Modifying the Mullard High Speed Valve Tester to read gm

[David Simpson]

Aye Lawrence, the BBC with their VT/4 were clearly initially worried that the basic DC formula gm = delta Ia/ delta vg was only true if the Valve's HT remained constant, instead of dropping as more Ia was drawn over the whole slope. Hence the use of a small signal being applied to the grid at a fixed Vg & a fixed Va.
RoeTest DC results are, I understand, obtained by a constant maintenance & monitoring of Va over the whole plotted curve. As are mine(using a built-in Marconi TF1041C Valve Voltmeter). I suspect that the BBC, ACWEECO, and others found this too burdensome in having to fit so much extra HT circuitry.
I've observed the dropping-off of Va in my Mk3 and other CT160's, when plotting graphs, at the top end of max Ia, with high slope power valves, thus compromising Ia - due to the mains T/F's secondary VA & Vs windings inability to cope. But then, for "Test" procedures, AVO VCM's & others, only give a Vg which works the slope much lower than someone plotting a graph.
So, for low & medium power amplifying valves, and big thumpers working pretty low on their slope, the signal method for gm will work fine. By measuring either the P - P or RMS values of a number applied grid signals & their corresponding anode signals & plotting a graph - hopefully the result will be the same as using the Ia/Vg method. Also, monitoring Va, I reckon, is essential in any homebrew tester.
Electronicing is great fun, Peter.

Regards, David

[Electronpusher0]

I mentioned that I had built and tested the oscillator.
I first simulated it in ltspice, then built it.
I attach the ltspice schematic, it will require a +/- supply, which unfortunately has to be a separate one to the one used to power the meter as the latter is floating at HT!
I am writing up the design as I go and will post a single document at the end with the info needed to repicate the mods if anyone wishes.

Regards
Peter

[Electronpusher0]

Originally Posted by David Simpson

Electronicing is great fun, Peter.

It certainly is!, it started as a hobby for me in school with the radio club, then became my career as design engineer. Inevitably I drifted into management but since I retired it has gone full circle and is a hobby again (including valves of course!)

Regards
Peter

[Electronpusher0]

re the oscillator, I should have mentioned that the 2 off 2.5K resistors represent a 5K pot, the junction of the resistors is the wiper. This is used to set the 100mV rms level.
Peter

[David Simpson]

Great job Peter, your professional background shines through. With your graphics skills, I bet you could easily rustle up a family of graphs, with software, for your tester once its finished. Old analogue me just plods on with hand plotting.

Regards, David

[Electronpusher0]

Problems, Solutions and extra functions.
With the wires brought out to the front of the tester and the covers back on I made some more tests and measurements.
This showed up a small but significant ripple current in the Anode resistor, this was predominantly 50Hz with some 100Hz components and is present on the HT line. This created a 10 to 20% error in the gm measurements as it was superimposed on the 1kHz signal I needed to measure. The magnitude of the error changed with the valve type, it was worse on power pentodes.
The obvious thing to do would be to eliminate the noise signal at source by adding further decoupling, however Rule 1 applies, “Nothing is to be done to the tester that prevents it being restored to its original condition” so I decided to extract the 1kHz signal from the noise.
I therefore calculated the components to make a single stage bandpass filter centred on 1kHz. I built this and tested it using 2 9v PP3 batteries for the supply since the filter circuit would be floating at HT potential.
This was successful in reducing the 50 /100 Hz component significantly. However I have decided to use 2 cascaded identical filters to remove as much unwanted signal as I could.
Since I was now committed to at least 2 op amps I needed a +/- 12 or 15V supply. I therefore decided to add a precision op-amp rectifier, since I had the supplies ready, and thus remove the need for a special 200mV ac meter, as I could now use a common dc voltmeter.
At this point I started to get worried as I now had a circuit of increasing complexity and associated power supply all at HT potential! A guaranteed recipe for shocks during testing.
I was inspired by reading about an Anolog Devices op-amp that offered a differential input, unity gain amplifier with +/- 600V common mode rejection, AD8479. This would allow me to measure the voltage across the 100 ohm Anode resistor but with reference to 0V. However this is a very expensive part and only available in surface mount. Consequently I decided to design my own amplifier using a TL071.
I modeled the whole circuit, Differential amp, Bandpass filter and rectifier in LTSpice and attach the schematic.
The principle of the Differential amplifier is to divide the input voltage by about 60 to get the voltage within the range the op-amp can handle and then amplify the differential by 60 again to get unity gain overall
The critical thing in such an amplifier is not the tolerance of the individual resistors but how closely matched they are. I decided to use a chain of 3 X 390K resistors in series using MR25 resistors with a voltage rating of 250V each, giving an overall voltage rating of 750V. I have a large quantity of these on a bandolier and used a multimeter to select 12 resistors as closely matched as I could. They were all pretty well matched anyway having been made in the same batch. I likewise matched the 2 X 22K resistors.
I assembled the resistor network for the op-amp “in air” with no pcb or vero bd and then encapsulated the whole network in a small potting box with the high voltage inputs on the top and the op-amp connections on the bottom. I attach a photo showing the resistors before potting.
On testing the network after curing I found the “arms” were matched to within 200ohms in 1Meg, not bad!
I have constructed the power supply and am now building the rest of the circuit, testing can now all take place at a safe low voltage.
I have also decided that since I have Anode volts, Grid volts and Anode current all available outside the tester I will put extra meters in to display these values, I now have 5 digital meters, to display the following:
• Mains Voltage
• Anode Voltage
• Grid Voltage
• Anode Current (dc)
• Gm
I you look at the schematic you will note a reference to 2 BNC connectors outputting Anode Voltage and Anode Current signals, these are marked as for a ‘scope. I am using a larger box than originally planned and have some ideas to add a curve tracer function at a later date……..