ADI ADMX3652 - 6½ Digit, ±20 V DVM

[Jendas]

Hi All,
some new ADC module or rather DVM with range ±20 V, 6½ digit resolution and UART interface with SCPI support from ADI. Maybe for some voltnuts here
More specs on ADI page:
https://www.analog.com/en/products/admx3652.html#product-overview
and ADI wiki:
https://wiki.analog.com/resources/eval/user-guides/admx3/admx3652


and is available on Farnell too, for 660 USD:
https://www.newark.com/analog-devices/eval-admx3652z/evaluation-board-digital-voltmeter/dp/72AK1816?st=admx3652

Jendas

[dobsonr741]

Wow. Could not wait to see a teardown. Or a guessing game what ADI parts are making it up.

The specs only suggests and did not state the two input channels are galvanically isolated, only hints with the 70V max rating. Does anyone know?

[jbb]

I think this will make test engineers somewhere happy; might be able to solve a problem without forking out for a whole new 6 1/2 digit DMM. Especially if there are multiple testers.

I wonder how the go/no-go decision was made. It’s not exactly an evaluation board… but I guess Analog does make some modules / mini bench gear.

[coromonadalix]

what a joke at this price

but i must admit,  they permit you to CAL  the unit directly if you have a reference source ....

very high baud rates speed

a few goodies,  but hard to justify this pricing for me

they should be honest from the start they use FTDI drivers  it seems, instead of a no name driver package  loll  have to read their spec sheets

and i would have made some provision say, 2x external relays controls pins, say to provide 200v and 1kv input, just add precision dividers and tvs / gaz tube discharge  etc ... 

[jbb]

what a joke at this price
...

Yeah, it's pretty expensive.  Still cheaper than a comparable multimeter though (assuming 2V and 20V ranges will be sufficient)!

...
and i would have made some provision say, 2x external relays controls pins, say to provide 200v and 1kv input, just add precision dividers and tvs / gaz tube discharge  etc ...

That sounds suspiciously like a multimeter to me...

I expect that by the time someone designs a multimeter-like front end for this sort of module it probably makes sense to go all-in and deploy a reference and ADC yourself.

[Andreas]

Hello,

for a multimeter I would at least expect a INL specification.
And the 70V Maximum rating between analog input and digital GND is also too low for a serious DMM.

My best guess for the parts used:
Some microcontroller with integrated sigma delta converter like ADUCM36x
A ADR1399 voltage reference divided by factor 3.

With best regards

Andreas

[guenthert]

I think this will make test engineers somewhere happy; might be able to solve a problem without forking out for a whole new 6 1/2 digit DMM.
[..]

But it's not a DMM.  It's a multi-range DC ADC + voltage reference only.  Building a DMM with commonly available features using that will double the prize I'd think.

I rather expect to find it to be part of very specialized (and very expensive) gear.

I couldn't find it stated, but I miss galvanic separation between input and digital output and power supply.

[Kleinstein]

The infos on the module are a bit sparse and in part contradictory (different accuracy specs in the datasheet, Wiki page and product page).
A point that is not clear is if the inputs and the power are isolated from each other.

The noise and claimed accuracy look good, but the input bias (1 nA) and input resistance are not that good. It may not be possible to just add an input divider divider.
Lower input resistance for the 200 mV range is odd. Noise for the high PLCs looks no longer that good - maybe some 1/f noise from a non AZ input amplifier.
Given the noise specs, especially for low PLC this suggests a relatively good SD ADC with some +-2.5 V FS range. It may well be separate µC and ADC chips.
There also seems to be a 2.5 V internal "supply" that seems to be critical to have an error message for it.

The protection specs also don't look that good - external connections may want added buffers and protection. To me this does not look like a very practical product.

[Roehrenonkel]

Hi all,
 
for 6.5 digits resolution 200 seconds are needed (datasheet),
limited to 2000m altitude (nothing for a mountain-man),
no isolation (at that price!) and limited range.
 
For someone who can affort it the small form-factor is nice,
else get a used HPAK/Fluke-DMM.
 
Best regards

[iMo]

..Could not wait to see a teardown. ..

Yep, at least we'll get how to design the AFE, finally.. 

At the front end, the ADMX3652 incorporates a precision ADC driver with integrated
precision resistors. The precision resistors can be strapped to achieve different gains for the
ADC driver, which allows the user to match with the input signal range.
All active and passive components in the circuit, including the thin-film resistors with precise
matching, are designed by Analog Devices, Inc., and are factory calibrated to achieve a high degree
of specified

With the ADUCM360 and such a trimmed AFE it is easily achievable, indeed..
The INL of the ADUCM360 itself is +/-10 - 20ppm of the FS based on the DS.

The 310mA power consumption indicates the module has got a part of the ADC ovenized, imho.

PS:
For example:
ADUCM360+others around 10mA@5V  ....  10mA@5V
Vref 30mA@15V      ....   100mA@5V (switcher)
AFE 15mA@+/-25V  ....       200mA@5V (switcher)

Total ~310mA@5V..

[coromonadalix]

i can understand this design or applications

but as other(s) wrote, specs are somewhat incomplete

and yes we should see that module as an advanced ADC  ....  not a dmm   loll    i thought so, my bad

You can get very good adc in the 12-14-16 bit ranges and more,  some can go at 125Mbs of sampling too or way higher  ...


As a DMM
You have many pci  pci-e  pxi pxie card / devices who can be bought cheaper in some models,  sure the form factor is not helpful .... and some rare usb dmm's from keysight or NI

There is some ideas or project here about an 6.5 digit front end ...   even the lm399 zenner  or the famous ltz1000 pricing has gone   bonkers



I can't go over this Analog  pricing  .... 

[miro123]

I've looked at proveded documentation.
My conclusion  - It is definitly not substitute of DMM even if you need limited DC range.
Spec is marketing pumped. No exception here  any semiconductor manufacture do it. Mixing typical with most likely and average parameters.
e.g. ADMX3652 vs HP 34465a DMM6500
 -  AD - 90 days accuracy: <17 ppm of reading range at 2 V range
 - HP34465a - 90 days 10V native range - 0.0020% + 0.0004% - Tacal =  +-2C
 - DMM6500 - 90 days 10V native range - 0.0020% + 0.0005%  - Tacal =  +-5C
Giving the noise parameters where all integrated SD ADC excels and not giving INL is another example of marketing trick.

Application that require decent DC accuracy and small size is the niche of this product.