Author Topic: Building your own voltage reference - the JVR  (Read 52492 times)

0 Members and 1 Guest are viewing this topic.

Online Vgkid

  • Super Contributor
  • ***
  • Posts: 2414
  • Country: us
Re: Building your own voltage reference - the JVR
« Reply #100 on: February 25, 2016, 05:22:20 am »
Thanks for the updates, especially the drift performance.
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 3341
  • Country: gb
Re: Building your own voltage reference - the JVR
« Reply #101 on: February 28, 2016, 02:27:59 am »
Odd bit of synchronicity here. This thread came back to life a few days ago and today I was hunting around for an old microphone preamp design of mine. In the process I tripped across the circuit for the AKG C451E microphone's internal preamp. This is phantom powered and lurking inside it is a JFET acting as a voltage reference for the on-board voltage regulator. It's T5 in the diagram below. Note that R15 is marked as select-on-test.

Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #102 on: February 29, 2016, 09:36:30 pm »
That is one of the best 12h runs I've had with this reference. It was taken on Saturday with the room closed and no heating, so the temperature was about 18C all day from 7:30 till 19:30. Vertical scale is 0.5ppm/div.

Cheers

Alex
« Last Edit: February 29, 2016, 09:38:40 pm by Alex Nikitin »
 

Offline zlymex

  • Frequent Contributor
  • **
  • Posts: 556
  • Country: cn
Re: Building your own voltage reference - the JVR
« Reply #103 on: March 05, 2016, 02:48:05 pm »
Anyone heard of the related story? Here is goes.

1. For a very long time, only LT's LTZ1000 claims to have 2uV drift @1kHr(@65C, roughly 0.3ppm @1kHr)

2. In late 90's, Analog annouced its XFET based ADR293 5V Vref claimed to have drift of only 0.2ppm @1kHr @25C (better than LTZ1000!)
http://pdf1.alldatasheet.com/datasheet-pdf/view/48847/AD/ADR293.html

3. There are people, without extended test, written an article in 1999 introducing this 'remarkable' device.
http://www.ti.com/lit/an/slyt183/slyt183.pdf

4. LT were getting mad at this and written a design note in 2000 to criticize this using phrase like 'THIS IS A DELIBERATE LIE!'
http://cds.linear.com/docs/en/design-note/dn229f.pdf

5. Analog made modifications to the datasheet from 0.2ppm to 50ppm, test conditions also modified
http://www.analog.com/media/en/technical-documentation/data-sheets/ADR293.pdf

6. Although this ADR293 is still in production, its not recommended for new design. Rather, Analog gives an alternative ADR3450, which is CMOS based and with much larger noise.
http://www.analog.com/media/en/technical-documentation/data-sheets/ADR3412_ADR3420_ADR3425_ADR3430_ADR3433_ADR3440_ADR3450.pdf
 
The following users thanked this post: kony, mmagin

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2460
  • Country: gb
Re: Building your own voltage reference - the JVR
« Reply #104 on: March 05, 2016, 03:59:14 pm »
@zlymex - Thanks for that.

So the XFET based ADR293 has a 1000 hour long term stability of just 0.2ppm @ 25C - but the latest datasheet has it at 50ppm @ 125C !!? ... and they recommend this for battery powered devices? Did they really burn this in at 125C? why? :palm:

ETA: Design Note 229 explains it!  :rant:
« Last Edit: March 05, 2016, 04:07:31 pm by Macbeth »
 

Offline zlymex

  • Frequent Contributor
  • **
  • Posts: 556
  • Country: cn
Re: Building your own voltage reference - the JVR
« Reply #105 on: March 05, 2016, 07:05:56 pm »
@zlymex - Thanks for that.

So the XFET based ADR293 has a 1000 hour long term stability of just 0.2ppm @ 25C - but the latest datasheet has it at 50ppm @ 125C !!? ... and they recommend this for battery powered devices? Did they really burn this in at 125C? why? :palm:

ETA: Design Note 229 explains it!  :rant:
That's right. Analog must have tested some devices at 125 deg C for 1000hrs that they typically show 50 ppm drift. And they derived at first that the drift at 25 deg C to be 0.2ppm @1000hrs.
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #106 on: March 07, 2016, 11:19:52 pm »
Well, back to the topic  ;) . Here are the measurements on HP3458A for my 10V JVR unit for two weeks combined, 1ppm/division and with some room temperature markers. There are some gaps in the timeline.
Cheers

Alex

« Last Edit: March 07, 2016, 11:23:21 pm by Alex Nikitin »
 

Offline Squantor

  • Contributor
  • Posts: 38
  • Country: nl
Re: Building your own voltage reference - the JVR
« Reply #107 on: March 08, 2016, 06:51:37 pm »
Just to share my experience.

When I was matching BF862's on VGS using the setup of gate connected to source and measuring the voltage drop at a certain current. The Vgs usually climbs as the device is heating up, this was prevalent on the higher side of the IDSS current range (10 to 25mA). But I noticed that devices around 15mA did not have this effect at all, and below this current it actually was the opposite.

When I have some time I will perform some tests with the BF862. Problem is that these are SMT devices and interchanging is not as easy. I do have a socket from my old matching setup that I can reuse, but dont know about the thermal EMF quality of this socket
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #108 on: March 15, 2016, 03:51:47 am »
I've checked the load regulation on the 10V unit today. 10K (1mA) load directly on the output connectors drops the voltage by 2uV or 0.2ppm, so the output impedance is about 2mOhm. There is no visible additional voltage shift for over 30min under that load.

Cheers

Alex
« Last Edit: March 15, 2016, 03:54:56 am by Alex Nikitin »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2125
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #109 on: March 15, 2016, 09:05:36 am »
I've checked the load regulation on the 10V unit today. 10K (1mA) load directly on the output connectors drops the voltage by 2uV or 0.2ppm, so the output impedance is about 2mOhm.

Hello Alex,

that is unbelievable for me.
I would expect a impedance at least 3 orders of magnitude higer (> 2-20 Ohms).
Are you shure that the cirquit is not oscillating?
(I had some problems with my FET voltage regulator until I put some bypass capacitors on input+output).

With best regards

Andreas
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #110 on: March 15, 2016, 09:42:44 am »
I've checked the load regulation on the 10V unit today. 10K (1mA) load directly on the output connectors drops the voltage by 2uV or 0.2ppm, so the output impedance is about 2mOhm.

Hello Alex,

that is unbelievable for me.
I would expect a impedance at least 3 orders of magnitude higer (> 2-20 Ohms).
Are you shure that the cirquit is not oscillating?
(I had some problems with my FET voltage regulator until I put some bypass capacitors on input+output).

With best regards

Andreas

With correctly designed output stage it is not difficult to get very low output impedance (at DC) and good stability. I have 220nF capacitor connected across the output and the circuit is completely stable. Here is the "live" recording of the loading experiment (unfortunately, at the usual NPLC100). You can see the points where the load connected and disconnected by the transient "spikes".

Cheers

Alex



« Last Edit: March 15, 2016, 09:53:32 am by Alex Nikitin »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2125
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #111 on: March 15, 2016, 04:22:38 pm »
Hello,

so its the (modified?) cirquit variant with the INA133 buffer from page 2
measured at the output of the INA and not directly at the FET?

with best regards

Andreas
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #112 on: March 15, 2016, 10:07:30 pm »
Hello,

so its the (modified?) cirquit variant with the INA133 buffer from page 2
measured at the output of the INA and not directly at the FET?

with best regards

Andreas

Hi Andreas,

No, all my latest data are for the version with 2N4391 JFET and Ultrohm resistors, from the end of November 2015. I've used OPA227 (actually, a dual OPA2227) for the 7V to 10V amplifier and output buffer. So the output impedance is for the buffer stage.

Cheers

Alex
 

Offline The Soulman

  • Frequent Contributor
  • **
  • Posts: 542
  • Country: nl
  • The sky is the limit!
Re: Building your own voltage reference - the JVR
« Reply #113 on: March 22, 2016, 09:26:17 am »
Hello,

so its the (modified?) cirquit variant with the INA133 buffer from page 2
measured at the output of the INA and not directly at the FET?

with best regards

Andreas

Hi Andreas,

No, all my latest data are for the version with 2N4391 JFET and Ultrohm resistors, from the end of November 2015. I've used OPA227 (actually, a dual OPA2227) for the 7V to 10V amplifier and output buffer. So the output impedance is for the buffer stage.

Cheers

Alex

Hi Alex, did you or maybe someone else did measurements on the relationship between vgs(off) and the current (and therefore resistor) required for optimum (as close to zero) tempco?
Could do it the hard (on the device..) way as in the first post using temperature extremes and a variable pot but would prefer a more scientific approach, at least to get in the ballpark.  :-+

 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #114 on: March 22, 2016, 10:52:01 pm »
Hi Alex, did you or maybe someone else did measurements on the relationship between vgs(off) and the current (and therefore resistor) required for optimum (as close to zero) tempco?
Could do it the hard (on the device..) way as in the first post using temperature extremes and a variable pot but would prefer a more scientific approach, at least to get in the ballpark.  :-+

There is no single formulae - the position of the zero tempco point depends on the JFET geometry, junction temperature and the Vds, so it is possible to make a rough prediction where that point should be only for a particular type/make JFET . For the lowest tempco at a certain temperature you need to adjust it at that temperature +/- few degrees. I will collect some data for 2N4391 but don't have any real statistics yet.

One really attractive point of the JVR is that the tempco can be adjusted directly for the main reference device and so tempcos of the rest of the circuit can be largely compensated for, around the temperature of interest - or even used to "flatten" the curve in 5-10 degrees range.

Cheers

Alex
 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3633
  • Country: au
Re: Building your own voltage reference - the JVR
« Reply #115 on: March 23, 2016, 05:44:54 pm »
The old Natsemi Discrete semiconductor book is pretty useful - it has diagrams of the chip layout and characteristic data for each family of device.

http://www.introni.it/pdf/NatSemi%20-%20Discrete%20Databook%201978.pdf

The 2N4391 is a type 51 process.


This layout changed between the 1974 version of this data book and the 1978 version, so there may be different versions of the 2N4391 that behave sightly differently.

I wonder if a type 58 process FET would be lower noise - it has a much bigger area and lower resistance.

The 2N5432 (Type 58) has a pinch-off between 4 and 10V and a channel resistance 6 times lower then then the 2N4391. The package is a TO52 that is similar to the TO5 metal can.

A bit harder to get and more expensive. More likely to find them at a smaller distributor or on Aliexpress then the major distributors. I can get them for about US$6 here in Australia.

Richard
« Last Edit: March 23, 2016, 06:00:54 pm by amspire »
 

Offline zlymex

  • Frequent Contributor
  • **
  • Posts: 556
  • Country: cn
Re: Building your own voltage reference - the JVR
« Reply #116 on: April 07, 2016, 05:35:34 pm »
Hi Alex, did you or maybe someone else did measurements on the relationship between vgs(off) and the current (and therefore resistor) required for optimum (as close to zero) tempco?
Could do it the hard (on the device..) way as in the first post using temperature extremes and a variable pot but would prefer a more scientific approach, at least to get in the ballpark.  :-+
I have an old 491-page Chinese book(in pdf) called "High Stability Power Supply"where on page 121 the author derived a formula:
Vg0=Vp-0.66V
which means the zero tempco Vg is the Vp(Vgs off) minus 0.66V. And the corresponding current is
Id0 = Idss * (0.66/Vp)^2
I applied this to my DIY 100V voltage reference and it works fine.
« Last Edit: April 07, 2016, 05:45:38 pm by zlymex »
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #117 on: April 12, 2016, 07:13:31 am »
I have an old 491-page Chinese book(in pdf) called "High Stability Power Supply"where on page 121 the author derived a formula:
Vg0=Vp-0.66V
which means the zero tempco Vg is the Vp(Vgs off) minus 0.66V. And the corresponding current is
Id0 = Idss * (0.66/Vp)^2
I applied this to my DIY 100V voltage reference and it works fine.

It is not a very accurate formula, even on the same make and type of JFETs the voltage difference between the cut-off voltage and "zero tempco" voltage varies according to my measurements (at the same temperature) and on top of that the cut-off voltage does vary with temperature, so this difference is temperature dependent as well!

Cheers

Alex
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #118 on: April 12, 2016, 07:33:49 am »
I am struggling to measure the drift on this sample of the JVR. In three months it is less than I can reliably measure with the HP3458A (opt 002) and Fluke 731B as the second reference. The difference between the Fluke and JVR stays about the same for the same temperature (at 24C it is about 90-110 uV) and it is the same absolute value (+/- 1ppm) when measured with the HP after auto-cal over these three months (at the same room temperature), despite several moves between my home lab and my work lab. Here are today's measurements at home of the difference between the JVR and the Fluke, done by the Keithley 2015 with the upgraded input opamp used as a null meter at 100mV range and NPLC 10 with 10 averages.

Cheers

Alex

« Last Edit: April 12, 2016, 08:00:49 am by Alex Nikitin »
 

Offline The Soulman

  • Frequent Contributor
  • **
  • Posts: 542
  • Country: nl
  • The sky is the limit!
Re: Building your own voltage reference - the JVR
« Reply #119 on: April 12, 2016, 08:04:27 am »
I have an old 491-page Chinese book(in pdf) called "High Stability Power Supply"where on page 121 the author derived a formula:
Vg0=Vp-0.66V
which means the zero tempco Vg is the Vp(Vgs off) minus 0.66V. And the corresponding current is
Id0 = Idss * (0.66/Vp)^2
I applied this to my DIY 100V voltage reference and it works fine.

It is not a very accurate formula, even on the same make and type of JFETs the voltage difference between the cut-off voltage and "zero tempco" voltage varies according to my measurements (at the same temperature) and on top of that the cut-off voltage does vary with temperature, so this difference is temperature dependent as well!

Cheers

Alex

In the Vishay journal AN103 (http://www.vishay.com/docs/70596/70596.pdf) they use nearly the same number 0.65volt (VGS(0TC)  VGS(off) – 0.65 V) that would probably get one in the ballpark,
however what Alex is saying does make sense and testing and trimming is the way to go for optimum thermal stability.

I need to get some parts ordered and start building one!
 

Offline quarks

  • Frequent Contributor
  • **
  • Posts: 815
  • Country: de
Re: Building your own voltage reference - the JVR
« Reply #120 on: April 12, 2016, 02:52:41 pm »
Very interesting.
Now bookmarked.
 

Offline Alex Nikitin

  • Frequent Contributor
  • **
  • Posts: 974
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Building your own voltage reference - the JVR
« Reply #121 on: April 19, 2016, 08:28:54 pm »
Another update, this time ~200 hours run in my home lab, with the top temperature about 24C and some night time drops. The Keithley 2015 measures the difference between 10V JVR and Fluke 731B, 100mV range, 10NPLC + 10 averages, so the 100NPLC equivalent. 1ppm/div vertical scale. Again, this JVR build is quite stable in time.

Cheers

Alex

 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3633
  • Country: au
Re: Building your own voltage reference - the JVR
« Reply #122 on: April 26, 2016, 06:22:19 pm »
I finally got around to a few tests myself. I had some MPF102 (plastic) so I decided to test them. The zero temp coefficient point was only 1.2V at 1mA, but this can be useful if you need a reference that can run from 5V.

The voltage sensitivity with a 5V supply was 0.003V/Vsupply. At 15V it reduced to 0.0013V/Vsupply. I am pretty happy with both numbers. As long as you maintain the voltage across the reference fet to 100mV, the voltage sensitivity even with a 5V supply is negligible.

Resistance sensivity is worse. For both 5V supply and 15V supply, the output varied by about 1/3 the resistor variation. This does not include the fact that as the resistor drifts, the FET current drifts and so it gets a temperature coefficient. The TO92 package is not good for temperature tests as you can get a positive or negative change depending on where the heat is applied to the package.

There are a number of 3ppm to 5ppm reference ICs for around $3. To equal this, I will need 10-15ppm resistors.

So on this score, it may not be worth the effort.

The other issue is the long term stability and temperature hysteresis. For reference ICs, it is not unusual to see a 50ppm drift spec  for 1st 1000 hour drift and another 50ppm drift if the chip is temperature cycled from 25degC to the min and max temperatures and back to 25 deg C. It would be interesting to know what kind of drifts and temp hysteresis you get with the JVR reference. It is certainly cheaper and easier getting a hermetically sealed JFET then getting a hermetically sealed voltage reference IC, but of course, the resistor needed for the JVR is probably not going to be hermetically sealed.

If it turns out the JVR is low drift/low hysteresis, it is interesting for custom jobs, especially if the JVR is put in an oven.

Anyone have some suggestions for economical and stable resistors in the 5ppm to 15ppm range?

 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 3341
  • Country: gb
Re: Building your own voltage reference - the JVR
« Reply #123 on: April 27, 2016, 01:43:13 am »
Anyone have some suggestions for economical and stable resistors in the 5ppm to 15ppm range?

For surface mount, the Panasonic ERA series has 10 and 15ppm resistors at relatively low cost. They' range from 20p to a pound (GBP) each in small quantities [RS pricing, packs of 5 or packs of 100, 10k ohms 0.05% 10ppm to 0.1% 15ppm, 0603 and 0805 sizes].
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Online Vgkid

  • Super Contributor
  • ***
  • Posts: 2414
  • Country: us
Re: Building your own voltage reference - the JVR
« Reply #124 on: April 27, 2016, 02:51:55 am »
Anyone have some suggestions for economical and stable resistors in the 5ppm to 15ppm range?
For through hole, TE Connectivity Neohm- YR1 series offer 15ppm tc, and are cheap.
If you own any North Hills Electronics gear, message me. L&N Fan
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf