Agilent U1733C 100kHz mode: unstable high readings normal?

[Per Hansson]

Got this Agilent U1733C on eBay around 3 years ago, I have not used it extensively but for my simple needs it has been fine.
But yesterday I was testing some ultra low ESR capacitors and everything looked fine first but then I realized all readings at 100kHz where an order of magnitude too low.
That is to say the datasheet says 8mΩ but the meter read 0.0008Ω, not 0.0080Ω as it should be, it might always have been like this.

So I used a Pomona MDP-S-2 shorting bar to take the below readings.
Note how in 100kHz mode it is arbitrarily high.
Doing a user calibration with the shorting bar I get better readings.
But they are still unstable at 100kHz.

I opened the unit up and it is evident someone had a go at it:
One of the plastic PCB supports was broken off.
And the PCB had soldering paste, dirt and solderballs around U18 & U6 (NXP 74HC4053D).
I cleaned it off with IPA but that did not affect the readings at all.
The SMD electrolytic caps C19 and C83 do infact look like they are bulging.
And C44 near the above mentioned area looks like it went through a war, they all read ok-ish in circuit ESR except maybe C4 that was twice of the others.

The unit has firmware version: Pro 00.25
Maybe someone else has a similar meter and could remake the tests to give me an idea what is normal?
Also maybe someone can explain how to switch from OS-Factory to OS-User without changing the startup default setup value to either one?

Code: [Select]

OS-Factory.
Pomona MDP-S-2 shorting bar:
DCR: 0.0012 Ω

R 100Hz: 0.0086 Ω
R 120Hz: 0.0086 Ω
R 1kHz:  0.0086 Ω
R 10kHz: 0.0078 Ω
R 100kHz:0.0365 Ω


Z 100Hz: 0.0086 Ω
Z 120Hz: 0.0086 Ω
Z 1kHz:  0.0086 Ω
Z 10kHz: 0.0085 Ω
Z 100kHz:0.0425 Ω

L 100Hz: 000.1 µH
L 120Hz: 000.1 µH
L 1kHz:  00.00 µH
L 10kHz: 0.055 µH
L 100kHz:0.035 µH


Code: [Select]

OS-User.
Pomona MDP-S-2 shorting bar:
DCR: 0.0002 Ω

R 100Hz: 0.0000 Ω
R 120Hz: 0.0000 Ω
R 1kHz:  0.0000 Ω
R 10kHz: 0.0001 Ω
R 100kHz:0.0012 Ω


Z 100Hz: 0.0002 Ω
Z 120Hz: 0.0000 Ω
Z 1kHz:  0.0000 Ω
Z 10kHz: 0.0002 Ω
Z 100kHz:0.0016 Ω

L 100Hz: 000.0 µH
L 120Hz: 000.0 µH
L 1kHz:  00.00 µH
L 10kHz: 0.000 µH
L 100kHz:0.000 µH


[Per Hansson]

Rewatching Daves teardown video #234 of the U1733C I noticed the same area has flux residue on his board too.
Maybe it is from hand soldering the through hole power connector in that area? If so it's a bit how'ya doing to not clean the flux residues! 
Yesterdays numbers where with an ambient temperature of around 30°C I redid them this morning at around 20°C
The numbers are then around 15% lower, except for the 100kHz ones that are ca 50% lower and DCR that goes in the opposite direction.
Reducing the temp to ca 10°C by putting the meter in the fridge for a couple hours brought another 15%.
And now that the temp is up to 30°C again the same numbers are back, so is this normal behavior? was the meter calibrated in a fridge at Agilent? 
Also can someone ID the brand and model of those SMD electrolytic caps?

Code: [Select]

OS-Factory.
Pomona MDP-S-2 shorting bar:
30°C:   20°C:    10°C:
DCR: 0.0012 Ω 0,0037 Ω 0,0098 Ω
R 100Hz: 0.0086 Ω 0,0071 Ω 0,0062 Ω
R 120Hz: 0.0086 Ω 0,0071 Ω 0,0062 Ω
R 1kHz:  0.0086 Ω 0,0071 Ω 0,0062 Ω
R 10kHz: 0.0078 Ω 0,0066 Ω 0,0058 Ω
R 100kHz:0.0365 Ω 0,0180 Ω 0,0060 Ω


[TheSteve]

Specified performance is at 23 degree +- 5 degrees.

[RoV]

I think the ESR accuracy of the U1733C @100 kHz, even in the lowest full-scale range, is much worse than 0.8 mohm.
Check annexed page of the User's Guide: if I am reading correctly, in the 2 ohm range, even ignoring Q (note 4), we get 1% of 2 ohm + 50 digits, or 0.02 ohm + 50 digits.

[Per Hansson]

Thanks for pointing that out TheSteve and RoV!
I still would really appreciate it if someone with a U1733C could reproduce my tests with a shorting bar: it would easy my mind
Just to give an example if I measure the actual ESR of a low ESR Chemi-Con PSG polymer capacitor in the OS-factory mode with no test leads it is all over the place:

Code: [Select]

OS-Factory.
Chemi-Con PSG 1000µF solid polymer capacitor rated 8mΩ (0.0080Ω) at 100kHz:
ESR 100Hz:  0.0207 Ω
ESR 120Hz:  0.0203 Ω
ESR 1kHz:   0.0064 Ω
ESR 10kHz:  0.0044 Ω
ESR 100kHz: 0.0244 Ω

Of course with a tolerance of 1% + 50 counts at 100kHz a reading of 0.0250Ω + the expected ESR of the capacitor is within tolerance.
But then with specifications like that I don't know that this meter is any better than the venerable $20 component tester!
But just to be clear it is that the values form a "bathtub curve" by significantly going up at 100kHz that concern me...

[RoV]

Uhm, sorry I was wrong. 1%+50 is 1% of reading + 50 digits, not 1% of full-scale. The full-scale portion is already managed by the "digits" part.
However, since ESR is derived from impedance, the reading to be used is higher and this is why there is the correction term based on Q.
However, the 50 digits alone give 0.005 ohm. Obviously, actual performance will likely be better: these are the values checked when the instrument is sent for calibration.
I have just tried with mine reading a 100 uF 900 V film capacitor (ICAR LNK-P2X-100-90) rated 0.55 mohm. At 100 kHz I get a 0.0343 ohm readout by connecting to the same pole; zero with delta-null is not allowed in ESR mode; when I connect one side to the other pole I read 0.0367-0.0369, so measured ESR is 2.4-2.6 mohm. Contact resistance and uncertainty plays a role here!

[Per Hansson]

Thanks for the clarification, agree that reading specification sheets can be dizzying
I guess it might be normal the behavior then, still would love if someone could repeat my tests with a shorting bar on their unit

[TheSteve]

I do have a u1733a but would need to use a diy short made from two banana plugs and wire. I also won't have the same capacitors to test.

[exmadscientist]

I will note that when I did a lot of LCR meter work a few years back, 100kHz was about the frequency when everything went to crap due to the leads. Yes, your meter might run up that high (or even higher!), but any measurements using leads will be so dependent on the leads themselves that you just can't trust them. To make it work you have to lock down the leads, conductors near them, then run the open-short calibrations (preferably twice in a row... not entirely sure why it makes that much difference, but it seems to). Then pray.

I'm convinced that it's in the nature of LCR meters that they just don't want to be precision instruments. Sometimes you can force them... but that goes better if you know what you're trying to get away with.

[Per Hansson]

Thanks for your comment.
Just to be clear the shorting bar is just that of course
And the PSG capacitor I referenced was measured with no leads.
I.e. the capacitor was directly plugged into the special port with spring clamps inside.
That thing does raise a few questions though, because you can't calibrate it!
A quality LCR meter really needs 4-wire leads to measure low resistances in my opinion!

[PushUp]

I.e. the capacitor was directly plugged into the special port with spring clamps inside.
That thing does raise a few questions though, because you can't calibrate it!

Don't you have a "Shorting Bar" for that? Before using the internal contacts of my Voltcraft LCR-400 (APPA-Rebrand), I always use the "Shorting Bar", which was part of the whole set:






https://www.conrad.com/p/voltcraft-lcr-400-digital-cat-i-display-counts-20000-1301292


Interesting, that one of the most expensive handheld LCR-Meter's (at least before covid and war) does not include this accessories...


Cheers! 

[Per Hansson]

Don't you have a "Shorting Bar" for that? Before using the internal contacts of my Voltcraft LCR-400 (APPA-Rebrand), I always use the "Shorting Bar", which was part of the whole set

Wow, now that is interesting, just goes to show you don't always get what you pay for!
I'm glad I only paid $325 for it a few years ago, the retail price for what you get is not really what I'd call a bargain!

[PushUp]

I'm glad I only paid $325 for it a few years ago, the retail price for what you get is not really what I'd call a bargain!

That was the reason why I referred to "covid19 and the war of aggression", because before this time it was less than half the price, what I did pay for the set: 213 Euro at the top before and 430 Euro at the bottom after covid19 and the war of aggression!







...the current prices of testgear are exorbitant and I have the feeling, that they will never be as low as before...and in addition to this, you cannot even buy testgear at all and have to wait up to 4 month to get a Rigol DM3068 as far as Batronix, Conrad/Digitalo and others are concerned, for example...

$325 is not cheap, but probably fair, however, without the decisive accessories, I think it makes sense to invest in a single "Shorting Bar".


Cheers!