choosing an affordable supply for 1mV resolution

[WastelandTek]

I have a calibration procedure I need to perform where I need an accurate 2.0 mV source.  I need several other voltages at this (really 100uV) resolution as well for other steps.

I have an HP E3617A here.  It only offers 10mV resolution.  I have tried dividing it down but it is just so cumbersome and isn't really cutting it.

What would be the hot ticket for a supply at this resolution on a budget?  I certainly have no bias against older gear, but I don't want to pay a collector premium for this particular piece either, just something affordable that I can trust and will do the job.

I have been poking around ebay myself, but I would really value this forum's input.

thanks

[JXL]

May not be affordable, but should meet your requirements:
http://www.ebay.com/itm/EDC-MV105-J-DC-Voltage-Standard-Source-Calibrator-1uV-10V-tested-good-/282535720231?epid=699181874&hash=item41c870f527:g:usMAAOSwol5Y0rpM

I picked up it's brother, an EDC-MV116, for about $120 a while back.

[alm]

Depending on your requirements (voltage, current, definition of affordable) the cheapest option would probably be an analog* precision power supply (e.g. HP 611x, some of the Power Designs units, Lambda LR-61x-DM). They are probably a couple of decades old, but generally well made. Obviously no digital programmability. Available with voltage ranges from 10 V to 200 V, and power generally 20-40W.

Other options that might be affordable is voltage sources. The EDC ones (e.g. MV-105/6) are available fairly cheap, but very limited current capability (50 mA?). The IET VI-700 is available with GPIB option (can not do remote range switch), slightly higher current capability (70 mA guaranteed, up to 140 mA with reduced accuracy). The Keithley 230 is more expensive, goes up to 100 mA and is fully programmable (including current limit).

The third option that comes to mind is old source measure units. Definitely more expensive, but I have seen the Keithley 23x series go for ~300 in as-is state (schematics are available), obviously more in tested state. Step size goes down to 100 µV. The 236/7 go up to 100 mA, the 238 up to 1A.

Keep in mind that if you want an accurate voltage with a decent current draw, you should generally use remote sensing. Precision power supplies / SMUs will generally have remote sensing connections, voltage sources usually not.

You should go through the datasheets and do the math on the accuracy at 2 mV. Offset error may contribute substantially. If you can adjust the value based on readback, then resolution and stability may be more important than accuracy. Especially if the source can also go negative.

*) I guess technically setting voltage via thumb wheels / rotary switches could be considered digital control.

[cellularmitosis]

How much current do you need?

If you only need about 10mA, this would be a great DIY project.  DAC -> 1000x divider -> opamp buffer.

Here's a DAC-controlled current source I made.  Your voltage source would be even simpler!  https://github.com/pepaslabs/prog-cc-100mA/blob/master/kicad/releases/v1/prog-cc-100mA-schematic.pdf

[Assafl]

I use a DVC8500 and a DVC350 from Datel.

Both have a 1mV resolution (actually, the handheld DVC 350 does 100uV).

They have push pull follower outputs so can be used both as sources and sinks within the 30-60mA they can deliver (e.g. You can use them to dial in a zener value or a bias point).

Surprisingly useful devices - I find.

[The Soulman]

If there is only a tiny current required (micro amps) a current source pushing current into a low value resistor could create very low voltages.
And is perfectly diy-able.

[The Soulman]

A 0,01 Ohm resistor at 10 mA gives 100 uV or 1 mV at 100 mA, your hp supply should work well.
However accuracy of the low voltage depends on the accuracy of the source current (the meters in these type of psu's aren't that accurate so hook up a decent dmm to verify) and the accuracy of the shunt resistor and any loading on the resistor also introduces a error.

A resistor like:

http://www.isabellenhuette.de/fileadmin/content/praezisions-leistungswiderstaende/PBV.PDF

Or

https://www.burster.com/en/calibration-instruments/calibration-and-precision-resistors/p/detail/1240/

Or:

http://www.ietlabs.com/esi-sr1-calibration-resistor.html

[Conrad Hoffman]

If small currents are needed, look up the old GR Microvolter.

[alm]

A 0,01 Ohm resistor at 10 mA gives 100 uV or 1 mV at 100 mA, your hp supply should work well.

I would not expect a standard bench supply to go down that low in constant current mode. Since they can not go negative, any offset (drift) could mess this up.

[pigrew]

Another method (if you don't need high accuracy) could be to repurpose an arbitrary waveform generator (like the 33120A). If 50-ohm output impedance is acceptable, they work well (within a mV of the target).  Their attenuator networks allow them to produce small voltages, both positive and negative).

Edit: additional measurement:

Warming up my 33120A, the output (high-Z) has an offset of 1.8 mV when set to 2 mV. This meets the Keithley 230's accuracy spec. Also, it has a resolution of 10 uV, so it can be tweaked to provide a more accurate output (for short-term use).

Of course, the voltage calibrator units would be more accurate, and should be used depending on the need.

[WastelandTek]

Thanks for all the replies.  Clearly I'm going to have to save up a bit more than I thought.

[The Soulman]

A 0,01 Ohm resistor at 10 mA gives 100 uV or 1 mV at 100 mA, your hp supply should work well.

I would not expect a standard bench supply to go down that low in constant current mode. Since they can not go negative, any offset (drift) could mess this up.

Mine do (lm723 or lm317 based), they happily push single mA's all day and would not drift more than ,01%.
And would expect this hp to even do better, but you're right:

From this manual: http://www.dudleylab.com/HP-E3617A-5959-5310.pdf

e. The converted output current should deviate less than
0.1% plus 10 mA from the current obtained in step d
over a period of 8 hours.

If it is for hobby use I,d build a constant current source with a lm317 and parallel a bunch of (cheap) ,1 percent resistors
and power it from battery's, connect a multi-meter in series with the battery's to adjust the current.
Could be below 10 dollars in parts.
It depends on your application and the current and precision required.

[The Soulman]

Less affordable but nice:

http://www.ianjohnston.com/index.php/onlineshop/handheld-precision-digital-voltage-source-v2-detail

[alm]

A 0,01 Ohm resistor at 10 mA gives 100 uV or 1 mV at 100 mA, your hp supply should work well.

I would not expect a standard bench supply to go down that low in constant current mode. Since they can not go negative, any offset (drift) could mess this up.

Mine do (lm723 or lm317 based), they happily push single mA's all day and would not drift more than ,01%.

I was more worried about it maintain good regulation when the voltage goes down to 1 mV or 100 µV.