Matching Transistors - Electronics Question

[Tim Hodges]

Hopefully someone can shed some light on this.

When matching transistors for Fender Rhodes electronics. Which is more important to match?

• HFE?
• VBE?
• Both?

The posts I've read elsewhere have say that it depends on the use of the transistor. In this case these will be the ones used in the power amp and also the preamp (not model specific).

Thanks

Tim

[Tim W]

This can end up being a pretty deep discussion of electronics.

But, for the most common Rhodes pianos, here is the deal:

Peterson Preamp:
Matching does not really matter except for the vibrato oscillator transistors, to keep the vibrato even.  But even here it is not that critical since the Rs and Cs that make up the oscillator are more likely to have drifted or have wide tolerances.  The audio part of the circuit is single ended.  No matching necessary.

Janus Preamp:
No transistors to match.

Peterson Power Module:
Output Transistors *ideally* will be matched for both Vbe and Hfe.  But no basic multimeter will run them at real world currents to get useful measurements.  One would need to construct a test fixture or use a curve tracer to properly do it.  In lieu of that, installing two transistors in an amp module and powering on into a dummy load will give you an idea of the matching.  Usually new devices from the same lot or date code are close enough.  We look for a low DC offset on the amp output and fairly symmetric clipping just at the onset of clipping.  Again, into a dummy load.  If both DC offset and symmetry track well when the amp is cold and hot, the devices are matched well enough.  Again, here, we also depend on the tolerances of the bias resistors and driver transformer.

Janus Power Module:
Small signal transistors Q1 and Q3 should be ideally matched, but there is an offset adjust pot to trim DC offset.  So matching isn't that critical.
Having matched output devices would be also ideal, but again, with an offset adjust it can be nulled.  But you are adjusting the offset of the composite circuit, not just the input or output devices.  So one needs to look for symmetry and again stability over temperature, like in the Peterson circuit.   

Super Satellites:
Input Diff Pair (Q1 and Q2) would be nice to have matched.
Output devices (Q6, Q7) and predrivers (Q4, Q5) for low DC offset, but there are trimmers here too...
This amp design is not particularly good by today's standards- be careful subbing in modern power devices for the older, slower ones.  The amp may oscillate at HF and overheat.

Good luck,
Tim

[Tim Hodges]

Thanks Tim, that's a perfect response and exactly what I needed.

In terms of building a matching circuit I did have a look at Ian Fritz's PCB and also the design by Bob Moog (where the schematic is incorrect) but I decided to buy an Peak DCA55 which should do the job for me (also it only cost £20 from Maplins instead of £48)

All the best

Tim

[JanneI]

For synth diy work I have used this measuring technique (VBE):

http://www.dragonflyalley.com/images/TransistorMatching/ianFritz-transmat0011_144.pdf

But I'm not sure if it's needed here. In synth diy these are used in the expo convertors,  in "volts per octave" -> current. For example, precisely controlling voltage controlled oscillators or filters, etc.

[Tim W]

The Ian Fritz or Moog test circuits are good for small signal transistors in low power applications, but won't realistically work for power devices.

Unfortunately the Peak tester also will not realistically work well for power devices.

To get a good idea of matching for power devices, you need to put them in a circuit that powers them with lots of volts and amps- the region where the device normally operates.  No battery powered multimeter or mini-analyzer will do this.  For where those testers exercise the devices (very low power) you may be able to match them, and if you are lucky, that will also translate to matching at the higher power operating regions.  But, it is unlikely.  You will need to use a curve tracer with some real teeth (Tektronix 576) or build a custom power circuit with a big power supply to test for matching power devices.  The Rhodes amp module itself is the best tester you have, since it is actually the real application.  Careful exercising of the circuit through multiple operating conditions can give you the data you need to determine if the transistors are matched closely enough.

To sum it up:

Just because the car runs fine at 20mph doesn't mean it won't fall apart at 70mph.

Best
Tim