Simple Voltage Reference Project

[EpicIntelGamer]

Hello!

I recently noticed I have 3x REF02 5v precision voltage reference IC's.

I am wondering if I can use one to make an accurate 5v reference to calibrate DMM's with.

I also saw a circuit in the data sheet for the REF02 that said the chip could be used to make a constant current supply and I wanted to know if I could do that easily and use it to calibrate DMM's.

Thanks

EDIT: Here it is currently (I need more solder to completely finish it but here it is)

[Lightages]

The REF02 is a low precision voltage reference. Yes you can make a simple circuit and have a reference within 0.3%. This would be good as a sanity check on a multimeter to see that it is reading OK. You would not be able calibrate any better than to maybe 1% with confidence. You really should have a reference that is an order of magnitude better than what you are trying to calibrate to.

The other consideration is if your multimeter has simple trim pots to make the adjustment or if it is software based. With a trim pot based one you just need to tweak the pot until you get the correct reading on the meter. With software based adjusting meters you rarely find then wanting to see a 5V input as their reference. Usually they will want multiples of voltages live 1.9V or 3.5V for example.

Why not build a couple circuits anyway. It will be very simple and they can run from a 9V battery very well.

http://www.ti.com/lit/ds/sbvs003b/sbvs003b.pdf

[EpicIntelGamer]

The REF02 is a low precision voltage reference. Yes you can make a simple circuit and have a reference within 0.3%. This would be good as a sanity check on a multimeter to see that it is reading OK. You would not be able calibrate any better than to maybe 1% with confidence. You really should have a reference that is an order of magnitude better than what you are trying to calibrate to.

The other consideration is if your multimeter has simple trim pots to make the adjustment or if it is software based. With a trim pot based one you just need to tweak the pot until you get the correct reading on the meter. With software based adjusting meters you rarely find then wanting to see a 5V input as their reference. Usually they will want multiples of voltages live 1.9V or 3.5V for example.

Why not build a couple circuits anyway. It will be very simple and they can run from a 9V battery very well.

http://www.ti.com/lit/ds/sbvs003b/sbvs003b.pdf

Sounds like a good idea to make one and use it as the sanity check.

I might actually use it to calibrate my current DMM because this DMM only claims like 1% accuracy. I think most of the DMM's I might buy in the future will probably all be manual calibration (older flukes) so we should be good there.

I have a question though, how clean does the power coming into this chip need to be? I guess if it needs to be really clean I will just use a 9v battery but I did also want to put a little DC power jack on it (I have 3 laying around) that correspond to some 9v 1a wall warts I have.

Also, do you know any high precision voltage reference IC's that you know off hand that would be precise enough to calibrate old flukes but have a simple circuit?

Thanks for your help.

[alm]

For real calibration, you will need multiple voltages ranging from say 100 mV to 1000 V. Also AC voltages up to 600 V at several frequencies and standard resistors. Read through the performance verification or adjustment section of a Fluke service manual to get an idea what's involved. The old Fluke 87(-I), for example, requires DC voltages from 0 V to 1 kV at 0.035% accuracy, AC voltages from 0V to 1kV at 60 Hz to 20 kHz at ~0.2% accuracy, currents up to 10 A both DC and AC up to 1 kHz and resistors between 1 Ohm and 100 Mohm with ~0.06% accuracy. Unless you have a small cal lab at home (there are some people guilty of this on this forum), leaving the meter alone unless it's obviously out of spec is usually the best strategy. The mini metrology lab series by Conrad Hoffman could give you an impression what's involved to derive multiple voltages from a single reference. But that would be limited to DC voltages.

If you just want to do a spot check, the voltage references sold by Voltagestandard and Geller Labs are a good choice. The advantage you get from buying a pre-made reference is that someone will have calibrated it with a high-precision DMM with traceable calibration to a much higher accuracy than offered by the IC manufacturer. The Voltagestandard DMMCheck (plus) would probably be suitable. But note this is only for checks: it's not sufficient for calibrating a serious DMM.

If you want to go the DIY route, then this thread might be of interest. There are at least half a dozen other threads about voltage references on this forum (Google is quite good at searching eevblog.com), but the general trend is to try to make a stable reference and then use a DMM to figure out the value, as opposed to trying to buy an IC with a very good initial accuracy.

[EpicIntelGamer]

For real calibration, you will need multiple voltages ranging from say 100 mV to 1000 V. Also AC voltages up to 600 V at several frequencies and standard resistors. Read through the performance verification or adjustment section of a Fluke service manual to get an idea what's involved. The old Fluke 87(-I), for example, requires DC voltages from 0 V to 1 kV at 0.035% accuracy, AC voltages from 0V to 1kV at 60 Hz to 20 kHz at ~0.2% accuracy, currents up to 10 A both DC and AC up to 1 kHz and resistors between 1 Ohm and 100 Mohm with ~0.06% accuracy. Unless you have a small cal lab at home (there are some people guilty of this on this forum), leaving the meter alone unless it's obviously out of spec is usually the best strategy. The mini metrology lab series by Conrad Hoffman could give you an impression what's involved to derive multiple voltages from a single reference. But that would be limited to DC voltages.

If you just want to do a spot check, the voltage references sold by Voltagestandard and Geller Labs are a good choice. The advantage you get from buying a pre-made reference is that someone will have calibrated it with a high-precision DMM with traceable calibration to a much higher accuracy than offered by the IC manufacturer. The Voltagestandard DMMCheck (plus) would probably be suitable. But note this is only for checks: it's not sufficient for calibrating a serious DMM.

If you want to go the DIY route, then this thread might be of interest. There are at least half a dozen other threads about voltage references on this forum (Google is quite good at searching eevblog.com), but the general trend is to try to make a stable reference and then use a DMM to figure out the value, as opposed to trying to buy an IC with a very good initial accuracy.

Well I don't really think it'd be nessary to go to those extremes. I mean if I buy a fluke and its seirously out of wack, I just want something that will get it back into acceptable range again. Theres no way I'll ever pay to have a DMM calibrated so whatever I can do at home is the best any of my DMM's will ever see.

I am still wondering how clean the DC needs to be that I'm feeding the REF02. Does it need to be clean at all, very clean, or can it be straight off a bridge rectifier?

[c4757p]

Regulated with an LM317 should be more than enough for this level of accuracy.

[EpicIntelGamer]

Regulated with an LM317 should be more than enough for this level of accuracy.

Wait do you mean use that as in input for my REF02 or are you saying I don't need the REF02?

[c4757p]

As the input to the REF02. LM317 can't even calibrate an old analog meter...

[EpicIntelGamer]

As the input to the REF02. LM317 can't even calibrate an old analog meter...

Okay that's what I thought but I was just making sure.

Well how would it do running off of just a 9v battery? I don't have any lm317's on hand and I don't want to get one just so I can use a wall wart on this if a battery will do fine.

[EpicIntelGamer]

I'm still waiting on whether a 9v battery is good enough before I solder this up......

[Lightages]

I mentioned right at the beginning that a 9V battery would be fine. All the information you need is in that data sheet I linked too.

[EpicIntelGamer]

I mentioned right at the beginning that a 9V battery would be fine. All the information you need is in that data sheet I linked too.

Well the data sheet doesn't specifically say how clean the power needs to be.

I'll go solder this up.

[c4757p]

Yes it does, it's right on the front page in the marketing blurb:

? EXCELLENT LINE REGULATION: 0.01%/V max

[AG6QR]

I mentioned right at the beginning that a 9V battery would be fine. All the information you need is in that data sheet I linked too.

Well the data sheet doesn't specifically say how clean the power needs to be.

It does, indirectly.  Unless you specify exactly how much output noise you could tolerate, it won't tell exactly how much supply noise you'll be able to tolerate.  But the "Typical performance curves" on page 4 tell you how noise on the supply voltage may be transferred to the output.  See in particular the "Line regulation vs. Frequency" graph and the "Line regulation vs Supply voltage" graphs.

[EpicIntelGamer]

Yes it does, it's right on the front page in the marketing blurb:

? EXCELLENT LINE REGULATION: 0.01%/V max

That's the output, no?

[Lightages]

That tells you how much the output changes based on the input voltage. The 9V battery will be fine. Just change it when it gets down to 8V.

[EpicIntelGamer]

That tells you how much the output changes based on the input voltage. The 9V battery will be fine. Just change it when it gets down to 8V.

Why change it at 8, it runs down to 7v......

[EpicIntelGamer]

Here's the finished product!

Well almost finished, I ran out of solder and was unable to finish the positive pad and also unable to make the negative pad.

I decided to make another way for the DMM to connect to the board besides banana jacks (I didn't want to use any of my remaining precious binding posts) so I thought I'd make pads on the underside of the edge of the board where I could clip on alligator clips connected to the DMM!

Here is how it looks!

[Lightages]

Why change it at 8, it runs down to 7v......

Yes it will run down to 7 volts but you try to avoid the extremes of the operation limits of devices

Another hint. If you are going to make another one it is better to not solder the chip to the board. It would be better to use a good IC socket instead. Soldering directly to the board introduces unnecessary thermal stress and physical stress to the chip. With the REF02 this is probably not a big deal but it is general practice to stress reference chips the least you can.

[EpicIntelGamer]

Why change it at 8, it runs down to 7v......

Yes it will run down to 7 volts but you try to avoid the extremes of the operation limits of devices

Another hint. If you are going to make another one it is better to not solder the chip to the board. It would be better to use a good IC socket instead. Soldering directly to the board introduces unnecessary thermal stress and physical stress to the chip. With the REF02 this is probably not a big deal but it is general practice to stress reference chips the least you can.

I actually have some 8pin sockets but I was too lazy to de-solder them from what they are on.

I completely understand the thermal stress but what do you mean physical stress?

[Lightages]

After you solder a device to the circuit board things cool down and the differential expansion of the materials cause physical stresses. Changes in ambient temperature can cause differential expansion and contraction of the different materials which then cause changing stresses and that can cause more drift.

[EpicIntelGamer]

After you solder a device to the circuit board things cool down and the differential expansion of the materials cause physical stresses. Changes in ambient temperature can cause differential expansion and contraction of the different materials which then cause changing stresses and that can cause more drift.

Wow that's actually good to know.

I really thought sockets were only used for ease of replacements of the parts.

Thanks for the info!

[PA4TIM]

I do it this way. these are the opamps following a LM399. But the 399 is mounted the same way.

By the way, a friend made a reference with a 7805 and a LM317, just for fun, he got it very stable. So do not say you can not even calibrate an analog meter with it (besides that, I have an analog meter that can do 1uV full scale ;-) )


By the way, this LM317 was build using shielding, ovenized, good connections to the meter etc but his first test with a standard non shielded but well build 7805 and LM317 based reference drifted also less as 500 uV

The reference , building it right and good components all are important. I made very good references with 4 cheap TL431 parallel.

[EpicIntelGamer]

I do it this way. these are the opamps following a LM399. But the 399 is mounted the same way.

By the way, a friend made a reference with a 7805 and a LM317, just for fun, he got it very stable. So do not say you can not even calibrate an analog meter with it (besides that, I have an analog meter that can do 1uV full scale ;-) )


By the way, this LM317 was build using shielding, ovenized, good connections to the meter etc but his first test with a standard non shielded but well build 7805 and LM317 based reference drifted also less as 500 uV

The reference , building it right and good components all are important. I made very good references with 4 cheap TL431 parallel.

Are those the things from single inline headers?

That's pretty cool what your friend did!

How good is your quad TL431 reference? Is it good enough to use to calibrate DMM's?

[c4757p]

So it looks like the lower voltage winding has potential!

Indeed it does.