Use differential ADC instead of single ended?

[byoungblood]

I am working on a project to add an Arduino to an old 1960s Hickok 539B tube tester. Unfortunately many of these old testers that have seen frequent service use are now experiencing issues with their analog meters and replacements/repair are either very costly or nigh impossible.

To the point, I have gotten most of the code squared away, however I am running into a problem when trying to measure the bias voltage in the tester.  Attached is a schematic of the way the circuit I've prototyped to interface the tester with the Arduino.

From left to right, L1 is the transformer that the bias voltage is derived from in the tester. The unmarked legs of the transformer go off to the remainder of the bias circuitry, which goes through a rectifier, dropping resistors, some other voltage dividers, etc.. The center tap is where the bias voltage eventually ends up connecting to the tube. The voltage divider there is to shift the -40-0V down to -5-0V, then goes to a summing amplifier to convert it to a more ADC friendly 0-5V.

V1 is what I am using to represent the voltage of the main meter of the tester. C1 is to remove the AC component (original meter has a cap across its terminals to deal with the AC) and the resistor in parallel to provide a way for the cap to discharge so the reading doesn't lag due to the capacitance.

Both the bias and main meter circuit are inevitably linked together though the remaining circuitry of the tester and the tube under test. So I am getting some interactions between the two readings when the Arduino. I corrected one issue where I had made a ground loop (doh!) when hooking everything up. But with *just* this circuit hooked up (Arduino isn't even in the picture) I can't adjust the bias below 1v any longer. Obviously I'm getting a current, probably through the voltage divider circuit on the bias measurement, that is causing the bias voltage circuit to offset by a volt. If I remove the circuit in the schematic, everything reverts to proper operation. If I try to measure either the bias or main meter circuit in isolation of each other, it works right.

So doing some reading, it looks like a differential ADC may be the solution to my issue, it would (hopefully) eliminate the interaction I'm getting between the two due to sharing a common ground. Or do I need to try another approach here?

Below link goes to a schematic for a military TV-7 tester, which uses the same basic testing circuit, so someone can get a basic idea of what is going on in the measurement circuits in the tester. Most of the differences between that tester and the one I have are generally immaterial with what I'm trying to accomplish here. Scroll to page 16 for the simplified schematic.

http://www.i1epj.ham-radio-op.net/Manuali/TV7U/TM%2011-6625-274-35%20hi%20res.pdf

[moffy]

If you could provide values for your R1 and R2 as well as the voltage range of L1 it would provide a better idea of what is happening. I think you need either an isolation amplifier or just a very high impedance amplifier. A differential ADC won't necessarily solve the leakage problem, and instrumentation amplifiers also provide differential measurements but I am not sure how that would help the backfeed issue. Also if you just have R1 and R2 in circuit but not connected to anything else do you still have a problem?

[byoungblood]

R1=100K and R2 is actually a 8.2K resistor in series with a 5k trimpot. I think the total resistance of R2 is just over 12K to arrive exactly at a 10:1 divider with real components.

L1 is 160V CT.

While brainstorming on the way home, I had the thought of taking the R1/R2 divider and having its ground attached directly to the common terminal on the cathode selector switch, as that is where the grid bias is referenced from. But I'm still left with how I'd reference all that back to the ground of the Arduino. That's the issue I still keep coming back around to. I can make a measurement exclusively of one or the other and it works OK, it is just when both are in circuit that I get the interactions.

If you want a challenge to trace the unit I'm actually dealing with, here is a redrawn version of the factory schematic. And this one actually makes a little more sense than the factory one (yeah, it is that bad, as are most tube tester schematics). Bias circuit is in the lower left hand side of the schematic. Main meter is in the upper right of the same page.

https://irp-cdn.multiscreensite.com/d8d319a5/files/uploaded/539B%20SCHEMATIC%20%20MUSEUM.pdf

This link shows (via a poorly scanned hand drawn diagram, but still ok to follow generally) how the main meter figures into this specific tester.

https://www.byan-roper.org/m_higgins/2017-539bc-Calibration.html#measuring-transconductance-how-did-hickok-do-it

[moffy]

Sorry I am not more familiar with tubes, but perhaps an isolation amplifier like the AMC1200 : https://www.ti.com/lit/ds/symlink/amc1200.pdf?ts=1709678959931&ref_url=https%253A%252F%252Fau.mouser.com%252F
and all you have to do is provide suitable supplies for both sides, the output from the Arduino, and the input could be from a small Tx or maybe even derived from L1.

[byoungblood]

Sorry I am not more familiar with tubes, but perhaps an isolation amplifier like the AMC1200 : https://www.ti.com/lit/ds/symlink/amc1200.pdf?ts=1709678959931&ref_url=https%253A%252F%252Fau.mouser.com%252F
and all you have to do is provide suitable supplies for both sides, the output from the Arduino, and the input could be from a small Tx or maybe even derived from L1.

Thanks, a little confused on what you meant by "the output from the Arduino and the input could be from a small Tx"

[moffy]

Sorry I am not more familiar with tubes, but perhaps an isolation amplifier like the AMC1200 : https://www.ti.com/lit/ds/symlink/amc1200.pdf?ts=1709678959931&ref_url=https%253A%252F%252Fau.mouser.com%252F
and all you have to do is provide suitable supplies for both sides, the output from the Arduino, and the input could be from a small Tx or maybe even derived from L1.

Thanks, a little confused on what you meant by "the output from the Arduino and the input could be from a small Tx"

Sorry, my abbreviated writing, the output power supply could come from what is powering the Arduino and the input power supply could come from a small mains transformer.

[byoungblood]

Sorry I am not more familiar with tubes, but perhaps an isolation amplifier like the AMC1200 : https://www.ti.com/lit/ds/symlink/amc1200.pdf?ts=1709678959931&ref_url=https%253A%252F%252Fau.mouser.com%252F
and all you have to do is provide suitable supplies for both sides, the output from the Arduino, and the input could be from a small Tx or maybe even derived from L1.

Thanks, a little confused on what you meant by "the output from the Arduino and the input could be from a small Tx"

Sorry, my abbreviated writing, the output power supply could come from what is powering the Arduino and the input power supply could come from a small mains transformer.

Ah thanks. Got it now. I thought that's what you were driving towards, but the shorthand had me a little confused.

For a follow up, when I connected the ground of the voltage divider for the bias back to another common point in that circuit and now it adjusts back to 0v. So I definitely need to pay attention to where my ground connections are made for all of these measurements.

A plus side of all of this is I'm actually understanding the inner workings of this thing a bit better. I spent some of my time overnight going over the schematic and trying to make the bias voltage not sag when testing power tubes. There's always another rabbit hole when working on things like this.

[moffy]

Modifying someone elses design or equipment can have a world of subtle traps and pitfalls, I hope I have learned to tread cautiously. I once had to modify some GE locomotives to fit my custom low speed controller, due to limited documentation and being thousands of kilometers from the locomotives, everything was designed and tested remotely, there were a few unexpected side effects that needed correction.

[byoungblood]

Modifying someone elses design or equipment can have a world of subtle traps and pitfalls, I hope I have learned to tread cautiously. I once had to modify some GE locomotives to fit my custom low speed controller, due to limited documentation and being thousands of kilometers from the locomotives, everything was designed and tested remotely, there were a few unexpected side effects that needed correction.

Ha, you and a buddy of mine should talk. He's been a locomotive electrician/air brake guru for the past 25-30 years. You guys could probably talk shop for hours.

So far the only previously untested modification I've made was to replace a resistor with a Zener so that the bias circuit isn't sensitive to variations in current. My testing this morning shows it working as intended.

[jonpaul]

OP:RE  Your original  motive :


Unfortunately many of these old testers that have seen frequent service use are now experiencing issues with their analog meters and replacements/repair are either very costly or nigh impossible.

Sorry: not true:

We have several Triplet and Hickock circa 1950s still going strong, and calibrated.

The D'Arsonval panel meters DO NOT go bad unless the tester is dropped or such.

The meter faces can deteriorate but can be reproduced, some have specialized to replicate and print old meter faces.

The meter plastic fronts are common to may as only a few USA firms made the meter movements.

Most tube testers need some new lytic caps, switch cleaning and pot lube.

As the ckt is simple easy to troubleshoot.

Between the line isolation (2 cond mains cable!) 60 Hz hum,   floating,  meter movement and current (not voltage)  meter movement (eg 200 uA) you will have many issues to solve to get an accurate and safe digital ADC  off any old tube tester.

Just my experience,

Jon

[byoungblood]

OP:RE  Your original  motive :


Unfortunately many of these old testers that have seen frequent service use are now experiencing issues with their analog meters and replacements/repair are either very costly or nigh impossible.

Sorry: not true:

We have several Triplet and Hickock circa 1950s still going strong, and calibrated.

The D'Arsonval panel meters DO NOT go bad unless the tester is dropped or such.

The meter faces can deteriorate but can be reproduced, some have specialized to replicate and print old meter faces.

The meter plastic fronts are common to may as only a few USA firms made the meter movements.

Most tube testers need some new lytic caps, switch cleaning and pot lube.

As the ckt is simple easy to troubleshoot.

Between the line isolation (2 cond mains cable!) 60 Hz hum,   floating,  meter movement and current (not voltage)  meter movement (eg 200 uA) you will have many issues to solve to get an accurate and safe digital ADC  off any old tube tester.

Just my experience,

Jon

Unfortunately I'm 1-4 on tube testers with working meters. I've got a Jackson 648-1 and 658-1 that the ferrite on both had turned to dust in the movement, and then this Hickok that has a sticky movement. I converted the 648-1 to a digital meter (as it just has a 0-130% scale, so easy to duplicate) and the 658-1 to another analog meter using an op amp to drive the less sensitive meter.

I had a Hickok 538 (not the 538A) that I sold last year that had a functional meter, but documentation for that particular tester is non-existent. Only mention of the correct manual and schematic for it was by someone on ARF that unfortunately passed 5-6 years ago.

So far, I've gotten measurements off of everything I want to measure (mutual conductance, bias, plate current) all done independently, it has just been trying to make them all three simultaneously that has been the big issue. Yeah, it is taking a little more work than anticipated, but nothing insurmountable. The isolation amplifier(s) should take care of the issue of not being able to use a common ground.

[moffy]

Just a thought, have you tried a 10Meg resistor between the center of L1 to ground to see if that influences the minimum voltage you can get? A 9.1M and 1M divider would be workable, if you don't have a 10M resistor lying around, most DMMs have a 10M input impedance on voltage ranges you would just need to measure the voltage between L1 center and gnd.

[byoungblood]

Just a thought, have you tried a 10Meg resistor between the center of L1 to ground to see if that influences the minimum voltage you can get? A 9.1M and 1M divider would be workable, if you don't have a 10M resistor lying around, most DMMs have a 10M input impedance on voltage ranges you would just need to measure the voltage between L1 center and gnd.

I'll give that a shot. I haven't had much time to work on it in the past few days, but I did place an order from Mouser for the isolation amps and should have them and the power supply modules this Thursday to try out.