CT160 movement op-amp

[glowinganode]

Hi all, though you might like to see the op-amp circuit I fitted to my CT160 due the the movement going o/c. New movement is a 100uA panel meter, re-scaled. Within reason any sensitivity movement could be used.
Rob.

[Brian R Pateman]

A nice neat job there Rob, especially the mounting of the bridge rectifier (I'll use that method my self in future).

I like the dedicated power supply, although it precludes fitting the amplifier board on the meter terminals I'm sure the benefits out weigh that small disadvantage.

Are you intending to post the schematic and constructional details?

Regards,

[Oldtestgear]

A neat solution to the problem of a dead meter!

Phil

[glowinganode]

Thanks for the interest. It certainly took some shoe horning in. I wanted to construct an stand alone unit, as I wish to replace the movement I fitted for a better quality item in the future. The circuit is pretty basic, a straight forward non-inverting amplifier with the new movement in the feedback path. "Jezzer" has done a very nice CAD schematic of a very similar circuit to mine, maybe he would be kind enough to publish it here.
The amplifier and power supply were built on standard perf board, layout is not too critical. The only thing to watch is insulation to earth of the transformer secondary and plenty of clearance around the board mounting screws, as this can be up to 440Vrms WRT chassis.
The transformer I used was one out of the junk box, the smallest I could find. The secondary voltage was too high, so the primary is connected between the 100V and 220V tappings of the main HT transformer. It was mounted 'drop through' style to reduce height and held in place with epoxy. It also keeps the mains connections on the reverse of the sub chassis.
Rob.

[Pamphonica]

Here's the circuit I drew out, taken from the units supplied by Gerry Horrox at Crowthorne Tubes (but now no longer available). I hope Gerry does not mind, but I have also scanned his fitting instructions.
These little op-amp units plus the nice meters he supplied have rescued countless CT160s. He told me last year that he could not source suitable meters at any reasonable price (they are very large - 100x120mm), so had stopped supplying these kits.
I hope his legacy will carry on as a homebrew. I know that Rob has added a few enhancements to the circuit. I will draw these up when I have time.
Jeremy

[deco-doctor]

Looks like a nice neat job you have done there Rob.

If anyone is using the Crowthorne ALPS replacement meter I made a CAD drawing of a new scale for it. PDF attached.

Martin

[Pamphonica]

Martin,
That's a lovely scale. What package did you use? Something very pro no doubt.
Jeremy

[deco-doctor]

Not really. Just a simple 2D CAD program I have been using for over 12 years, very easy to use and cheap to buy too.

www.deltacad.com

I have no affiliation with them at all, just a satisfied customer.

Martin

[glowinganode]

Martin and Jeremy, thankyou both very much for your help and support.
Martin, how are you getting on with your CT160?

[glowinganode]

Some observations I'd like to share.
If you do replace the movement, don't use a high sensitivity one for the sake of it. Originally I used a 50uA movement, but found it hard to get a decent zero reading. I assume it's because of the very fine hair spring required on a high sensitivity meter. I now use a 500uA movement, shunted up to 1mA fsd (more about that later) which is a lot more robust and is a lot less sticky around the zero.
The (original) movement has a fair amount ac across it, if you put a scope across it you will see why they go o/c! WARNING both terminals of the movement are at anode potential. Use two scope probes, and a diff input to measure this. You will need a good CMRR to get a meaningful trace. I've found a 1uF poly cap across the movement reduces 'needle rattle' and limits the rate at which the needle traverses the scale. I would recommend doing this to prolong the life of the movement and will reduce the chance of bent needles.
The original movement is 30uA, shunted up to 39.8uA fsd. This shunt resistor helps to dampen the movement, whilst the unit is being transported and switching between ranges. I used the value I did because it was to hand. This can be compensated for in the op-amp circuit, so is worth doing.
The two EB91's play a critical part in the correct operation of this unit. If you replace them, then expect a change in calibration.
After 50 years, expect some drift from the original values. Ultimately this is an indicating instrument, the best you can hope for is 5% accuracy, even 10% will still give useful results.
The operation of the tester makes the assumption that the valve under test will display the same characteristics operating under mean rectified voltages / currents as true dc conditions. This is a gross assumption, and the error may well be a lot greater than 10% anyway.
When testing high current devices especially at low anode voltages, the voltage drop across the 200 ohm anode load resistor can compound errors. For example testing a 6080 series regulator at 70mA, at Va=150 the 100% mark corresponded to an increase of 1mA (correct). At Va=90 the 100% mark corresponded to an increase of 1.4mA ie 40% error.
If you want really accurate results, then a test rig operating under true dc conditions is the way to go.
Please don't think I'm belittling the CT160, I'm not, but it is important to be aware of its limitations.
Thanks for reading, Rob.

[deco-doctor]

Originally Posted by glowinganode

The two EB91's play a critical part in the correct operation of this unit. If you replace them, then expect a change in calibration.

I can confirm this point. Replacing the original EB91's in my CT160 with new M8079 (Mullard special quality EB91 http://www.r-type.org/exhib/aaa0131.htm) increased the rectiifed DC voltage which upset the balance one one side of the meter bridge, resulting in an inaccuracy of the anode current settings across the whole range of about 6%, in other words I had about 6% more anode current flowing than indicated when the meter was balanced at zero. To correct this error I had to adjust the value of the resistor R6 from 700 ohms to 795 ohms. (I replaced R6 with a ten turn 2K trim pot).

After 50 years, expect some drift from the original values. Ultimately this is an indicating instrument, the best you can hope for is 5% accuracy, even 10% will still give useful results.
The operation of the tester makes the assumption that the valve under test will display the same characteristics operating under mean rectified voltages / currents as true dc conditions. This is a gross assumption, and the error may well be a lot greater than 10% anyway.
When testing high current devices especially at low anode voltages, the voltage drop across the 200 ohm anode load resistor can compound errors. For example testing a 6080 series regulator at 70mA, at Va=150 the 100% mark corresponded to an increase of 1mA (correct). At Va=90 the 100% mark corresponded to an increase of 1.4mA ie 40% error.
If you want really accurate results, then a test rig operating under true dc conditions is the way to go.
Please don't think I'm belittling the CT160, I'm not, but it is important to be aware of its limitations.

After building a small DC test rig for myself I can see the limitations of the CT160. some of the results I have had by comparing the DC test results with the CT160 can vary quite a bit, I got a 30% error in the GM reading of a 6SL7GT from the CT160, but an almost identical result from a 6SN7GT.

Also it seems to make quite a difference with some valves which method adopted to take a reading. ie set the expected anode current to the book value and adjust the negative grid volts for meter balance, or set the grid volts to the book value and adjust the anode current for meter balance.

Martin

[glowinganode]

Hooray I've sorted it. Can now test 6080's at 100mA with 75 V on the anode. Calibration spot on, no more rattling needle and the movement no longer buzzes.
I've got a breakdown to attend right now so I'll tell you all about it when I get back.
Rob.

[glowinganode]

As mentioned above, I was suffering from lack of meter sensitivity when testing valves on the higher current ranges.
When I designed the op-amp circuit I did a quick sum 30uA x 3,250 ohms= 100mV. I originally used a pair of back to back diodes on the input for clamping, I thought this would allow sufficient head room. Slightly puzzled I disconnected them temporarily, this improved matters but only up 70mA. I had trouble scoping the input due to the high CMRR required, so I rigged up a voltage doubler out of a couple of signal diodes and .47uF caps to see just how large the signal across the original movement was. Testing a 6080 at 100mA, I was very suprised to get a reading of 1.57V, add to this 1.4V drop across the two diodes makes over 3V pk-pk across a 30uA movement! Like I said, no wonder they go open circuit.
With the circuit shown on an earlier post, I would require nearly 100V headroom from the op-amp to accomodate this. These pulses make up an important part of the signal, and to filter them out would lead to errors. Equally, I wanted to reduce the pulse content going to the movement, as this can only be doing harm.
In the end I added a .22uF capacitor from the op-amp output to the inverting input, as shown below. This reduces the AC gain, and acts as an integrator. The resistors maintain the DC conditions.
As an aside, I connected my inputs the opposite way round. My reasoning is this. One input goes directly to the op-amp, the other input also connects to the transformer secondary which will have stray capacitance to earth. Of the two connections to the original movement, one is a rectified sine wave which is fairly clean, the other is a mixture of pulses and sine waves. I felt this one should go to the lower capacitance input. By reversing the meter polarity, the correct sense in maintained.
The calibration is now spot on, from an EF86 to a TT22. The needle no longer rattles and the movement no longer makes that irritating buzzing.
It must be worth considering what can be done to protect and reduce the stress on the original movement, if still fitted. I suggested earlier fitting a 1uF cap across the movement. I'm now unsure if this will leave the calibration un affected, I'll leave that for someone else to discover.
I would be interested in any other input before this thread is closed, please.
Regards, Rob.
p.s. I standardised a strapped CV455 at 200V / 16mA as described in the calibration, but measured 10.0mA/V not 4-5mA/V as stated.

[deco-doctor]

Originally Posted by glowinganode

It must be worth considering what can be done to protect and reduce the stress on the original movement, if still fitted. I suggested earlier fitting a 1uF cap across the movement. I'm now unsure if this will leave the calibration un affected, I'll leave that for someone else to discover.
I would be interested in any other input before this thread is closed, please.
Regards, Rob.
p.s. I standardised a strapped CV455 at 200V / 16mA as described in the calibration, but measured 10.0mA/V not 4-5mA/V as stated.

Rob

That's really nice work. Thank you very much indeed for sharing your findings with the forum.

I would be very interested in some way of protecting the original movement as I seem to be quite fortunate in having an intact original meter. Is there any reason why your op amp meter driver could not be adapted to drive the original meter?

Martin

[glowinganode]

Martin, the op-amp circuit can be modified for any sensitivity movement within reason. The op-amp will do all it can to maintain the same voltage at the inverting input as the non-inverting input (signal). The current required to do this must flow through the feedback circuit (i.e. the 'new' movement). By making the resistance between the inverting input and 0V rail the same as the input resistance (that between non-inverting input and 0V rail i.e. 3250 ohms), the same current flows through the 'new' movement as through the original.
For more on op-amp theory see http://www.play-hookey.com/analog/
Shunting the movement with a 1uF cap is certainly easier, the effect it may have on calibration is probably best found by experimentation. I would try this first before adding any un-necessary complexity, unless anyone else has any bright ideas.
Rob.