Empathy List Archives

FS

Frank Stellmach

Sat, Oct 26, 2019 8:09 PM

Hi Philipp,

you definitely have to clean the fan filter of your unit.. standing free
on a table, and with a clean filter, the internal temperature rise is
12..13°C at most.

Defining a metrology grade 3458A would explicitly require to define and
control the environmental conditions in the specification, like reduced
temperature range, specifying the ventilation requirements, like no use
in a rack, by using a different filter, and also by regularly checking
the internal temperature by firmware.

The 3458A is designed and specified as a system DMM, inside a rack, in a
hot manufacturing environment, like in a electronics manufacturing line,
but never was intended and built as a metrology DMM, as hp and KS still
try to advertise.

Your argument, that a sample distribution is root cause of these wide
specification limits, does not convince me:

The annual and temperature drift (with ACAL) is mainly determined by the
drift of the two internal references, due to the ACAL technique, the
rest of the circuit does contribute only marginally, but on the w/o ACAL
specifications .

Therefore, the 8508/88A simply beat the 3458A, because they use an LTFLU
running at 45..55°C only, instead of 95°C, and due to the heavy usage of
stable Vishay BMF resistors, which also contribute to the superior
short- and midterm stability of these instruments.

So only the sample distribution and yield of the LTZ and 40k resistor
affect the annual drift specification limits for the 3458A.

Concerning the LTZ reference, that is already covered by their
monitoring / selection process, where they have for sure have yield
problems already.

If they would simply reduce the temperature for metrology purposes, like
in practice done on the FLUKE DMMs, they would directly get near 100%
yield for 2..3ppm/year (@65°C), due to the ACAL DCV feature..

The 8508/88 on the other hand have to account for the drift of other
components, to achieve e.g. these 4ppm/year, although the LTFLU very
probably performs more like 1..2ppm/year (inside the 732B, @ 47°C, for
example).

I think the 3458A would at least be on par with the FLUKE DMMs if KS
would simply chose the same metrology grade resistors and oven
temperature as DATRON/FLUKE had done...

Frank

Hi Philipp, you definitely have to clean the fan filter of your unit.. standing free on a table, and with a clean filter, the internal temperature rise is 12..13°C at most. Defining a metrology grade 3458A would explicitly require to define and control the environmental conditions in the specification, like reduced temperature range, specifying the ventilation requirements, like no use in a rack, by using a different filter, and also by regularly checking the internal temperature by firmware. The 3458A is designed and specified as a system DMM, inside a rack, in a hot manufacturing environment, like in a electronics manufacturing line, but never was intended and built as a metrology DMM, as hp and KS still try to advertise. Your argument, that a sample distribution is root cause of these wide specification limits, does not convince me: The annual and temperature drift (with ACAL) is mainly determined by the drift of the two internal references, due to the ACAL technique, the rest of the circuit does contribute only marginally, but on the w/o ACAL specifications . Therefore, the 8508/88A simply beat the 3458A, because they use an LTFLU running at 45..55°C only, instead of 95°C, and due to the heavy usage of stable Vishay BMF resistors, which also contribute to the superior short- and midterm stability of these instruments. So only the sample distribution and yield of the LTZ and 40k resistor affect the annual drift specification limits for the 3458A. Concerning the LTZ reference, that is already covered by their monitoring / selection process, where they have for sure have yield problems already. If they would simply reduce the temperature for metrology purposes, like in practice done on the FLUKE DMMs, they would directly get near 100% yield for 2..3ppm/year (@65°C), due to the ACAL DCV feature.. The 8508/88 on the other hand have to account for the drift of other components, to achieve e.g. these 4ppm/year, although the LTFLU very probably performs more like 1..2ppm/year (inside the 732B, @ 47°C, for example). I think the 3458A would at least be on par with the FLUKE DMMs if KS would simply chose the same metrology grade resistors and oven temperature as DATRON/FLUKE had done... Frank

PC

Philipp Cochems

Sat, Oct 26, 2019 8:39 PM

Hi Frank,

I use my private 3458A mostly as a ratio meter. Therefore, I'm not that
much worried about the temperature (changes), but I should have a look at
the filter ;).

I would also have no doubt, that the 3458A could compete with the Fluke
meters, if HP had installed expensive components. But the 3458A is a box of
cheap parts (my opinion) which are regularly calibrated against the two
internal standards. It is a great marketing stunt to make the customers
believe that ACAL is something great. In my opinion it is only to
cover/compensate the mediocre ingredients. Many meteorologists don't like
such self adjusting boxes, but that (in combination with the superb ADC
linearity) made it possible to create a "cheap" metrology DMM.

The 8508/88 are specified with 2.75 ppm/year not 4 btw. Our 8508A drifted
by 0.8ppm in the first year and I assume this rate will go down with age.

I don't think Keysight will offer a second (metrology) version of the
3458A. I also think that 10V are good enough on all the meters. If you
really need less uncertainty, one should use a 10V standard in combination
with the 3458A. But there are other differences between the 3458A and the
Fluke meters. Especially low and high ohms are much better on a 8508/88A.
Real ratio measurements, low TC and no Auto Zero are other points which
lead me to buy a 8508A and now a 8588A instead of an additional 3458A (we
already have more than one). But I don't want to start a Fluke vs. 3458A
discussion again ;).

Best regards
Philipp

Am Sa., 26. Okt. 2019 um 22:12 Uhr schrieb Frank Stellmach <
frank.stellmach@freenet.de>: