[volt-nuts] Keysight 3458A vs Keithley 2002 8.5 digit multimeters.

Illya Tsemenko illya at xdevs.com
Mon Jul 18 07:11:04 EDT 2016

Well, it's the topic many can go on and on about.
As with everything there are specific cons and pros of both units, but 
here are my 5c:
+ K2002 is smaller, lighter (4kg vs 10kg), less powerhungry (sometimes 
it's important if you send unit for cal overseas, or take to remote site).
+ K2002 can support scan cards (standard, or low-EMF ones) via expansion 
+ K2002 DCI/ACI goes up to 2Amps
+ K2002 goes bit higher on ACV frequency (15MHz, but performance there 
is so-so)
+ K2002 natively supports thermocouples, RTDs, SPRT thermal sensors and 
custom settings
+ K2002 supports rare rainbow unicorn nanovolt preamplifier Keithley 
1801 (substitute can be bodged up but that's not beginner's project)
- Old K2002's require capacitor replacements (units with S/N 06xxxxx and 
05xxxxx). Mandatory! Meters from 2005+ year have S/N starting with 11xxxxx
- Readings noisier than 3458A
I have two units in use, and have worklogs for both :
https://xdevs.com/review/kei2002/ - newer K2002 from year 2007
https://xdevs.com/fix/kei2002_u2/ - older K2002 from 1994 which was 
repaired after capacitor leak damage.
Service of these Keithley meters (and 2001, which are very similar 
design) is not complicated, but require attention and knowledge what you 
doing. There only few custom parts used.

Now regarding industry standard 3458A. Reason why you saw only 3458A in 
metrology lab is simple:
Superior ADC (many rightfully think it's best on market, ever), amazing 
linearity (JJA was required to test linearity with good confidence), 
very easy calibration (just need 10V reference and 10KOhm standard 
resistor to get 3458A calibrated to uncertainty of used standards), 
Artifact calibration (ACAL) to significantly remove impact of 
temperature variation on accuracy, proven performance. Many labs use 
3458A as main transfer standard due to it's ADC performance.

It comes at cost though, as key components of meter's A/D are custom 
hybrids. Meaning if you got sour pill, and A/D drifting, you will have 
to replace expensive A/D board. Newer Agilent/Keysight branded 3458A's 
are essentially same boxes, just with newer hardware and likely with new 
FPGA-based digital brains. Performance is same though, unless you really 
getting early 1990's meter. Many design issues and items I got covered 
in my repair worklog: https://xdevs.com/fix/hp3458a/

Brief 10V comparison : https://xdevs.com/datalog_tcomp/ - sampling 10V 
connected to 2001+2002+3458A
Some of the A/D noise comparisons (meter input shorted) - 

So decide what fit best your needs. 3458A is overall higher performance 
instrument, but in lot of applications outside of metrology it's 
performance benefits may not cover up for higher price and size. Both 
3458A and 2002 are still supported and sold new by manufacturers. 
Regular calibration of these is likely be over 1000$USD, and lab-grade 
calibration for 3458A is usually in range of 2600$USD.

P.S. there are also few other 8.5d meters, but much less information 
available on those, as less people have them. Only "modern" 8.5d meter 
is 8508A, which cost even more than 3458A without much performance 

18/07/16 17:54, Dr. David Kirkby (Kirkby Microwave Ltd) пишет:
> I am interested to hear people's view on the relative merits  of the
> HP/Agilent/ Keysight 3458A vs the Keithley 2002.
> I noticed that when I recently visit the standard lab for voltage at NPL,
> the place was littered with Agilent 3458As, but I did not notice a single
> Keithley 2002. But maybe NPL get better discounts from HP/Agilent Keysight
> than Keithley/Tektronix.
> I notice that the used prices of 3458A is higher than 2002s.
> Also, are there any things ons should be looking out for when buying a
> 3458A or 2002? I have in the past thought an Agilent front panel was a
> better than an HP one, as it would not be as old, but this is not always
> true - see my comment later.
> What S/N are best avoided? Does anyone have a record of S/Ns of Keighley
> meters with time?
> Very recently I had a somewhat scruffy HP 6674A PSU with a damaged LCD. It
> cost me $50 or so to pick up a clean Agilent front panel with a new
> display.  The display actually turned out to be unsuitable,  yet my Agilent
> badged 6674A is actually a fairly old HP unit.  So white my intention was
> not to rebadge the PSU, that happened as a result of a legitimate reason to
> get it working. So this really indicates how an old HP 3457A could be made
> to look a newer Agilent one.
> Dave.
> _______________________________________________
> volt-nuts mailing list -- volt-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
> and follow the instructions there.

More information about the volt-nuts mailing list