a tube-limiter from a cheap chip. You are working for your own satisfaction, not for the crowd.

Put it together like it shows, un-selected new-made parts. Run the trim. -You- have enough experience to know if the module runs properly or not. Build several, and you'll know if any are not-right, and can work back to find what the differences are.

Can you still buy the PUT programmable unijunction transistor today? In craft-production quantity/quality, not barrel-dregs like we get Germanium transistors now? Yeah, you could fake a PUT with 6 or 8 parts.

[quote author="PRR"]> for a living since 1987

I must bow to your superior.... wisdom? experience? persistence? stubborness? [/quote]

Bad choice of a career with zero transferrable knowledge, such as your obvious technical skills?

Technically I "made a living" maintaining and teching the ARP and Other Duties As Needed (hanging rugs, running PA system, pouring musicians into their cars). But we put the ARP aside about the time you were getting into it as a living.

Click to expand...

I began making a "living" at it when computers were first commercially grafted into synths: ESQ-1, etc. Did you know that some folks at electro-music.com are cloning the Odyssey right now? At least, the separate sections as individual PCBs...might want to keep a low profile if you don't want those questions occuring here

You are working for your own satisfaction, not for the crowd.

Click to expand...

Exactly, unless one markets a product. As no one wants something like this, I'm making it for myself and sharing the artwork.

Put it together like it shows, un-selected new-made parts. Run the trim. -You- have enough experience to know if the module runs properly or not. Build several, and you'll know if any are not-right, and can work back to find what the differences are.

Can you still buy the PUT programmable unijunction transistor today? In craft-production quantity/quality, not barrel-dregs like we get Germanium transistors now? Yeah, you could fake a PUT with 6 or 8 parts.

Click to expand...

I'll be dropping transistors in the existing modules to determine usability, but for other things, I still wish to learn to use the B&K for matching and IDSS. I'll need matching capability upon building The Filter. (I -could- just use THAT arrays, with .5uV matching, but want to do it the old-fashioned way, with original components first, as a labor of love. THEN I'll use THATs.)

Yes, I have all of the parts I need for both the Moog and Buchla projects, excepting IDSS selection knowledge and the CL1 rectifier diode. The earliest versions of both Moog and Buchla oscillators used something that I cannot find, the "U147" FET, alongside the 2N2646.

> zero transferrable knowledge, such as your obvious technical skills?

Tech-chops got me $0.50/hour more than a basic secretary. And stuck me there for 29 years. It hasn't been the path to wealth; and as for "happiness", there's ups and downs.

> I still wish to learn to use the B&K for matching and IDSS.

It must have a manual. Stick a transtor in, read the reading. Stick another one in, read the reading. Just for education, stick a very different transistor in, such as a fat power device.

For "matching" one parameter, you rough-check a bunch to see what the value-spread is. Divide that spread by 12 and mark the cups in an egg-carton with each sub-range. Now measure carefully and toss into the approriate cup. Then when you need two matched parts, take them from the same cup.

To match two parameters, get a 144-egg tray. Mark Hfe down the side and Vbe across the top. Measure each part for both parameters, and toss into the cups. Obviously it is possible to sort 100 transistors and get no matchers at all. You may not need such fine sub-ranges. It may help to know that an "X" error does to the circuit: at the 1V/oct point, an 80mV error is a semi-tone, an 8mV error is a gross off-tune, 1mV or even 2mV may be inaudible. And usually the 1V/oct node has adjustments, so such errors may be totally trimmable. The details of error-budgets for musical pitch control can be quite tedious. You may get further faster by picking a 10mV MIS-match and seeing how bad it really is, for that socket.

> he "U147" FET

Wow, that's a past-blast. I know that part only as cross-ref to newer parts; 2N5020 may replace U147. Ah, 2N2050 is also obsolete.
http://www.interfet.com/pdf/DS_2N5020_21.pdf
It is low Vgs low Idss. In a pinch, the FET in a Panasonic electret mike element might work (ah, and P-sonic has obsoleted most of that line...)

> alongside the 2N2646.

Then buffering the hi-Z ramp node. How old is this stuff.... ah, 44 years this month. Explains much. Yes, in 1964 we were just getting really-low leakage BJTs, the go-to part was still an FET. And by the time production was solid, the 2-BJT thing was a better idea.

Was that first decade the best or the worst? "Giorgio Moroder helped to shape the development of disco music by incorporating the Moog synthesizer in the 1975 Donna Summer hit "Love to Love You Baby"." Maybe Disco was good to you; acid-rockers saw it as the end of the world.

I suppose you know the movie "Island of Death". Mastorakis was inspired by the profit/cost ratio of "Texas Chainsaw..." and needed money, but made a rather different film. Quite shocking for 1975 and still a bit eye-peeling today. It's cheap, but he has Vision which many directors lack. Anyway, if you blank the video, there's music of several types, including a passage of piano and Moog-like synth, and another with synth, and maybe more (the visuals are quite distracting). Classic synth, not Disco-synth. (That site's store is broken; Amazon.com has it.)

Thanks a bunch, I greatly appreciate your time and insight!

The B&K does have a manual; set beta infinity, set beta cal=1, etc. It assumes much knowledge that I simply do not have. There are a few interacting parameters which I cannot follow: Do I decide to test for 10mA, then set Beta and then swap out DUTs for similar readings, etc. It's sadly beyond me. I give up for the moment...because a simpler approach appears:

http://tinyurl.com/63notw

With what I read on the internet is a mistake in the second test bed, a missing 10K between transistors.

It appears that (at least for the Minimoog) they didn't test for both HFE and VBE. The earlier, target device, the Moog 904a filter, has many matched pairs, and voltage targets for each location... I really should just use THAT arrays; at least one boutique manufacturer is doing so in their clones. I don't remember if it's Club of the Knobs or Mos-Labs.

Just as there is a market for clones of pro audio, many are making some sort of income from clones of vintage synth gear. There is very little ARP product out, aside from filter variations and the upcoming Odyssey oscillators. Phil Cirrocco does potted modules and replacements, but that's a bit different...the DIYers buy bare PCBs as you well know, and if you're self-destructive enough, you might wish to investigate cloning sections of the 2500. I believe one person was looking into at least the oscillators, and was asking after other aspects, but it might be a wide-open opportunity. It's a can of worms...

> a simpler approach appears:

Sure. Maybe. Could be simpler still. Didn't think you were open to that path.

Two 9V batteries. One C to B to hold the base-collector junction biased. Resistor from emitter to another battery to base.



Two transistors with different Vbe will take different current; so what? Two devices with same Vbe will take same current, which is the critical fact.

The current is not exact; we don't need it to be: if two devices match at 0.9mA they match just as well at 0.8mA or 1.0mA. So don't claw for an exact resistor value. Do watch your battery voltage; if it drops to half that is a 30mV error, so you want to stay within a half-volt for the whole session. If you have +/-15V ARP power or +12/-6V Moog power, that will do fine, adjust the resistor in proportion.