3457A - should I preserve SRAM contents, or purposely corrupt them by removing power?
As noted in another thread, I want to replace a battery in a 3457A that has
not been replaced in at least 11 years. The instrument has not been
calibrated in that time either. At the time it was purchased from a dealer,
I was told it was within specification, but if I wanted it calibrated I
would need to pay. I never did have it calibrated, so have no historical
data about this. I've no idea when it was calibrated, and have no cal
certificate, so have no idea of the errors at the time of calibration.
I now want to change the battery, and send it to Keysight for calibration
- I have a healthy skepticism of 3rd party labs, so would rather pay more
and get the job done by Keysight.
I'm wondering if I would be better purposely removing the battery, and
putting a short across the SRAM so I ensure the contents are definitely
lost. My logic is that
-
If I send it to Keysight and the cal data is corrupt, they will set the
meter correct. -
If sent to Keysight, with the cal data in tact, then if its within
specification they will not adjust it. So if the limit on some parameter is
1%, and the error is 0.5%, then it will not be adjusted. But if the SRAM is
corrupted, the error will be huge (if it will read at all), so it would
force Keysight to adjust it to the correct value. At that point the error
should be effectively zero given its a 6.5 digit multimeter, which means
the uncertainly in Keysight's measurements should be much lower than the
uncertainty of my meter. So by corrupting the SRAM, I should get a meter
returned to me that is more accurately calibrated than if I take the
trouble to preserve the SRAM contents. -
The adjustments are I believe software, so there's no risk that
adjusting potentiometers will cause drift to increase. -
I don't have any historical data from cal certificates, so even if I a
range is in error by 0.5%, I can't make any attempt to estimate the drift
over time. Clearly if I had cal certificates over a period of years, I
maybe able to get some idea of how the instrument is drifting, so possibly
correct for that, if it is drifting in one direction.
Thoughts?
Dr. David Kirkby Ph.D CEng MIET
Kirkby Microwave Ltd
Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Essex, CM3 6DT,
UK.
Registered in England and Wales, company number 08914892.
http://www.kirkbymicrowave.co.uk/
Tel: 07910 441670 / +44 7910 441670 (0900 to 2100 GMT only please)
In message CANX10hALwjXT+8eV8LYWdqy9EEZ+AnaWihgqEyRzntsY80v9RQ@mail.gmail.com
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
I'm wondering if I would be better purposely removing the battery, and
putting a short across the SRAM so I ensure the contents are definitely
lost. My logic is that
I would save the content before doing that, so that I could compare it
to whatever I got back from calibration.
My experience so far says that there probably is a GPIB command which
can read out the memory, but of course you still need to know the
address-space layout.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
On 25 January 2016 at 15:40, Poul-Henning Kamp phk@phk.freebsd.dk wrote:
In message <
CANX10hALwjXT+8eV8LYWdqy9EEZ+AnaWihgqEyRzntsY80v9RQ@mail.gmail.com>
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
I'm wondering if I would be better purposely removing the battery, and
putting a short across the SRAM so I ensure the contents are definitely
lost. My logic is that
I would save the content before doing that, so that I could compare it
to whatever I got back from calibration.
My experience so far says that there probably is a GPIB command which
can read out the memory, but of course you still need to know the
address-space layout.
I've not looked myself, but I did see a note on the web there was no
documented command to do read out the cal data on a 3457A. But in any case,
unless I know how to interpret those values, a bunch of meaningless 0's and
1's is not going to be a lot of use. Of course, if I could save them I
would, then in the event someone ever figures out what the numbers mean, I
could do a comparison.
It would be nice to know if the meter is in/out of spec, but to me, I think
having it put as close as possible to correct is more worthwhile, even
though a metrologist who gets their meter calibrated regularly would no
doubt see it different.
Dave
In message CANX10hBS-pkJ_0KsF6FQYv7A4izmRF+k-hi+5rBHk03ZbDRKHA@mail.gmail.com
, "Dr. David Kirkby (Kirkby Microwave Ltd)" writes:
I've not looked myself, but I did see a note on the web there was no
documented command to do read out the cal data on a 3457A.
Apart from the HP8568 I have not seen them documented on any instrument.
But a good first shot: Try sending the command "MREAD 0" and see
what happens...
And it's not like disassembling the firmware of instruments of this
vintage is a big deal anyway...
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
On 25 January 2016 at 11:12, Dr. David Kirkby (Kirkby Microwave Ltd) <
drkirkby@kirkbymicrowave.co.uk> wrote:
As noted in another thread, I want to replace a battery in a 3457A that
has not been replaced in at least 11 years. The instrument has not been
calibrated in that time either. At the time it was purchased from a dealer,
I was told it was within specification, but if I wanted it calibrated I
would need to pay. I never did have it calibrated, so have no historical
data about this. I've no idea when it was calibrated, and have no cal
certificate, so have no idea of the errors at the time of calibration.
I now want to change the battery, and send it to Keysight for
calibration - I have a healthy skepticism of 3rd party labs, so would
rather pay more and get the job done by Keysight.
I'm wondering if I would be better purposely removing the battery, and
putting a short across the SRAM so I ensure the contents are definitely
lost. My logic is that
-
If I send it to Keysight and the cal data is corrupt, they will set the
meter correct. -
If sent to Keysight, with the cal data in tact, then if its within
specification they will not adjust it. So if the limit on some parameter is
1%, and the error is 0.5%, then it will not be adjusted. But if the SRAM is
corrupted, the error will be huge (if it will read at all), so it would
force Keysight to adjust it to the correct value.
At that point the error should be effectively zero
Thoughts?
Someone sent me a private email, saying that if by purposely removed the
power from the SRAM caused the 3457A to fail its self-tests, which is may
well do, then Keysight may consider the 3457A needs repair, and so offer a
fixed-price repair. Hence my "trick" to get this calibrated as accurately
as possible would have backfired, and I'd be out of pocket. In fact, on a
3457A, the fixed-price-repair would probably cost more than the instrument
is worth, as these are not worth a lot now.
Hence I would not do this before getting clarification from Keysight. It
might be possible to get them to set it right, even if within spec, but I
think their calibration routines are very automated, so its probably not so
easy to get a special calibration, whilst paying for their cheapest
calibration service.
Dave