ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

TUBE MATCHING PROJECT

:

9/16/1997 7:52 PM

TUBE MATCHING PROJECT

I came up with an idea for a tube matching contraption today that I wanted to get some feedback on. It's just a guitar amp power supply(heater, B+, Bias) with several octal sockets. No output trans. The plates would hook up to the B+ taps you'd normally tie the output center tap too. 1 ohm resistors on pin 8 of each socket to get a current reading from. 1K screen resistors. All the grids running to the bias supply with their own grid leak resistors. I'd like to find a way so I could switch the B+ by about 50 volts so I could get my "delta"ep. By being able to change Ep along with Eg, I should be able to figure gm, AC plate resistance, Rp, idle current and amplification factor. Any problems hooking the plates up like this? I'm not sure if running several tubes at once would interact and give me strange readings. I could use a variac to change B+, but I'd like to keep it self-contained so I was thinking of using a switch to take a dropping diode in an out of the circuit. Any thoughts?

9/17/1997 9:45 AM

Something Nathan Stewart and I discussed at length was a tube tester that was a kind of computer driven tester that could vary the applied voltages and measure the resulting voltages and currents. This would offer the ability to do a more in-depth testing than just manual meter reading.

This needs a few things to work correctly - obviously an ability to do A-D measurements and D-A output from the computer, and some outboard circuitry to cope with the conversions of the higher tube voltages to more computer-friendly ones. This can be simple, probably 8 bits of resolution is all that's needed, and also it can be fairly slow. I think that two channels of A-D and two channels of D-A would do it.

This kind of setup would allow you to match tubes on a dynamic as well as a static basis, as well as doing more exoteric things like running a full curve trace on a per tube basis.

What's needed?
- a high voltage D-A; this is probably best addressed as a B+ voltage regulator with a reference voltage driven by a garden variety 0-5V D-A
- a low current A-D with overvoltage protection, which would read cathode current through a cathode sense resistor
- a low voltage D-A providing a negative voltage to the grid of the device under test, possibly as much as -30V or so.
- A high voltage A-D to read actual plate voltage, probably just a resistive divider on the plate.

That would cover triodes. You'd need some way to measure screen voltage and current, the hard one being screen current.

Since these are all being done essentially at DC, sampling speed doesn't matter that much.

9/17/1997 11:15 AM

For what you want to do, what you describe should work just fine. Not bad either, the ability to characterize tubes with not much more than a power transformer and a volt meter.

There's nothing wrong with connecting the plates as you described, however, you would need to be a little careful on how things powered up. You want to make sure that your bias voltage is there before the B+. If the bias is being generated by a voltage divider, it will probably have a time constant associated with it. So, you would want some sort of 'Stand-by' style switch to allow you to set the bias voltage way negative before applying the high-joltage.

Switching a dropping diode in and out would be an easy way to get a delta on the Ep however, you probably wouldn't want to switch it in and out while current is being drawn. The standby switch would help here.

A thought I just had, if you wanted to check how
well matched a pair of tubes are, you could connect the volt meter across the two Cathode resistors and vary the bias voltage. If the tubes are perfectly matched, you should read 0 volts over the entire range, if they are not you will read a differential voltage between the two. This would allow you to quickly check the matching over a continuous range. Unfortunately the tubes are referenced against each other so it doesn't really give you quantitative data for matching tubes. It would only tell you if they are matched.

If you do build it, I would be interested in hearing how well it works for you.

--john

9/17/1997 6:01 PM

Re: Thanks, but...

It looks pretty easy on paper, but did you guys see any problems when trying to test several tubes at the same time? I can get each tubes current from the 1ohm resistors on the cathodes and the plate voltage from the B+, but I was just thinking if I would need to check tubes one at a time to get things right. One other thing that I thought of today would be it sure would be nice if I could take the B+ down to 300vdc so I could also use it to check 6V6 and 6BQ5 tubes. I was thinking of useing a fil. trans with its own AC cable so I could keep that steady and hook the mains to a variac, but I would still like to make this thing just one box without any other connections needed. Ideas?

9/17/1997 6:42 PM

There shouldn't be any problem with testing several tubes at the same time. The differences in the tubes will show up on the Cathodes. It's not much different than having two resistors of different values in parallel. The current through the two will be different but the voltage drop across them will be the same.

The only problem I could see is drawing too much current from the power supply but I'm sure you would make sure it could handle it.

Taking the B+ down internally will require some sort of regulator. A hefty one at that. If you are wanting to drop 100V or more it's going to be generating some heat and you have to come up with an adjustable control voltage. You will have increased the complexity considerably, however, it can be done.

Ideally you would want a multitap power transformer so you could just switch from tap to tap and get the voltage you would like. Someone gave me one of these several years ago that their father-in-law used for a purpose very similar to this. That would make it self-contained, otherwise your option is to use a variac as you mentioned.

--john

--john

Page 1 of 1