100VDC reference circuit

[king.oslo]

Yes. I have a transformer I was going to use for a PSU, which was ditched. It has secondary: 2x18V = 36V. Perhaps this can be multiplied by three to get 108VAC? I will search google to figure out how that works.M

[amspire]

Use a voltage multiplier like this:

http://en.wikipedia.org/wiki/Voltage_multiplier

That transformer can supply 100V p-p. So the circuit in the first diagram will give 200V. 4 diodes and 4 capacitors. You will not need that first 22uF cap I put in the circuit (C1) since this multiplier circuit has its own capacitors.

The current is low - diodes can be the cheap 1N4004 are more then you need. Even a small signal diode rated at well over 100V is fine. The capacitors need to be something like 47uf to 100uF electros. Rated at over 100V obviously.

Richard.

[alm]

About running straight off the 240V.  Yeah it can be done, but I don't like it.

Agreed.

If a socket has the neutral and active leads swapped, then you are connecting your earthed meter straight to an active mains.

Earthed meter? I've never seen a DMM which connected one of the inputs to PE, although that doesn't mean they don't exist.

[amspire]

The meter is earthed. The inputs are not earthed, but there is a limit to the voltage above earth that you can apply. The floating parts of the circuit internally are capacitively linked to earth through the transformer windings.

Even if the meter can easily float the input with a 240V AC signal, that is not really the point. The point is you are trying to check the calibration of a 6 1/2 digit meter (or whatever it is), and you are applying potentially hundreds of volts of common mode noise. When you are making precision circuits, it is all about doing everything you can to eliminate sources of errors. Adding a few volts noise to ground via the mains neutral, or hundreds of voltage via the mains active is a very bad start.

Richard.

[alm]

The meter is earthed. The inputs are not earthed, but there is a limit to the voltage above earth that you can apply. The floating parts of the circuit internally are capacitively linked to earth through the transformer windings.

Is there any meter where the max. voltage from low to ground is less than 240 VAC? It's 500 V peak for the HP 3456 A that is the DUT in this case as far as I know.

Even if the meter can easily float the input with a 240V AC signal, that is not really the point. The point is you are trying to check the calibration of a 6 1/2 digit meter (or whatever it is), and you are applying potentially hundreds of volts of common mode noise. When you are making precision circuits, it is all about doing everything you can to eliminate sources of errors. Adding a few volts noise to ground via the mains neutral, or hundreds of voltage via the mains active is a very bad start.

Fair enough, your comment "connecting your earthed meter straight to an active mains" (emphasis mine) suggested to me that you considered it a safety issue.

[king.oslo]

I built it up now. I am pretty excited about the output stability. I am looking forward to showing you.

My question today, it seems that it is possible to trim the LT1031DCH voltage reference by driving the ground pin. The example in the datasheet has a potentiometers connected between V_OUT and a -15V supply. The wiper is connected to ground.

I suppose it is best to use a really stable -15V supply. Can you give me advice on how to achieve this?

Thank you for your time.

Kind regards,
Marius

[amspire]

That looks like a pretty old design of reference. Doesn't mean it is bad, but all the modern 8 pin references have a much easier way to add trimming that does not need a separate supply or as much stability.

It is not just that you need to negative supply - it is also that after adding the negative supply, you may not actually need it. It is only there for the case that the reference voltage needs to be adjusted down. If your regulator is 9.998V, you could adjust it with just a positive supply only.

To answer the question, you do not need -15V.  -10V is better. I bet they used -15V for the simple reason that the engineer writing the app notes up had a -15V supply already available. To get a very stable -10V, you use a standard inverting opamp circuit connected to the 10V reference out. Since the power in the two opamp resistors will be identical, they will track thermally pretty well.

A better option might be something like the Maxim    MAX6350CSA+  (about $11). 1ppm temp stability.

But you are probably going to say it is only 5V.

Ok no big problem. Get two of them and two plastic cases. Power each one up from a 9V battery  (They need 8V or more for the supply) and put the two in series to get 10V. Plus you can compare them against each other to check for drift.

Richard.

[king.oslo]

Ok, guys.

For the AC reference I need a negative and positive supply. I thought I would use two 17V zener diodes plus positive and negative LDO regulators.

Attached is my circuit schematic. When I simulate the two zener regulators in LTSpice, it simulates correctly, but when I built it up, it is not.

The problem is that I cannot get the correct measurements across one of the zeners. The drop is only circa 0.9V across D11 (the lower one). The drop across tje other zener is showing 18V, which is ok.

Any ideas?

Thanks.M

[amspire]

Sounds like you have a dual 56V RMS output transformer.

Use each winding for each different reference - then the two are isolated.

You can use half wave rectifiers for the +/- 17V. and a voltage doubler for the supply for the 100V reference.



Make sure the power rating for the resistors are high enough. If you want say 20mA peak output on the 17 rails, the divider resistors will have to be about 2W power rating.

Richard

[king.oslo]

Thanks Richard,

Are you able to work out why my circuit doesnt work like it does whilst simulating?

Kind regards,
Marius

[amspire]

The wire from the negative of the lower filter capacitor should be connected to the positive of the top filter cap, not the bottom. R16 should be in the wire near the -69V and replaced with a short in its current location.

Edit: The big problem with your circuit is the "ground" for the sinewave circuit is different from the one for the 100DC circuit.
You could not directly connect both to mains ground at the same time which is nasty. With my suggestion, the two references are seperately isolated so you can connect them how you like without any problem.

[king.oslo]

Hello again,

I have been working on this on and off. My motivation took a blow when I one evening wrecked yet another one one of the LTC1052s. Coupled with the frustration of building on Veroboard, I decided I would take a break. But I am feeling fresh again. But I wish I knew why these opamps keep on short circuiting inside.

Amspire, I hope you are well. I remember I had at least one concern. When SW1 was closed, I measured 10VDC at the output relative to V_REF. That was great. The problem was that the measurements on the DMM would jump around about 300ppm max for each new reading. Output of the voltage reference was stable with 1ppm fluctuations. What do you reckon it could have been? Diode? Potentiometer?

And everyone else, I hope you are well too. Have tested the accuracy of the Geller Labs SVR boards against a traceable source? On their website, they say that stating that it has 10ppm absolute accuracy is conservative. Sounds interesting. http://gellerlabs.com/SVR%20M.html

Thank you for your time.M

[quarks]

Does anyone have experience with the Geller Labs SVR board? http://gellerlabs.com/SVR%20M.html

I do have two with calibration data from Joe Geller and two more boards to experiment. I can highly recommend it and communication with Joe Geller is always very helpful.
Attached are two graphs (TEK DMM4050 was running the recommended >1h to warm up, set to 100 NPLC, the x-axis represents 4:41h time)
edit: the one spike in the middle was me moving things while measuring was still running