YATMA - Yet Another Transistor MAtcher ( for lazybones )

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

Moderators: Kent, Joe., luketeaford, lisa

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

YATMA - Yet Another Transistor MAtcher ( for lazybones )

Post by elektrouwe » Tue Jan 19, 2016 11:27 am

I did not like the known single DUT (device under test) transistor/diode matching methods for several reasons :
- they require precision resistors (0.1%) and/or precission DVMs (>=4.5 digit)
- I often need pairs, quads of matched diodes/transistors or even more for ladder filters and is a pain i.t.a to keep temperature constant in serial measurements
- I don't like writing down lots of mV - values and subtract them later (yes, I'm a lazybone)

here is a circuit that uses cheap standard parts to measure delta-Vbe of 2..8 DUT at a glance with an oscilloscope or a sound card based software scope.
Pic.1 shows the schematics:
Image[/img]

Basic principle is to scan the forward voltages of all DUT with a CD4051 octal analog switch U3, amplify the resulting (sub-) mV step pattern with U4, a high gain (x1000) AC-amplifier and visualize the voltage steps with a scope.
U2, another CD4051, is used to switch a single reference current to each DUT while its voltage is scanned. U1, a CD4060, generates a 1.5KHz scanning clock , a 3 bit switch address and a narrow scope trigger spike which is added to the amplified output.
Screenshot Pic.2 shows a typical measurement :
Image

a LM3046 transistor array containing 5 NPN transistors Q1..Q5 is connected as DUT1..DUT5. Because only 5 of 8 possible DUT are measured, unused DUT6,7,8 are connected to DUT5 to avoid fake delta-VBE steps for these unused channels.
note: scope Y-setting is 200mV/div and YATMA gain is 1000 : delta-Vbe resolution is 200uV/div !
The trigger spike indicates the begin of a scan sequence and ensures stable triggering even for (hopefully) small or zero delta VBE.

NPNs and PNPs can be compared even simultaneously. Schematics shows how to connect different types of DUT.
Pic.3 shows the breadboard prototype with another LM3046 connected to an oscilloscope and parallel to the microphone jack of a netbook running XOSCOPE under LUBUNTU:
Image

If , for example, VBE of Q1 is definded as reference, the scope's Y-position knob can be used to move Q1's voltage level to the 0V scope base line.
Then you can see at a glance how much the other voltages differ from reference Q1 !

I made a PCB with a ZIF test socket and a patch field. ICs or discrete semiconductors can be patched with short DuPont female wires.
next picture shows 8 diodes with anodes patched to the 8 DUT pins (right sideof ZIF socket) and cathodes jumpered to GND (left side)
Image

PCB design files to download / order: http://www.seeedstudio.com/service/RDOmfuY6

additional notes:
- Because only voltage differences between DUT are important, U4's amplified offset voltage does
not matter. A TL071 or similar opamp without offset trimming is good enough.
- C7 depends on how big the trigger spike must be to trigger your scope or soundcard-scope. 100p works in most cases
- for reference currents requiring Rref < 10k , delta-RON of switch U2 could introduce a reference
current modulation . This can easily be ruled out by connecting 1 DUT to all inputs DUT1..8 and
checking if the resulting delta-VBE pattern is flat.
- when measuring discrete semiconductors, keep the whole setup close in a small plastic box to keep all DUT at same temperature
- Rref1..3 can be selected by jumper or rotary switch or you can a pot to adjust the reference current

happy matching !

User avatar
yan6
Ultra Wiggler
Posts: 973
Joined: Sat Jun 15, 2013 7:14 pm
Location: Ontario, Canada

Post by yan6 » Tue Jan 19, 2016 3:25 pm

This looks really neat! I'm a little lost on how a transistor would mount on the ziff socket to match you DUT diagram. It looks like the ziff has 3 pins all connected together and the outermost is a common ground.

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Tue Jan 19, 2016 4:47 pm

yan6 wrote:... how a transistor would mount on the ziff socket to match you DUT diagram...
like the diodes in the picture:
NPN : B and C together on the right ZIF side, E jumpered to ground on the left side.
PNP : E to the right side, B and C to left side and grounded.

it's also shown in schematics : all "anodes" = anode of a diode, emitter of a PNP, base of a NPN) are patched to the positive reference current, all "cathodes" = cathode of diode, emitter of NPN, base of PNP are connected to GND via jumpers
so 8 transistors would fit in a ZIF16 socket.
yan6 wrote:... 3 pins all connected together...
1 is the ZIF socket, the 2 others are used for patching the reference current to DUT1..8.
In case of matching less than 8 DUTs, unused DUT lines should be patched together to the last used DUT; therefore I need 2 pins.
Outer GND lines are used for jumpers from "cathodes" to GND.

my english is not the best; hope I could explain it good enough

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Wed Jan 20, 2016 1:41 am

Seems like a good idea. You might consider using a DIP switch instead of all of those jumpers as they can be a pain to change in such close quarters like that.

frijitz
Super Deluxe Wiggler
Posts: 1646
Joined: Mon Aug 31, 2009 3:37 pm
Location: NM
Contact:

Post by frijitz » Wed Jan 20, 2016 11:08 pm

Very nice!

Ian

klangbauer
Common Wiggler
Posts: 79
Joined: Wed May 29, 2013 3:52 am
Location: Germany

Post by klangbauer » Tue Oct 18, 2016 12:03 pm

build, tested and select transistors

Image

User avatar
diablojoy
Super Deluxe Wiggler
Posts: 1619
Joined: Fri Jan 22, 2010 9:03 pm
Location: melbourne

Post by diablojoy » Wed Oct 19, 2016 2:05 am

interesting :tu:
It probably wont happen today but if it does it definitely wont go smoothly.

User avatar
keninverse
Super Deluxe Wiggler
Posts: 1013
Joined: Mon Jul 26, 2010 4:12 pm
Location: mission control htx

Post by keninverse » Wed Oct 19, 2016 8:34 am

any chance you can post the gerbers for this here? Seeed link is dead.

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Wed Oct 19, 2016 9:31 am

keninverse wrote:any chance you can post the gerbers for this here? Seeed link is dead.
can't find my YATMA data even after login @ Seed - strange...
here is the ZIP file with gerber data and docs
You do not have the required permissions to view the files attached to this post.

specialunderwear
1-Post Wiggler
Posts: 1
Joined: Wed Feb 21, 2018 3:51 am

Post by specialunderwear » Wed Feb 21, 2018 12:24 pm

Great work! I'm building one.

KSS
Super Deluxe Wiggler
Posts: 3784
Joined: Mon Jan 25, 2016 7:28 am

Post by KSS » Sat Feb 24, 2018 3:14 am

4051 has 8 matched channels? Doesn't this need per channel compensation to be accurate?

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Sat Feb 24, 2018 5:02 am

KSS wrote:4051 has 8 matched channels? Doesn't this need per channel compensation to be accurate?
as described above, you can check 4051 "quality" by measuring a single component connected to all 8 channels. Then you should see a straight line on the scope.
4051's R_on channel mismatch is <=5 Ohms which is about .1% of the DUT feeding resistor.
So, as long as you don't want to match delta Vbe to < 10uV you are fine ;-)

User avatar
gabba@gabi
Learning to Wiggle
Posts: 29
Joined: Sun Jan 28, 2018 2:58 am
Location: Berlin/Maputo

Post by gabba@gabi » Tue Feb 27, 2018 12:48 pm

Just great idea!

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Fri Jun 01, 2018 7:28 am

UPDATE ! there was some request for a tiny YATMA version, so YATMA2 was born: same principle, but only 2 chips required for a 2 transistor matcher.
My FB buddy Jim made a nice layout , built a prototype and now offers all design files here. Thanks a lot, Jim !
Image

how it works:
both DUTs (NPNs,PNPs or diodes) have their anodes connected and share a pullup resistor (R10,R3 selected by jumper) that determines the measurement current. DUT cathodes are pulled down to GND alternately by a square wave oscillator with complementary open drain outputs (74HC125).
This generates a small square wave voltage with amplitude Vdut1 -Vdut2 which is AC-amplified by an opamp and watched on the scope.
A gain of ~500 shows Vbe differences of 100uV as 50mV signal on the scope.
The circuit ist powered by USB 5V.

Jim's pictures show a random selected transistor pair and a very well matched one:
Image
Image

picture of the prototype:
Image

and attached Gerber & drill files.
Happy matching !
You do not have the required permissions to view the files attached to this post.

Veryfungi
Learning to Wiggle
Posts: 13
Joined: Sat Jan 20, 2018 12:15 pm
Location: USA Forest Floor

Post by Veryfungi » Fri Jun 01, 2018 8:01 am

I'm the Jim guy. :) This tester only needs a scope to work and I'm blown away at how precise it is. Thanks for allowing me to test and implement your smaller version tester. I hope it helps many people!

User avatar
gabba@gabi
Learning to Wiggle
Posts: 29
Joined: Sun Jan 28, 2018 2:58 am
Location: Berlin/Maputo

Post by gabba@gabi » Sun Jun 24, 2018 7:42 am

Allow me to ask:
how importand is it to have two different measurement currents?

User avatar
wackelpeter
Veteran Wiggler
Posts: 528
Joined: Fri Apr 17, 2015 5:56 pm
Location: Germany, MD

Post by wackelpeter » Sun Jun 24, 2018 12:25 pm

very neat, Uwe and Jim... much appreciated effort you made

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Mon Jun 25, 2018 10:04 am

gabba@gabi wrote:Allow me to ask:
how importand is it to have two different measurement currents?
depends on what you want to match. For transistors a measurement current between 10 and 100 uA is often used ( 68k => ~60uA).

For matching LEDs - I used them as VC-resistors in VCFs and VCAs - R10 was in the range 3k3..6k8, as far as I remember (1k is a just a place holder value). In general, measurement current value is not too important; keep it small enough to avoid self heating of the DUT and big enough to get a clean output signal.

User avatar
cygmu
Super Deluxe Wiggler
Posts: 1564
Joined: Mon Feb 15, 2016 3:33 pm
Location: UK

Post by cygmu » Sun Aug 19, 2018 2:24 pm

I just wanted to say how happy I am that you shared this device with us all. I built one today and I’m delighted with it. Everyone in my family enjoys watching the scope trace move when you put a finger on one of the transistors under test!

consumer
Learning to Wiggle
Posts: 31
Joined: Sun Apr 24, 2011 11:36 am
Location: United States

Post by consumer » Sun Dec 08, 2019 9:50 pm

Thank you guys for this nifty project.

OSH Park doesn't like the the gerbers uploaded for the YATMA2: It errors out because it can't find a Board Outline file.

That said, if anyone has a spare PCB of the YATMA2, please PM me.

User avatar
d.simon
Common Wiggler
Posts: 149
Joined: Wed Mar 04, 2015 11:06 pm
Location: San Diego

Post by d.simon » Tue Dec 10, 2019 1:58 pm

@consumer

I might have one...let me check. I believe I had to add some things to the gerber. I think I used gerbv to add the outline, but maybe I used a text editor.

I made both the YATMA and YATMA2 btw. I believe on YATMA2 I added a silkscreen to the bottom layer (a duplicate of of the top silkscreen, which means it was reversed as viewed on the bottom). The response from the fab house was kind of strange - I sent them a reverse image for that layer, but they tried to correct all the numbers and letters (by re-reversing them). Anyway it doesn't affect the functionality.

da6353
Learning to Wiggle
Posts: 3
Joined: Mon May 20, 2019 6:28 am
Location: PA, USA

Post by da6353 » Thu Dec 12, 2019 6:47 pm

Files seem to work when uploading to jlcpcb.com:

Image

User avatar
Alfrede
Learning to Wiggle
Posts: 44
Joined: Wed Dec 20, 2017 3:43 pm
Location: Germany

Post by Alfrede » Sat Dec 14, 2019 3:35 am

Hi one question is the design Creative Commons or what must i note when i want to rebuild it which kicad?

consumer
Learning to Wiggle
Posts: 31
Joined: Sun Apr 24, 2011 11:36 am
Location: United States

Post by consumer » Mon Dec 16, 2019 10:05 pm

@ d.simon: Thank you, I PM'd you.

Maybe I'm doing it worng but Gerbv (installed through Homebrew on MacOS, iirc) only lets me delete things: I can't change an object's layer or anything like that. And even that delete thing is a little fraught, now that I'm poking around at it... in any event, kudos!

Alfrede wrote:Hi one question is the design Creative Commons or what must i note when i want to rebuild it which kicad?
The image in the post just before your post shows CC BY NC, so that seems to be a safe assumption.

elektrouwe
Common Wiggler
Posts: 152
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Re:

Post by elektrouwe » Mon Dec 23, 2019 6:59 am

Alfrede wrote:
Sat Dec 14, 2019 3:35 am
Hi one question is the design Creative Commons or what must i note when i want to rebuild it which kicad?
yes, the design is "CC NC BY". What is your intention? If you want the Kicad files send an email.

Post Reply

Return to “Music Tech DIY”