VCM163 AVO Valve Tester

[Mike Watterson]

I'm fixing up one of these for a friend. Originally I replaced the meters, creating new scales a couple of years ago. Those were big round ones sitting outside panel. Then small modern ones fitted. Then he found a couple of SIFA 50uA meters similar in size to the original. I then made third and fourth set of scales. The ones I did first were not quite right.

So obviously he wanted all the calibration re-checked.

Anyway, at "5" on 0 to 10 scale of grid volts the voltages per range should be about 0.78, 2.6, 7.8, 26. Adjusting RV2 the pot goes to one end of travel to get not quite as high as 2.6, 7.8, 26 on three upper ranges, but then the lowest range is over 0.9V. You have to put pot to nearly the other end of travel to get 0.78V, then the other ranges are far too low, like about 2.4V and 23.5V.

The resistors on the range switch seem OK. We replaced the 60uF 250V smoothing cap on the Grid supply. We replaced the 10K wirewound pot as it was bad on the first third of the range, intermittent connection, so needles dancing. With the new pot the voltages are the same, except it's now stable on the lower third of the scale.

No doubt the 0.52 ratio to actual volts from scale setting is to do with the cunning way the AVO works. The voltage at the junction between RV2 and the resistor dividers etc is supposed to be -52V. Why is the ratio 0.52 rather than 0.5 or 0.707? I must re-read the Radiomuseum article.

I traced the schematic for nearly an hour last night examining how the AC modulation from the oscillator is added to the slider of grid voltage pot (RV4) on panel and gets to the thumbwheel switches / valve sockets. The final switches & sockets are not on the schematic.

I THINK I might have spotted the faulty part, I'll verify next time I go over. Probably next week.

Anyone else any ideas or encountered this issue of mismatch between 3, 10, 30 and 100 grid voltage ranges?

[Mike Watterson]

Worked on this most of today.
1) The lack of tracking on 3 / 10 /30 /100 Grid voltage range:
I realised there is a 15nF cap to decouple the gm measuring Osc signal on to grid voltage. It was on the mA/V gain switch and a Hunts! Leaky, the only Hunts in the entire machine. Doesn't look like a "waxy". Plastic sleeve. Replacing that with a Polyester(metpoly) fixed that.

2) Calibration of AC mains.
In the absence of a variac and/or stable 240V AC the solution is to measure AC on the transformer used for Anode, Screen & Grid. It's fed from taps on the Heater transformer that has the switch and fuseholder. So put fuse in holder that allows front panel switch to be closest to mid point. In theory the 230V place, but my friend's house is rural and voltage drops a lot. It needs 220V position though here is 230V or a little more. Then adjust RV3 so left meter needle as at Cal while watching mains across the transformer kept at 240V by front panel switch. The 60uF cap isn't in circuit during normal test.

3) Calibration of Grid supply.
You need an AVERAGE reading meter. Analogue works by default, a DMM might work. A True RMS Fluke will give "wrong" reading! With 240V on the Grid/Anode/Screen transformer (see point 2), adjust RV2 for 52V where it feeds the resistors on the 3/10/30/100 range switch. Both sides of the pair of 2M2 resistors are handy. Then on 100V range put cursor of RV4 pot so when at "5" there is 26V. The cursor will slightly go past 0 and a good bit past 10. Then as you change switch you DO get 0.78, 2.6, 7.8 and 26.

While we were doing this the voltage dropped to 47 V approx and would not adjust! Intermittent connection on the wiper of the pot. Replaced the pot.

Note that RV4 grid voltage on front panel must be 10K 1% or better otherwise the calibration won't work. The pairs of resistors on 3, 10 and 30V ranges in conjunction with RV4 always give a 10 K Ohm load to the grid supply. On the 100V range the load is the 10K pot. Why is it -26V when scale is -50V, because the Screen and Anode are half wave raw voltages, no smoothing. So the Grid voltage on the "Test" position is negative half wave. Hence the -26V AVERAGE reading on a meter is equivalent to testing the valve at -50V. The calibration of the scales isn't the actual average Anode current a test meter will see, but the equivalent. So no actual voltage or current reading will match the knobs or readings. A true RMS meter is even more misleading than an average reading meter and can't give the readings specified in the Service Manual.

Next visit I will add the connector for the Tektronix curve tracer. The AVO VCM163 will provide valve bases, screen voltage, g3 voltage (if wanted and not 0V) and heater supply. The Tektronix has a Depletion J-FET mode, so bias is up to 0 to -20V and HT is up 1500V at loads of Watts! The Tektronix had dodgy switches, which operating and cleaning has mostly sorted, so I've managed to test n-jfets, enhancement MOSFET/VFET and regular bipolar NPN on it. It's very heavy!

[Craig]

Dennis Tillman (he's in the US) has produced an excellent description, and supplies a small circuit board to adapt a valve tester to connect to a Tektronix curve tracer. That uses the transformer taps in the valve tester to produce heater, screen grid voltage and and control grid steps via an amplifier from +10V to -50V.

Blurb is here. http://www.ke5fx.com/An%20Inexpensive%20...racers.pdf

I've just taken delivery of one, and plan to use an AVO MkII with a stuffed meter as the valve tester mule.

Craig

[Mike Watterson]

It's not complicated. All that's needed is connections to the ends of the rotary switches for the heater and screen. The Tektronix can do the grid (to -19) and Anode. Add a 4 pole change over switch and a socket beside the rotary switches. I'm using a 7 or 8 (forget which) way connector of the style used on CB and Ham rigs.

Perhaps some of the Tektronix testers don't do Depletion mode N JFETS, like my friend's tester that has 0 to -19V for gate and over 1300V for drain.
No amp needed. Also unless the Tektronix is doing the control grid and Anode (or screen, you can swap) it's not a full curve tracer.
Admittedly some valves do need less than -19.

There is an issue that the screen voltage needs smoothed and regulated as it's half wave pulses.

I've downloaded the manual / blurb and the only thing in it that's needed (for only a few valves) is the amplifier:
"Semiconductor curve tracers do not have voltage steps large enough to drive a low-mu triode from full
conduction to cutoff. The VTCT adapter includes a Grid Voltage Step Amplifier that has gains of X1,
X10 and X100 and an output range of +5VDC to -50VDC for this purpose."
Except the Tracer in question set up for N JFETS seems adequate.

I guess the design is to accommodate inferior Tektronix curve tracers and inferior valve testers.

Valves should be fully tested for shorts & leakage, cold and hot, then tested normally for Ia and gm before curve tracing. The Curve tracing is pointless for restoring gear and the best test of valves (quoting Tektronix) is in the intended equipment, after verifying it has the correct voltages etc.
Valve testers were needed in factories or wholesale for QA, or big service depts. They aren't actually needed by a restorer of a radio or amp.

The curve tracing is really only of use for designers or the curious.

[Craig]

Not sure what you mean by "Inferior Tektronix Curve Tracers". But whatever, I was just attempting to be helpful.

[Mike Watterson]

Not sure what you mean by "Inferior Tektronix Curve Tracers". But whatever, I was just attempting to be helpful.

Earlier ones that don't do -Ve gate with +Ve drain, ie BJT only, no depletion mode FETs up to -19V in and +1400V out! I did have to read the manual twice to realise that there was a magical combination of settings suitable for valves.

No, I appreciate the info and I've saved it. Interesting review of valve testers and curve tracers. The circuit he has shows a 0 to +5V input (or even 0 to 0.5V) for +1V to -50V out which means it suits ANY transistor curve tracer.
Obviously with almost all valve testers the most useful bit is the heater supply, though a bench PSU is possible. The 2nd most important thing for tetrodes, pentodes, hexodes etc is a +HT screen voltage. Few people have a HV bench supply. I keep meaning to make a variable one for up to 1000V. Or at least 250V using and old radio transformer HT winding. Though a shaver transformer will do 250V @ 15W, or about 500V using a doubler.
I do have a fixed 85V HT (smoothed but unregulated, that's at 10mA load) test unit.

I have Dekker's uTester which does valve curve tracing and can do Depletion mode FETs. It's not good for heater supplies below 6.3V generally and mine is worse because the large number of sockets are AVO style wired. I should have had a single connector and plug-in adaptors. I will probably change it so that I can also have a plug-in for LC feedback to create oscillators and measure mixer/osc conversion gain and LO leakage at RF grid. It's because the uTester uses PWM of a laptop 19V to 20V PSU for the heater. It's also unsuitable for direct filaments, so I use a bench PSU or a separate 0.7 to 2.2V supply. It only provides pulse voltage for screen and anode via 2x SMPSUs from the Laptop PSU, however continuous current is enough for a Magic Eye. There is a mod for the uTester to do +Ve grid, needed for valves used at 13V HT in car radios and to test enhancement FETS or BJTs.

The Tektronix uses continuous voltages so a 10 amp 1200V FET or BJT needs as much heatsink as the application! The uTester can test transistors at hundreds of volts without a heatsink as  it's using pulses, if they are depletion mode, or any N type if the positive grid modification is added. Unlike the Tektronix, the uTester is inherently unable to provide -ve voltages other than on the grid socket.
   
I've had it a little over 6 years. It's rarely used.
The supplies from the uTester and separate low voltage filament supply are the 2mm sockets on two rows, bottom right.
Small stackable patch cords are used to the circular array of sockets and two sockets for top caps that are wired in parallel to all the valve bases and TC sockets. All the connecting wires are in a complete loop like an AVO panel with ferrite beads on the wires. That's AVO's theory to avoid oscillating valves.

[Craig]

I guess you are thinking about the 576 curve tracer. With a heater and screen grid supply I agree that this superb tracer can do valves, and quite a number of people around the world have done that. It is a fine instrument.

But the 1960 575 (a semiconductor curve tracer that uses valves internally!) can do opposite sense collector/drain and steps. And if you get the mod 122C version that adds a voltage range up to 400V collector. The only downer is that the base steps that only go to 0.2V/step (so 2.4V maximum). Although you can get higher voltage steps by (1) setting base current steps to 1mA (2) connecting a 1k resistor across the base-emitter terminals. That gives 10 steps of 1V. Still nothing like the flexibility of the 576, but certainly better that using voltage steps.

I have a 90% finished uTracer; must get around to finishing it.

Craig

PS I have a 575 with 175 high current unit, a 575 mod122C, a 576, a 577D2 and a 7CT1N. Plus an AVOII (the valve curve tracer mule), AVOIII (the most often used), AVO CT160 and Taylor 45C. Pass the pills, dear chap....

[Mike Watterson]

:D

Space can be an issue!

[Craig]

I have a long suffering wife, a lockup, and the kids flew the coop quite a while ago - so there is a bit of overspill ;-)