Again I have to vote for Prema meters and their service.
Had my unit out for service and calibration in Mainz. The unit was sold to Bosch in 2001. After 17 years without calibration the values on DCV only in 300mV range where slightly out of spec. ACV fully in spec, ACV+DCV fully in spec, DCI fully in spec, ACI fully in spec, ACI+DCI fully in spec, 2W and 4W ohms where selective out of spec.
Only 6 electrolytic capacitors where replaced. Saftey inspection passed wihtout any issues. Scanner is fully working. So this gear is reliable as hell and received without any additional errors today. The price was below half of what Keysight asks for calibration of a 3458A, so maybe you pay them for the extra errors they integrate to your gear?
-branadic-
The accuracy is also below half of what the 3458A will do.
Sure, but Prema understood service from a customers point of view as LT did, while Keysight is more like a vampire, hungry for your money.
-branadic-
If you think Keysight is hungry, try Fluke Calibration
Well, calibrating a Keithley2002 at Tektronix/Fluke/Keithley is 480€ for the calibration + 105€ for the report + 50€ for transport.
Keysight asks for 712€ without telling you what you pay in detail for, a vampire policy.
The price list of Prema is public and says 55€ for the as found report + 262€ for the adjustment and calibration including report.
Since calibration is a fully automated process, everything else would be silly, Keithley and Keysight ask to much money, that's for sure. But the business model works since this crazy quality management system forces you to blow out this money and there are only a few competitors out there with sometimes unknown or suspect quality.
-branadic-
712€ is much and ~1500€ for the 8508A is much more and perhaps it is much more than it is worth, BUT you can't compare a 3458A or 8508A with the lower accuracy meters. There is no calibrator which can directly calibrate a 3458A or 8508A with a good TUR. There is some more work to do to lower the uncertainties. In contrast to that a 20ppm meter is really as easy as you described it.
The quality management don't force you to bring the 3458A to keysight. You have to have a unbroken chain to your NMI, but another cal lab is also ok. We pay much less than 712€ for our 3458As.
That said, you will always find someone who is more barefaced, no worries whatsoever. It's pretty simple, put a label like "baby", "marry" or "metrology" on it and you can ask any price you want. As long as noone complains, nothing will change.
-branadic-
The quality management don't force you to bring the 3458A to keysight.
e61_phil So true
One can calibrate 3458A even to 9V battery and 10K resistor 0.1% from Digikey, and it will be calibrated indeed. Question is, to what?
Without exact datapoints (we didn't see any of it for above-mentioned 8017 nor 3458) talking about calibration prices is just waste of bits and bytes. Even 100$ for 10V 20ppm calibration might be too much, while 2000$ for 10V 2ppm calibration - justified if needed. And automated calibration does not mean it's free. Maintaining calibrator/standards to their best spec is lot of time, money and effort, automated or not.
Without exact datapoints (we didn't see any of it for above-mentioned 8017 nor 3458) talking about calibration prices is just waste of bits and bytes.
So is your comment on that.
And automated calibration does not mean it's free. Maintaining calibrator/standards to their best spec is lot of time, money and effort, automated or not.
Noone asked for something free, but the bucks they ask for is definitely to much. And the price for printing what they have measured is simply barefacedness. You can have your opinion, I have mine. Thank you very much.
-branadic-
There are currently two Prema 5017 units sold as defect / spare parts
on ebay. At least on of them has the scanner build in. Maybe someone wants to start a repair thread?
-branadic-
I couldn't resist and since noone of this board took the chance I bought both units. They arrived today and both of them came with the scanner
One is hanging, which simply indicates a communication issue between the MCU board and the analog board, maybe due to some bad power rail. Need to check that, which shouldn't be of an issue, since I received the schematics from Prema Semiconductor and since almost all parts are standard components available today.
The second one didn't power up. The issue was pretty simple to find. The POWER ENTRY MODULE EMI FILTERS (Delta Electronics 06AK5) had a broken 110V/220V switch. So I simply used the one from the first unit and bingo, it's working. Ordered two of them today and will replace them in both unit.
By the way, I got some insider information why they stopped selling and developing meters. The 5017 and 8017 have been the first meters using SMD components. They only sold about 400 unit per year. In the very beginning of the year 2k it was hard to almost impossible to get even standard passive SMD components. Prices for precision opamps doubled. TI controller discontinued even though they where told the that they will be in production for a very long time like the 8086. So most of the developer had nothing to do. This was the time when the management decided to stop developement and production. After SMD crisis, when management wanted to continue, all developers had left the comany, so there was noone left except some technicans. Pretty sad.
-branadic-
Thanks for the story. Now it's the opposite situation, that you can't find thru-hole packaged parts. For example direct replacement for LF356BN ain't easily available, you either have to go expensive hermetic TO-packages or bodge adapters for SOIC8 or SSOP parts.
Sure there are other parts in DIP that can be subbed, but I like to keep repairs as 1:1 when possible , unless I'm convinced in improvement.
Also have a fully working Prema 5017 that I wouldn't mind selling, if the price is right.
I want $800 or 800Swiss Francs. Local pickup or shipped to Switzerland/Germany.
branadic, did you had the chance to compare the noise level on your different units ? also I'm wondering if servicing improved the noise problem.
Recently got a Prema 5017 and compared to my 34401a it's really noisy. Pretty sure something is wrong inside. probably some caps.
branadic, did you had the chance to compare the noise level on your different units ? also I'm wondering if servicing improved the noise problem.
Recently got a Prema 5017 and compared to my 34401a it's really noisy. Pretty sure something is wrong inside. probably some caps.
I bought his recently serviced Prema 5017. I calibrated a couple of power supplies but I've had it connected to my DMMCheck Plus(aged about 6000 hours now) and it's been sitting between 4.9999365 and 4.9999512 for a couple hours. Seems ok considering the conditions here. When the AC isn't going off and on it doesn't usually move more than 1 uV a step when it does(10s time).
I bought his recently serviced Prema 5017. I calibrated a couple of power supplies but I've had it connected to my DMMCheck Plus(aged about 6000 hours now) and it's been sitting between 4.9999365 and 4.9999512 for a couple hours. Seems ok considering the conditions here. When the AC isn't going off and on it doesn't usually move more than 1 uV a step when it does(10s time).
Interesting. It confirm my suspicion. 2s integration time ? filter on or off ?
Thanks a lot for the info.
I bought his recently serviced Prema 5017. I calibrated a couple of power supplies but I've had it connected to my DMMCheck Plus(aged about 6000 hours now) and it's been sitting between 4.9999365 and 4.9999512 for a couple hours. Seems ok considering the conditions here. When the AC isn't going off and on it doesn't usually move more than 1 uV a step when it does(10s time).
Interesting. It confirm my suspicion. 2s integration time ? filter on or off ?
Thanks a lot for the info.
2s, filter on it changes by 1uV when it changes.. Can't be sure if it's the meter or ref though. dmmcheck plus isn't the best especially looking at uV. I think it's a couple uV p-p noise.
EDIT: My rs232 card died(found out when I tried to hook up meter) when the ne one comes I can graph it.
So I spent some time yesterday doing a setup to compare my 34401a to the 5017.
The setup:
- 10V LT1021 based voltage source.
- 34401a in High impedance mode, NPLC 10, 10V range.
- 5017, 2s Integration, Filter ON.
- Boths meter connected to the same source.
- 2 hours warmup time (source is always hot).
- Sampling every 3 sec for 7 hours.
This is the deviation in ppm of each meters
Considering the integration time of the 5017 vs the 34401a (2s vs 166.66ms), I was expecting the 5017 to be a lot more stable. So my 5017 is definitly sick.
The curve really looks like a sick meter.
There are a few more test that could be done: like check the noise with a short at the input in a few ranges. This would eliminate the meters reference from the test.
One could also directly check the noise of the 5017 internal reference.
A possible other cause of trouble could be a relay contact at the input. When not used for a long time contacts can have contact problems.
branadic, did you had the chance to compare the noise level on your different units ? also I'm wondering if servicing improved the noise problem.
Recently got a Prema 5017 and compared to my 34401a it's really noisy. Pretty sure something is wrong inside. probably some caps.
Did some minor comparisons but nothing I really documented to show here. Your noise looks really high and I would suggest checking the hardware. If you sent me an email I can send you the schematics including the upgrades they made. You can also contact Prema for the software update that all my units got when they were for service and calibration.
-branadic-
I'm going to log overnight. After about an hour I was getting peaks of 2ppm/-1.6ppm and average of -9.34E-10ppm, so most likely longer sample time would make that better. Running 2S, filter on. Not logging temp but it's still getting colder. Should stabilize soonish.
I will probably stop the tests right away on my unit and put it in the repair queue for a complete disassembly
Thanks a lot for your feedback guys.
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.