Prema 5017 / 8017

[Andreas]

Hello,

would be interesting to have a Allan Deviation Diagram on the sleeping part.

with best regards

Andreas

[maginnovision]

Here is one from alavar. Thought I'd add that was WITH math ax+b to get 100nV digit adding extra noise. Going to do another log without it tonight and compare.

[Kosmic]

I know you're gonna get it fixed but here is what I got. 12 hours from 11:00pm - 11:00am, and one graph when I was actually asleep from 2:00am - 8:00am. The heater would have been running off/on during the night as well. Still 2 seconds integration with filter on. I threw a sweater over the DMMCheck Plus while I slept to try to insulate it from the heater but the DMM was fully exposed.

Really usefull data for me. Give me a target to declare my meter fixed 

Thanks a lot!

[Kosmic]

So I finally opened up my 5017 to test some caps and do a visual inspection. I ended up changing all the electrolytic on the analog board (tested bad) and some in the power supply.



I didn't really check the dimension of the replacement capacitors I ordered, so I had to install 2 of them on the side.



Did another comparison with my 34401a and the 10V LT1021 source. Integration time was 200ms (averaging disabled) on the 5017 and NPLC 10 on the 34401a.



Standard deviation
5017:  0.0000447757571
34401a: .0000172379238

Look like changing the bad caps helped a lot. I need to check the other measurement mode though. I think there still a problem with 4Wires resistance mode.

[maginnovision]

Very good, when I had problems not long ago with some severe "noise" it turned out to be one of the power supply caps as well. One of them was very very hot the other was only warm. I replaced both of the big ones near the edge and it came back to normal.

[exor]

Is this a known issue also with 8017? I’m just about to send my 8017 to Prema for calibration and input socket replacement. I wonder if they replace these/bad caps by default?

[maginnovision]

I'm not sure if it's a known issue but I'd say that supply rails are important on any device and if the caps are sat next to the regulators they're eventually going to fail in a way you notice. I'm not sure if they do replacement by default, branadic might know better. I believe he said they replaced them on both of the 5017's he sent in but he may have asked for that.

If they update the software perhaps you could upload a drive image?

[chronos42]

I'm not sure if it's a known issue but I'd say that supply rails are important on any device and if the caps are sat next to the regulators they're eventually going to fail in a way you notice.
If they update the software perhaps you could upload a drive image?

The caps are a well known issue of these devices, all caps should be replaced.
The latest 5017 firmware is rev. 12, I have attached it.

For a firmware update of a 8017 I recommend to ask Prema for the updatesoftware.

[exor]

Thanks to both for the information, I will ask them about caps. The FW update I already wrote to Service Request form.

[jaromir]

Did another comparison with my 34401a and the 10V LT1021 source. Integration time was 200ms (averaging disabled) on the 5017 and NPLC 10 on the 34401a.

Please correct me if I am reading the graph wrong, but does both 34401 and 5017 read roughly 50-100uV peak-to-peak of noise from LT1021?

[Kosmic]

Please correct me if I am reading the graph wrong, but does both 34401 and 5017 read roughly 50-100uV peak-to-peak of noise from LT1021?

Yes around 100uV p-p. The source is powered by a bench power supply (linear).

[Kosmic]

I'm not sure if it's a known issue but I'd say that supply rails are important on any device and if the caps are sat next to the regulators they're eventually going to fail in a way you notice.
If they update the software perhaps you could upload a drive image?

The caps are a well known issue of these devices, all caps should be replaced.
The latest 5017 firmware is rev. 12, I have attached it.

For a firmware update of a 8017 I recommend to ask Prema for the updatesoftware.

Yeah, kind of disappointing that they were installing cheap capacitors  I left some of them in place since they tested ok but I should have replaced them all.

Thanks a lot for the firmware! will update mine.

[e61_phil]

Please correct me if I am reading the graph wrong, but does both 34401 and 5017 read roughly 50-100uV peak-to-peak of noise from LT1021?

Yes around 100uV p-p. The source is powered by a bench power supply (linear).

100uVpp is way to high for a 34401A. Could it be the reference source?

[Kosmic]

Please correct me if I am reading the graph wrong, but does both 34401 and 5017 read roughly 50-100uV peak-to-peak of noise from LT1021?

Yes around 100uV p-p. The source is powered by a bench power supply (linear).

100uVpp is way to high for a 34401A. Could it be the reference source?

Yes, I don't think it's coming from the 34401a. I checked the power supply and it seam to be ok. The regulator in the voltage reference seem to do a poor job at filtering noise.

Anyway, my goal was to try to fix the 5017. So, in the end it seem noisier then the 34401a but nevertheless accurate over time.

[maginnovision]

Please correct me if I am reading the graph wrong, but does both 34401 and 5017 read roughly 50-100uV peak-to-peak of noise from LT1021?

Yes around 100uV p-p. The source is powered by a bench power supply (linear).

100uVpp is way to high for a 34401A. Could it be the reference source?

My results weren't much different powering an LT1021 with a rigol dp832. About 10ppm p-p over 12 hours, 5ppm p-p while it was mostly left alone. Mine was a DMMCheckPlus completely unshielded using pretty crappy leads and alligator clips.

[e61_phil]

I attached a measurement of my 34401As connected to my Fluke 5440B at 10V. The temperature varies a lot on a hot summer day. But even with that kind of temperature variations it is far from 100µVpp (10ppm)

[maginnovision]

My ltz1000A and non-A are also way lower with temp variations but those aren't LT1021's.

[Kosmic]

Captured some data over night again, but this time with 10s integration (averaging off). And as expected, noise was reduced at around 50uV p-p (over multiple hours). I double checked the power supply and it seem to be in spec in term of ripple and noise (1mV p-p).



I'm also done testing the PM5017 and it seem to be working perfectly in all mode. So changing the caps was the only thing he needed.

[branadic]

For those of you wanting a Prema 5017 SC, there is currently a unit on ebay 274080785055

-branadic-

[Kosmic]

Captured some data over night again, but this time with 10s integration (averaging off). And as expected, noise was reduced at around 50uV p-p (over multiple hours). I double checked the power supply and it seem to be in spec in term of ripple and noise (1mV p-p).



I'm also done testing the PM5017 and it seem to be working perfectly in all mode. So changing the caps was the only thing he needed.

did a quick test with a Fluke 332a and noise was around 10uV. I guess I need to find a better supply for my 10V ref 

Will try to power the voltage ref from batteries. The good news is that it's definitely coming from the source and not the meter.

[maginnovision]

The biggest changes I see are over days with my 5017 and 8017. I usually log 2 weeks at the beginning of the month. My biggest problem is everything is in the path of the AC airflow.

[Kosmic]

A bit more data on my side.

I shorted the input of my Prema 5017, HP34401a and Solartron 7061 and logged the data for 22 hours. Resolution of all 3 DMM was set to 6 digits. Each meter were not exactly at 0 so I removed the DC offset in post processing.



Interesting how the HP 34401a is a lot less noisy than the 2 other meters.

[iMo]

I've got 45nVrms noise with my 34401A when shorted.
Try to calculate the rms in your excel:
=STDEV.P(B:B)
where ie. B is your column with measured data (no need to remove DC).

[Kosmic]

Here's the results:

HP 34401a:
3 hours = 49nVrms
22 hours = 65.2nVrms

Prema 5017:
3 hours = 235nVrms
22 hours = 242nVrms

Solartron 7061:
3 hours = 138nVrms
22 hours = 153nVrms

[jaromir]

Each meter were not exactly at 0 so I removed the DC offset in post processing.

Just curious - what was the average difference from zero for those meters with shorted input?