A special valve tester

[Ed_Dinning]

Hi Folks, repairing some valve audio gear I found that some nominally new and tested valves gave high levels of hum and distortion (especially AC/DC valves).

How would a simple valve tester that applied the appropriate voltages to the valve and allowed injection of a signal into the grid, at say 1KHz be?

Output could be monitored on a scope for distortion and for 50Hz component (with a modern DSO the waveforms could also be quantified.

Ed

[Radio Wrangler]

I thought The Sussex applied about 1kHz for the Gm measurement.

Something in the anode supply to sense current with an output for a scope would be needed. Either a current transformer, a bit of active magic (with high precision common mode rejecting resistors) or elase a separate anode supply that could be floated to put the sense resistor at ground potential.

David

[julie_m]

Or a HCPL7800-alike device in the anode supply?

[David Simpson]

A "Lineolite" Shaver Transformer is the answer. My LeyMarcAvo DC Valve Tester has one in the Grid supply. I've been carrying out dynamic testing for about 18 months using 1KHz. With the Sec connected in series to the Grid, and the freq source to the Pri. The losses at 1KHz are less than 5%, and the wee T/F doesn't start to drop off till about 1.5KHz.
A much larger 1:1 Isol. T/F is connected in the Anode, again with little loss when subjected to 1KHz, and both i/p & O/P waveforms studied on a 2 channel scope.
Gain "G" (a valve's Vout/Vin)can be studied visually on the scope, and also on my VT's built-in Marconi TF1041C VTVM.
The Pri & Sec of the O/P T/F present an approx. Xl of 14KHz, and a particular valve's Ra can be obtained from a valve data book. A non-reactive resistance box can be used to adjust the overall "Z" . Fannying about with the maths takes a bit of time.
For quite a few O/P valves, roughly expect G to be about 10 - 15% of mu,
(ie.Ra/(ra + Ra) is mostly a low fraction.
For just a "quicky" test, depending on a valve's "ra", - just the 1:1 T/F's o/p waveform will display approx. x2(+ or - 20%) of the correctly matched (Z-wise) "G".
All this needs to be determined by hours of trial & error, and mathematics, and studying an Illiffe Data Book's tabulations for a number of valve's Ra.
Maybe I've just been lucky in my choice of transformers, a lovely big exGPO non-reactive resistance box, etc., but scoped, VTVM'd, and the maths have all resulted with values of Gain "G" within 5%. So I'm happy.

Regards, David

[Ed_Dinning]

Hi Folks, my main intent was to operate the valve at realistic signalpower levels to look at distortion with a scope; I believe the drive levels used in conventional testers are quyite low. I like David's idea with the transformer, I have some LEM transducers so may try those.

Ed

[David Simpson]

I've often thought & hoped, although its way beyond my old-fashioned hands-on analogue skills, that some high tech Forum member would devise a computer graphics program to display a valve's true performance. Signal wise - as Ed requires.
Although it can take ages, one can draw up a graph of a valve's Ia/Vg, and then superimpose a theoretical sinewave on the grid's working point. Then a corresponding Ia sinewave. This could then be compared with actual scoped displays of working Grid & Anode signals. This in fact was one of the main functions of Oscillograph Cameras in the late 30's & 40's. However, its a lot of fannying about. Particularly scaling your graph paper to the graticule V/div scaling, or visa-versa.
I'm not knocking AC VCM's, as a vintage item needing loads of tlc, I love them dearly. The Sussex & other homebrew DC testers are the way forward. But for those vintage enthusiasts who want to put to good use their stock of NOS valves, then perhaps modern technology could help.
I've graphed quite a few valves over the years. Many NOS older than myself. Even taking into account VCM errors, eyesight errors, drawing errors, etc., significant "kinks" often reveal themselves in the resultant Ia/Vg plot. Particularly on the lower regions.(Valves - not myself).
I've never owned or used one - but a keen young electronics engineer once showed me his flat-screen computer and oscilloscope function software. 4 channels and a lovely colourful display. He couldn't understand why I was playing about with renovating Cossor 339's, or even using a more modern Hameg605.
Surely, one of you younger thrusters could devise a program that would help Ed & many others.

Regards, David

[Ed_Dinning]

Hi David, one of the TEK 4 channel DSO's now goes on the second user market for under £600, not cheap but value for money. It would be calibrated and with a set of probes and manuals for that as well. (I have a TDS 3054 bought this way as well as the more normal earlier 400 series Tek)
Quite easy to learn to use and with the FFT software it can reveal a lot about a waveform.
My original thoughts were just to look at the high power AC signal on the scope, but using the TEK would give much more quantifiable results.
The other approach would be to capture the waveform on a PC sound card and do the analysis that way.

I've also posted in the components section to see if there is any theory behind the observations.

Ed

[julie_m]

I designed and built a little amplifier using an ECL80 with the help of a computer!

I used a Velleman K8055 experiment board, with a dual op-amp to invert one of the DAC outputs to apply negative volts to the grid and also to amplify the cathode current to feed to one of the ADC inputs. I wrote a Perl program to measure the current as the voltage was swept over a range, created a file with the data and used gnuplot to plot graphs.

It was slightly less labour-intensive than doing it all by hand, especially as I would only have spent the whole time thinking I should be making the computer do some of the donkey work .....

[Diabolical Artificer]

Saw this Ed and thought of you. It's simple and I think could be adapted to show sinals at high levels - www.valvewizard.co.uk/curvetracer.html

Andy.

[Ed_Dinning]

Hi Andy, that looks a useful bit of Kit, certainly worth thinking about.

However my OP was for something much simpler.

Power an audio valve at its normal DC operating point with PSU's and use a power resistor as the anode load.
Inject a good sine wave into the grid at operating drive levels and monitor the volts (or amps) at the anode
Compare waveforms on a scope for distortion; if you have a modern scope with FFT available analyse output waveforms for distortion products.

A similar test could be performed to look at the output waveform for 50/100Hz products.

This would test the valve under full power Dc and full signal conditions.
Most testers do these tests with low levels of applied signals


Ed

[jimmc101]

If you have a spare computer around why not use the soundcard as the sig-gen and 'scope / spectrum analyser / distortion meter.
Free software such as Visual Analyser will do the job.

Simple buffers on input and output will protect the soundcard from possible transient damage.
Alternatively add a cheap USB soundcard, something like one of these in with the buffers to give an all in one adaptor.

Taking it one stage further using a 2 or even 3 of these USB soundcards connected via optical cables would give complete isolation from hum loops and DUT power supplies (of course floating PSUs would be required).

I've used the software and these soundcards with surprisingly good results but I haven't tried using the optical links.

Jim

[jimmc101]

Sorry I confused myself in my previous post, I meant to say use the optical ports of the soundcard with an A to D and D to A converter to give isolation.

Jim

[David Simpson]

I'm with Ed on this. The Dynamic ad-on unit for my DC Tester, as I've previously mentioned, contains a big 1:1 mains transformer connected in the valve's anode circuitry. Thankfully, its kind to 1KHz, exhibiting very little loss. I've also used thumping big resistors, as Ed suggests. Thus replicating working conditions for EL37's, KT66's, 807's, etc. The T/F would take 200mA, I reckon, but I've only ever gone as high as 117mA.
However, bearing in mind what I've said on someone's "Soak Testing" thread last year, I try & keep full Ia conditions to a minimum. Just enough time to study i/p & O/P waveforms on my scope. Being an analogue rig, its pretty straightforward. However, I can appreciate that the display results would be even better with digital equipment & associated software. Just a matter of personal choice, really.

Regards, David