matching transistors - DIY

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: lisa, luketeaford, Kent, Joe.

Post Reply
skautkurt
Common Wiggler
Posts: 62
Joined: Sat Dec 08, 2018 8:36 am
Location: Germany

Post by skautkurt » Fri Feb 15, 2019 12:51 pm

fuzzbass wrote:
You just have an unusually strong aura. :zen:

If it was your body heat, and not from contact with either transistor, it would show up in both of them at the same time. You don't need a box. A fan helps. So do plastic tweezers for handling the parts. My little matching rig (Fritz method, built on stripboard) has a ziff socket so I can drop the parts in with tweezers, and clamp them with the little lever. If you can keep your fingers off the transistors between resting and testing, things speed up a lot. I also use an hfe tester to rough sort them, which results in fewer tests per match.
Tehehehe, it indeed feels a little on the esotronic (electronics + esoteric) side :hihi: I am using a pair of ceramic tweezers (they have ceramic tips), and I am a ziff-socket aswell. So no direct contact with the transistors.

User avatar
phisynth
banana nabab
Posts: 1788
Joined: Wed Sep 07, 2011 5:06 pm
Location: right above my feet

Post by phisynth » Fri Feb 15, 2019 4:00 pm

emmaker wrote: The transistors are extremely sensitive to temperature changes. I wouldn't use a fan to blow air over them. Reason is that even though the volume of air going over the transistors is going to stabilize the temperature somewhat there could be changes in that air's temperature going into the fan. I think that you'd be better to insulate the transistors so you have no air movement. I have a rag or box I put over the testing circuit and let it stabilize before I take readings.
Ray Wilson from MFOS must have been wrong then :

Image

User avatar
Kevin Mitchell
Common Wiggler
Posts: 230
Joined: Tue Jun 27, 2017 10:55 am
Location: Springfield, Massachusetts

Post by Kevin Mitchell » Fri Feb 15, 2019 4:53 pm

Using a fan is your best shot for consistent ambient air flow. But that should go without saying.

I'm considering boxing up the board I've posted on the previous page and somehow mount a zif socket to the enclosure. Possible update on that soon.

Gandalf
Common Wiggler
Posts: 125
Joined: Thu Nov 29, 2018 8:59 pm
Location: Australia

Post by Gandalf » Fri Feb 15, 2019 5:22 pm

You should really put the whole thing in a large box/container with a small heater element so that you can raise the temperature a little above ambient (25C is a good temperature). The fan, which should not be pointing at the transistors, is meant to circulate the air around the box to 'mix it' and remove any thermal hotspots.
Place untested transistors in the box as well so that they reach the same temperature.
Wear some cotton gloves to handle the transistors (easier than tweezers if using regular ic or transistor sockets for mounting).
A large box will allow you to open it, get your hand in and out with minimal disruption to the temperature, but will take long to get up to operating temperature.
All the above is a bit clumsy. I have a smaller container with the test boards in it, heater element and fan (not pointing at the transistors) and I an get stable readings within 30 seconds

User avatar
Kevin Mitchell
Common Wiggler
Posts: 230
Joined: Tue Jun 27, 2017 10:55 am
Location: Springfield, Massachusetts

Post by Kevin Mitchell » Mon Feb 18, 2019 2:01 pm

That's another reason why I very much prefer the GO-NOGO circuit. Since the transistors under test are next to each other, once they stabilize you wont be getting any bad data. Less variables to worry about since your testing 2 transistors back-to-back instead a load of them individually.

It's more fast to test a load of transistors this way as well. You can record the voltage difference with a fixed transistor (only swapping one transistor out to test) and see about matching the ones that are close in measurement.

I think you all are missing out :hihi:

existenz81
Learning to Wiggle
Posts: 22
Joined: Mon Apr 30, 2012 9:19 am

Pcboards for sell?

Post by existenz81 » Sat Mar 23, 2019 8:24 am

Who sells the Ian fritz pcbs? I’m interested in acquiring one if it’s possible!
Thanks

TOPLEL
Common Wiggler
Posts: 110
Joined: Sun Apr 28, 2013 12:49 am

Post by TOPLEL » Sat Mar 23, 2019 9:33 am

Did the Fritz "two transistor difference method" on a recent (bought in the last 3 years) batch of 2N3906 and the largest difference between the two outliers was 0,3mV but generally it was between 0,0 and 0,1mV.

existenz81
Learning to Wiggle
Posts: 22
Joined: Mon Apr 30, 2012 9:19 am

Post by existenz81 » Sat Mar 23, 2019 10:20 am

TOPLEL wrote:Did the Fritz "two transistor difference method" on a recent (bought in the last 3 years) batch of 2N3906 and the largest difference between the two outliers was 0,3mV but generally it was between 0,0 and 0,1mV.
So you are saying that once I use 2 Transistors from the same ammo belt stock, I can go without matching them, they’re quite already matched?

existenz81
Learning to Wiggle
Posts: 22
Joined: Mon Apr 30, 2012 9:19 am

Post by existenz81 » Sat Mar 23, 2019 10:42 am

TOPLEL wrote:Did the Fritz "two transistor difference method" on a recent (bought in the last 3 years) batch of 2N3906 and the largest difference between the two outliers was 0,3mV but generally it was between 0,0 and 0,1mV.
So you are saying that once I use 2 Transistors from the same ammo belt stock, I can go without matching them, they’re quite already matched?

User avatar
wackelpeter
Veteran Wiggler
Posts: 508
Joined: Fri Apr 17, 2015 5:56 pm
Location: Germany, MD

Post by wackelpeter » Sat Mar 23, 2019 11:20 am

seems unlikely, These would be a Batch of super matched trannies…

Guess it's more pointing towards a mistake in the measurement process… less than 2mV is very often and also smaller differencies, but 0,0 and 0,1mV difference and 0,3mV at worst looks a bit suspicious to me...

And well let's assume that this is correct, you can't expect all Batches and those of the different manufacturers to be of the same Quality and thus a Transistor matcher is always the way to make sure that they are really matched...

TOPLEL
Common Wiggler
Posts: 110
Joined: Sun Apr 28, 2013 12:49 am

Post by TOPLEL » Sun Mar 24, 2019 11:56 am

wackelpeter wrote:seems unlikely, These would be a Batch of super matched trannies…

Guess it's more pointing towards a mistake in the measurement process… less than 2mV is very often and also smaller differencies, but 0,0 and 0,1mV difference and 0,3mV at worst looks a bit suspicious to me...

And well let's assume that this is correct, you can't expect all Batches and those of the different manufacturers to be of the same Quality and thus a Transistor matcher is always the way to make sure that they are really matched...
I matched BC547C NPNs before the 2N3906s (i changed the circuit between them) and there were bigger differences there, i only found two pairs within 0,1mV and the biggest diference was around 2,3mV.
Since my DMMs resolution is only 0,1mV there can be some errors in the measurement.

On the other hand on the good pairs the individual Vbe measurements were equal.

bootsy52
Learning to Wiggle
Posts: 2
Joined: Tue Jun 17, 2014 4:45 pm
Location: Germany

Post by bootsy52 » Sat May 25, 2019 7:55 pm

whoop_john wrote: If there is a demand for these I could post PCBs out to people at modest cost. Speak up if you are interested and I'll do a batch to cover everyone's needs. Failing that I will get about ten PCBs made and sell on 9.
Do you still got a PCB? I would like to get one

bootsy52
Learning to Wiggle
Posts: 2
Joined: Tue Jun 17, 2014 4:45 pm
Location: Germany

Post by bootsy52 » Sat May 25, 2019 8:09 pm

Kevin Mitchell wrote:I've designed my own layout for the MFOS circuit. Works like a charm. Let me know if you build your own! You could translate the layout for vero board if that's your style.
Do you have also Gerber files (or .brd, .sch) for it, so that we can fabricate a PCB?

User avatar
Altitude909
Super Deluxe Wiggler
Posts: 3392
Joined: Wed Aug 24, 2011 5:13 pm
Location: Meesheegan

Post by Altitude909 » Sun May 26, 2019 8:18 am

About the cooling/heating etc thing: It makes no difference how its done as long as it's consistent. You're measuring relative values to compare to other relative values. As long as the VBE stablizes, that's all that matters. I use the MFOS circuit and a little rig I used with an old computer fan. They will flatten out after about 30sec but more importantly, I can get consistent results with same part where I can take it out, put it back in and get to almost the exact same value

User avatar
Europa313
Learning to Wiggle
Posts: 35
Joined: Mon Jan 18, 2016 10:43 pm

Post by Europa313 » Mon Jun 17, 2019 3:47 pm

I got 10 "Ian Fritz" transistormatcher made at pcbway as I couldn't find any online.

If anyone's interested please pm, I'd check shipping cost if worthwhile

User avatar
synthcube
you will be assimilated
Posts: 2082
Joined: Wed Dec 21, 2011 6:42 pm
Location: Boston
Contact:

Post by synthcube » Sun Jun 30, 2019 1:17 pm

friends and current/future tranny matchers-- with thanks to ian fritz and fonik for their permission, we're happy to make the pcbs and full kits available now in the store.

https://synthcube.com/cart/ian-fritz-fo ... -match-pcb

our kits include the pcb, power cable and right angle header, 0.1% tolerance 100K resistors, 3-pin inline sockets, compact test points for DVM probes, etc.
facebook: https://www.facebook.com/2SynthCube

synthcube webstore www.synthcube.com/cart
music from outer space: www.musicfromouterspace.com
blacet research: www.blacet.com
MOTM DIY Analog Synthesizers: www.motmsynthesizers.com
music from outer space euro smt: www.mfoseuro.com

User avatar
rolfdegen
Common Wiggler
Posts: 81
Joined: Sat Jan 12, 2019 6:48 am
Location: Wuppertal (germany)

Post by rolfdegen » Tue Aug 20, 2019 3:15 pm

Hallo

This is my matching tester. The 100K resistors R1 + R2 in the circuit have 0.1%. A small fan cools the transistors very fast. After 2 minutes, the values on the multimeter are stable.

Matching Circuit
Image

Hardware
Image

Results in mV
Image

Greetings from germany. Rolf

Gandalf
Common Wiggler
Posts: 125
Joined: Thu Nov 29, 2018 8:59 pm
Location: Australia

Post by Gandalf » Tue Aug 20, 2019 4:49 pm

Having a fan blow directly across the transistors is not a great idea. Any thermal transients on the input side of the fan will be sent straight across the transistors.
The better solution is to build a large box, thermally stabilised and heat that up. Build up internal walls from polystyrene foam that will allow air to be moved around yet create minimal drafts. A fan (or fans) is used to raise the temperature above ambient (25C is perfect) and provide a draft area in which the tester sits.

User avatar
EATyourGUITAR
has no life
Posts: 4794
Joined: Tue Aug 31, 2010 12:24 am
Location: Providence, RI, USA

Post by EATyourGUITAR » Tue Aug 20, 2019 7:06 pm

heat the clean room to 98.6F body temp. that way you can handle them without robots.
WWW.EATYOURGUITAR.COM <---- MY DIY STUFF

User avatar
emmaker
Veteran Wiggler
Posts: 621
Joined: Sat Mar 10, 2012 5:07 pm
Location: PDX

Post by emmaker » Tue Aug 20, 2019 10:15 pm

rolfdegen wrote:Hallo

This is my matching tester. The 100K resistors R1 + R2 in the circuit have 0.1%. A small fan cools the transistors very fast. After 2 minutes, the values on the multimeter are stable.

Results in mV
Image

Greetings from germany. Rolf
If you just picked a transistor at random and keep that transistor as the test reverence for the testing you've done I'd use one of the -0.4 mv as the reference. In your sample selection that seems to be the normalized value. Better yet sample a large number of transistors and the value that comes in with the most transistors use one of those for the reference transistor.

User avatar
rolfdegen
Common Wiggler
Posts: 81
Joined: Sat Jan 12, 2019 6:48 am
Location: Wuppertal (germany)

Post by rolfdegen » Wed Aug 21, 2019 2:21 am

Thanks for your tips :)

consumer
Learning to Wiggle
Posts: 27
Joined: Sun Apr 24, 2011 11:36 am
Location: United States

Anything to this idea?

Post by consumer » Tue Dec 03, 2019 12:22 am

ejr27233 wrote:As I've only got a 3.5 digit multimeter is there any point in using a x100 gain op amp in the circuit to measure the offset mV?.
I read through this post, but never saw anyone respond to this idea: Is there any merit to multiplying the gain in order to exaggerate the offset?

Or would that just introduce more uncertainty?

jakobprogsch
Common Wiggler
Posts: 103
Joined: Tue Jul 31, 2018 7:48 pm
Location: Switzerland

Re: Anything to this idea?

Post by jakobprogsch » Tue Dec 03, 2019 7:02 am

consumer wrote: I read through this post, but never saw anyone respond to this idea: Is there any merit to multiplying the gain in order to exaggerate the offset?

Or would that just introduce more uncertainty?
I'd say no merit since at least the garden variety opamps that aren't specifically marketed as low offset probably have input offset specs in the mV range themselves.

I have to admit all this talk about the DMM digits in this thread has me confused anyway. Isn't the whole point of these "differential" circuits that you don't have to measure absolute voltage but only relative one? How many digits the DMM has is pretty irrelevant as long as it has a mV range setting.

User avatar
Revok
Ultra Wiggler
Posts: 763
Joined: Sun Jan 15, 2012 12:11 pm
Location: Phoenix

Post by Revok » Tue Dec 03, 2019 7:15 pm

I'm possibly wrong but I think the 4.5 digit meter is only specified for matching the resistors by hand.

jakobprogsch
Common Wiggler
Posts: 103
Joined: Tue Jul 31, 2018 7:48 pm
Location: Switzerland

Post by jakobprogsch » Wed Dec 04, 2019 10:04 am

Revok wrote:I'm possibly wrong but I think the 4.5 digit meter is only specified for matching the resistors by hand.
Right, which really makes this a non issue though? You can trim out the difference by adjusting the trimmer until swapping two transistors only changes the sign of the measurement. It's slightly more work than just measuring lots of resistors with an expensive DMM but you only have to do it once.

Post Reply

Return to “Music Tech DIY”