Author Topic: Building your own voltage reference - the JVR  (Read 127202 times)

0 Members and 1 Guest are viewing this topic.

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Building your own voltage reference - the JVR
« Reply #75 on: October 29, 2015, 01:16:36 am »
Why NPLC1? Run NPLC5 ;).
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #76 on: October 29, 2015, 09:53:06 am »
Why NPLC1? Run NPLC5 ;).

OK, I'll try. I am building an amplifier with 0.1-10Hz filter to measure the p-p noise on a scope, that should give a much more accurate result.

Cheers

Alex
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 3221
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #77 on: October 29, 2015, 08:52:37 pm »

OK, I'll try. I am building an amplifier with 0.1-10Hz filter to measure the p-p noise on a scope, that should give a much more accurate result.

Which of the cirquits / application notes do you use for this?

With best regards

Andreas
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #78 on: October 29, 2015, 11:00:50 pm »

OK, I'll try. I am building an amplifier with 0.1-10Hz filter to measure the p-p noise on a scope, that should give a much more accurate result.

Which of the cirquits / application notes do you use for this?

With best regards

Andreas

I took the response graph from the OP27 datasheet and simulated a suitable circuit using the OPA2227 opamp with a couple of 9v batteries for the P/S.

Cheers

Alex
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 3221
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #79 on: October 30, 2015, 08:08:15 am »


I took the response graph from the OP27 datasheet and simulated a suitable circuit using the OPA2227 opamp with a couple of 9v batteries for the P/S.

Cheers

Alex

Hello,
we are talking of figure 31 in the AD datasheet?
http://www.analog.com/media/en/technical-documentation/data-sheets/OP27.pdf

With this cirquit the input impedance is 10 Ohms.
The DUT has to be below 0.1 Ohms without having too much errors.
(And the output voltage has to be very close to 0 Volts without saturation).

A standard Zener has around 10 Ohms. Your cirquit will have several 10s of Ohms.

Have a look on AN124
http://cds.linear.com/docs/en/application-note/an124f.pdf

Similar results can be get by a 2 stage design with LT1037 and LT1012 + selected (low leakage current)
standard electrolytic capacitors.

And dont forget the (metal) cookies box (when using LT Parts) or the (metal) paint can (when using TI parts).

With best regards

Andreas

 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #80 on: October 30, 2015, 08:42:04 am »


I took the response graph from the OP27 datasheet and simulated a suitable circuit using the OPA2227 opamp with a couple of 9v batteries for the P/S.

Cheers

Alex

Hello,
we are talking of figure 31 in the AD datasheet?
http://www.analog.com/media/en/technical-documentation/data-sheets/OP27.pdf

With this cirquit the input impedance is 10 Ohms.
The DUT has to be below 0.1 Ohms without having too much errors.
(And the output voltage has to be very close to 0 Volts without saturation).

A standard Zener has around 10 Ohms. Your cirquit will have several 10s of Ohms.

Have a look on AN124
http://cds.linear.com/docs/en/application-note/an124f.pdf

Similar results can be get by a 2 stage design with LT1037 and LT1012 + selected (low leakage current)
standard electrolytic capacitors.

And dont forget the (metal) cookies box (when using LT Parts) or the (metal) paint can (when using TI parts).

With best regards

Andreas

Hi Andreas,

I only took the graph as a reference, the circuit there is not suitable. I also have an option to adapt one of my FET phono stages circuits for this filter-amplifier, the simulation shows that I can get a gain of 1000 with the right frequency response, however I am not sure how quiet the input FET will be at 0.1-10Hz. I'll probably try it anyway, as I have plenty of nice pcbs with an excellent low signal level behaviour and even some good 24V DC power supplies (including some rechargeable batteries) left from the time I was building and selling these plus these fit perfectly into an aluminium Hammond enclosure. If I can get <100nV p-p self-noise, I'll be quite happy.

Cheers

Alex
 
The following users thanked this post: Roman oh

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #81 on: October 31, 2015, 09:10:45 pm »
I've received some Vishay Bulk Foil 5K+5K resistor dividers - I plan to use these for 5V to 10V amplifier in the "ultimate" version of the JVR. I've decided to check how accurate they are. With 0.05% stated tolerance I was not expecting a much better performance, however the measurements show that the resistor pairs are matched very closely - the worst one out of ten is about 15ppm off, and the best two were good to 1ppm or less, as that was the limit how accurate I could measure the error of the divider in my simple improvised DC bridge, using Keithley 2015 as a null-meter (the inconsistency in the results with plugging/unplugging the resistors and repeated measurements were about 1ppm).

Cheers

Alex



« Last Edit: October 31, 2015, 09:35:13 pm by Alex Nikitin »
 

Offline BravoV

  • Super Contributor
  • ***
  • Posts: 7547
  • Country: 00
  • +++ ATH1
Re: Building your own voltage reference - the JVR
« Reply #82 on: November 13, 2015, 07:48:00 am »
Will 2N4392 TO-18 (datasheet) work decently ? Our local store carries it.

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #83 on: November 13, 2015, 08:48:50 am »
Will 2N4392 TO-18 (datasheet) work decently ? Our local store carries it.

Yes, it looks like a very suitable device, with a good Vgs(off) between 2 and 5V., and a metal case. Another possible option available is 2N4857A/2N4856A.

Cheers

Alex

P.S. I've ordered some of 2N4392 and 2N4391 from Mouser, it looks like the cheapest TO-18 metal can JFET available new right now. Thank you for the pointer.
« Last Edit: November 13, 2015, 01:06:51 pm by Alex Nikitin »
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #84 on: November 17, 2015, 10:22:27 am »
1) I've received 2N4391 and 2N4392. The 4391 devices look really interesting, as the zero tempco point is very high (from 6V to 8V) and the current at that point is low (1-1.5mA), that makes for a very low internal dissipation (only 2 to 6mW with 10V supply) and very good stability. The gate is connected to the case and is grounded in the circuit, which is also good. I will build one JVR sample with the 2N4391 now, it looks very promising.

2) I've done a small mod on the Fluke 731B. As I am running it without a battery, the needle meter on the front panel is essentially useless. I've installed a temperature sensor (LM35) and wired the meter so it shows the temperature of the internal screen around the reference board. It did not require any significant changes, only a cut in one of the power supply board tracks and a couple of resistors. Now I have an accurate internal thermometer with the scale roughly from 18C to 27C (the green sector is from 20C to 24C) which gives me a very good idea of the output voltage deviation.

Cheers

Alex
 

Offline ltz2000

  • Regular Contributor
  • *
  • Posts: 103
Re: Building your own voltage reference - the JVR
« Reply #85 on: November 17, 2015, 01:24:09 pm »
I've done a small mod on the Fluke 731B.

Another mofication worth doing, which actually improves the stability, is replacing the crappy pre-regulator. I used the good (enough) old LM723.
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #86 on: November 20, 2015, 09:03:25 pm »
I've built the first 10V prototype with the 2N4391. The reference voltage was around 7.228V, so I've used OPA227 and some wire-wound resistors to scale it up to 10V, the same type wire wound resistors were used for the reference source as well. Without much tuning I've got around 2ppm/C and very low noise level. This is the first time I've measured the JVR against the Fluke 731B, using Keithley 2015 on 100mV range. The variations are around 1ppm, the noise is well below 1ppm p-p.

This is a step towards a properly built reference - I've used a good aluminium case, good resistors and Tellurium Copper terminals, though the circuit is built on a prototyping board.

Cheers

Alex

10V JVR on 2N4391 measured for 30 min against the Fluke 731B by Keithley 2015, 100mV range, NPLC1, no filtering, 1ppm per division:

 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #87 on: November 21, 2015, 10:45:10 pm »
Here is another graph of the 2N4391 JVR v Fluke 731B. This time I've used a better cable (a screened pure copper twisted pair with copper plugs which I've just received), a foam cup on the 2N4391 FET and NPLC10 on the Keithley 2015. 30min run, 1ppm per division.

Cheers

Alex

« Last Edit: November 21, 2015, 10:48:20 pm by Alex Nikitin »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 3221
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #88 on: November 21, 2015, 11:11:12 pm »
Hello,

ist the lesser noise mainly from NPLC10 or also from the additional shielding?

with best regards

Andreas
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #89 on: November 21, 2015, 11:19:15 pm »
Hello,

ist the lesser noise mainly from NPLC10 or also from the additional shielding?

with best regards

Andreas

The noise is less because of NPLC10, however the slow voltage variations are less due most likely to smaller thermal effects from pure copper cabling and connectors. I've used just standard cables previously and at least some of the drift was due to thermoelectric effects.

Cheers

Alex
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #90 on: November 22, 2015, 10:53:49 am »
Here is another graph, this time for an overnight measurement, where the lab temperature varied from 22.5C in the evening down to 19C early in the morning before the heating was switched on. Vertical scale is 1ppm/div, so the tempco is about 1ppm/C, on par with the Fluke 731B. I may try to tweak it a bit more to get even lower figures. The circuit allows for a two pot trimming, one for the tempco and another for the voltage, without affecting the tempco, however as there are drifts in the other parts of the circuit, some iterations are required to get closer to zero in a 5-6 degrees C range. I may try to do it later today.

Cheers

Alex

« Last Edit: November 22, 2015, 08:02:42 pm by Alex Nikitin »
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #91 on: November 24, 2015, 08:54:38 pm »
It looks like I've got the tempco under 1ppm/C, though it is still positive and combined with a bit over 1ppm/C negative from the Fluke 731B the tempco of the differential voltage between two units is about 2ppm/C, on the first graph. The combined noise is, however, very low, a small fraction of a ppm, as visible on the first 24h graph (2ppm/division). On the second 24h graph the JVR is measured by Keithley 2015 and the noise of the meter dominates, however the tempco is just under 1ppm/C. The 2N4391 costs about $2 from mouser.com and probably the cheapest new metal can JFET you can buy - and quite suitable to use in the JVR. A very simple and cheap reference with potentially a good long-term stability, plus low and adjustable tempco can be built with this JFET, a quality opamp (i.e. OP27) and few good resistors (actually, you only need three resistors with the right values, but as JFETs parameters have a large spread it is necessary to build up the right values from several standard ones).

Cheers

Alex



« Last Edit: November 24, 2015, 11:04:34 pm by Alex Nikitin »
 

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2571
  • Country: gb
Re: Building your own voltage reference - the JVR
« Reply #92 on: November 24, 2015, 10:22:25 pm »
731B? meh, we need a 734A at least when measuring a cheap FET standard, providing a JJA is not immediately at hand  :palm: (ignore me, just bookmarking!  ;) )
 

Offline uski

  • Frequent Contributor
  • **
  • Posts: 295
  • Country: us
Re: Building your own voltage reference - the JVR
« Reply #93 on: November 25, 2015, 12:14:47 am »
:popcorn: keep going, I've been looking for a cheap reference good enough to roughly calibrate my 5 1/2 digits multimeter
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #94 on: November 27, 2015, 10:48:02 pm »
The weather was warmer today and I've pushed the room temperature up to 24-25C. Interestingly, it looks like my current version of the JVR with 2N4391 has a near-zero tempco between 22C and 24C - the voltage changes are less than 1ppm in that range, and at 25C the voltage starts to drop, so the tempco changes to negative. The Fluke 731B voltage changes about 4ppm in the same 3 degrees range, still negative.

Cheers

Alex

P.S. - this version runs untouched for about 150 hours and there is no voltage change at the same room temperature I can reliably detect (~below 2ppm).
« Last Edit: November 27, 2015, 11:03:07 pm by Alex Nikitin »
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #95 on: January 12, 2016, 10:11:59 pm »
Here is a graph of the JVR voltage v room temperature over one month (from 12th of December till today). As I've recorded the voltage mostly from the display of Kethley 2015, there is at least 1ppm uncertainty of that readout. When the display did flicker between two points, I've recorded a half-ppm point. It looks like the JVR + Keithley combination drifted not more than 2-3ppm over that month. Either both are quite stable or there is a remarkable similarity in the drift character of two units  :-// .

Cheers

Alex

« Last Edit: January 12, 2016, 10:39:41 pm by Alex Nikitin »
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #96 on: February 19, 2016, 04:53:45 pm »
I've measured the same 10V JVR unit with HP3458A over three and half hours today. The temperature in the lab was between 21-22C all that time. Vertical scale is 0.5ppm/div .

Cheers

Alex

 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #97 on: February 24, 2016, 04:34:11 pm »
More measurements with HP3458A - this time for 29 hours. At night time temperature in the lab drops steadily to about 18C in the morning, and during the day the temperature is mostly reasonably stable around 21-23C. The day to day stability looks very good, even with the unit travelling with me and so experiencing a much colder conditions (around 3-7C). 1ppm/div vertical scale.

Cheers

Alex

 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #98 on: February 24, 2016, 05:24:40 pm »
For such a low current reference the performance looks really good.

With the HP3458 measurements there might also be some contribution from it's internal drift. With changing temperature you are supposed to use ACAL routine from time to time.
 

Offline Alex NikitinTopic starter

  • Super Contributor
  • ***
  • Posts: 1149
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #99 on: February 24, 2016, 06:09:22 pm »
For such a low current reference the performance looks really good.

With the HP3458 measurements there might also be some contribution from it's internal drift. With changing temperature you are supposed to use ACAL routine from time to time.

Thank you. I've just left the setup running unattended. The HP3458 has option 002 though, so I hope the internal drift (especially on 10V range) should be minimal (ACAL gave me a difference below 0.5ppm when I've run it at different temperatures, i.e 18C and 23C.

Cheers

Alex
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf