Building your own voltage reference - the JVR

[Cerebus]

Hi Alex,

Is J2 used to start up the circuit? I'm assuming it pulls up the input of the op amp when power is applied, in order to generate a voltage to allow J1 to begin regulating.

Thanks in advance

Ha! You beat me to it. I was just about to type out the very same question.

[BNElecEng]

  you know what they say about great minds

[iainwhite]

Here is the latest schematics (for the unit running for 2 years now).

I appreciate you making this available - Thank you.

[Cerebus]

  you know what they say about great minds

Yeah, and also, sadly, about fools. 

[vindoline]

Here is the latest schematics (for the unit running for 2 years now). R1 is selected for the zero tempco point, R5 and R6 for 10V output.

Cheers

Alex

Hi Alex, thanks for posting your schematic. I'm thinking of building one up and I have a question. R3 on the output is marked "k12" I'm assuming this is a typo and it should be 1k2? Thanks!

[Alex Nikitin]

Hi Alex, thanks for posting your schematic. I'm thinking of building one up and I have a question. R3 on the output is marked "k12" I'm assuming this is a typo and it should be 1k2? Thanks!

R3 is 120 Ohm .

Cheers

Alex

[stijena1973]

Hi Alex, thanks for posting your schematic. I'm thinking of building one up and I have a question. R3 on the output is marked "k12" I'm assuming this is a typo and it should be 1k2? Thanks!

R3 is 120 Ohm .

Cheers

Alex

Any new developments? Statistics maybe?

[Alex Nikitin]

Any new developments? Statistics maybe?

Not much , the unit is still running 24/7 and didn't drift more than 5ppm over last four years.

Cheers

Alex

[BravoV]

Any new developments? Statistics maybe?

Not much , the unit is still running 24/7 and didn't drift more than 5ppm over last four years.

Cheers

Alex

Looking at the price/performace, I call it boringly good. 

[Kleinstein]

Any new developments? Statistics maybe?

Not much , the unit is still running 24/7 and didn't drift more than 5ppm over last four years.

Cheers

Alex

Looking at the price/performace, I call it boringly good. 

The long time performance very much depends on the stability of the source side resistor. A very high stability resistors can be quite expensive.
In addition to being stable the resistor value has to be right for the given FET. So it may take more than just 1 resistor.
If it just comes to short time stability / noise the JFET version can still be good and cheap.

[Alex Nikitin]

The long time performance very much depends on the stability of the source side resistor. A very high stability resistors can be quite expensive.
In addition to being stable the resistor value has to be right for the given FET. So it may take more than just 1 resistor.
If it just comes to short time stability / noise the JFET version can still be good and cheap.

The changes in the source resistor are attenuated by a factor of ~30, so a 100ppm change in the resistor value creates ~3ppm change in the output voltage, essentially even a quality metal film like RC55 series may be sufficient as a source resistor. As usual, the main problem is with the 7V to 10V amplifier, as resistor changes are not attenuated much. In my prototype old wire-wound resistors are used and holding surprisingly well as the <5ppm change in 4 years is for the 10V output (the change looks smaller but 5ppm is the reasonable limit that I can measure it to).

Cheers

Alex

[guenthert]

The changes in the source resistor are attenuated by a factor of ~30, so a 100ppm change in the resistor value creates ~3ppm change in the output voltage, essentially even a quality metal film like RC55 series may be sufficient as a source resistor.

  While it might not be nearly as tolerant as a LM399 (with its impressive 0.1Ohm dynamic resistance), contrasting this with, say, an old temperature stabilized Zener like the 1N829 with r=15Ohm, fed from self-referenced 10V supply via a ~515Ohm resistor at about 7.5mA for which  a 100ppm change in the current setting resistor value causes a voltage change of about 10ppm, I'd say that's a very good value. 

  So, when can we buy it? 

[Alex Nikitin]

  So, when can we buy it? 

I have no intention to produce it for sale. As the thread title says: build your own. It is not that difficult  .

Cheers

Alex

[TERRA Operative]

Have the schematics been taken down? I can't see them anywhere.

[jaromir]

This one?
https://www.eevblog.com/forum/metrology/building-your-own-voltage-reference-the-jvr/msg775190/#msg775190

[Kleinstein]

The circuit with the better choice 2N4391 is just some 12 post before:
https://www.eevblog.com/forum/metrology/building-your-own-voltage-reference-the-jvr/msg1343014/#msg1343014

Jaromir linked the old one with an J211.

[TERRA Operative]

Hmm, well that's weird. I can't see any attachments at all...

I'll see what's up with my browser etc.

[EDIT] Dove into the HTML stuffs and found the link. Got it now, all good.

[BravoV]

No problem here what so ever, all images are fine at Firefox.

[exe]

I don't see the image. Inspecting the element, I see computed image width and height are both zero. I suspect onload="pagespeed.CriticalImages.checkImageForCriticality(this);" for some reason doesn't work. May be it doesn't interract well with one of my browser extensions (disabling uBlock Origin didn't help).

[TERRA Operative]

I figured out the issue in my case, if I disable the HTTPS-Only mode if Firefox settings, the images work.

Seems like the OP doesn't have any form of HTTPS enabled on his website so Firefox just says a big nope to any crosslinks from there if HTTPS-Only mode is enabled.

Here's a direct link to the schematic on the website.
http://www.ant-audio.co.uk/Test_Gear/10JVR_2N4391_001.gif
(R3 = 120 ohm)

[SilverSolder]

Reposted circuit as a local image, to reduce dependencies:

[Kleinstein]

For a practical implementation the resistor R1 would likely be a combination of a few resistors, like  4.7 K in series with 300 Ohms and than maybe some larger ones in parallel to the 300 Ohms to do the fine trim. This way the resistors for the fine trim don't need to be highest stability and one does not need to order a  highly stable resistor at a specific value (it may be practical in cases - e.g. custom wire would ones)

The OPs can be lower grade: for a start the good old OP07 should be OK.

[exe]

As I understand, J1 is the jfet that generates reference current. What is the purpose of J2?

[Gyro]

As I understand, J1 is the jfet that generates reference current. What is the purpose of J2?

To provide bias for reliable startup (J1 is biased from the opamp output).

[exe]

Ah, got it, it stops conducting when the circuit is active.