DIY Calibrator attempt

[wolfalex]

Hello,

 I am in the middle of building my first DIY calibrator which will be used with my two 3478A multimeters. The both lost their calibration and I thought why not try to build something that maybe will allow me to
 "calibrate" the meters myself for hobbiest use of course. Goal of this attempt was not to select/order/use a buch of high performance components and use them to get the best possible performance...my challenge
 was to open my parts box, take whatever I find which could be good enough and then see what I get at the end

 I certainly learned a lot already while building the first building blocks of the unit. What I have achieved already: generation of a trimable 300V, 30V, 3V, 300mV and 30mV reference voltage. Generation of a
 1A and 100mA constant current (again trimable).

 The current schematic and some results are shown under the following link:

http://wolfalex.bplaced.net/electronics/CAL/calibrator.htm

[Vgkid]

Nice job.

[wolfalex]

Nice job.

Thank you very much It was way more work than I have expected. I will continue with the assembly and of course I will also keep the information about the attempt up to date. I can't await the moment when the
 thing is completely finished and I try it out on the 3478A. Will also be interesting how much my reference will drift over time.

 I forgot one topic before: I still need to built the signal generator for the AC calibration of the 3478A. My information is that the signal I need to generate is a square wave with a frequency around 1kHz, amplitude
 level +/- 3V and a duty cycle of perfect 50%.

 Is this information true? 

[sync]

I think the AC calibrators listed in the 3478A manual produce a (low distortion) sine wave. Fortunately you only need one voltage for AC calibration: 3Vrms +/- 0.1% at 1kHz. A square wave might work too with a bit of additional error.

[RobK_NL]

Nice job, certainly.

I merely wonder why you have chosen to use 300mV as the reference for the 30 and 300V outputs, rather than the 3V from your main reference device.

[Vgkid]

Agreeing with robK. I would have used the 3.0V as the default instead of the 300mV. One less divider network.

[wolfalex]

I agree, it would probably have been better to use the 3V directly as the reference for 30V and 300V. I think I did what I did after seeing that the input common mode voltage range of the ICL7650CPD was specified realy low with worst case 1.6V for a supply voltage of +/- 5V. So when I saw that the opamp works better for low input voltages I decided to stay lower to be at the safe side.

But when I revisit the datasheet again and also think about my supply voltage of +8V or even +15V I think a reference of +3V would also have been possible

Thanks for pointing that out.

@sync: thank you for the AC calibration info: I was hoping that I do not need a precise sine signal, but now I saw that your info is correct, and that a sine signal is needed. At least I know now what the next challenge will be My current challenge is to find enough big resistors to  built the needed 300 MOhms resistor needed for resistance calibration

[sync]

Using a square wave for AC calibration is probably not too bad. Just out of curiosity I made a test with my bench meters. I used a cheap DDS function generator to produce a square wave. At 0.1Hz I measured the high level and the low level on DC mode. Then changed the frequency to 1kHz and measured it on AC. I calculated the Vrms for the DC measurement: (V_high - V_low) / 2.

Of course there are a lot of unknowns. I don't know how accurate my meters are and if my function gen have the same levels at 0.1Hz and 1kHz. But except the Fluke the differences are less than 0.1%. I didn't expect that it's so close. So it looks like that these bench meters are working very well with square waves.

Meter           Vrms AC@1kHz    Vrms DC@0.1Hz   diff [%]
HP 34401A       10.0129         10.01190         0.01
Keithley 2000   10.0074         10.01140        -0.04
Keithley 199     3.00542         3.00676        -0.04
Fluke 8840A      1.91634         1.90614         0.54
Siemens B1046    2.30617         2.30580         0.02

[bagnoz]

Very interesting project.


Might I suggest the following change:

This should apply to all the uses of (what I assume are) potentiometers-not-just-trimmers in your circuit.

Assuming no bill-of-materials change, I think this would be a strictly superior circuit.

[sync]

I would replace the trimmer with a slightly too large fixed resistor and a much higher value trimmer parallel to it. Than the influence of the trimmer drift to the divider is reduced.

BUT for 10:1 or 100:1 I would use a hamon divider anyway.
http://conradhoffman.com/HamonResistor.html

[bagnoz]

I would replace the trimmer with a slightly too large fixed resistor and a much higher value trimmer parallel to it.

Than the influence of the trimmer drift to the divider is reduced.

My suggestion is to use the trimmer as a "true potentiometer", which theoretically eliminates drift due to heating or aging (assuming no current through the wiper) instead of just reducing it.

Genuine question: Am I wrong?
My head says it is obvious, but I keep seeing (especially adjustable voltage divider) circuits using just the two terminal connection.
Thus, it is either not obvious or not better.  Which is it?


Of course the values need to be changed a bit, to have the same adjustment range.
In this case paralleling 100 ? with the 100 k? potentiometer and replacing the 1 k? with 900 ? lead to the same adjustment range (±5 %)

[wolfalex]

Hello,

 A big thank you very much to you all for the ideas and inputs. Sorry that you havn't heard back from me until know but unfortunetly something unexpected happend, and now I was not at home at all today, and most likely I will also not be there tomorrow. Therefore no forum access

For the trimming capability I have used 25 or at least 10 turn trimmers. I think the trimmers I have used are not high quality, but I still hope the will do a descent job as soon the whole unit is assemled. You can probably see in the different schematics that some values and positions of the trimmers were mainly decided on what values I had available at this point in time. If my trimmer value was relatively high it ended up in the upper path of the resistive divider. If it was relatively low value it ended up in the lower half of the divider and so on. Thanks for the ideas, I will think about them.

What I will do for sure when the unit is fully functional is to "calibrate" the 3478A, and then do not touch the calibrator at all, and remeasure it with the same 3478A once every week or so for a certain time to see how much the calibrator changes over time. will be very interesting to compare the different reference voltages and their stability over time.

For the AC calibration: I have to think about either using the square or sinusoidl signal approach. I will probably just try to generate a good sinusoidal signal.

 

[giovannirat]

Any news about this interesting project ?

The home page disappeared.......

[Bryan]

Ditto...404, page is gone.

[ManateeMafia]

I see it here  http://wolfalex.bplaced.net/electronics/cal/calibrator.htm

[wolfalex]

Hi,

 I'm very sorry for the late response, I should really log in more often as I was not even aware that I was being asked a question here It seems when I worked on my website I probably ruined
 the link. @ManateeMafia: this link is correct, thank you very much. In the mean time the following happened:

   1, I (think) I was able to use this circuit to move the "calibration" from my Fluke 87 to the 3478A. "Calibration" because of course the accuracy of the Fluke87 is not enough as a calibration
      standard, but at least it made my 3478A useful again for working on hobbyist stuff. And the 3478A accepted the calibration

   2, Still missing is the AC calibration part. I used my 3325A instead because I was too lazy to develop the circuit until now.

    3, After one hour of continuous operation my initial power supply was overheating and the unit stopped working. I found that a bridge rectifier stopped working because of overheating. I then 
     decided to add a buck +5V power supply which delivers the high current to the constant current sources. This helped to reduce the power dissipation a lot. The website contains the updated schematic.

   Other than that not much happened with this project I hope I can finish my second 3478A soon, and then I will reuse this circuit again to "calibrate" it. Thanks to all for the interest in this
  project. It is really a nice feeling when people are interested in your own work. I will try to log in more often in the future to answer questions in a more closely manner.

[wolfalex]

Some pictures of the current state of the unit. Let me just put it this way... It will not win the price for most beautiful construction