A Commutating Transistor Vbe Matcher

[mediatechnology]

Its been years since I posted here but my work on a low noise moving coil preamp using the ZTX851 has led me to the subject of transistor Vbe matching. When I thought of Vbe matching I thought of electro-music. So here I am.

I'm certain I don't need to introduce Ian Fritz and his design for a transistor Vbe matcher to this audience as I'm sure his work is well-known here. My design draws on Ian's work but removes the requirement of matched emitter resistors by commutating the emitters of the DUTs into a single emitter load. By using a single common load there are no resistors that can be mismatched.

An additional benefit of commutation is that the DC Vbe error is made AC to simplify low-level measurement using a Soundcard A/D. The peak-to-peak value is the delta-Vbe. This permits direct measurement of the error to very low <<100 µV levels.

This is the schematic for a device I've used to match many pairs of ZTX851:


A Commutating Transistor Vbe Matcher

Circuit Description

A CD4093 is used to provide a 1 kHz oscillator. A CD4013 is used to provide an exact 50% duty cycle for the commutator to minimize errors of self-heating and average Vbe.

A DG413 switch commutates the bases of the DUTs. The RdsOn errors of the switches are minimized by the hFE of the transistors. More important than RdsOn is the difference in RdsOn between switches which is very small typically <<1Ω. For a 100µA Ie and an hFE of 100 Ib is only 1µA and the error of delta-RdsOn x Ib is <<1 µV. The ratio match of RdsOn is what is important. I've taken near-perfect pairs and swapped their base drive with no change in measurement.

What matters far more than RdsOn is the emitter contact resistance of the test socket. I'm using a ZIF socket with a lot of clamping force. 0.1Ω in the emitter represents a 10 µV error when Ie is 100µA.

The delta-Vbe of the DUTs appear at the emitter node as an AC component. A pair with a large error (about 14 mV) is shown here. (A pair with low Vos isn't very compelling to look at since its a flat line.)


A Commutating Transistor Vbe Matcher Showing a High Vos Pair

The top waveform is the delta-Vbe; the bottom waveform is the sync signal coincident with Q1 conduction. Q1 is more negative with respect to Q2.

A OP07 voltage follower buffers the AC component at the emitter and provides an average Vbe to be used to establish cutoff. To reduce switching noise the bases should not be connected directly back to the emitter. A 10µF at the "off" node provides a low AC impedance to reduce noise.

The OP07 stage provides three outputs:

1) An AC-coupled output to feed a Soundcard A/D
2) A 'scope monitor output. (Not level accurate. The 100K build-out provides a crude LPF when combined with the probe capacitance.)
3) An average Vbe of the pair.

There are also two utility outputs:

1) A 15V P-P sync signal to trigger a 'scope.
2) A 1.5 mV P-P signal (1/10,000th the supply V) to use as a calibration signal for the soundcard.

By using a soundcard's A/D (and AudioTester) to perform an FFT on the fundamental AC-component it is possible to read delta-Vbe with <<100µV precision.

Here are a couple of examples:


ZTX851 transistor matching using a soundcard and FFT to read delta-Vbe.

You can read more about the Commutating Vbe matcher here: https://proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=1153

Enjoy;

Wayne Kirkwood

[mediatechnology]

Forgot to mention that I epoxy the transistor pair in 1/4-20 acorn "cap" nuts once I'm done.

The ones shown below are chrome-plated steel that I got at Home Depot. I've found aluminum ones available at Fastenol which have better thermal conductivity.


The ZTX851 Moving Coil Preamp With Vbe-Matched Pairs In An Acorn Nut.

The ZTX851 and ZTX689B (both NPN) have exceptionally-low rbb'. The ZTX851 is about 1.75Ω the ZTX689 about 5.5Ω. They should have good log conformity at high currents but I've never tested them as log amps. The ZTX689B has much higher hFE around 500 or so.

[elektrouwe]

I had a similar idea some years ago:
https://www.muffwiggler.com/forum/topic-153845.html

[mediatechnology]

I like the fact that yours can do more than 2 devices.

Thank you for the link. You should re-post it here. It never turned up in any of my searches.