[volt-nuts] volt-nuts Digest, Vol 12, Issue 11

[Marvin E. Gozum]

Hello Dick,

Thanks for sharing your insights.  It is extremely helpful.

I fear the exact same fate for myself, in that I anticipate any 
'working' eBay 3458a sold 'as is' that is not tested by a competent 
vendor or calibrated by the seller to some spec, or moreso NIST 
traceable, will be bound for some expensive repairs in the near 
future, after purchase.  The going rate for working meters is 
$5k.  Thereafter, there isn't much DIY I can forsee I may be able to 
do myself that can keep the 3458a factory accurate, so professional 
metcal costs should be factored in unless the meter starts to 
'settled' in and stabilize.


At 03:26 PM 8/25/2010, Dick Moore wrote:
>Hi, Marv -- RE my 3458 -- It was very clean externally, leading me 
>to think it had been used in a test or lab environment, but it 
>needed a display board and a ROM board, both of which I installed. 
>Then it went to Loveland for calibration. That was last September. 
>Because the ROM board was replaced, there was no "before" data to 
>compare the cal to. It goes back to Loveland this September, and now 
>there will be "before" and "after" data which will provide an 
>estimate of drift in parameters. It is not an -02 high-stability option unit.

Great!  If you don't mind sharing the results of this cal results, it 
would be much appreciated.  Theoretically, an aged 3458a should start 
to approach the stability of the high stability board but without the 
factory blessing and testing, but one can't tell unless the cal 
numbers come rolling in over time.


>The ROM board change essentially made this a good-as-new unit, at 
>least according to Loveland -- it passed all pre-cal tests with 
>flying colors and cal'ed without problems.
>
>I believe that Loveland uses a Fluke 572x calibration system, which 
>for DC is at least as good as the 732B. As to AC -- well I think it 
>probably is as good as there is right now for cal use outside of 
>labs. The specs for the 3458 reflect the uncertainty of the 
>calibration standards, and the AC specs definitely show that. I 
>bought that 3458 for $1600 on fleabay -- didn't want to pay too much 
>more because of the ROM battery problems on older units with the 
>Dallas ROMs that had built-in battery back up. The new ROM board has 
>ROMs that have a snap-on battery which is apparently easy to 
>replace, though I wonder how long they will be available via Agilent....

I read of those battery ROMs issues, even more maintenance costs.  Is 
there an anticipated replacement interval or a system test to 
determine its remaining life?  Are they specialized batteries that 
cannot be substituted? are they costly?

>In any case, the display board was $400+, the ROM board was $500+ 
>and the cal was about $500, so my total investment to this point is 
>around $3100 -- a lot, but worth it. I just trust the 3458 without 
>wondering if it is right. It is the only high-res measurer in the 
>shop now, so it is right by definition. As Steve Rooke likes to say, 
>"A man with one clock knows what time it is. A man with two clocks 
>isn't ever sure."

Yes, a statistician would look at both clocks and know what it means ;)

I use the 3456a stat functions to reduce the uncertainty of the last 
digit of measurement, and target as small a variance as 
possible.  Likewise, will multiple meters, you can compare the means, 
and the variances.  You may not have a rock solid number, but a 
better handle of the uncertainty of the LSD of the scale.


>Best,
>Dick Moore
>
>



Sincerely,



Marv Gozum
Philadelphia, PA