VCM 163 setting up procedure.

[Radio Wrangler]

Thanks for the nudge, Bazz.

The VCM163 applies a sinusoidal modulation on top of the grid bias voltage to the valve under test. It then uses a transformer-isolated voltmeter to measure the small voltage this modulation of the anode current generates across a rather low value anode load resistor.

The problem is that the valve is already powered by AC, the good old 50Hz mains.So there is already huge (100% ! ) modulation on the anode current in addition to that added for the Gm measurement. The fix for this issue is simple, set the Gm test modulation frequency significantly higher than 50Hz, and use a tuned amplifier in the voltmeter circuit. Easy to say, and quite easy to do.

The tuned amplifier isn't quite a highpass as most people would think of one, tuned circuit filters have the resonance Q effect. They have a peak. With the topology used, there will be no response down near 0Hz (well, there's a transformer blocking DC for a start) The characteristic will fall more rapidly and further on the low frequency side of the bandpass peak. This is good, this is what we need. Maximum attenuation of 50Hz, from a simple circuit. Could it have been a plain highpass? Yes but the use of a resonant bandpass lets a low-order filter fall faster (proportional to its 3dB width, rather than the inversely proportional to its cutoff frequency you get with a trad highpass.) Also the limiting of the sensitivity of the voltmeter to higher frequencies reduces the amount of noise it picks up and adds to the reading and it gives some protection against instability effects, redio pick up and other things that go bump in the night.

The frequency of the Gm test isn't precious, we can move it around a bit and not get into trouble. The resonated transformer has two capacitors. One is heavily damped by the low value current sensing resistor. This resonance will be low Q and rather broad. This capacitor value will be the less touchy of the two. The other capacitor will be controlling a tighter resonance, so it's worth being fairly close to the ideal value, but not busting a gut over it.

Anyway, the stray C in the windings of a new transformer will shift things a bit, so some cut and try with parallelled capacitors would be needed to get things on song (meter reading peaked for a defined test valve and carefully maintained settings).

Method B is to say that the detector tuning is a bit off, and to fiddle with the oscillator frequency to set it on the peak.

What is important is that the source and the meter are tuned in agreement with each other. As they are two boards, make sure they stay as a pair.

So, I wouldn't get too upset. Make up something close with E12 values and two or three standard parts in parallel, then twiddle the oscillator to peak it up. Let the thing thoroughly warm up and check the peak. After a couple of months to settle, check the paek again and you'll know whether things are stable enough.

It doesn't have to be terribly narrow (that would need greater stability) it just needs to be narrow enough to get enough rejection of 50Hz.

Don't worry and over-think things just now. Get something going and then do a few tests to look for trouble. There isn't likely to be much.

David

[Bazz4CQJ]

Thanks for the comments David. I'm in the fortunate position of having a spare transformer, made by Ed after he had rewound HBWOODY's item.

What I think would be interesting to do would be to create a PCB based upon the input circuit of the amp and carrying all the relevant components up to the first transistor. I'll make it rugged so things can be swopped and changed as required. Then we can think about how to evaluate what's "adequate for purpose" (as we oft said during my short time at Plessey!).

I'll order the 1uF polyprops from China; by the time they arrived, I'll try get to a point where we can test.

If anyone has ideas about the way forward with this, please feel free to chip in .

B

[Dekatron]

Originally Posted by Bazz4CQJ

Thanks for that info Martin, but is it necessary? I guess they would have a minimum quantity level, so that would be an issue.

Is there a real problem with two off-the-shelf polyprop 1uF caps?

B

He only ordered 10 capacitors and they were just 10% over the regular cost as they just selected capacitors from within their usual range, 1.99uF would probably fall within their 2uF +/-1% capacitor range.

You can of course chose any capacitors you want and just let them add up to the specified value of the original circuit.

[Dekatron]

For those interested the design of an almost identical oscillator and amplifier circuit can be found in the AVO TT-537 Transistor Tester with full description of its design and inner workings, either look for the original manual or in chapter 6.6 of my AVO circuit comparisons document here: https://www.vintage-radio.net/forum/...9&d=1342875214 and here: https://www.vintage-radio.net/forum/...5&d=1343473814

There are some small differences but the description does explain how AVO thought when designing these circuits.

[Bazz4CQJ]

Originally Posted by Dekatron

He only ordered 10 capacitors and they were just 10% over the regular cost as they just selected capacitors from within their usual range, 1.99uF would probably fall within their 2uF +/-1% capacitor range.

Ah, so they did not make specials (which I'm would have been very expensive) but supplied their standard 2uF selected to suit.

I ordered some BC549C's and hopefully they will be here before Xmas, so I'll start re-populating my two boards. I've now designed the "Filter Mock-Up Board" to play with. Whether it's vital to have 1.99uF (_+2%), I'm not sure.

Wrangler has suggested that we should consider starting a new thread (as this one is specific to HBW's project), which could be titled something like "Maintenance of the VCM Circuit boards" (?) as a place to discuss generalities; I think that would be good; how say you?

B

[HBWOODY]

I think a new thread "Maintenance of the VCM Circuit boards" would make sense.

[Bazz4CQJ]

Ah, the 163 boards! I didn’t start a thread because some related work I was hoping to complete has been going very slowly, so no results yet.

The two boards have not aged well. So we are not just replacing electrolytics but other components too, including transistors. I think that Martin has more experience in maintaining the 163’s than anyone, and he is of the opinion that we simply do the work of re-populating the boards. Over a period of time, I have wondered about a project to replace the boards with new circuitry. There are some significant pro’s and con’s with this idea and in recent times, I’ve probably been moving more to adopting the strategy favoured by Martin.

If there were a few owners interested in pursuing the ‘new circuitry’ approach, I’d be happy to participate, but I’m not sure whether I’m going to do much more on this just working alone. It occurs to me that my 1968 “pass” in the Radio Amateur’s Exam, may not qualify me for the task .

B

[Dekatron]

In my experience the unobtainable components like the thermistors and transformers very rarely break, the only damaged ones I've come across were accidentally damaged by people refurbishing their boards - and my transformers which I unwound to get the details from. Poor storage resulting in corrosion is not good for any electrical components but it seems it damages the rotary switches and meters more than the circuit boards.

Replacing resistors/trimmers, capacitors, diodes and transistors is usually enough to get a fully working oscillator and amplifier board. At first I tried to find the faulty component but nowadays I replace them all as it is cheap and it will probably keep the unit working longer until a new fault appears than leaving old components in place.

Making new boards is interesting when we come to the rare cases when the thermistors and/or transformers have failed. Right now I think that the thermistors are the hardest to replace, we know how to make new transformers as we now have all details of the original ones: https://www.vintage-radio.net/forum/...&postcount=157. All of the manufacturers that I have spoken to say that they don't have any direct replacements, partly due to that the way of defining the working points for thermistors have changed over the years with the old resistance normally defined at 20°C is nowadays usually defined at 25°C which makes it hard to find thermistors with the same range/curve (i.e. the STC/ITT original A14 thermistor have values defined like 10K at 20°C, 8.25K at 25°C and Rmin at 75 Ohm and a B-value of 3400K whereas the closest modern type that I have been able to find has a resistance of 10K at 25°C and a B-value of 3250K, or one with 20K at 25°C with a B-value of 3450K - paralleling two 20K and shunting them with 47K is the best match I've found so far). The simplest solution that I can think of is making a design where each new thermistor is adjusted with resistors in series and parallel to adjust for the resistance of the thermistor, that will keep its curvature almost identical to the original ones - unless someone can find a better match somewhere. I guess simulating it in LTSpice is a good way forward for the thermistors.

[Bazz4CQJ]

I suspect that when Avo was designing the 163, they anticipated that one of the major customers would be the military services. The military would probably want them to operate over a wide range of ambient temperatures (you may have the spec) and Avo met that need by utilising the second thermistor which controls level (the other one controlling feedback).

In a situation where (I would guess) most users now operate their 163's in a much narrower range of ambient temperatures than the military, might it be possible that the specs for any thermistors do not need to be matched exactly with the originals?

Alternatively, let's consider a situation where someone decided to make a completely new oscillator for use in the 163, realistically, what range of ambient temperatures would it need to meet? I guess it would be spec'd in terms of maximum % variation of Gm reading between Tmin and Tmax? I wonder if it could be a "non-issue". Many published circuits for Wien oscillators use bulbs of one kind or another rather than thermistors in the feedback loop. I think there are some circuits using op amps which do not use any discrete thermistors.

B

[Dekatron]

I see I was a bit unclear, I meant that it is hard to replace the thermistors in the original circuit.

Making a new circuit with modern op amps will probably not need much to have stable signals over a more "normal" operating temperature range. I've been fiddling around with the Wien Bridge circuit designed by Jim Williams, see page 29 in app-note AN43F here: https://www.analog.com/media/en/tech...otes/an43f.pdf .

What's also interesting in that app-note is the "Diode Bridge-Based 2.5MHz Precision Rectifier/AC Voltmeter" found on page 32 for the amplifier portion.

These two circuits might be overkill but they work really nice, which of course is an understatement when it comes to circuits designed by Jim Williams!

I did build these circuits many years ago and fiddled around with them and they work very well in the 163, but at that time I didn't know as much about electronics as I do nowadays (which is far from as much as I would like to understand) so I never passed beyond the experiment stadium, might be time to dig them up if I can find them and try them out again - however there are probably a hundred other more pressing projects to finish first.......

[Bazz4CQJ]

Originally Posted by Dekatron

I see I was a bit unclear, I meant that it is hard to replace the thermistors in the original circuit........

No, in my first two paragraphs, I too was talking about replacing the thermistors in the original boards. The big question is how close a match do they have to be? And it may be the case that depends on what range of temperatures do you want to work within.

I agree with you that this a problem that ill-defined, and it's difficult to commit to spending what could be a lot of time to deal with it .

B

[Radio Wrangler]

Some thoughts:

Thermistors for Wien bridge audio oscillators are a general problem. All the favourites are no longer available. Filament light bulbs are on their way out too.

But for the Gm oscillator in the VCM163, why bother with a Wien bridge? Why not do a square wave oscillator and then an active bandpass filter? No thermistors, no settling time or bounce. A couple of poles will make as low distortion as that Wien bridge likely did. You can also get beautifully stable amplitude.

David

[Bazz4CQJ]

Originally Posted by Radio Wrangler

But for the Gm oscillator in the VCM163, why bother with a Wien bridge? Why not do a square wave oscillator and then an active bandpass filter? No thermistors, no settling time or bounce. A couple of poles will make as low distortion as that Wien bridge likely did. You can also get beautifully stable amplitude.

David

Interesting comments. I had not thought of a square wave oscillator and filter. Earlier I posted the comment that, "It occurs to me that my 1968 “pass” in the Radio Amateur’s Exam, may not qualify me for the task", referring to the design input for this task.

And this is where I am on this, that I'd be happy to collaborate in such a project, but I'm sure it needs someone much more skilled in the art than me to lead it.

B

[Radio Wrangler]

There are many ways of doing it.

The worst way is to get indecisive and never manage to pick one. There is no 100% right way for most jobs. One that's good enough will do nicely. I can get myself in a fix, vacillating between several candidate approaches with no clear winner. I've learned to spot the situation and to kick myself into action. If the damned things are so similar that I can't choose the best, then just maybe it doesn't matter which one I choose?

Sometimes the way out of an engineering quandry is to toss some coins. Don't let the boss see you doing it though.

OK, I'd use a canned crystal oscillator, one of the 5v CMOS jobs. Then I'd have a chain of CMOS (74HC family) dividers down to the wanted frequency.... 10kHz is it or 20kHz? So the crystal oscillator (a couple of quid) gets chosen so that the division factor is easy... a round binary number times the wanted frequency would be good.

Those little crystal oscillators are a couple of quid, far cheaper than a thermistor.

The power consumption is so low that a 3-terminal regulator off of the supply that previously fed the wien oscillator should be fine.

Designs are done in stages, so get this going, then we'll move on to filters and driver amplifiers. Finally, we'll add some protection so a grenading valve still leaves you a working VCM.

You can't now twiddle the oscillator frequency to suit the tuned amplifier for the meter. No problem, we'll just have to arrange to peak the tuned amp. BUT we now have an accurate frequency oscillator, so the probable disagreement is going to be less, and more stable to boot.

David

[Dekatron]

I'm taking the other approach with trying to find components that will make the original circuit work.

Hunting for thermistors wasn't that hard but finding types that would fit both the STC/ITT A14 and R14 doesn't seem to be possible, however using types with similar parameters and putting a few in series and parallel will accomplish something that is not far from the original types when it comes to the resistance curve.

I think that it won't actually do that much with these minimal differences as we are looking at a limited span of operating temperature for normal use, say 10-40°C where these thermistors are quite linear in their resistance change compared to lower and higher temperatures.

Simulating this in LTSpice XVII with models from Vishay shows that we can use five (5) 10K thermistors in series and then paralleled with an identical 10K thermistor to get a thermistor that results in one thermistor that have a resistance/temperature curve that is not far from the original A14/R14 types even though the curve differs by a small amount as long as we use types with similar B-values, there aren't any with identical B-values but quite close and we'll have to use the same types for both thermistors. Using just one 10K thermistor and using ordinary resistors in series/parallel doesn't reach the same fit to the curve but it might work just as well as we are well within tolerances of the old types. Looking at Vishay datasheets they take into consideration tolerances and other parameters and generally specify +/-1%-5% tolerances whereas the original STC/ITT thermistors are specified at +/-20% according to the datasheet - so I don't think we have to worry about the circuit to much as it is calibrated with a potentiometer on the output signal and a trimmer on the driver side of the transformer. The potentiometer is also accessible through the hole in the side so as long as the oscillator itself has been calibrated with the trimmer to the correct value (with a temperature span of some 2°C) I think any necessary calibration can be performed as usual. Some of these thermistors can be bought as surface mount types but they generally have a much higher B-value, but there are a few through hole mounted types that might fit on a small PCB that can be put into the holes of the aluminum block on the oscillator board, however it might be necessary to make a new block to accommodate them so that the hole can be filled with thermal paste.

Now, the hard part is to source these thermistors nowadays with the component shortage we experience worldwide right now. I'll try to see if I can find any of these types for sale anywhere and do some experiments with them by fastening a Peltier element with an RTD element to the original aluminum block and see how the circuit behaves when I vary the temperature - I will however have to have my transformers rewound first.....

[Radio Wrangler]

Is the thermal mass/timeconstant suitable?

The STC ones were tiny beads in an evacuated glass bulb. The thermistor isn't a temperature compensator, it's an amplitude to gain feedback element.

David

[Dekatron]

According to Vishay and some other manufacturers there are no similar types as the old STC models manufactured today for the general market, only on special order, the closest types are glass encapsulated beads most manufacturers sell but they all say that they need to be mounted in a manner that emulates the original types - how that would be done no one wants to answer unless I require their professional help - I guess they need an order for at least a few hundred thousand to start to help out with that.

Googling some shows very few people who have replaced similar old thermistors with modern types and they just used thermal grease to fill the hole the original ones where mounted in, but I can't find any references to anyone who has done any proper development nor any tests of this.

Most modern manufacturers mention a Time Constant, but I can't find any similar figures for the old STC types and I only have one datasheet for the old types, there should be some data books I know as I've seen some portions of pages but I can't find those anywhere.

[Bazz4CQJ]

Originally Posted by Radio Wrangler

Is the thermal mass/timeconstant suitable?
The STC ones were tiny beads in an evacuated glass bulb. The thermistor isn't a temperature compensator, it's an amplitude to gain feedback element.

David

There are two thermistors on the 163 board. My understanding was that one is indeed a feedback element in the Wien oscillator. The second thermistor is in the output circuit which I think attempts to keep the AF level supplied by the board constant wrt temperature. Attached CCT highlights thermistors. Perhaps Martin can confirm?

B

[Dekatron]

AVO have described the workings of the two thermistors in the AVO TT537 instruction book, which can also be found in my circuit comparison document in chapter 6.6: https://www.vintage-radio.net/forum/...ad.php?t=86262.

The text on the TT537 says:

"Transistors VT8 and VT9 form a low gain amplifier with negligible phase shift at the oscillator frequency of 1kc/s. Transistor VT10 is an emitter follower to handle the current requirements in the output circuit. Feedback is taken from the junction of resistor R76 and thermistor TH2, the values of these components being chosen such that constancy of output is maintained for any changes other than ambient temperature changes. The operating current through thermistor TH2 is set during manufacture, by adjustment of potentiometer RV8 connected in the emitter circuit of VT8, to give a volt drop across R76 of 400 millivolts r.m.s.

The effect on the amplitude of the oscillator output due to ambient temperature changes is compensated for by the parallel connected components, thermistor TH1 and resistor R77. To ensure the correct degree of compensation the calibration network is designed such that the load on the oscillator is maintained approximately constant at 3.3k ohms."

Fortunately AVO kept the numbering of the thermistors the same in the VCM163, so the description for what the thermistors in the TT537 compensates for holds true for the VCM163.

[Bazz4CQJ]

If taking the option of staying with the existing boards, can you not gain any advantage Martin, if you relax the original range of acceptable ambient temperature (as required by the military) to a narrower range?

I doubt if my 163 often gets used outside of the range of 19 to 23'C (which is the range of temperature where the operator is most reliable!). Would doing that make the task of finding a usable modern thermistor easier?

I suspect that even without any temperature compensation, over a narrow range of temperatures, the change in measured Gm would be very small indeed!

B