AVO CT 160 Improved Anode Current Measurement

[Karsten]

Improved Anode Current Measurement
Anode current is measured as the voltage drop across a 200 ohm shunt resistor. The shunt increases the internal resistance of the C.T. 160 anode circuit which can cause considerable error when measuring mutual conductance of valves with low dynamic resistance delta Ua/delta Ia.
The shunt resistance and anode circuit internal resistance can be greatly reduced by replacing the 200 ohm shunt with e.g. 4.7 ohm followed by an op amp with V=200/4.7=42.5. The output voltage swing of this circuit must go up to >24 V in order to measure the 120 mA maximum in rectifier mode correctly.

Circuit description
Fig. 1 shows the circuit diagram. The circuit operates from a 29 V isolated supply and uses a standard LM358 dual op amp. Virtual ground is at 880 mV. At this level the amplifier output has sufficient current sinking capability. Max. output voltage is >26 volts above virtual ground, limiting measurable current to approx. 130 mA. This is sufficient as the overcurrent relay will trip at slightly above 120 mA.
The 68 kOhm / 2.2 uF network smoothes anode current pulses to values within the 26 V output swing. Individual trimpots on the amplifier output enable independent and precise adaption to the A and D measuring circuits.

Together with this circuit the original power supply for the backing off resistor network should also be replaced by a supply feeding well smoothed DC to the network. This greatly reduces dancing of the panel meter needle particularly when measuring mutual conductance.

Installation
Fig. 2 is a section from the C.T. 160 circuit diagram (Fig. 2a: modified original by Martin Forsberg) together with the rather simple and straightforward wiring modifications. Fig. 3 is the realization on a breadboard. The photo also shows the new RV4 A circuit balancing potentiometer. The worn out original was replaced by a robust 100 Ohm military type which was shunted to align it with the mA scale on the panel.

Alignment
A. Basic alignment
Connect a DC voltmeter between amplifier output (pin 1) and – input.

• Set Null adjust trimpot to 0 V on output.

• Apply 120 mA DC between + and – inputs. Set V adjust trimpot to 24 volts on output.

• Increase DC input to 130 mA. Output should read 26 volts.

B. Adaption to A measuring circuit
Set Anode current controls to 0 (fully counterclockwise), Anode volts to 20 and Electrode Selector to A1. Connect a 25 kOhm resistor (variable from 15 to 25 kOhm in series with a DC mA meter and a silicon diode (anode to A1) between A1 and cathode. Alternatively use a suitable valve and proceed as below.

• Turn Circuit Selector to Test and adjust current to 0.5 mA. Now advance mA/V disc to "Set Zero" and adjust A1,A2 trimpot to 1 mA/V mark on AVO meter scale (center of green "good" area).

C. Adaption to D measuring circuit
Set Anode current control to 1 mA (inner dial) and Electrode Selector to D1. Connect a DC mA meter in series with a silicon diode (anode to D1) between D1 and cathode.

• Turn circuit selector to Test. Measured current will be close to 1 mA. Adjust D1,D2 trimpot until AVO meter reads 72*(measured current). Example: measured current = 1.05 mA => AVO meter should read 72*1.05 = 75.6.

 

Results
The following internal resistances of the anode circuit were measured:
Anode / Internal
Volts / Ohms
400 / 150
300 / 108
250 / 85
200 / 69
100 / 63
Method: for each anode volts setting vary anode current and plot measured values (Ua, Ia). The internal resistance is the linear regression gradient of the plotted curve.
The C.T. 160 manual states 375 Ohm internal resistance for Va = 400 V. This complies quite well with the now measured 150 Ohm of the improved circuit: 375 Ohm – 200 Ohm (deleted shunt resistor) + 4.7 Ohm (new shunt resistor) = 180 Ohm.
A graph of the approximate correction factor for a measured mutual conductance can be drawn for these results. As can be seen from the graph in Fig. 4, mutual conductance values for valves with Ra >3000 Ohm need not be corrected even when measured at 400 V anode voltage.

[mikko rintala]

Mikko hi. Great and big work but
How much does this change in the rest of the games affect the tube measurement result?

Best regairds Mikko from Finland

[Karsten]

Mirko,
the only valve measurements in which the R10 shunt resistor is involved are mutual conductance and anode current (amplifier and rectifier valves). There are two potentially critical issues:

• Clipping by limited shunt amplifier output swing. The RC low pass input filter transforms the half wave rectified sinus input into a triangular shaped symmetric signal (0 V DC) which appears superimposed on the DC output. Its amplitude can be given by
  
  triangular symmetric Uapeak)/(half wave rectified Uepeak) = 0.55/(2*pi*fRC) = 0.012 (50 Hz, 68 Kohm, 2.2 uF)
  
  130 mA mean DC half wave rectified input (10 mA above the 120 mA for the biggest testable rectifier) would cause an equivalently shaped 26 V mean DC output voltage caused by a single peak amplitude of pi*26 V = 81.7 Vpeak per 20 ms (50 Hz) period. This is transformed by the low pass network into 0.012*81.7 = 0.98 V ~ 1 V peaks on top of the 26 V mean output. To avoid clipping in this extreme case the output must go up to 26 + 0.98 >= 27 V above virtual ground. For 120 mA this maximum drops to 24*(1+0.012*pi) = 24.9 V which is well within the output span. The ripple amplitude is 0.012*pi = 3.8 % of the mean DC value.
  Meanwhile I have increased V+ to 32 V and replaced the 68 Kohm filter resistor by 20.5 Kohm to improve transient performance. This increases the ripple amplitude to 12.3 %.

• Clipping by the panel meter protection diodes. When measuring rectifier diodes the shunt voltage (and ripple amplitude) is reduced to the 97.5 mV range of the meter, so the ripple stays well below diode forward voltage. When backing off anode current during the measurement of amplifier valves the maximum ripple at 50 mA mean DC (100 mA on the backing off controls) is 10 V * 3.8 % = 380 mV. This is reduced to 73 % = 277 mV by the backing off resistor network. Measuring gm adds up to 97.3 mV giving max. 374 mV peaks across the meter. Problems might begin here when Schottky protection diodes are used. Simply increase the shunt amplifier time constant in that case.
  Panel meter clipping is not an issue with the meter amplifier described earlier.

[Karsten]

I have accurately oscillographed and measured ripple amplitudes on shunt amplifier and meter amplifier outputs. For meter amplifier see https://www.vintage-radio.net/forum/...t=ct160&page=4.

Please find results in the attached .pdf.

[Karsten]

There is an error in the .pdf: pi/2 interval is designated as 50 ms instead of 10 ms. The document has been corrected.

[Karsten]

T/2 (half of 50 Hz period) instead of pi/2 of course.

[David Simpson]

Karsten, You've certainly done some excellent & extensive circuitry to protect the meter & give a stable mA/V readings. Anything that helps protect the CT160's delicate, old & expensive meter is always welcome.
How accurate is the Ia readings from the coarse & fine Ia pots when carrying out AVO's recommended "Standardised Valve" calibration test? With AVO's range of VCM's & Testers, for some time now, I've recommended use of a much higher source of DC Standardised Ia from a valve. As the CV455/491's are just small doubled-up triodes. Something like a DC Standardised 6AQ5 or 6AU5 pentode, for example, drawing 40 to 50mA.
With a CT160, its easy enough to work out mA/V by just altering Vg by a fixed small amount either side of the recommended Vg - then reading off the resultant coarse & fine Ia pot readings for zero meter resolution. Simple maths will give you mA/V. If calibration is fine, this result should be jolly close to the dialled mA/V procedure. I.e. Delta Ia by Delta Vg(RV2) should be the same9near enough) as delta Ia by delta vg(RV1) in that recommended Vg region of the mA/V graphed slope.

Regards, David

[Karsten]

Tested with a 6L6 at Ua=250 V; Us=250 V:
@Ug = 13 V: Ia(backoff) = 74.1 mA; Ia(measured) = 73.64 mA; mA/V(scale) = 5.7
@Ug = 14 V: Ia(backoff) = 68.3 mA; Ia(measured) = 68.70 mA; mA/V(scale) = 5.7
@Ug = 15 V: Ia(backoff) = 63.5 mA; Ia(measured) = 62.70 mA; ma/V(scale) = 5.2

Because I have fitted a digital panel meter to my CT160 for Ug, its setting is regarded as accurate. Mean DC Ug is about 4% higher (AVO's "fudge factor").

mA/V values calculated from these measurements comply quite well with those on the mA/V (gm) scale. Some guessing is required with the mA/V scale as there are no marks between 5 and 6 mA/V. Also, the measurement results are in line with the gm value from a Hickok TV-7 which measures 5.3 mA/V for this valve.

[David Simpson]

That's a decent source of Ia - using a 6L6, Karsten. And certainly using a separate Vg meter is ideal. Might upset the AVO purists, but it certainly helps with accuracy. I incurred their wrath about 10 years ago when I hybridized a C160 out of its clamshell & into an upright wooden cabinet, containing additional Vg & Ia metering.
I see your tabulated results over a 2V Vg change results in a Gm of 5.3mA/V - which is jolly close to the AVO VDM's value of 5.2mA/V. And your result for Vg of 15V is spot on. Whereas the lower Vg/higher Ia results are a wee bit optimistic at 5.7mA/V, but still very good considering all the % variations in voltages which AVO allow in their 60 year old service manual's calibration spec.
The AVO VDM recommends a Va of 350V, Vs of 250V, and a Vg of 18V - the same as the ILIFFE Valve Data Book for a Mullard 6L6. It would be interesting to see an Ia/Vg Graph of your 6L6's tabulated results over a range of say 15 to 25V Vg. Being such a good 6L6, you could use it & its graph as a reference standardised valve for the future. Not only for yourself, but for other valve testing enthusiasts who live near you.

Regards, David

[Karsten]

Attached are measurement results for my 6L6 bogie valve. Two things have turned out as a result:

• Measured Ia = f(Ug) curves can be approached by 3rd order polynomials with almost no error. Remaining errors are in the sub mA range and can be attributed to backing off mA scale reading errors. Therefore the gm graph (delta Ia)/(delta Ug) = f(Ug) can be expressed with good accuracy as the differential of the regression polynomial.

• Comparisons of the thus computed gm values with measured values at several arbitrary points show good agreement.

[David Simpson]

Talking of additional metering to improve a CT160's accuracy, mA/V - wise, I've sometimes thought & mooted the prospect of fitting a wee meter to monitor the 55V T/F winding. After all, the whole caboodle of initially ensuring the "Set AC" locus before proceeding with all the tests - is to mainly make sure that RV2's source of working -ve half-waveform is spot-on. However, if one is involved with studying or tabulating Ia & Vg & taking several minutes to do so, then one cannot keep going back to "Set AC"(& in doing so - switch off the heater supply) to ensure that the incoming mains hasn't altered. That single transformer'd CT160, in my experience, is more susceptible to domestic mains supply changes. Lets face it - it was designed for field or ship usage by semi-skilled servicemen who followed the old adage - if in doubt - take it out, if the valve's mA/V indication failed to reach the Green zone.
These days, CT160's are a collectable item of vintage test equipment, for use in enthusiast's workshops. But in truth - they are just a 60 odd year old, expensive to repair, AC operating portable valve tester. Not really a workshop/laboratory VCM(Valve Characteristic Meter).

Regards, David

[Karsten]

David,

basically you are right of course. Accurate measurements do require a reasonably stable AC mains. Fortunately this is less an issue where I live. Attached is what I have measured last night when the world around me was asleep and AC was stable then. The single line red and blue curves are the measured Ia=f(Ug). The thin black lines inside the two curves are their 2nd order approximation polynomials. The straight double line red and blue curves are their respective differential functions (calculated gm)=f(Ug). The closeness between measured and calculated gm values provides an assessment of the measurement accuracy. Not too bad for a 60+ year old instrument.

I have also tried to assess the sensitivity of the anode current backing off equilibrium against small power variations. As a test I switched on my hot air gun plugged into an outlet next to the CT160. Its meter needle danced nicely +- ca. 10 millivolts with the on-off rhythm of the air gun thermostat. Next I will borrow the magnetic power stabilizer from our son's hifi and try it as a remedy. Doing that I hope that the stabilizer will not create another problem by distorting the AC sinus to the CT160.

[David Simpson]

The thing is, Karsten, back in the day(for CT160 usage) - late 1950's to the 1970's the UK Military bases & ships would have had better stable AC supply sources than the general public & business premises. These days, CT160's are just owned by private vintage radio enthusiasts, & a number of audiophools with high expectations & deep pockets. Folk have their workshops/shacks in garden sheds/garages/lofts/etc. either sharing their house's ring mains or on a spur. So supply fluctuations are a common occurrence. Problems relating to the UK's sinusoidal accuracy & domestic interference from a multitude of switch-mode appliances have been well discussed Forum-wise(see "Search").
From my experiences back in the 60's - servicing & repairing CT160's, and more recent ownership of a couple, I reckon that accuracy beyond 10 to the minus 1 for Gm & a couple of % for Ia is more than can be expected for these 60 odd year old valve testers even if they are kept in good calibration. Particularly as the don't give a decent meter indication of Ia - like a MK3 or MK4(after all, the CT160 is more or less just a MK3/4 squeezed into a clamshell).
There might well be a handful of experimental/design folk who require true & accurate static valve parameters, or even venture beyond Barkhausen's Law & study dynamic properties - to these folk I say - build a pure DC Tester. From what Martin Forsberg tells me - the RoeTest is extremely accurate. As is the VCM163, and the "Sussex", as other Forum folk tell me. And I've heard that Tektronix made a super accurate Curve Tracer back in the 70's/80's, which is hellish expensive to acquire these days.
Me(an old analogue guy), & quite a number of Forum/Vintage Radio folk, are now of pensionable age. So we have to consider very carefully what we spend, regarding expensive test equipment. Hence, in my experience, building a decent DC tester out of redundant/surplus parts/equipment, was a very attractive project - coming in at less than £200. Some folk ask 4 times or more than that amount these days for a CT160! Jesus - a genuine AVO 10K RV2 or a 30uA meter can cost £200!

Regards, David