RANT - Buying a 3458A why is that hard?

[TiN]

TiN there is "repair per incident" that is "just" 2300 USD so you are not going to spend others 4000 for any repair.

Repair usually does not include calibration, which is in case of A3 board replacement is required. Of course, one can go with regular calibration, which is much cheaper than standards one for ~2.5k, so there are options for any requirements.

Even in business you don't have unlimited budget and all you spend must produce a ROI. This is especially true if you are very small.

Sure, that's why I said cost of ownership meter like 3458A, rather than price of having the box. Just after few years of use and maintenance, cost of calibrations alone will exceed MSRP of meter itself.

HighVoltage

The guy that does my official calibrations claims that the older HP units are much more stable because of those ceramic chips.
May be we should put an older and a newer model next to each other and compare?

This is case which needs a photo first , too see those magical ceramic parts  . Reader's unit I was fixing recently was the oldest I ever saw, 1989-1990 parts with SN 04xxx, and yet it uses all the same plastic opamps as ones made in 00's.
So before we speculate any more, let's get solid info what is different. So far even photo of newest A3 2015 year from KS site shows same ceramic ADC hybrid and plastic opamps around.

Old units are likely more stable just due to aged parts and reference .

CAL?72 test shows only ADC stability. Meter still can have other issues/drifts on other functions/ranges, but show perfect CAL?72 data.

[mimmus78]

TiN this "repair per incident" includes also calibration to upper standards. After this official refurbishment you can also apply to warranty agreement for few hundred bucks per year.

Anyway I actually need more the stability than absolute accuracy. I'm building 3 or 4 LTZ1000 references and I will send the best to be calibrated. I'm also planning to buy a 10k entry level standard resistors or some of this ultrastable Vishay "cans resistors" that where proven to drift only fraction of ppm per year.

This will be fair enough for what I need in my business, and it will also cure my volt nutting needs for some time.

I don't even need AC calibration as it's only needed for high frequency > 100kHz and for me 30kHz is more than enough.

I prefer to spend this other 4000 Eur to buy other equipment that can verify the meter than just the thing.

[Andreas]

I'm building 3 or 4 LTZ1000 references and I will send the best to be calibrated.

You have a "man with 2 clocks" problem:
How can you determine wether the HP3458A reference (LTZ1000) or your built references
drift faster (usually in the same direction).

The only way out is a Fluke 732B with frequent calibrations.
But in this case you might not need a further LTZ1000 reference to be calibrated.

Or the other way round:
If you cannot tell which of your 4-5 (including the HP3458A) references really drifts the least
you could also take the difference voltages between the LTZ1000 references (in 100mV range)
with a much cheaper (6.5 Digit) instrument.
You will only need 2 additional calibration points of one of the LTZ1000 references to determine the best.

With best regards

Andreas

[mimmus78]

Andreas:

as I write before I don't need absolute accuracy to ppm as 50 pmm will be more than adequate to check calibration
of instruments I sell. I know for this I can also use the 6.5 digits meter, but nobody can ensure me that all the ranges
will stay in specs for the needed time on this type of instruments.

A 3458A is not only capable of calibrate itself but in doing this you can also monitor calibration constants of each
range and spot any excessive drift from last autocal. If you find excessive drift somewhere you can usually tear
down the instrument, find what's happening and eventually fix it.

So my plan is initially to use this 4 LTZ1000 just to check with a reasonable confidence that the 3458A is not
drifting too much. Than I hope to find at least one reference that I trust enough so that I can send it some cal lab (or
someone I trust) for calibration and import this 7V with a reasonable uncertainty to my lab.

As for the 10K I think it is more easier as any calibrated Vishay VHP202 or VHA518-7 will be more than enough for
indirect checking that this 40K resistor inside the 3458A has not drifted too much. Those resistor have proven 2ppm
x 6 years drift, it wouldn't be so much hard in a second time to characterise a couple of this resistors for temperature
and use this as entry level standard. Even with a drift of 3ppm per 6 years I will be good for more than a decade.

I know this is quite excessive for my requirements but this is a quite enjoyable path too.

[TiN]

That's almost the very path I went year ago with my rusty 3458.

Yesterday I received an offer from a dealer in the same ballpark of the other dealers I contacted.

So we will soon to see pics of new toy at your home?

[plesa]

Andreas:

as I write before I don't need absolute accuracy to ppm as 50 pmm will be more than adequate to check calibration
of instruments I sell. I know for this I can also use the 6.5 digits meter, but nobody can ensure me that all the ranges
will stay in specs for the needed time on this type of instruments.

A 3458A is not only capable of calibrate itself but in doing this you can also monitor calibration constants of each
range and spot any excessive drift from last autocal. If you find excessive drift somewhere you can usually tear
down the instrument, find what's happening and eventually fix it.

So my plan is initially to use this 4 LTZ1000 just to check with a reasonable confidence that the 3458A is not
drifting too much. Than I hope to find at least one reference that I trust enough so that I can send it some cal lab (or
someone I trust) for calibration and import this 7V with a reasonable uncertainty to my lab.

As for the 10K I think it is more easier as any calibrated Vishay VHP202 or VHA518-7 will be more than enough for
indirect checking that this 40K resistor inside the 3458A has not drifted too much. Those resistor have proven 2ppm
x 6 years drift, it wouldn't be so much hard in a second time to characterise a couple of this resistors for temperature
and use this as entry level standard. Even with a drift of 3ppm per 6 years I will be good for more than a decade.

I know this is quite excessive for my requirements but this is a quite enjoyable path too.

It doe not look like you needs 3458A, but it is quite good meter.
Before buying used one, check it can run over 24h on NPLC >10, this can highlight convergence errors ( A3 replacement )
Ask seller to enter CAL? 2941 from front panel - overloads counter.

[razberik]

Sure, that's why I said cost of ownership meter like 3458A, rather than price of having the box. Just after few years of use and maintenance, cost of calibrations alone will exceed MSRP of meter itself.

Thats questions me, for what anybody need the 3458A for when he dont use its capabilities ?
I wonder how many 3458A are out there for hobby use like storing it for the feeling of "just owning it".
I dont say I am different, I want 3458A too because it is a King of All Multimeters. I dont really need it, but I want it. 

[TiN]

Having 3458 sitting idle is a voltnut crime Mine almost always logging Vrefs or resistors , even if I'm out of the city, like now.

[HighVoltage]

I have my 3458A for about a month now and it has been running and measuring something every day.
It is finally something that I can rely on and compare my other multimeters against.
And I finally know for sure, how bad the 34401A is in comparison in the low ppm range.

[tszaboo]

Flinstone
But who knows, may be Keysight as a new 8 1/2 digit meter ready?

I specifically asked an engineer from Agilent on the electronica show about an upcoming new 8.5 digit meter. He said, that they have issues, that they dont make anymore accurate analog parts. As far as I understand, they dont have anything comparable in development, not in the near future anyway.
Quite sad, if you ask me. Would have loved to see what they can do with new components.

[mimmus78]

@plesa

"CAL? 2941" interesting but what number is safe? I suppose 0 ... than even if 1 shows up how can you assume it was a very dangerous or not?
I don't know if this can be a "realiable" parameter to check the meter.

Maybe is more useful to ask seller to do autocal, wait some days (7 will be the best) redo autocal and than check drift of all calibration variables.
Anyone has ever produced typical drift for all calibration variables?

>> Before buying used one, check it can run over 24h on NPLC >10, this can highlight convergence errors ( A3 replacement )
On the service note they say to test drift for one week and than divide by 7 to get an average value ppm x day.

[mimmus78]

@ TiN
>> That's almost the very path I went year ago with my rusty 3458.
Hope I don't end up buying a 57X0 in a year :-)

>> So we will soon to see pics of new toy at your home?
It seems more likely. It will be not as cheap as one can imagine, but still with a reasonable price and with its replaced new A3 board I should be covered for the most scaring problem it may have.

[plesa]

@plesa

"CAL? 2941" interesting but what number is safe? I suppose 0 ... than even if 1 shows up how can you assume it was a very dangerous or not?
I don't know if this can be a "realiable" parameter to check the meter.

Maybe is more useful to ask seller to do autocal, wait some days (7 will be the best) redo autocal and than check drift of all calibration variables.
Anyone has ever produced typical drift for all calibration variables?

>> Before buying used one, check it can run over 24h on NPLC >10, this can highlight convergence errors ( A3 replacement )
On the service note they say to test drift for one week and than divide by 7 to get an average value ppm x day.

For me it is 0 = OK, !=0 possible problems. All parameter including this one will give you more information about unit. Check also calibration string CALSTR?, cal labs store there last calibration date.
You do not needs to run whole acal, just ACAL DCV and check the CAL? 72 value, similar for ACAL OHMS.

You mixed together two things, drift and convergence errors. For convergence errors, just keep meter running over 24h on higher NPLC, it is more probable that convergence error will occur than on fast NPLC settings. The drift is monitoring of CAL? 72 value over time after ACAL DCV.
Did you asked Keysight remarketed/premium used  guys for 3458A?

BTW check this http://www.gellerlabs.com/HP%203458A%20DMM.htm

[CalMachine]

If you purchase one that you're sketched out with, you can send it to me and I'll give it a legit 17025 accredited cal with uncertainties for a fair price!  If you get one off ebay and it comes as-is with issues, I can also get some repairs done.

Also, generally the older HP labeled 3458A's are better than Agilent or Keysight.  They were made with higher quality references and are more stable.... as long as they've been used.  If they've been powered off for any significant periods of time, it could take months for the stability to be achieved.

[HighVoltage]

Also, generally the older HP labeled 3458A's are better than Agilent or Keysight.  They were made with higher quality references and are more stable.... as long as they've been used.  If they've been powered off for any significant periods of time, it could take months for the stability to be achieved.

CalMachine
Can you be more specific in regards to what components really were better on the older models?

[mimmus78]

@plesa:
My unit will have new A3 board from Keysight and it will be covered by official warranty so I don't think I have to worry of short term drift problems for now.
I will have one mount of warranty for all the unit and this is plenty of time to check everything and considering I also own a multifuction calibrator my first project will be to check it in all ranges.

As for "CAL? 2941" let say I would be happy if is zero, OK if is a small number. If is a reasonable bigger number I need to study what this really means ...

@CalMachine
Anyway based on serial number my unit will be one of last unit made by HP before they become "another company" ... and hope to get the very best aged components (that fails more often).
I still think that a 5yo unit is the optimal, who really cares if drift 3ppm instead of 1ppm in a year (and if you really care, you still need to calibrate it).

[CalMachine]

Also, generally the older HP labeled 3458A's are better than Agilent or Keysight.  They were made with higher quality references and are more stable.... as long as they've been used.  If they've been powered off for any significant periods of time, it could take months for the stability to be achieved.

CalMachine
Can you be more specific in regards to what components really were better on the older models?

Usually its the voltage reference board.  That's where most of the stability comes from.  You can see service notes put out by agilent and perhaps even keysight regarding unstable voltage references.  This was almost never an issue with HP labeled 3458As.  I've only been in the field for almost 3 years now, but this is what I have experienced and learned through the contacts I've made.

@plesa:
My unit will have new A3 board from Keysight and it will be covered by official warranty so I don't think I have to worry of short term drift problems for now.
I will have one mount of warranty for all the unit and this is plenty of time to check everything and considering I also own a multifuction calibrator my first project will be to check it in all ranges.

As for "CAL? 2941" let say I would be happy if is zero, OK if is a small number. If is a reasonable bigger number I need to study what this really means ...

@CalMachine
Anyway based on serial number my unit will be one of last unit made by HP before they become "another company" ... and hope to get the very best aged components (that fails more often).
I still think that a 5yo unit is the optimal, who really cares if drift 3ppm instead of 1ppm in a year (and if you really care, you still need to calibrate it).

Yeah aged units are always going to have a higher probability of having a failing component.  However, I feel like having the longer proven stable voltage reference would be more sought after, though.  If you don't need 1 ppm or lower drift, then for sure, any 3458A will work just fine.

[mimmus78]

Don't tell me ...

I brought a K2000 and a K2001 for parts. K2001 had a problem in an analogue switch in current measurement circuit (few hours of work to repair it).
K2000 I had to replace the ASIC from a donor unit because a "GPIO" was dead (here I spent a lot of time trying to fix it because I didn't wanted to change the ASIC and trying the most esoteric stuff that at the end didn't worked).

30 days agoo K2001 failed again: found an analogue switch was leaky. This time on the current source for ohm measurement. This required other 10 hours of work to fix it (and calibration lost in high ohm ranges was for free).
In the mean time on K2000 a diode bridge went wonky causing it to go crazy after 60 minutes of use.
I repaired it again (this time it was 5 minutes of work) and after I run self test I discover that is has also others small problems (didn't have yet time to check).

I hope to be more fortunate with this 20 years meter ... how was your experience with the 20 yo boys?

[pelule]

Posted by: mimmus78...Don't tell me ...

A really bad experiance. "Sometimes you have no luck and misfortune in addition."
I repaired my Datron 1281 (build in 1988) ~three years ago and is working without any problem since that time.
My HP3458A (build in 1989) I bought two years ago and is also working properly.
I assume both have well aged references.

[plesa]

@plesa:
My unit will have new A3 board from Keysight and it will be covered by official warranty so I don't think I have to worry of short term drift problems for now.
I will have one mount of warranty for all the unit and this is plenty of time to check everything and considering I also own a multifuction calibrator my first project will be to check it in all ranges.

As for "CAL? 2941" let say I would be happy if is zero, OK if is a small number. If is a reasonable bigger number I need to study what this really means ...

@CalMachine
Anyway based on serial number my unit will be one of last unit made by HP before they become "another company" ... and hope to get the very best aged components (that fails more often).
I still think that a 5yo unit is the optimal, who really cares if drift 3ppm instead of 1ppm in a year (and if you really care, you still need to calibrate it).

OKI, new A3 is good but if you purchase it from Keysigh it is not tested, also do not forget to check the A5 version. Because if seller is going to invest into new A3 and there is old A5  the investment is close to Keysight repair agreement 2k USD/2.8k USD ( which contains calibration report).

[mimmus78]

Plesa I didn't understand what you are trying to tell me. The A5 is the one with batteries? He will replace those chips and if they are still alive also data will be preserved.

Inviato dal mio Nexus 6P utilizzando Tapatalk

[mimmus78]

Interesting ... directly from 3458A manual ->

DEFEAT ON
disables (defeats) the input switch sequencing that protects the
multimeter's input circuitry from overload voltages. If input
protection is disabled and an overload situation is detected on
the 10V range or below, the multimeter will enable input
protection and internally tally the overload for instrument
warranty considerations.

[Alex Nikitin]

Interesting ... directly from 3458A manual ->

DEFEAT ON
disables (defeats) the input switch sequencing that protects the
multimeter's input circuitry from overload voltages. If input
protection is disabled and an overload situation is detected on
the 10V range or below, the multimeter will enable input
protection and internally tally the overload for instrument
warranty considerations.

Well, I don't know the history of our 3458A, CAL?2941 returns 1505  , however I don't have any problems with it, Keysight did calibrate it and put it on a 3-year service agreement without any comments.

Cheers

Alex

[mimmus78]

Wow 4 digits!!! I imagine is kinda of World Record!

Inviato dal mio Nexus 6P utilizzando Tapatalk

[Lowpass30Hz]

In general I wonder what the exact reason is why KS terminates the sales of the 3458A in europe ? Most likely due to RoHS regulations

That is exactly correct. Keysight is not updating the 3458A to be EU RoHS compliant. They will no longer sell one with a final destination inside the EU.
It is still fully supported, I imagine for a long time in to the future.