CGS64 VCA dual transistors

[sonic]

With LM394 now as rare as hen's teeth, is there an affordable dual transistor with the same pinout which would work in the CGS64? I've heard various substitutions suggested around the forum, but none seem to be any cheaper or easier to find.

Or does someone know of a source of affordable LM394s?

[andrewF]

The only long term and cheapest solution is to learn to match transistors. Then you can use $0.05 parts such as BC547.
It is unlikely supermatched pairs will ever be available again in thru-hole format.

There are a few matching circuits around, Ian Fritz's is #1 on my list:
http://electro-music.com/forum/download.php?id=25772
It only needs a power supply and about 4 parts, not difficult, and suitable for matching transistors to be used in VCOs as well.

[sonic]

Thanks Andrew. Yes, that was the response I expected, but being a lazy sod I was trying to avoid matching transistors myself. Mind you looking at that circuit and guide from Ian Fritz I have to admit it might actually be quite a useful and interesting learning exercise.

[The Bad Producer]

Thanks Andrew (and Ian!) I was looking for that recently and couldn't see it...

One thing though, would you always use BC547/557 or 2N3906/4 as the transistors for the VCA, VCF, VCO or something else depending on the application? Also does the B or C suffix have any relevance here?

[andrewF]

I used 2SC1085 and 2SA1015 on absolutely everything for many years as I bought a few 1000 of them for about $5 from Superjunk in Akihabara, so IMHO any general purpose trannie will do the job in most synth diy ccts.

These days I use BC547/557 for everything, simply cos they are cheap and are easily available.
The A,B,C suffixes refer to the typical current gain values.
No suffix means unspecified; could be anywhere between the min and max values. These seem to be the most common from retailers.
'A' will have the lowest typical value thru to 'C' having the highest. I don't recall any synth circuits where the current gain of the transistor was a critical point, tho it may be interesting to find out how much difference it makes.

[The Bad Producer]

Thanks Andrew... Superjunk! Sounds like a cool place!

[sonic]

I'm just getting around to matching some BC547s to use in these Ken Stone VCAs (and maybe for some VCOs too) and I've built Ian Fritz's circuit on a breadboard. I wonder if someone would be kind enough to give me a couple of pointers and let me know if I'm in the ball-park to getting some decent matches.

I used +/- 1% 100K metal film resistors in the circuit. My meter was only able to measure their accuracy on the 200K setting to one decimal place and all of them measured as 100.0K. Are these precise enough?

The readings I get across the two emitters range from -5mV to 5mV (assuming I am reading my meter correctly - it's on the '200m' setting). When I swap the leads, the reading always settles down to the polar opposite of the first reading, so this stage seems superfluous. Does this mean my resistors are just well-matched, or do I need a more precise meter to measure the mismatch?

The first pair of transistors I measured had a dVbe of well under 1mV (I forget what it was exactly). Since then I haven't had a pair with such a low differential. How close a match should I be aiming for, and just out of interest what kind of methods do you use when matching quantities of transistors? Right now I'm just randomly swapping and hoping for the best!

[sonic]

Is there anyone who might be able to help me with this? I think I'm probably on the right path, but any input would be greatly appreciated.

Thanks,

Richard

[prgdeltablues]

I recently used Ian's elegant circuit to match transistors for an exponential converter. Can I remember how good the matches were?... I think I found pairs within 0.1mV of each other within a batch of 20 or so. The SSM2210 supermatched pair is spec's at 0.2 mV dVbe max.

Like you I used 1% resistors hand matched to one decimal place ie 0.1% matched. I followed Ian's advice and didn't bother swapping transistors over, just kept Transistor A is place all the time, and measured all my others in position B, noting the dVbe values, and sorting them into bins - regardless of any system errors, if all you change is the one transistor in B, and two show the same reading, then they're a good match. Tranny A just goes back in the box.

hope this helps.

peter

[sonic]

Hugely helpful. Thanks Peter!

[Cynosure]

I can't remember if it was in the forum or chat, but someone mentioned an interesting method for sorting them that i made a mental note of.

After you test one, tape it to a sheet of paper and write the value above it. Fill the sheet with all your transistors with values above. Then you can easily select the closest matching pairs after you test them all.

[sonic]

Thanks Cyno. That's a nice simple solution that makes a lot of sense.