Hi everyone,
I am building a Kassutronics transistor ladder filter, and am in the process of matching some bc547c transistors.
It looks like there’s a few different ways… how accurate is the CGS method?
I know there are a couple of circuits (covered here), but I’ve got transistors on a roll and a multimeter already…
Yes, that’s an easy way of matching them. Just be really careful with the temperature and stick to the instructions. Don’t touch them and if you messed up one measurement, skip that transistor since the readings are garbage from that point on (until you do the next batch which you leave on your table for a couple of hours in a preferably temperature-stable environment).
I can’t stress enough how important the temperature is 
don’t exhale air onto the transistor and also don’t sit at the window where some reflections of the sun might fluctuate. Keep your fingers (body heat) as far away from the transistors as you can, use long test-leads and use the counting technique like described in the instructions to make sure that the amount of current is similar for each measurement since that will heat up the transistors too.
I usually do this at night, without any heating or open windows and then I spend like an hour or two for matching transistors and diodes. They come in different boxes and that’s it.
Don’t know. I used the Ian Fritz method. Dual power supply is no problem, matching 100k resistors is easy, and I had no stability problems.
Granted you have to take the transistors off the tape. But I’d worry anyway about temperature differences between transistors several inches apart. And you’ll have to take them off the tape sometime anyway, right? Just put matching transistors into separate piles and put each pile into a little bag when you’re done.
I also am not convinced you can do this very well without waiting for temperature to stabilize. Waiting makes for a much more prolonged process but you can do other stuff while waiting. I did 23 transistors over the course of an hour or so.
But maybe the CGS method is fine. You could try both methods on a dozen transistors and compare results and then come back here and tell us…
I also had good results with the method from Ian Fritz. I used two 9V batteries for power.
Thanks for the input everyone. I rang through 22 earlier with the CGS method. I marked down the differences so they can be paired. There wasn’t much difference to be found though really.
I’ll try the Ian Fritz method too, then we’ll see if the pairs line up the same. I’ll build up the stripboard sometime this week…
got one of these a few years ago , if you plan on doing more than one project its a great time saver and works with resistors , diodes , capacitors , transistors .
there are a few brands out there not sure of their price point though .
and handling the components will heat them up enough to change readings .
Never used one of those but I have the impression what they measure (for transistors) is the gain, hFE. But for most matched transistor usage in synth circuits, according to Fritz, that isn’t what you want. What needs to be measured is base-emitter voltage, V_BE. Actually you’d like to measure leakage current but that’s hard and V_BE is a good enough proxy.
The CGS method evidently tries to measure that directly. The Fritz method, in his words: “Since you want to know the difference between two transistors, why not set up a circuit to measure that difference directly, rather than having to make two separate measurements and then subtract the results?” The point is that if you want to match them at the 50 µV level then with the CGS method or any method that just measures one transistor at a time you’d have to measure V_BE at a precision of about 1 part in 10^4 which is beyond the abilities of most inexpensive multimeters. With the Fritz method you’re just measuring the difference so a 10% or so measurement is fine.
Then again, for synth DIY you probably don’t need anywhere near 50 µV. I’ve read a match at 2 mV is good enough, that’s a 0.3% measurement which is manageable.
Thanks, I do actually have one of these.
They do give a mV reading alongside a hFE reading… however I can’t find any info on what that mV reading actually is measuring. Do you happen to know what it is?
I don’t think it can be Vbe, as the readings it gave were consistently around 930-980mV, but in the CGS method, which measures Vbe directly, they note that the expected value should be somewhere around 0.6-0.7V. Which is roughly what my multimeter was giving (they varied around 0.595-0.6).
In this discussion
it’s claimed it is V_BE:
reference: https://www.darc.de/uploads/media/Transistortester_df1rn_rev_1.1_2013_03_11.pdf
“Für das Beispiel von Bild 5 ist dies der Typ des Transistor NPN, die Zuordnung der Anschlüsse Basis, Kollektor und Emitter zu den Anschlusspunkten 1, 2 und 3, die Stromverstärkung hFE sowie die Basis-Emitter Flussspannung Uf in mV.”which roughly translates to:
“The example in image no. 5 shows an NPN-type transistor; the mapping of the pins ‘base’, ‘collector’ and ‘emitter’ to the tester connectors 1, 2 and 3; the amplification hFE as well as the base-emitter forward voltage Vf in mV.”
but as you note the value it’s showing is higher than is reasonable for V_BE. So maybe the above claim is wrong. Or maybe it’s a crappy measurement.
Hmm, well… it would be interesting to see if the measurement was bad but in a consistent enough way to produce adequate matching.
I’ll add it to the list of methods to try with the same transistors for comparison.
Well, first set of comparisons.
The CGS method got me very consistent readings between 0.595 and 0.599.
The transistor tester didn’t go well - the readings increased from 0.919 to 1.04 progressively as I moved from transistor to transistor. The increasing pattern continued neatly even when re-testing earlier ones.
I was quite careful, I kept me and my breath away from them, used nylon pliers, rather than metal ones, which had been in the same room. Maybe it is temperature related, but I wonder if it might be the tester itself or the contact pins heating up?
On to the circuit tomorrow…
Ok, so here’s a comparison of the CGS method (1st number, V) and the Fritz method (2nd number, mV) across the first 14 I tested.
I paired off 1+2, 7+9, 10+11, 3+4, 6+12. I only needed 5 pairs so stopped at that point.
While I wouldn’t have have paired up 6+12 using just the CGS method, any of the obvious pairing choices I would have made using just the CGS method would have been tolerable according to Kassutronics’ 2mV max variation.
Whether the CGS method would hold up if you have a grab bag of random transistors with a lot of variance, I don’t know…
If you find yourself in a position where you are using the CGS method, as long as you choose those registering as exact matches you should be pretty safe.
Here’s the scatterplot. Looks pretty much uncorrelated to me.
For synth DIY purposes, you can probably get away with pairing any of these to any other though. Or just about.
It’d be interesting seeing this for a larger number of transistors with more of a spread in values.
thanks again for your nice post and the stripboard layout. I’ll have fun matching a bunch of transistors 
I am having to do a little matching for the Transistor ladder filter so I went ahead and built this little guy on a scrap of leftover stripboard
Seems to work well, but I realized that my voltmeter only goes down to two decimals. Is it worth investing in this?
I figure as a bonus I could use it to start testing module power draw as well.
See posts starting here:
I think it’s very likely your multimeter is precise enough, as discussed in that thread. (As stated above, the Fritz method only requires measuring at around the 10% level.) If not, no, a clamp meter is not what you want. Not for testing synth module circuits. It’s more for house wiring etc. Among the posts starting with the above one is:
I was going back over the testing procedure and figured as much.
I have time - apparently there are a lot of holidays going on right now that are slowing down shipping.
Best way to do that is with a multimeter/voltmeter. If you use 10Ω resistors on your power rails measure the voltage drops across them. I don’t, so I built a thing on stripboard that has two power headers with 10Ω resistors on the power rails. I connect it to my bench supply with one ribbon cable and to the module under test with another, and then measure the drops across the resistors. Then the current draw in mA is the voltage drop in mV divided by 10.
Amp measurement with a multimeter isn’t so useful for this because it requires getting the meter in series between the power supply and the module, and that’s cumbersome. And because if you then forget to switch back to voltage measurement before trying to measure a voltage, you short out what you’re trying to measure.
