Looking for a solid state replacement for this rare tube voltage stabilizer

[Bargainhunter]

I would like to repair an old A/C voltmeter.  I've come across an old tube that I'd like to replace with a solid state regulator.  I've attached two pdf files.  One pdf shows the specs and the other is the schematic of the voltmeter.  The tube is a XC12 or CV6004.

Any suggestions would be a great help.
Thank you,
Chuck

[DC1MC]

https://www.vishay.com/docs/85607/bzx85.pdf

[Bargainhunter]

Thank you.  I thought of using a zener but looking at the xc12 specs, I'm not sure I can use a zener.
Maybe I'm reading the specs incorrectly.  Is the XC12 taking an input of 85volts and regulating it to 1volt or should I be reading it as regulating the voltage at 85volts =/- 1volt?

Thank you.

[DC1MC]

Thank you.  I thought of using a zener but looking at the xc12 specs, I'm not sure I can use a zener.
Maybe I'm reading the specs incorrectly.  Is the XC12 taking an input of 85volts and regulating it to 1volt or should I be reading it as regulating the voltage at 85volts =/- 1volt?

Thank you.

The second part, is kind of a crappy steampunk Zener  , but a good stabilizer for its time. Make sure that you won't go over the  maximum power of the diode, otherwise either buy a higher power one or put a couple of low voltage ones in series to reach the desired voltage. In any case, at the tube rated current of 1mA you got just 85mW, but make sure that the old carbon resistors didn't change their value in time, they do have this annoying habit.

 

[srb1954]

Thank you.  I thought of using a zener but looking at the xc12 specs, I'm not sure I can use a zener.
Maybe I'm reading the specs incorrectly.  Is the XC12 taking an input of 85volts and regulating it to 1volt or should I be reading it as regulating the voltage at 85volts =/- 1volt?

Thank you.

I would interpret the spec as meaning that the regulated voltage varies by 1V maximum as the diode current varies from 0.5mA to 1mA not that the absolute voltage tolerance is +/- 1V.

You should be easily able to replicate or better the performance of this tube with Zener diodes. Since 85V is not a standard Zener voltage you could use two 43V Zeners in series  to obtain the required voltage. A BZX85B43 would be a good choice - the B suffix is for 2% tolerance.

If you are only running 1mA through the Zener the voltage will be a little lower than the nominal voltage since datasheet voltages are normally specified at higher test currents.

[T3sl4co1l]

Could also use a neon bulb, which is basically what that is, a fancy neon bulb -- but I think the supplies of voltage-regulating bulbs will be very small indeed these days, and just any random bulb is likely to have much more noise and drift (and worse line regulation) than the zener.

Tim

[jdragoset]

https://www.ebay.com/itm/Amperex-8228-subminiature-voltage-regulator-81VDC-3-2mA-NOS-ZZ1000/264496911845?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

[jdragoset]

Put  a 4 volt V zener or six  forward Si diodes in series with this 81 volt neon regulator

[srb1954]

Could also use a neon bulb, which is basically what that is, a fancy neon bulb -- but I think the supplies of voltage-regulating bulbs will be very small indeed these days, and just any random bulb is likely to have much more noise and drift (and worse line regulation) than the zener.

Tim

I would caution against using a standard neon indicator bulbs as these are not designed for long term stability of their breakdown voltages or for low noise. Also, long term operation on DC will cause damage to the electrodes which will result in considerable drift or random jumps in the operating voltage.

A pair of series connected 43V zeners will provide better initial performance, better long term stability, better reliability and be a whole lot cheaper than using the older technology neon regulator bulbs.

[jklasdf]

The temperature coefficient of high-voltage zeners is fairly bad. Many of these high voltage regulator tubes had temperature coefficients of only 2mV/degC at their regulated voltage of 80 something volts or higher. See the attached PDF.

You can use a circuit similar to the one shown on the third page, with a lower voltage zener (with a much better temperature coefficient) and a circuit to multiply the regulated zener voltage (+Vbe) up. Plus I think the temperature coefficient of the zener and transistor Vbe should cancel somewhat.

[bob91343]

I agree that a simple solid state replacement does not exist.  If you insist, you can make up a series string of low coefficient zener diodes and pay the penalty of lousy dynamic resistance.  Or use a single zener and have poor temperature stability.  Neither is optimum.

Many years ago I had to engineer this sort of thing and am very aware of the problems.

[basinstreetdesign]

I've used a 1N5267 75 V zener in series with a 1N4733 5V zener to make a 80 V reference voltage for a tube bench power supply.  I used it in a -100V regulator.  This was my version of the -150V regulator that Tek used in the 545 scope.  It normally used the 5651 reference tube but I decided not to spend it on my version.

My schematic is here:

The zeners are just left of center and were operated at about 1 mA and have been fine for years showing no appreciable noise.

[bob91343]

Apparently the temperature coefficient of voltage of the zeners is not bad enough to bother your system.  That's the main issue.  If someone wants to retrofit a modern part into a legacy circuit, I'd usually want the new system to perform as well as the old.

As I recall, a 75V zener has a positive coefficient of about 0.12% per degree C, so if the internal temperature of the unit increases 30 degrees, that results in the regulated voltage changing 3.6%.  On a 150V regulator that translates to a drift of 5.4 Volts (all other things not changing) so maybe that's not too bad.  To be more precise, the 5V zener diode reduces the drift slightly but probably less than the uncertaintly of the temperature drift of a random diode.  If the internal temperature changes more than 30 degrees so will the output change more than 5.4 Volts.

I am not criticizing this design but I find it great fun to make such calculations as I type.

[SeanB]

Stack 16 5V6 zeners in series, which will give about the closest to a zero TC zener as possible. and which will probably fit in the same volume as the regulator tube as well. As you are running them at low current you will find they probably regulate around 5V2, so you probably will want to reduce the dropper resistor value slightly, but otherwise it will work well as is. Be a lot more stable as the zener diodes around 5V6 are close to zero TC, but still might have some, but 16 in series will average out to close to zero. 16W power dissipation will probably survive a tube flashing over as well, which is not something the original would do easily.

[mawyatt]

Here's a simple circuit that might prove useful. Any small signal npn transistor with a BVceo of > 100V should be fine. Low current 6.2V zener diodes have a TC ~+2mv/C and the transistor Vbe has a TC of ~ -2mv/C. A shunt cap across the zener and/or R2 will help reduce noise if required. Should be easy to derive a result with the base current included and adjust the resistors accordingly.

Best,

Edit: Added some SPICE simulations.

[jklasdf]

Apparently the temperature coefficient of voltage of the zeners is not bad enough to bother your system.  That's the main issue.  If someone wants to retrofit a modern part into a legacy circuit, I'd usually want the new system to perform as well as the old.

As I recall, a 75V zener has a positive coefficient of about 0.12% per degree C, so if the internal temperature of the unit increases 30 degrees, that results in the regulated voltage changing 3.6%.  On a 150V regulator that translates to a drift of 5.4 Volts (all other things not changing) so maybe that's not too bad.  To be more precise, the 5V zener diode reduces the drift slightly but probably less than the uncertaintly of the temperature drift of a random diode.  If the internal temperature changes more than 30 degrees so will the output change more than 5.4 Volts.

I am not criticizing this design but I find it great fun to make such calculations as I type.

Agreed, the zener linked by DC1MC does have a slightly better temperature coefficient of 0.055 to 0.095%/degC to be fair, but it's still an order of magnitude worse than the original voltage regulator tubes, which have temperature coefficients as low as -2mV/degC at 82V, which corresponds to a temperature coefficient of only -0.0025%/degC.

The circuit I posted (along with the circuit mawyatt posted, which is nearly identical, but with a better explanation) is I think the lowest temperature coefficient easily achievable with just a few solid-state components. Even then though, the performance isn't necessarily any better than the voltage regulator tubes. If the temperature coefficient curves of the zener diode and Vbe don't cancel out exactly (which in practice they won't), the error gets "multiplied up" by the circuit. Still though, I think this is the best solid-state high-voltage regulator circuit with a relatively small parts count (it may not be as simple as just stringing a bunch of high voltage zeners together, but I think it's worth it, especially for Bargainhunter's application of an AC voltmeter). Any voltage regulator tubes you can still find now a days (even new old stock) are nearing their end of their lives (or more likely already past their end of useful life, even if unused).

[factory]

Philips had a modification procedure to replace a 103V neon regulator tube with a solid state circuit, using a low voltage zener and a pair of transistors, adjust the resistors as required for the tube being replaced, see here, pdf is in post #1, more info in post #4;
https://www.vintage-radio.net/forum/showthread.php?t=162689

David

[srb1954]

The temperature coefficient of high-voltage zeners is fairly bad. Many of these high voltage regulator tubes had temperature coefficients of only 2mV/degC at their regulated voltage of 80 something volts or higher. See the attached PDF.

The temperature coefficient of these tubes might look good on the face of it but this is considerably offset by the high dynamic resistance, which means the circuit line regulation is likely to have a greater effect than the temperature coefficient on the voltage stability of a typical reference circuit. Also, the drift of the reference voltage in the first 250 hours of operation is considerably more than any likely temperature variation effect, although this could be countered by frequent recalibration.

To achieve anything like voltage stability that the temperature coefficient figures might suggest you are going to need to fed a very stable current through these tubes either by using a constant current driver or a resistor from a pre-regulated supply. Both options add to the circuit complexity and cost.

[bob91343]

Yes most of the comments ignore the dynamic resistance.

[jdragoset]

This is $11 free shipping.

https://www.ebay.com/itm/Amperex-8228-subminiature-voltage-regulator-81VDC-3-2mA-NOS-ZZ1000/264496911845?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

[jklasdf]

For even more demanding applications, there is the "Maida-style" high-voltage regulator in Michael Maida's application note for National Semiconductor at the end of the PDF I linked. But this is no longer a simple replacement.

[Bargainhunter]

Thank you for all your help.
Meter is working now.

[drussell]

Thank you for all your help.
Meter is working now.

What solution did you end up with?!   

[Bargainhunter]

I found a schematic for the VM77 which showed a zener in place of the tube.  It must have been an updated schematic.  I didn't find the parts list for Z1 but I assumed any 81-85v zener would work.