NEW PDVS2mini DC Voltage Reference from Ian Johnston - Reviewed.

[Grandchuck]

It is an amazing product.  Mine does not seem to be drifting at all.  The reason I must use the word "seem" is due to an interim calibration/adjustment for my 3458A.  I wasn't careful enough and should have made more measurements and records.  I confess to being a bit of a sloppy volt nut

[IanJ]

i am not sorry for kicking this topic back alive :-)
wow amazing product you made there Jan, all looks super professional

I really like to see some closeups of the mecanical implementation of the LM399
you say it is under the display ?
and inside 3D printed isolation case ?
the display looks like it can be taken out, so this detail could be revealed,

Hi,

Here's the LM399 cover.........3D printed in conductive PLA, and grounded to the pcb GND via it's mounting screw, a pad and a ring around the LM399.

[IanJ]

Spec of the conductive PLA from the datasheet:-

How Conductive Is It?
● Volume resistivity of molded resin (not 3D Printed): 15 ohm-cm
● Volume resistivity of 3D printed parts along layers (x/y): 30 ohm-cm
● Volume resistivity of 3D printed parts against layers (z): 115 ohm-cm
● Resistance of a 10cm length of 1.75mm filament: 2-3kohm
● Resistance of a 10cm length of 2.85mm filament: 800-1200ohm

Ian.

[Anders Petersson]

Are there any long-term stability measurements? For example calibration over several years.

[IanJ]

Are there any long-term stability measurements? For example calibration over several years.

The closest I have is a customer who bought one in Oct. 2019, and then damaged it in Oct. 21 and it was sent to me for repair. A protection diode was replaced and then a full re-cal.
Unfortunately, back in 2019 I didn't produce as full a cal cert as I do now, however, you can see the BIT COUNT changes I had made on the re-cal which will give you some idea.

As follows:-

[Anders Petersson]

The closest I have is a customer who bought one in Oct. 2019, and then damaged it in Oct. 21 and it was sent to me for repair. A protection diode was replaced and then a full re-cal.
Unfortunately, back in 2019 I didn't produce as full a cal cert as I do now, however, you can see the BIT COUNT changes I had made on the re-cal which will give you some idea.

Thanks. Comparing a few of the data points, I only see 1-2 counts correction.
If you place a few units in different storage conditions and record their calibration every few months, that could answer the "tba" placeholder for drift in your specifications to increase the (already high) value of your product.

Nitpicking: The certificate specifies 11 digits for voltage output which seems optimistic. Also, including some calibration/accuracy estimation of your 3458A would be helpful.

[guenthert]

The closest I have is a customer who bought one in Oct. 2019, and then damaged it in Oct. 21 and it was sent to me for repair. A protection diode was replaced and then a full re-cal.
Unfortunately, back in 2019 I didn't produce as full a cal cert as I do now, however, you can see the BIT COUNT changes I had made on the re-cal which will give you some idea.

Thanks. Comparing a few of the data points, I only see 1-2 counts correction.
If you place a few units in different storage conditions and record their calibration every few months, that could answer the "tba" placeholder for drift in your specifications to increase the (already high) value of your product.

Nitpicking: The certificate specifies 11 digits for voltage output which seems optimistic. Also, including some calibration/accuracy estimation of your 3458A would be helpful.

First of all, those numbers are most impressive. 

     There's nothing wrong per se with 11 or more digits, but w/o estimate of the measurement error, they are a bit questionable.  Sure, we can look up the uncertainty Agilent promises to better with the 3458A, but that tells us little about your unit nor how much additional error is introduced when measuring the calibrator (I'm sure there are codified / documented best practices, but an ad-hoc approach might be to use an example unit, measure it's voltage a number of times over multiple days, inverting polarity and calculating the standard deviation from that -- that at least gives a hint about the quality / reproduce-ability of the measurement procedure).

[IanJ]

Thanks guys, some good points there.

My automated procedure just logs directly what the 3458A output gave at the time with an NPLC of 100.
Are those numbers reproducable?......no, certainly not at 11 digits, but i figured if thats the numbers the 3458A sees then thats what i record. To be honest, i didnt think twice about it.

I'll look into pulling those digits back on the cal certs on future ones, or maybe adding a statement to the certs.

Thanks,

Ian.

[TiN]

Or just remove everything after 7th digit and call it a day.
I wonder if Ian considering to use new 1399 reference in PDVS "3" mini version? 

[IanJ]

I wonder if Ian considering to use new 1399 reference in PDVS "3" mini version? 

Hmmmmmm!

Ian.

[TheDefpom]

Thanks guys, some good points there.

My automated procedure just logs directly what the 3458A output gave at the time with an NPLC of 100.
Are those numbers reproducable?......no, certainly not at 11 digits, but i figured if thats the numbers the 3458A sees then thats what i record. To be honest, i didnt think twice about it.

I'll look into pulling those digits back on the cal certs on future ones, or maybe adding a statement to the certs.

Thanks,

Ian.

I would be inclined to leave all the digits there, might even help identify minor drifts with the 3458A... there are "only" 5 decimal places on the PDVS2mini, with the last one known to have a little rounding due to DAC conversion, I really wouldn't worry about it, people will only be looking up to the 5th/6th decimal place I expect, the rest is for your own records.

[IanJ]

I would be inclined to leave all the digits there, might even help identify minor drifts with the 3458A... there are "only" 5 decimal places on the PDVS2mini, with the last one known to have a little rounding due to DAC conversion, I really wouldn't worry about it, people will only be looking up to the 5th/6th decimal place I expect, the rest is for your own records.

NIST would have trouble with 11 digits......heading into superconducting JAWS territory.
9dp's on my certs are not stable, not reproduceable so kinda pointless being there.
If i run the autocal again the numbers wouldnt be the same.
The PDVS2mini isnt that stable......so TiN is right, 7 digits will do.
It's a Precision Voltage Source, not some high grade piece if metrology gear.

Gotta be totally honest/clear when selling.........

Ian.

Update, it's fixed (example):-

Code: [Select]

volts10.Text = Format(Val(txtr1aBIG.Text), "#00.0000000")

[J-R]

Has anyone experimented using the PDVS2mini and a resistor voltage divider to generate much smaller voltages or with an external circuit to make a precision current source?

[skander36]

Hi @IanJ !
Two quick questions.
1. Why aren't you used AD 5791 for DAC?
2. The schematics can be obtained, in order to make an output amplifier that can be driven by PDVS2mini?

Thank you!
 

[IanJ]

Hi @IanJ !
Two quick questions.
1. Why aren't you used AD 5791 for DAC?
2. The schematics can be obtained, in order to make an output amplifier that can be driven by PDVS2mini?

Thank you!

The AD5791 is very expensive, and requires a more complicated power supply circuit. Easy to do but affects the BOM.
The schematic diagram is not available I am afraid.

Ian.

[skander36]

Indeed, is more expensive and now is "unobtanium".
There is a project with AD5791 as a calibrator but max 20 uV step.
I'll see how I will extract output of the calibrator for interfacing an external amplifier.
Thank you!

[deepfryed]

The AD5791 is very expensive, and requires a more complicated power supply circuit. Easy to do but affects the BOM.
The schematic diagram is not available I am afraid.

Ian.

On that topic, another question is why not use a discrete PWM DAC ? It looks like some of the old HP voltage sources use that design to get 6.5 digit accuracy. I'm only a hobby tinkerer, so if that's a silly question feel free to say so.

I'm thinking of making a hobby grade programmable source as learning exercise with a decent 10v reference like LT1021C, with a discrete PWM DAC driven by a STM32, analog switch like ADG1611 and a low pass filter. 

[deepfryed]

Heh, DAC schematic from 3245A. This might explain it

Still going to tinker with a simpler version and see how I go

[skander36]

About a month Ian has posted on Youtube a repair video of 3246A  - https://www.youtube.com/watch?app=desktop&v=0EPc8ttbU3Q

[p.larner]

is the schematic+fw available for this to homebrew one?

[skander36]

is the schematic+fw available for this to homebrew one?

https://www.eevblog.com/forum/metrology/new-pdvs2mini-dc-voltage-reference-from-ian-johnston-reviewed/msg4287775/#msg4287775

[J-R]

Just saw this on Ian's PDVS2mini page (https://www.ianjohnston.com/index.php/onlineshop/handheld-precision-digital-voltage-source-2-mini-detail):

"I have decided for logistical and health reasons to stop making PDVS2mini’s permanently.
If you’d like to purchase the rights to the design, all design files and my remaining stock of parts then please contact me.
The PDVS2mini is a very successful and well respected product, it deserves to live on.
Business enquiries only"

[deepfryed]

Oh noes, I've always wanted to get one and missed out on the last round. Too bad, hope you're ok Ian. Take care.

[IanJ]

Yeah I am fine.....health wise nothing drastic just can't hover over the workbench for days-on-end anymore.
I've had a few enquiries thus far and options to look at.......hopefully one of them will work out and the PDVS2mini will continue.

Ian.

[J-R]

Whew, glad to hear your health issue is nothing drastic!

Maybe you could find someone to do the harder parts like populating the PCB. I also think it would still sell at a higher price point, so you could take your time and make fewer?