Replacing Zener with vreg?

[TERRA Operative]

I'm currently in the process of recreating an unobtanium option board for my HP LCZ meter and I am stuck on one part.

In the top left corner of the schematic below, there is a 6.3V 1%, PD=0.4W, TC=0.001% zener diode.
I'm having trouble sourcing a zener of this type, is it possible that I could stick a vreg here instead?

I think it shhhoouulldd be relatively simple to drop in a vreg set to 6.3V, but I just want to get someone more knowledgeable to make sure there's not a trick hiding in there.

[Whales]

TC=0.001%

That sounds crazy grade.  0.001% = 1e-5.  Is the opamp it connects to even that good (eg in terms of offset voltage, tempco)?  I don't see any trimmers on the resistors either, so I can't imagine this needing that high of a spec.   

Typo?  I could still be wrong, but I'm suspicious given the schematic has no further info.


Sidenote: there are variable voltage references available.  Cheapest is something like a TL431, but it doesn't meet the above tempco spec

[TERRA Operative]

Yeah, I was wondering about that... The part is CR1 in the pic below.

The AD7530LN is good for 2ppm of FSR/^oC according to the datasheet.

I haven't looked at the TL074CN, but here's the datasheet: http://www.ti.com/lit/ds/symlink/tl074.pdf

[iMo]

It could be an 1N828/829 there, for example (0.001-0.0005%/C).
I think any regulator with tempco <50ppm/C will do..

[TERRA Operative]

It could be an 1N828/829 there, for example (0.001-0.0005%/C).
I think any regulator with tempco <50ppm/C will do..

That looks like a 6.2V zener, would that be ok to use instead of a 6.3V?

It looks like the HP part number (1902-0692) cross references to the JEDEC part number DT790107A but I can't find anything on that number online beyond HP cross references.

[SiliconWizard]

If you want to replace the zener diode and not change the inverting arrangement, you will have to use a negative regulator (just making sure )

I'm guessing that U6.4/U6.5 are mutually exclusive open/close?
Seems like U5.2 is buffering either -6.3V or +6.3V depending on U6 position.

I was pondering whether the zener diode didn't serve any kind of temperature compensation purpose, but at first sight, given that the zener has pretty tight specs (1%, TC = 0.001%), it's probably not, and a reference voltage with a precision regulator should do the trick.

Maybe someone experienced with this kind of instrument has another take on this arrangement, but I'm tempted to believe that this was chosen just because it was probably one of the most stable yet cost-effective solution at the time to get a stable reference voltage.

If you're willing to modify the board a bit more, I personally would be tempted to "invert" the topology: using a positive regulator (because it may be much easier to find an appropriate positive voltage regulator than a negative one). Make sure you modify the connections to U6.4/U6.5 s accordingly so you end up with the same polarity as before.

Also take care of the feedback resistors you use if you select an adjustable regulator, as you may end up with a worse temperature stability than with the zener setup.

[iMo]

The 1N828 or 1N829 are not exactly 6.3V.
All those zeners are +/-0.3V. They are expensive because of the tempco.

PS: there are 1N828-1 or 1N829-1 with 1% tolerance.

[SiliconWizard]

TC=0.001%

That sounds crazy grade.  0.001% = 1e-5.  Is the opamp it connects to even that good (eg in terms of offset voltage, tempco)?  I don't see any trimmers on the resistors either, so I can't imagine this needing that high of a spec.   

Yeah, that doesn't seem to make sense to  have such specs for the zener diode and then buffering it with an opamp (unknown specs?), the worst being the inverted position were it would travel an inverting amplifier with two 20k resistors and then the opamp buffer. I'm really wondering what kind of precision and stability you end up with.

[ogden]

Yes, obsolete zeners could be replaced with shunt voltage references. You can try two series (5V+1.25V) LT1634 which have required 10ppm TC. Thou I would try mentioned (trimmed to 6.3V) TL431 shunt reference first and see how it does, maybe do some cold-spray test (using "canned air") as well. TL431 is in your junk bin for sure, it's very popular in isolated switching supplies.

[ogden]

Yeah, that doesn't seem to make sense to  have such specs for the zener diode and then buffering it with an opamp

It makes a lot of sense to BUFFER voltage reference/zener using opamp.

[TERRA Operative]

Yeah, I'm not sure why they went with such a precision part. Maybe it was just a jellybean part that was in their internal catalogue.
The circuit is designed to set a DC offset across a capacitor under test. The inputs go through the AD7530LN DAC that set the reference voltage then the transistors amplify and output to the front panel terminals.
There is two ranges, 0v - +/-9.99V (10mV resolution) and +/- 10.0V - +/- 40.0V (100mV resolution).

I've included the two pages from the manual that explain how it works.


I have TL431's here so they are a viable option.

[SiliconWizard]

Yeah, that doesn't seem to make sense to  have such specs for the zener diode and then buffering it with an opamp

It makes a lot of sense to BUFFER voltage reference/zener using opamp.

Of course you have to buffer it. What it does to absolute precision (which is not what they seem to be after here) and temperature stability is another matter. We can suppose they selected the opamps carefully and mounted them appropriately. Still, the "inverting position" poses even more questions here IMO.

[iMo]

..Thou I would try mentioned (trimmed to 6.3V) TL431 shunt reference first and see how it does, maybe do some cold-spray test (using "canned air") as well. TL431 is in your junk bin for sure, it's very popular in isolated switching supplies.

LM431 claims 50ppm/C - that fits well..

[SiliconWizard]

Yeah, I'm not sure why they went with such a precision part. Maybe it was just a jellybean part that was in their internal catalogue.
The circuit is designed to set a DC offset across a capacitor under test. The inputs go through the AD7530LN DAC that set the reference voltage then the transistors amplify and output to the front panel terminals.

Maybe they just use it as a reference in other parts of the design?
Again when you look at the whole chain, I kind of doubt you get such a temp. stability in the end anyway?

[exe]

you will have to use a negative regulator

Or a shunt regulator.

Final voltage can be trimmed, if needed. So, it doesn't need to be precisely 6.3V. But replacing resistor  and adjusting for zero tempco can be a challenge.

For me it's unclear how precise and stable the circuit should be. If it allows some deviation, I'd probably just put lm329 or even a simple zener. May be trimming even not needed if it can be re-calibrated to the new range.

[ogden]

Yeah, I'm not sure why they went with such a precision part.

Could be so that they just had loads of that zener in stock and copy-pasted it. Anyway stability of test signal amplitude bias generator most likely will not impact measurements. I would just put TL431 here and see how it goes.

We can suppose they selected the opamps carefully and mounted them appropriately. Still, the "inverting position" poses even more questions here IMO.

Opamp is TL074CN, it's offset drift is 18 uV/C - way better than required. Also opamp inverter could be quite temperature-stable if low-TC resistor array used.

[TERRA Operative]

Well I reckon the TL431B (0.5% version) sounds the way to go. 

I'll just have to figure out the resistor values now and see how it goes.

[GigaJoe]

so zener feed reference input for a  DAC that produce a voltage output.
means the long term precision for measurement defined by this particular zener diode.

I would suggest to replace it by LM399 + voltage divider, just that , R1-6 we disconnect and we can use to feed LM399, and voltage divider,
make sure about heater:
https://www.eevblog.com/forum/projects/negative-m399-voltage-reference/
resistors for voltage divider 5ppm PTF56, should work.

as switch U6.5 connecting positive and negative , producing a precision meander , i guess,  to feed input of opamp, so current are minimal and do not need buffer for resistor divider+LM399  ....

[ogden]

so zener feed reference input for a  DAC that produce a voltage output.
means the long term precision for measurement defined by this particular zener diode.

Well, it's not even measurement, but bias voltage applied to DUT. Most LCR/LCZ meters does not eve have bias option, yet they are fine.

I would suggest to replace it by LM399 + voltage divider

I would not. LM399 is huge overkill for bias voltage gen.

[GigaJoe]

"bias" - ?   
precision zener , 10bit DAC - ?   for bias ?
I'm skeptical,  I rather say  it a precision meander, generated by switch and then feed ADC to measure

[ogden]

"bias" - ?   
precision zener , 10bit DAC - ?   for bias ?
I'm skeptical,  I rather say  it a precision meander, generated by switch and then feed ADC to measure

You can be believer or not, but this is what it is. Get service guide from HP site, check schematics, page 303. Pay close attention to name of the module. You can check file name of the attachment in the OP as well.

[edit] DC bias specs (two ranges): +/- 40V, 0.1V step, +/- 10V 0.01V. Quite a bits needed for that.

[exe]

precision zener , 10bit DAC - ?   for bias ?

Well, that's what in the original schematic . I still feel no shame suggesting lm329 as it is cheap (from TI).

But I agree, it's unlikely DC bias circuitry needs precision.

[GigaJoe]

well, ok,  probably a reason of famous HP bulletproof design.
lm329 - still large tempco,  LM369 - also old one , 

how about some modern 10V +  divider ?

[ogden]

how about some modern 10V +  divider ?

Just noticed R7 (2K) trimmer named "FS adj". Modern 5V with adjustment of particular trimmer could be option as well. If trimpot range does not reach 40V with 5V reference, then R1-8 could be increased.

[Andreas]

I think it shhhoouulldd be relatively simple to drop in a vreg set to 6.3V, but I just want to get someone more knowledgeable to make sure there's not a trick hiding in there.

Hello,

before LM399 the 1N829A zeners were state of the art for up to 6.5 digit instruments.
These zeners are selected for T.C. (zero T.C. current) and usually also for noise.
Typical you have around 2-3 uVpp with these so you cannot replace them by a standard
bandgap (like TL431) which has minimum a factor 5 more noise.

I also guess that R1 (resistor array)  is specially paired/selected (with low T.C.)
together with the zener to get accurate output.
So it will be difficult to use a standard voltage divider (with 50-100ppm/K T.C.)
without degrading the cirquit.

1N829A are still available (e.g. at Farnell) even if they have now much higher price
as they used to be at the time when they were state of the art.
Eventually you will have to adapt the 1K resistor to get optimum
zero T.C. current for a different device.

So before replacing the temperature compensated zener with
a different device I would carefully check if I can maintain the specs
of the instrument with a different (low noise, low drift) voltage regulator.

With best regards

Andreas