EEVblog Electronics Community Forum
Electronics => Metrology => Topic started by: oz2cpu on January 14, 2022, 05:08:16 pm
-
i am monitoring a 34465A DC voltage, about 4V
running constantly on, voltage is in parallel with other meters,
so i can easyly double check it is NOT the input that jumps !
but once every 2-4 days i see intermittent reading, it is even the last digit
but the 2nd last digits that is 1-2 counts off !!
when i push a button, the error goes away,
i did send it for firmware upgrade (was locked for upgrades) but still it does this !!
anyone else see this ??
-
Hello,
thats about 120 ppm deviation. This is clearly not within spec when the environment conditions did not change largely.
I would consider some "popcorn" noise up to 1 ppm as limit for a 6.5 digit instrument.
See a measurement of my K2000.
Is the going back to "normal" value only related to a "button push" or is there also a jumping back without actuation?
Is the "offset compensation" still active or did you deactivate it for the measurement? (this would give a contiouous slow drift).
with best regards
Andreas
-
you mean Auto zero ? see attached
i turned that off as you see, that is why i love to take pictures of the screens, this way i can remember better what is going on.
next time i see the problem, i better try record a video, and then press a button.
this is a sw bug, that no normal mortal will discover,
it takes days to show up, and goes away when i touch a button, very wierd
display update / refresh / saver ?
the unit looks like it is alive
-
Turn autozero back on and then try it again for a few days.
-
you mean Auto zero ? see attached
this is a sw bug, that no normal mortal will discover,
Yes auto zero is what I meant.
Setting auto zero off is only intended to increase measurement rate for short measurements.
But also measurement error is increased.
For longer measurements you have to use the default setting with auto zero = on.
Accuracy specification is also with auto zero = on.
It is not a bug it is a feature (read the manual).
with best regards
Andreas
-
The non Az mode could in part explain the behaviour. For a long term measurement one should use the AZ mode.
The jump after a long time in the non AZ mode is still not correct behavior. It looks like that after a long time the meter decides it needs to have a new zero adjustment even without the AZ mode active. This is still a problem, though less critical and with rare occurance and easy to circumvent. It can effect a measurement done after having the meter idle for 2 days and than just looke at a short sequence with the meter by chance already in the right settings. When using the non AZ mode it is still good practice to force a zero adjustment before the actual sequence, e.g. by switching range or short time activate auto zero or call for an AZ once (via computer). The non AZ mode is anyway mainly usefull via the computer interface, as it is mainly for reading faster than the screen can handle (e.g. < 1 PLC).
-
He is not complaining about the drift. He's complaining that meter was showing something on screen that was largely in error and then he touched the button and then meter has show different (accurate) value.
Like it stopped refreshing screen.
Or did I understand wrong?
-
The wrong reading after a long time may be due to drift in the non AZ mode. Some 1 mV off still looks quite high for this, but maybe OK if the meter was not fully warmed up at the start.
The problem and possibly still a bug is than the jump after hitting a key, which may trigger a new zero correction cycle. Depending on which key is pressed, this may acutally also be OK and the intendent side effect / secodary function of that key.
-
If it is due to a missing AZ the OP should observe a constant drift downwards, not an instant jump down.
-branadic-
-
Hello from another proud owner of 34465A.
Never seen this behavior.
1/ Perform self test
2/ Place source on digital scope, set to trigger on any slight AC HF pulse check for some days or weeks if source has spikes
3/ If in warranty, contact Keysight, do RMA for repair.
Kind Regards,
Jon
-
I don' think this is a broken meter. It is more like a user error in doing a long time measurement with Auto zero turned off.
Than there seem to be a more minor software/ firmware bug, that may cause a unintended AZ cycle one in a while - still not clear which keys were pressed and if this may be OK. Anyway the non AZ mode is usually not really useful for with manual reading - though there is the rare case of choosing the non AZ mode to get less disturbance from current spikes at the input with a very sensitive signal source. With a shorter run time, they may still be some jump in the curve, just not as visible.
-
you mean Auto zero ? see attached
i turned that off as you see, that is why i love to take pictures of the screens, this way i can remember better what is going on.
next time i see the problem, i better try record a video, and then press a button.
this is a sw bug, that no normal mortal will discover,
it takes days to show up, and goes away when i touch a button, very wierd
display update / refresh / saver ?
the unit looks like it is alive
Hello,
using AZ OFF during long term measurements is definitely a false usage of the instrument. I wonder, why you don't observe a continuous drift of the reading of your source.
Anyhow, are you sure to use a very stable source, so you would have any chance of detecting such effects? You might use the data logging feature of the 34465A to monitor the drift behavior.
The sudden return to the "correct" value at key press might be due to the description in the manual:
AZ OFF: the instrument uses the last measured zero measurement and subtracts it from each measurement.
It takes a new zero measurement each time you change the function, range or integration
time.
You also use 10MOhm input impedance, instead of AUTO, which would be High Z. (Why?)
There once existed a 3ppm difference between both modes and which was corrected with FW 2.11.
Another deeply hidden >10ppm error was corrected with 2.14, that was connected to the HV compensation of the 100:1 divider.
I did not find any further such errors later on, especially I did not see such a dip during my usual stability measurements on 34465A and 470A, so I doubt that your finding is another one.
Which FW version do you use?
Hopefully 3.0 at least, where all these former errors were corrected, and the DIGI option is available for free.
I also do not understand, why your unit should be locked for FW updates, another user error, evidently.
There's a specific procedure to unlock the instrument with a standard pass code (see manual), and then you are surely able to do FW updates yourself, either over USB connection, or via memory stick.
The update utility is required.
A good description video from Keysight is available online, performed by Jim Durr. Search for: 34460A, 34461A, 34465A, 34470A TrueVolt Updating Firmware from USB, IOS, or Linux
Jim supported me many years ago to notify Keysight about the two above mentioned faults, and get FW updates, I think he's retired as of today, what a pity.
Last, you can better make screen shots of your settings and put it here as a bmp or png file. Search manual for 'screen shot'.
You can also use the WEB interface to steer and virtually display your 34465A instrument, and make a screenshot there.
Anyhow, I would be interested in your findings, provided you're using correct settings of your instrument.
Frank
PS: There once was another Keysight video: Updating Firmware From the Front Panel of Truevolt Series DMM
It's currently not available, due to maintenance operation on the Keysight knowledge database.
-
I thought that was a temporary error, a spike. If that's the case, I wouldn't think lack of AZ is the issue (that I'd expect to cause a drift rather than a spike). I would rather guess EMI and the other meter was just too slow to pick it up.
-
NEWS !!
finally i got this behavior on a video
hope this explain in more detail what is going on (sometimes the error value, is a bit worse, but you get the point)
https://youtu.be/DBspfWwQTJo
-
NEWS !!
finally i got this behavior on a video
hope this explain in more detail what is going on (sometimes the error value, is a bit worse, but you get the point)
Yes, it makes it crystal clear that this is happening because you inappropriately have Auto Zero turned off, just as everyone here has speculated. See the second paragraph of Dr. Frank's reply for the most detailed, exact description of what is going on.
-
thanks a lot, last time i "complained" about this, the advice was the oposite as far as i understood it,
now i turned auto zero on, and lets see in a few days, if this is all gone or not..
to Frank : FW is A.03.03-02.40-03.03-00.52-03-01
this unit was send back to my instrument responsible, he performed the fw upgrade
i dont mess with brand new calibrated company property (at least not with out asking first)
he showed me the firmware lock screen, it was asking for a password,
he also tried another 34465A same lock screen, he was in contact with our supplier, got the code, and could continue.
SEC option not installed
DIG Included
MEM not installed (pm me, if you know the trick ?? since i might get a similar unit,
for my self, once i find a place that sell them in white, i really hate the new black only)
Thanks for all the deep tech info, I did not see any difference from 10M to Auto, it bothers me a little bit, it dont say 10M/10G
but it only say 10M/Auto, even in Auto, how do the user expect to know, what it chooses as its Auto ?
My test : i keept looking at the displays 4 times pr day, and dont see any change at all, so i expect my voltage source is at good as the meter.
then suttently, the 34465 jumps 1-3 mV and stay wrong.. it is not a slow climb in error as you suggest.
i think the worst i have seen was 8mV of this suttently apearing error.