Keithley 2015 teardown

[bingo600]

Well i think i goofed in the above measurements ....
When you use 6.5 digit you can't be sloppy ...

Above i had all 3 meters connected to the REF at the same time , and was using 3 different wire sets.

I have just dropped that approach , and am now only "the same" wireset , and connecting one meter at a time to the REF

The results seems a lot more as i would have expected.

LM399 / 22.3 C :

2015        : 06.83068/69
3401a      : 06.83070/71
3478a      : 06.8314

Geller / 22.4 C:

2015        : 09.99998
3401a      : 09.99993
3478a      : 10.0008

I have to get my 3488A switches going , and add an automatic Temp. measurement to the measure set.

But now both the 2015 & the 34401a seems so be close to the expected 10v.

The 3478A (ebay) has always been "high" , and i wonder if i can calibrate "just the 10v" my self.
I don't have any other decent references ready yet , my EDC E100C isn't restored yet , and it will still only give 10v & below.

I think the 2015 is the one i have to call my "REF meter" , as it was calibrated on 20-may-2011.

I can hope my arriving 3457a has a recent/valid calibration (ebay seller wrote ... pulled from a working lab) ,
but i don't expect much.

Wonder what a 3457a cal costs here in Denmark or Germany.


Ohh ... I wonder why the 2015 is higher than the 34401a on the Geller , and just the opposite on the LM399 


/Bingo

[KJDS]

Glad it's working well.

I've still got a few left, if anyone wants one then PM me.

[bingo600]

Well after 24'hours warmup of the Geller referece & The Keithley 2015 i have hit the magic number   

My 34401A (2003 factory cal) is 7ppm away fro the 2015 that has a 2011 calibration , not that bad .... 

/Bingo

[nukie]

I have one example of the LM399 even though rated at 20ppm at 1000 hours it barely drifted after 1 Year thats 8760 hours and still counting. This is a late manufacture I got more than a year ago. Unfortunately I got only 1 so I don't have a lot of data.  If my finding is true then it is bullshit that Agilent 'hand select' parts because I happen to own an example of a LM399 that perform like a well aged reference, or perhaps I was lucky. Attenuation resistors are more likely to contrinute to long term of multimeter drifting.

It makes me wonder, if LM399 is so damn good, then the LTZ1000 is more than enough for 6 digits multimeter calibration.

[mimmus78]

Hi guys,

you all seem pretty expert in references.

I would appreciate if anyone can help me on this other >> POST << to check the reference of my keithley 2000.

Sorry for the offtopic.

[Christe4nM]

What do the experts think of this: Precision Voltage Reference Achieved <1ppm/deg C Performance. Site of the manufacturer
It's even on the Maxim website.

The LM399 and LTZ1000 have quite the pedigree and documentation to back their performance. Yet, if you think about it, shouldn't there be a modern day's ultra-low tempco reference? Might those Rejustor-guys be on to something? (Taken into account that we engineers are sometimes quite conservative in hanging on to our proven methods ) Would this be a modern day's method to get a (Temperature) stable reference? I have no clue, so I'm interested in what you guys think.

[Harvs]

I'm intrigued as to how in a commercial product like this they scale down the LMx99 reference to a useful voltage for the ADC (I'm assuming the ADC takes <5V.)

With quick search on digikey it becomes pretty apparent that a single sub 2ppm resistor is at least as expensive as the LM399.

[Harvs]

Wow, I thought there must have been a more economical way.

It sure makes the previously mentioned MAX6350 or LTC6655 seem like a much easier option for normal applications (i.e. just drop in one of those directly connected to your ADC of choice and your done.)

[Harvs]

Thanks for the info.  I'm just asking out of interest, I don't have a specific application at the moment but you've summarized if quite well.

Cheers

[PTR_1275]

Hi guys,

Hoping that this is the right place to ask, I couldn't see another post about the capacitor replacement specifically and seeing as it was mentioned in here, I thought I would ask.

I have just bought a 2015, the display board has a manufacture date of August 2003, yet the date codes on the main board are as late as 0541.

Was it a bad batch of Nichicon capacitors that will ruin them, or is it best to just replace them anyway? I saw someone else asked about top and bottom boards, but didn't see a reply to it. I can see 7 electrolytics on the top board and 6 on the bottom board.

I don't know much about capacitor selection, is there a better range of the nichicon caps that I should be putting in, or just leave it with the series that they came as?

Thanks
-Chris