Agilent U1733C LCR Meter FAIL at 100 kHz mark

[Kiriakos GR]

On December 12, 2012 I am working on the evaluation of the DER EE DE-5000 LCR, and I am exploring it from top to bottom.
Previously I was had one quick experience with the Agilent U1733C LCR meter about the exploration of it main functions.
This time I got it again back as loan, so to explore it further and compare it in depth against the DE-5000.

Unfortunately for the Agilent camp this U1733C LCR meter has an issue which are not justified in any way.
 
The headline of this message is that it fails to calibrate correctly it self at 100 kHz (open / shorted leads) and the display never gets a zero reading at the shorted leads.

It did take me some time to realize the concept of Agilent about setting up this meter correctly.
For example you have to set the OS-User option as boot up one, so your last open/short calibration to be saved with out any password.
Also I did realize after some deep read of the User manual, that the complete Calibration cycle (open / shorted leads) it can be performed only by a Power up option (Press Cal and power up the meter).

I did everything by the book, and my finding is that all the frequencies’ from 100Hz up to 10 kHz are capable to calibrate and display at the Resistance in Series mode lots of nice zeros at the 2 Ohm range (lowest one) except the 100 kHz which loads at about 22 counts with just sorted alligators clips.

How does this effect the measurement?   
Well the Resistance in Series mode is what every dedicated ESR meter use so to measure the ESR of one capacitor.
By using the single Resistance in Series mode instead of the dual display (capacitance + ESR) which the ESR is displayed at the secondary LCD display, there is some gain about better visibility or even more counts, depending on the design.
In this case the U1733C as 19.000 counts meter at the 2 Ohm range it is capable to offer four digits after the decimal mark

When a low ESR measurement requires three digits after the decimal mark it becomes obvious the need for perfection even at the last count.   

One inspection of the firmware revision of this U1733C showed: Pro 00.21
I hope Agilent to investigate my report, and even better to offer one solution for it.
This meter as is it does not considered as reliable at 100 kHz ESR and low Ohm measurements. 

This meter has a serial number MY52210035.
And it does make me to wonder how in earth this behavior was not detected upon the factory certificate of calibration which came with it?

Certificate No: U1733CMY52210035
Date of Cal: 19-May-2012
Procedure: CELQ-056-AT-57
The Certificate does not have any calibration data printed on it.

Looks in a way that Agilent holds the detailed data in their computers and they are the only one who can follow the traceability if the device goes back in one of their own labs.

This new inspiration about the handling of the calibration data makes the printed certificate to have no value at the End-User, and in this case it is best to stop wasting trees about it.

[Kiriakos GR]

Further investigation on this issue shown that the meter is not capable to maintain the last full calibration (open / shorted leads) more than 30-40 seconds.
The last count starts to add +2 or +5 (it is very unstable). 

And the most important is that with such bad stability at the last counts I can not use this meter for side to side comparison in my review of the DE-5000.
There is no need to say that the DE-5000 is rock stable in this area.

Some pictures of the U1733C after successful calibration ( with the problem to be there).
My time is precious and I am not going to investigate more this subject, but I have one comment to make,
if the extra LCD display resolution is not accurate and stable, then it is not worth having it. 

[Kiriakos GR]

Finally one last picture from the DCR test in which the meter looks to also have an issue in comparison with the others.
I do consider the measurment of the U1273A as acceptable at the 12M Ohm.
The Brymen BM869 and the DE-5000 gave 100% identical measurments.
But the Agilent LCR it is far out from the expected measurment.