measuring uV

Why do you need matched transistors?

You can buy cheap dual transistors that are already pretty well matched.

You could set up a 2-transistor diffamp and measure its output with your meter. Apply an imbalance voltage between the bases until the diffamp output is zero. The trick is that the imbalance could be generated by a voltage divider, 1000:1 or something, and your meter could measure the *input* to that divider.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation

You could use an 'instrumentation' amplifier. Fairly stable, set the gain at 100, and buy a new meter. or....

Your soundcard, when properly used, can easily go down below microvolts. It is AC, but you could 'infer' your match? Make a small test fixture, or possibly use the one you mentioned?, plug transistors into that test fixture, watch the PC screen, and voila!

If you're not into 'rolling' your own, check out DAQARTA. Bob Masta DAQARTA v7.50 Data AcQuisition And Real-Time Analysis

Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE Signal Generator, DaqMusiq generator Science with your sound card!

Or, come back and people will talk you through the basics of doing it yourself.

roject.I have diagrams for a transistor matching cct but need to measure di fferences in the region of 50uV.My digital meters lowest DC scale is 200mV. I know you can get more sensitive DVM's but these seem to be quite costly a nd would defeat the object of matching your own transistors.Is there a way of increasing the sensitivity of my meter.At the moment the lowest it shows is 0.1mV, another decimal place would be better. Cheers Colin

Hi Colin, Welcome. What are you matching? (Vbe @ some current) or Beta? Maybe a link to your matching circuit would help.

As Robert said an instrument amp would be easy. I also wonder if a Vbe multiplier would work? Can you run these at x10 gai n?

I was also thinking you could run the two transistors at the same Vbe and t hen measure the collector current. (I'm a little vague on exactly how to d o that.)

George H.

Good idea. Use two transistors as a differential amplifier, to amplify their own offset voltage. Given the emitter current and collector resistor values, the gain is predictable.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

ain?

then measure the collector current. (I'm a little vague on exactly how to do that.)

Hi George,I'm matching gen purpose npn or pnp pairs(2N3904/2N3906)for use a s exponential current generators in VCO's used in an analogue synth I want to build.Ideally the better they are matched when used with a tempco in cir cuit,the more it will stay in tune during temp changes. I've used the cct h ere at

but I've also built a much simpler one but both need to measure to uV leve l. Colin

You can just buy a "zero drift" op-amp and set it up as a gain of 100 amplifier, which would give you 1uV resolution on a typical inexpensive multimeter. For example, the LTC2051. It can be operated from +/-5V.

--sp

You might want to look at the LM3046, which has five monolithically matched NPNs. They aren't that great at low current, like below a microamp, but above there they'll do a good job. They're about 50 cents in onesies from Arrow.

As JL says, you can also get cheap multiple transistors that work pretty well right out of the box, as long as you keep the power dissipation low. (The thermal coupling between dice is poor, about the same as two devices next to each other on a board.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

project.I have diagrams for a transistor matching cct but need to measure d ifferences in the region of 50uV.My digital meters lowest DC scale is 200mV .I know you can get more sensitive DVM's but these seem to be quite costly and would defeat the object of matching your own transistors.Is there a way of increasing the sensitivity of my meter.At the moment the lowest it show s is 0.1mV, another decimal place would be better. Cheers Colin

I once tried to sort pnp's by Vbe. I first discovered you can't pick them up. (hot hands) So I'm sitting there, trying to pick transistor leads (to-220) out of a parts bin with grabbers. I "sorted" a bunch.. and discovered later... that the room temp was changin g. (or I breathed on them, or something...) And finally realized that I didn't just need Vbe, but also some small fract ion of beta...

George H.

"Colin Butterfield"

Hi there,I'm new to the group and need to match transistors for a synth project. I have diagrams for a transistor matching cct but need to measure differences in the region of 50uV.

** The Vbe of a silicon transistor changes by around 2000uV per degree C.

Think it through.

.... Phil

Essentially, that cannot be done. Just look at how much Vbe changes for a measly 1 degree C temperature change. Pick up a transistor that you "matched" and measure again; the temperature change from handling will kill the "match".

I used to make opamps from discrete transistors, well, a long time ago. It's not too hard to get below 10 uV/degC with a pair of TO92s glued together.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation

roject.I have diagrams for a transistor matching cct but need to measure di fferences in the region of 50uV.My digital meters lowest DC scale is 200mV. I know you can get more sensitive DVM's but these seem to be quite costly a nd would defeat the object of matching your own transistors.Is there a way of increasing the sensitivity of my meter.At the moment the lowest it shows is 0.1mV, another decimal place would be better. Cheers Colin

That's a requirement from someone who doesn't know anything about electroni c design...

I don't suppose you know of a selection of discrete opamp designs that can be reused? Might be useful for 3rd world electronics.

NT

Many thanks for your helpful input, Mr Bloggs.

Well, you assume great care taken when measuring each separate transistor. With experience,it is not too difficult - but the way the OP submission read, my take was the proposed measurement scheme would leave a LOT to be desired.

Check.

I recently bought some ADA4528-1 devices. The first one I soldered down seems to have an offset of less than 100nV. I built a x101 amplifier inside a piece of copper water pipe, to keep it all isothermal, and the offset is too small to measure easily (

The OP-07 maybe a nice chip to work with but, they still supply a NULL set of terminals to balance things out.

We use those in some older precision circuits due to their stability. I know newer types can do better these days but even so, we still need to employ the null circuit to trim it.

Also, I seem to remember the supplies +/- need to be very closely matched if you don't have a method of balancing.

I'd be curious as to how you are supplying the rails to this circuit that is inside a copper pipe?

Jamie

Thank you for the link to the site. Ray Wilson's VCO

looks gorgeous (at least compared to my far older DIY VCO). It almost looks like copper with decals and a layer of varnish. If it is indeed copper it is a fine example of what I call "cupric design language."

Wilson uses an N channel JFET and a comparator to generate a ramp that functions as the heart of his VCO. He also credits three guys for his ideas.

These ideas were pioneered by the Alan Pearlmans, Bernie Hutchins, and Robert Moogs of the world I am merely a student of their landmark work.

My own older VCO was bought as a kit. It used a UTJ as a relaxation oscillator to generate (you guessed it) a ramp that functioned as the heart of its VCO.

I only recognize Moog's name in Wilson's list. No doubt the original idea to use a ramp wave as the heart of a synthesizer VCO goes back to one of those three guys. And that's the way it's been ever since.

Thank you again for the link.

--

Don Kuenz