Fluke 8558A/8588A 8.5-digit DMMs

[TiN]

About 18 years after release of first Fluke's 8.5-digit DMM based on acquired technology from Wavetek merger and Datron 1281 project.
Yes, I'm talking about Model 8508A that compete with respected ultralinear industry golden standard - Keysight 3458A. Now in 2019 we finally get an update:




Fluke started to release some docs on subject:

Fluke 8588A/8558A operators manual
Fluke 8588A/8558A Remote programmers manual

So now we know that 8588A is a replacement for 8508A Reference Multimeter and has specifications suited for the calibration and metrology applications.

Cheaper 8558A 8 1/2 Digit Multimeter is more in line for general lab work, that need high resolution meter, but relaxed on accuracy.

Fluke 8558A specifications
Fluke 8588A specifications

Common features from manual:

* Visual Connection Management active 5-way low thermal binding posts terminal illumination, like Fluke 5730A/5790B
* Versatile resolution and reading rate settings: 8 1/2 to 4 1/2 digit resolution
* Now can adjust aperture time settings from 0 ns to 10 seconds (200 ns min resolution)
* Speed up to 100 KSPS at 4 1/2 digits (18-bit) resolution in remote to rival 3458A
* Digitizer up to 5 MSPS at 18 bits with up to 20 MHz bandwidth
* 30A range for 8588A, but only 2A range for 8558A model.
* Programmable front/rear inputs, automatic ratio ohms, voltage, and more using the front/rear inputs. (no more overpriced Option 001? yay?)
* Analyze, with graphing, trending and statistics
* Capacitance measurements to calibrate multi-product calibrators
* RF power meter readout for R&S NRP series power sensors
* GPIB SCPI, Ethernet, and USB remote interfaces
* PRT and thermocouple readouts
* Front and rear USB memory ports for data transfer
* Software emulation of the Fluke 8508A and HP/Agilent/Keysight 3458A remote interfaces
* Analog Zero to remove residual offsets, for example, from thermal EMFs.

Based on quick browsing thru the manual few notes I make:

* Fall for modern trend with colorful LCDs with poor color/font choices
* Added 8-digit up to 100Mhz frequency meter, added support for R&S power sensor for RF stuff
* Fixed issue with 8508A when ranges to only to 199900000 counts. Now full-scale overrange is 202000000 counts.
* ...But not for ACV functions! ACV functions limited to 1.212x FS. Old hotty 8508A had 1.9990x at least Keithley 2002 is still the champ on FS range.
* Still no specified linearity for ADC
* No thermal sensor for AC/DC transfers. I was kinda hoping for this, since Fluke have the technology for decades (used in 57xx/792A).
* Marketing pitch of "no need ACAL" is present as well.
* Now meter have a fan. Yay! I was not happy about having hot 8508A near my other gear, offseting the measurements. Had to place it on separate location due to that.
* Remote protocol is now SCPI-compatible, as expected.

Anyone getting a demo unit to play with?

[Dr. Frank]

TiN, thanks for the headup.

Seems that these are just technical updates of the 8508A, like done on 732C and 5730, as probably:
- update of component technologies and obsolete parts
- through hole => SMD
- replace Vishay BMF by Fluke precision Thick Film Arrays.
- additional faster 18Bit A/D for 100kHz and faster (like KEI7510)
- new, updated user interface
- improved data handling, like memory and fast transfer over GPIB, USB, Ethernet

A quick scan shows me, that specification did not change much compared to 8508A, especially A/D seems still to be the old one, as 8 digits (DCV, Ohm) still require 2-10sec acquisition time.. see page 32 of operation manual.. and linearity is again not specified, only again indirectly by transfer accuracy of (0.05 + 0.05) ppm.
OHM mode really has been improved further, great accuracy for high ohm values, due to 20V/200V current sources.
 
So price may be more competitive to 3458A. Let's see, what other volt-nuts here find out, when they join the official hands-on presentations by FLUKE.

Frank

[HighVoltage]

Thank you TiN for an early preview.
The official release date in Germany is in abut 2 weeks.

[splin]

Hmm.  8588A resistance specs for 100 ohms to 100k (page 26) have better 2 year Tcal +/- 5C absolute accuracy than 2 year Tcal +/- 1C relative accuracy! Probably got all the right numbers,  just not necessarily in the right order....      **

The 8558A specs have similar issues but different numbers. I wonder how many people proof read or reviewed the documents?

Also curious is the DC 10V 365 day Tcal +/- 1C relative accuracy spec is 2.7ppm rdg + .05ppm range,  compared to the 365 day Tcal +/- 1C absolute accuracy spec of 2.8 +.05. Does that imply that their calibration standard uncertainty is only 0.1ppm? Seems incredible - the 8508A adds 0.3ppm and the 3458A adds 2ppm for NIST traceability.

Why don't they specify absolute 24h and 90 day specs?  They don't provide those for the 8508A either.

[EDIT] The Datron 1281 specified an additional 2.5ppm typical calibration uncertainty.

** For the youngsters or non-UK readers - Eric Morecombe 'playing' Grieg's piano concerto, conducted by the late Andre Previn.

[e61_phil]

and linearity is again not specified, only again indirectly by transfer accuracy of (0.05 + 0.05) ppm

Not even indirectly. The transfer is only allowed within 10% of the initial value. Therefore, 0.05 + 0.05 does not apply for a 1V to 10V transfer.

[PTR_1275]

The million dollar question, how much can I not afford it by

[quarks]

just heard about it a short time ago

thanks for sharing

I will update my comparison charts and post it asap

[quarks]

here is a DCV comparison chart

Looks strange, but if I did not make a mistake, the 8508A is still better overall, exept for 8588 linearity in 10 (20) Volt range.

[Kleinstein]

The range much below 10 mV does not really matter. For the rest the 8588 is pretty close to the 8508 and the "cheap" 8558 is a little high in uncertainty.

The difference is so small that it's no significant, as the uncertainty levels are only some kind of best guess or even aims set by the marketing department. Especially for a new instrument the specs may be conservative. For an only slightly modified old design the new values may be just updates estimates due to experience even if the internals have not changed.

There is not much sense in absolute specs for very short time - the 24 h would be essentially over when delivered and warmed up. Not many would run a 90 days or similar cal cycle - and if so they should know how to calculate the numbers from the cal lab uncertainty and relative accuracy.

The added 2nd faster ADC makes absolute sense, especially if used for digital RMS anyway.

What is the purpose of a 1 M input impedance mode ? It looks rather odd to me.

[CalMachine]

What is the purpose of a 1 M input impedance mode ? It looks rather odd to me.

Some high voltage probes have a 1 MOhm output impedance.  A lot of Ross Engineering probes do, I believe

[e61_phil]

Some high voltage probes have a 1 MOhm output impedance.  A lot of Ross Engineering probes do, I believe

Does it make sense to calibrate such probes with an 8.5 digit meter? Most of them have a couple of % as spec.

[CalMachine]

Some high voltage probes have a 1 MOhm output impedance.  A lot of Ross Engineering probes do, I believe

Does it make sense to calibrate such probes with an 8.5 digit meter? Most of them have a couple of % as spec.

The Ross Engineering VD probes are 0.01% at DC.  https://www.rossengineeringcorp.com/products/measurement/hv-voltage-dividers.html

Normally I've seen then used with a little impedance matching box, too.  The 1 MOhm function negates that.  I'd admit it is a little overkill, though.  I'm sure there are probably other applications that might call out for a specific input impedance

[macboy]

What is the purpose of a 1 M input impedance mode ? It looks rather odd to me.

Some high voltage probes have a 1 MOhm output impedance.  A lot of Ross Engineering probes do, I believe

Do they have a 1 MOhm output impedance, or are they specified for a (i.e. expecting a) 1 MOhm input impedance instrument? Not the same at all.

[CalMachine]

What is the purpose of a 1 M input impedance mode ? It looks rather odd to me.

Some high voltage probes have a 1 MOhm output impedance.  A lot of Ross Engineering probes do, I believe

Do they have a 1 MOhm output impedance, or are they specified for a (i.e. expecting a) 1 MOhm input impedance instrument? Not the same at all.

Touche!

Here is what Ross Engineering states 

In the standard 1000/1 or 10,000/1 Ross voltage dividers the low voltage capacitors and resistors are selected to provide correct ratios when shunted by 1 megohm, 20-50pF oscilloscope (or other load), plus 2, 3, 6, 15, or 20 feet (please specify the shortest practicable length) of low capacitance RG59 or special Ross Engineering Corporation coaxial cable unless otherwise required

[try]

From the manual of the 8588A:

4 1/2 digits (18-bit)

Pretty strange.

[Inverted18650]

Hopefully this will allow the rest of us to get the "old" Fluke 8508A's for less than $10K.

[e61_phil]

Hopefully this will allow the rest of us to get the "old" Fluke 8508A's for less than $10K.

I don't think so. Unfortunately, the documents are no longer available due to the links above.

If I remember correctly the relative spec of the 8588A (20V 1 year) was 2.7ppm and the relative uncertainty of the 8508A is also 2.7ppm. There is no improvement in the new meter. The adder for the range has improved (0.05ppm vs. 0.2ppm) but I can imagine that is due to some experience over the years.

Our 8508A was calibrated with an uncertainty of 0.7ppm. If we assume the noise is stable over the range that will also lead to 2.8ppm/year.

I would expect the new meter has to show some years of really good performance before anybody with a 8508A will change to the new one. Especially if they changed the gain setting resistors to something new instead of the expensive ones used in the 8508A.
The highspeed stuff isn't an argument imho. If I want to have something like that I would go for an 7510 or something like that additional to my 8.5 digit meter.

[Dr. Frank]

Yes, 95% and 99% confidence DCV specifications are nearly identical, maybe the absolute specs are better due to usage of better standards like JJ array.
And transfer specs are much better.

It's always good to directly download such leaked documents.. 

Frank

[e61_phil]

Are the transfer specs really better? I think the old ones aren't limited to 10%

[splin]

Unfortunately, the documents are no longer available due to the links above.

If I remember correctly the relative spec of the 8588A (20V 1 year) was 2.7ppm and the relative uncertainty of the 8508A is also 2.7ppm. There is no improvement in the new meter. The adder for the range has improved (0.05ppm vs. 0.2ppm) but I can imagine that is due to some experience over the years.

Here you go:



But how are we to interpret the 0.1ppm difference between the highlighted values? They make no sense to me; it would be surprising if the factory calibration uncertainty was much better than 2ppm. Clearly Fluke have the capability for sub ppm 10V calibration, but in that case why not specify absolute uncertainties for periods less than 365 days?

If I were paying $17k+ I'd want it to be freshly calibrated, preferably within 48 hours - not from a distributer with potentially only a few weeks left of a 12 month calibration. I guess that would be an expensive option like the 3458A meteorological calibration option.

I see TEquipment are showing the 8508A as discontinued.

[Dr. Frank]

The 8588A is better specified than the 8508A, when making near 1:1 transfers, like 10V against 9.0...11V (+/- 10%).

The 8508As transfer specifications do not contain such a limitation, and in the explanations of the transfer specification, there's even an example of a 5V => 10V transfer/comparison. So you're right about the 8588A. Maybe the 8508A would also give 1:1 transfers as good as the 8588A (same / similar A/D ?)

In contrast, the 8588A now is not specified any more concerning transfers of random ratios, instead you have to use the 24h specification, which is much worse than the 8508A transfer spec for ratios > 1.1.

Frank

[e61_phil]

But how are we to interpret the 0.1ppm difference between the highlighted values? They make no sense to me; it would be surprising if the factory calibration uncertainty was much better than 2ppm. Clearly Fluke have the capability for sub ppm 10V calibration, but in that case why not specify absolute uncertainties for periods less than 365 days?

If I were paying $17k+ I'd want it to be freshly calibrated, preferably within 48 hours - not from a distributer with potentially only a few weeks left of a 12 month calibration. I guess that would be an expensive option like the 3458A meteorological calibration option.

What is the problem with the relative specs? That is what the meter does and it is what HP also specified for the 3458A. If you want to have fully traceable specs you have to use absolute specs from Fluke or add the 2ppm for the 3458A. Or you RSS the traceability uncertainty with the rel specs like described in the Fluke datasheet. The nice thing on Fluke specs is, they give you absolute values which you can take without using your brain. But they cannot know how you calibrate the meter and therefore you have the rel specs.

As we ordered our 8508A is came directly from Fluke and was calibrated with 0.7ppm uncertainty for 10V. Which gives 2.8ppm for 10V exactly as the new meter. The new better absolute value is no improvemt of the meter, it is improvement of the calibration.

They calibrated the 8508A with a characterized 5720A. There is no need for a JJA to calibrate such a meter.

[ScoobyDoo]

Hello folks - here is already a short commercial of the 8588A



ScoobyDoo

See you all soon on 20th of March at Fluke event in Kassel !

Herzliche Grüße/Meilleures salutations/Best regards

ScoobyDoo

[e61_phil]

Who would be in Kassel on 20th of March? I planned to go

[Krampmeier]

I'll be there, together with a colleague. It would be great to have a chat with some other EEVblog forum members - look out for Andreas and Johannes (me)!

[splin]

But how are we to interpret the 0.1ppm difference between the highlighted values? They make no sense to me; it would be surprising if the factory calibration uncertainty was much better than 2ppm. Clearly Fluke have the capability for sub ppm 10V calibration, but in that case why not specify absolute uncertainties for periods less than 365 days?

If I were paying $17k+ I'd want it to be freshly calibrated, preferably within 48 hours - not from a distributer with potentially only a few weeks left of a 12 month calibration. I guess that would be an expensive option like the 3458A meteorological calibration option.

What is the problem with the relative specs? That is what the meter does and it is what HP also specified for the 3458A. If you want to have fully traceable specs you have to use absolute specs from Fluke or add the 2ppm for the 3458A. Or you RSS the traceability uncertainty with the rel specs like described in the Fluke datasheet. The nice thing on Fluke specs is, they give you absolute values which you can take without using your brain. But they cannot know how you calibrate the meter and therefore you have the rel specs.

Ah there is the answer - thanks. What I'd missed was that Fluke tell you to RSS the relative and absolute uncertainties so .7ppm uncertainty for 10V plus 2.7ppm relative gives 2.8ppm as you state. This is in contrast to the 3458A datasheet which tells you to use the algebraic sum. If Fluke had simply stated their uncertainty in the datasheet, like most other manufacturers do (or used to do), then there would have no uncertainty. 

It still does't exuse them because the 365 day spec is not consistent with the 2 year spec. - subtracting 1.2ppm from the 2 year +/- 5C absolute figure (TC is .3ppm/C) gives 6.8ppm which implies an absolute 10V uncertainty of 4.1ppm RRS'd with 5.4ppm. That can't be right so either I've missed something else or it's another error.

As far as I can see neither Keithley nor Keysight specify which method they use to specify the absolute uncertainty for the DMM7510/34470A, nor their uncertainty for 10V. It makes quite a bit of difference so it would be helpful if all meters were specified consistently. I guess you find out from the cal certificate after you've bought one?

As we ordered our 8508A is came directly from Fluke and was calibrated with 0.7ppm uncertainty for 10V. Which gives 2.8ppm for 10V exactly as the new meter. The new better absolute value is no improvemt of the meter, it is improvement of the calibration.

Is that standard for the 8508A or an extra cost option?

[e61_phil]

First of all: I'm not an metrology expert in terms of uncertainty calculations. Therefore, I might be wrong on some points here.

What I'd missed was that Fluke tell you to RSS the relative and absolute uncertainties so .7ppm uncertainty for 10V plus 2.7ppm relative gives 2.8ppm as you state. This is in contrast to the 3458A datasheet which tells you to use the algebraic sum.

Fluke did it in the extended spec datasheet of the 8508A. Perhaps they add it in the final version of the datasheet? I don't know.

I think one cannot the compare the 3458A datasheet. HP says one have to add 2ppm for traceability, but that is many years old. In older datasheets Fluke also adds uncertainties algebraic. Is that changed due to the GUM in the last years?

If Fluke had simply stated their uncertainty in the datasheet, like most other manufacturers do (or used to do), then there would have no uncertainty. 

Could you explain what do you mean by that? They should remove the absolute specs?

It still does't exuse them because the 365 day spec is not consistent with the 2 year spec. - subtracting 1.2ppm from the 2 year +/- 5C absolute figure (TC is .3ppm/C) gives 6.8ppm which implies an absolute 10V uncertainty of 4.1ppm RRS'd with 5.4ppm. That can't be right so either I've missed something else or it's another error.

I'm also confused by that. I also can't reproduce the exact values from the 8508A datasheet. My results are <0.5ppm lower.
What I can imagine is another (higher) calibration uncertainty if you order a 2year calibration.

These are things one could probably discuss in Kassel with Fluke.

As far as I can see neither Keithley nor Keysight specify which method they use to specify the absolute uncertainty for the DMM7510/34470A, nor their uncertainty for 10V. It makes quite a bit of difference so it would be helpful if all meters were specified consistently. I guess you find out from the cal certificate after you've bought one?

I think in contrast to the 3458A specs, the 34470A specs are absolute specs (directly traceable). But I can't find any uncertainty calculations in the service manual. They only suggest calibration equipment. In the 34401A service manual they talk about at least 1:4 TUR imho.
I can have a look into the calibration sheet from Keysight for the 34470As. If I remember correct they are tested against 24h specs with an 5720A. But I will check that.

Is that standard for the 8508A or an extra cost option?

That was no option. I can imagine that they improved their uncertainties over the years and no one has changed the 8508A datasheet.

[TiN]

If I remember correct they are tested against 24h specs with an 5720A. But I will check that.

I doubt any commercial calibration lab maintain and characterizes 5720A every 24 hours for customer with 34470A. 

[e61_phil]

If I remember correct they are tested against 24h specs with an 5720A. But I will check that.

I doubt any commercial calibration lab maintain and characterizes 5720A every 24 hours for customer with 34470A. 

The 5720 was used to check the 34470A 24h specs.

There shouldn't be any need to characterize the 5720A

[TiN]

Perhaps it's possible to see data points from report to make sure of that? 

[e61_phil]

Perhaps it's possible to see data points from report to make sure of that? 

I will search for that paper tomorrow. But I can't see any problem here. Even the 1 year 10V spec of the 5720A delievers a TUR of nearly 1:2 for 34470A 24h specs.

And it isn't a big deal to bring the 5720A daily in 24h specs with artifact calibration.

[e61_phil]

The certificate of calibration lists a 5720A and a 5725A with their trace numbers and cal dates. But no further uncertainty values. The report shows readings for severeal ranges and the error and how much that error is in percentage of the 1year spec. Biggest error in that list is 1.1% for 10V ACV. As there is no spec which is in 1 year 100x worse than for 24h the meter was in 24h specs. But you don't know that for sure, because there are no uncertainty statements.
It might be is not fair to compare a manufacturing cal report of a 34470A against the report delivered with the 8508A, but if you do, the Fluke one looks much more like metrology

[TiN]

What I expected. To be exact testing against 24 hour specifications would be only valid right after adjusting all functions and ranges on the meter, within same time interval.
It's just bit strange how you say "not a big deal to bring 5720A into 24h due to ACAL", but keep saying that 3458A's ACAL is drawback and does more harm than good. You don't have to use 3458A's ACAL in which case you get same behaviour as 8508A and use external full calibration for all ranges/points every time. 
But I agree, this is bit offtopic in this thread.

[e61_phil]

What I expected. To be exact testing against 24 hour specifications would be only valid right after adjusting all functions and ranges on the meter, within same time interval.
It's just bit strange how you say "not a big deal to bring 5720A into 24h due to ACAL", but keep saying that 3458A's ACAL is drawback and does more harm than good. You don't have to use 3458A's ACAL in which case you get same behaviour as 8508A and use external full calibration for all ranges/points every time. 
But I agree, this is bit offtopic in this thread.

You are right, if one would argue completely straight I should also don't like the artifact cal of the 57xx. My point was just, that you are able to calibrate anything with a daily adjusted 57xx if you like. It isn't a big deal, even for smaller labs.

I doubt that a 3458A behaves like a 8508A if you don't use ACAL. And you are not able to adjust every single range on a 3458A. But, I think we should discuss that in another thread, as you already suggested .
I modified my logging scripts for the continous monitoring of severeal 10V standards against the 3458As. It logs all CAL? values until now. That should show the stability of a 3458A without ACAL in a couple of months. If you are right and a 3458A is as stable as a 8508A the CAL? values shouldn't move.

[splin]

First of all: I'm not an metrology expert in terms of uncertainty calculations. Therefore, I might be wrong on some points here.

I'm certainly not either, but I guess that's pretty obvious!

Fluke did it in the extended spec datasheet of the 8508A. Perhaps they add it in the final version of the datasheet? I don't know.

I think one cannot the compare the 3458A datasheet. HP says one have to add 2ppm for traceability, but that is many years old. In older datasheets Fluke also adds uncertainties algebraic. Is that changed due to the GUM in the last years?

I don't know, but since it gives better marketing numbers using RSS I guess all manufacturers use it for their more recent products.

If Fluke had simply stated their uncertainty in the datasheet, like most other manufacturers do (or used to do), then there would have no uncertainty. 

Could you explain what do you mean by that? They should remove the absolute specs?

The latter part was just a (poor) joke which should have read "... then there would have been no uncertainty". What I meant was that Fluke (and all manufacturers) should include in their datasheets their 10V uncertainty to traceable standards, ie. the 0.7ppm they use for calibrating these meters. That would eliminate the guessing and allow you to use the most appropriate method for calculating the absolute uncertainty for your purposes.

The Tcal +/-1C relative specs are the most important for understanding the meter's stability and I believe they should be provided for all high end meters - eg. you can use them to compare the drift of other instruments rather than the absolute errors (especially if you can control the temperature to +/-1 C). Fluke is very good in this respect and dissapointing that Keithley only provide 24hr relative specs for the DMM7510. Keysight presumably don't consider the 34470A a metrological instrument and point to the 3458A if you are that concerned with accuracy.

Absolute numbers are helpful of course but you can't necessarily calculate the relative figures from them, even if you know the cal standard uncertainty, as they don't tell you how they calculated the absolute numbers. For example, they could use a lower TC when calculating the Tcal +/- 5C absolute specs than the TC for the wider temperature range they specify in the datasheet. They might not do this of course but it could allow them to publish better numbers.

It still does't exuse them because the 365 day spec is not consistent with the 2 year spec. - subtracting 1.2ppm from the 2 year +/- 5C absolute figure (TC is .3ppm/C) gives 6.8ppm which implies an absolute 10V uncertainty of 4.1ppm RRS'd with 5.4ppm. That can't be right so either I've missed something else or it's another error.

I'm also confused by that. I also can't reproduce the exact values from the 8508A datasheet. My results are <0.5ppm lower.

It would be interesting to see how you calculated that.

What I can imagine is another (higher) calibration uncertainty if you order a 2year calibration.

Seems a bit of a stretch - I would expect a meter to be within the 2 year specifications after 2 years of ownership from new.

These are things one could probably discuss in Kassel with Fluke.

Right; it's unreasonable to criticse Fluke for specifications they haven't officially released yet. If I were spending $17k+ on a meter then I would certainly be talking to Fluke to resolve my uncertainties about the uncertainties in their specs!

I am unlikely to buy one but it is interesting (for me) to analyse the numbers anyway to help my understanding of how the specifications are derived and what they really mean.

I think in contrast to the 3458A specs, the 34470A specs are absolute specs (directly traceable). But I can't find any uncertainty calculations in the service manual. They only suggest calibration equipment. In the 34401A service manual they talk about at least 1:4 TUR imho.
I can have a look into the calibration sheet from Keysight for the 34470As. If I remember correct they are tested against 24h specs with an 5720A. But I will check that.

Is that standard for the 8508A or an extra cost option?

That was no option. I can imagine that they improved their uncertainties over the years and no one has changed the 8508A datasheet.

Ok, thanks. A bit surprising as marketing are generally very keen that any potential purchasers know about any improvements thay make.

[TiN]

A bit surprising as marketing are generally very keen that any potential purchasers know about any improvements thay make.

I'm sure they will spend few pages on benefits of color TFT screen and power sensor support. As for DC/LF metrology and performance improvements over 8508A (if any exist) - they will keep it all shut to avoid giving hints for competitors. After all this is very niche market and you can count remaining players by fingers on one hand.

[e61_phil]

The new meter looked really nice and the linearity is improved over the 8508A. (short check with a 10V and 1V source whose ratio I later compared on a 3458A)

They also said, that the transfer specs are valid for the whole range not only within 10%

[TiN]

linearity is improved over the 8508A.

Sorry, I find it very hard to believe just on words. I'd like to see linearity specified or measured, just like it was on 3458A.
Call me Fluke hater or whatnot, but from such expensive meter with "reference" marketing the bar must be set accordingly.
Fluke has ability to do such measurements, they don't need to go NIST today, like HP had back in 1989.

Again, I don't say linearity itself is bad on 8508A/8588A, but completely omitting this important specification is bad.
So saying that unspecified parameter on new meter is improved is no different to saying that unicorns have rainbow hair. 

[e61_phil]

@Tin: Linearity was briefly tested on a 8588A and the measured 1V to 10V ratio was within 0.05ppm of the ratio measured by my 3458A. They said there are a lot of mistakes in the leaked documents. Therefore, let's wait until the final version is published and start the discussion again.

@splin: I only saw updated specs for the HP 3456A after some time in the market. I think the 3458A is also much better than specified and there is much experience with the unit. However, nobody has updated the spec sheet.
I calculated it simply as shown in the 8508A datasheet (there is an example for the temperature range).

[TheSteve]

How can it be a "leaked document" if it was on the Fluke website for all to see?

[e61_phil]

How can it be a "leaked document" if it was on the Fluke website for all to see?

I can only repeat what they said today: "wasn't finished, should have been set to private and so on". I don't know why they published that and if the story is "true".
I'm looking forward for a final version. There were also some different numbers in his presentation.

I have no contract with Fluke

[Magnificent Bastard]

{...snip...}
@splin: I only saw updated specs for the HP 3456A after some time in the market. I think the 3458A is also much better than specified and there is much experience with the unit. However, nobody has updated the spec sheet.
{...snip...}

And THAT will never happen.  What they COULD do, is release a 3458B (same as the 'A' model with some firmware mods to identify as a 'B')-- and then release an updated spec sheet for THAT.

The reason they could release updated specs for the 3456A, is that the new specs applied to all known 3456A's in the wild--- which is definitely NOT the case with the 3458A.

[TiN]

[Krampmeier]

The noise level / speed of the 8588A was quite impressive. The Fluke rep claimed that the accuracy at 1 second aperture would the same as with the 8508 at 26 seconds. Indeed, it did look quite stable at 1 s aperture, but default was 10 s for 8.5 digits.

The 8505 had ranges like "20 V". The ranges of the new meters are now slightly wider, but they are now called differently, e.g. "10 V", even though the full scale value is acutally more than double of that. When the measurement uncertainty is calculated, the _nominal_ range must be used, so the "ppm of range" numbers in the specifications must now be divided by 2.02 when comparing with the 8508 specs. Quarks, did you take this into account in your graphs?

At 5 MS/s, the digitizer should still provide 16 effective bits.

We got to play with the UI a bit and it felt a bit outdated to me. You really have to learn which options hide behind which button and soft key. The graphing feature does not feel very well thought out, for example you cannot easily delete the buffer, or change between "graph only" and "graph + statistics". People who have used the new Keithley meters will be disappointed by the UI implementation.

However, we saw an engineering sample and everything written above may change.

The official introduction of the new meters will be on March 26.

We got some information about the price point of the meters already:
8588: 15 - 16 k€
8558: 10 k€

They did not comment on the strange limitation in the safety notes of the leaked specs, which says that voltage sources must not be able to provide more than 200 mA. Probably an error or requested by lawyers?

[dr.diesel]

It's gonna be tough to out the legacy of the mighty 3458A, but glad to see another option!

Keysight is going to be forced at some point in the near future to revamp the 3458A, if they still have the engineering resources.

We need a good reliable evaluation of Fluke's new contender, maybe we need to group buy one for TiN.

[ScoobyDoo]

It's gonna be tough to out the legacy of the mighty 3458A, but glad to see another option!
Keysight is going to be forced at some point in the near future to revamp the 3458A, if they still have the engineering resources.

Keysight Germany has commissioned the design of the successor of the 3458A to a German engineering house - three years ago -
their new unit will be launched soon - in term of weeks but no longer ... several sources inside Keysight told me such.
Let's cross fingers that their new unit lines up with the mighty 8558A/8588A duo ...

Herzliche Grüße/Meilleures salutations/Best regards

ScoobyDoo

[e61_phil]

The reason they could release updated specs for the 3456A, is that the new specs applied to all known 3456A's in the wild--- which is definitely NOT the case with the 3458A.

Could you explain your statement in a bit more details? Do you think not all 3458A are much better than specified?

[e61_phil]

You don't have to use 3458A's ACAL in which case you get same behaviour as 8508A and use external full calibration for all ranges/points every time. 
But I agree, this is bit offtopic in this thread.

I found a few full lists of CAL constants for our 3458As. Not much data until now, but one can see a 3458A doesn't behave like a 8508A if you don't use ACAL. I will go on with collecting data..

[HighVoltage]

I found a few full lists of CAL constants for our 3458As. Not much data until now, but one can see a 3458A doesn't behave like a 8508A if you don't use ACAL. I will go on with collecting data..

What behavior of the 8508A do you mean?
I am using ACAL on my 3458A once every day and it has shown to be working perfectly, based on resistance measurements.

[e61_phil]

What behavior of the 8508A do you mean?
I am using ACAL on my 3458A once every day and it has shown to be working perfectly, based on resistance measurements.

I meant the stability of the 8508A. The 8508A is stable without daily ACALs. The fact that the CAL? constants move quite a bit and the reading is stable on a 3458A (if you use ACAL) means that the internals of the 3458A drift by the same amount as the "constants" drift.
If you don't use ACAL anymore, the 3458A will drift like the shown constants (which are constant without running ACAL). My point was only that a 3458A isn't nearly as stable as a 8508A. The 3458A (at least ours) really need ACAL to compensate for drifting components.

[TiN]

e61_phil, now I see where your disappointment in 3458 ACAL comes from. If your data indeed correct and base DCV gain drifted +3 ppm in just 3 months, then your meter's ADC is very broken.

CAL72? data from rusty 3458A:



And I beg you, please, create separate thread. This is so offtopic in this one. Also answer to

Do you think not all 3458A are much better than specified?

Is obvious as No, many old 3458A's cannot even meet current published specifications, leave alone better ones. And good ones - how better would they specify "updated" spec? Entering realm of Vishay specmanship on this one, since Keysight cannot guarantee anything about meters already in the field. Take simple example: 10VDC TC. Current spec is 0.5 ppm. Better spec, let's say 0.25 ppm/K? My three meters meet that spec, Dr.Frank's does not. If he sends his meter to calibration and pay lots of gold for it, he will expect to see nothing else than "calibrated, in specification" without notes "apply old specification from 1989, not the new one from 2019".

Btw, here's actual real benefit of having ACAL in meter = you can see drift/errors and correction factors for every function/range in 3458A (K2002 can give you constants too, but have ACAL only for ACV) without ANY external ultra-super-stable standards. But Fluke 8508A (and I expect new 8588/8558A too) which goes for double of the price hides all that, so if you want to plot similar 10VDC range drift you will be stuck with 732's and need to connect stuff everytime and hope errors from cabling/user does not creep into your sub-ppm measurements.  My point here is that I want to know or at least to have reasonably easy ability to characterize my DMM (buying and maintaining Fluke 734A/C bank is not reasonable), not just trust manufacturer's marketing department with their "our meter is super-stable enough not to need ACAL".

[e61_phil]

If the ADC alone is the problem I would expect that all ranges behave equal. But they don't. But, you are right let's start a new thread for that kind of discussions.

Has anyone an example cal sheet for a 3458A? I saw one online which doesn't include values. Only "passed".

I also think no cal lab will check the temperature coefficient of the meter. But I understand the problem with newer specifications. It confuses more than it helps.

[e61_phil]

Btw, here's actual real benefit of having ACAL in meter = you can see drift/errors and correction factors for every function/range in 3458A (K2002 can give you constants too, but have ACAL only for ACV) without ANY external ultra-super-stable standards. But Fluke 8508A (and I expect new 8588/8558A too) which goes for double of the price hides all that, so if you want to plot similar 10VDC range drift you will be stuck with 732's and need to connect stuff everytime and hope errors from cabling/user does not creep into your sub-ppm measurements.  My point here is that I want to know or at least to have reasonably easy ability to characterize my DMM (buying and maintaining Fluke 734A/C bank is not reasonable), not just trust manufacturer's marketing department with their "our meter is super-stable enough not to need ACAL".

That isn't true. The only thing the 3458A gives you is what it means relative to its LTZ1000 and 40k references. That is not more than a Fluke 57XX will give you with calibration check or what you can do on your own with some stable sources. There is no need for ultra-super-stable standards.
If you own a controllable calibrator and a linear DMM you can run through this ACAL also fully automated with any meter and gain these constants.


But it's on you if you like it better if manufacturer say: "Our ranges are that drifty, you have to adjust them daily to get good performance. We also cannot check if the ADC still behaves linear, but the whole concept relies on that fact". Muuuuucccchhh better
And of course is it cheaper to have a drifty meter which correct itself daily.

[TiN]

If 8508A ADC is drifty or got bad linearity, you'll never know without high-end external gear.
If you get 3458A with bad drift/linearity, all you need to do is run bunch of ACALs procedure. Don't even need calibrator.
What I really meant is that you can verify 3458's A9 reference stability or A3 ADC stability, without needing 57xx/other good 3458A/reference DMM. But yes, choose your poison.

No amount of ACAL will fix broken meter, that is last thing I'll say about this in thread about 8588/8558A. I'm lost already on what we disputing about, anyhow. More meters to everyone, with or without ACAL.

[e61_phil]

I don't take it as a dispute. It is interesting to discuss the pros and cons, I think.

A function that compares the ranges against each other like ACAL is great, no objections! But you need a lot of knowlegde and effort to gain that information. The 3458A don't give it soo easy. But it is an old instrument and one cannot compare it to newer ones with a lot of software like the 5730A calibrator. And I don't see how ACAL can warn you if your ADC isn't linear anymore. The results are just false. But if you have a constant non-linearity that wouldn't show up in the CAL values.

The only thing you can detect on the 3458A is a drifty ADC. If the reference will drift the CAL values will not show that. And you have to read out the data and monitor that, there is no warning lamp on the meter. Neither an Error message.

And having a just stable 100mV, 1V and 100V source isn't ultra-super-stable stuff. That is everything you need for such a test. If you don't have that, you probably don't need it

But I don't want to get stuck on the 3458A. My point is more in genereal ACAL vs. a stable meter. Perhaps that isn't that offtopic anymore, because Fluke says they don't need ACAL


My points are (Perhaps a wischlist for the next generation of meter ):
- I would like to have a function that runs something like ACAL, but don't adjust anything on the meter. Just reports "largest drift: 4% of spec on 1kV" range, like the 57XX does. And of course give access to all the gained data. With this data you can calculate corrections if you want, but nothing disturbs your calibrated (by a proper calibration) ranges.

- I don't like the concept of adjusting everything automatically with no chance for corrections. That has several drawbacks:
1. Nothing can be better adjusted as the transfer concept is able to transfer it from the artifacts. That means you are not able to charactize any range much better than the transfer can do, because ACAL will destroy that every time you run it. That doesn't matter, because the 3458A ranges aren't better than the transfer (at least our 3458As), but that don't need to be true for better meters (stability wise) like the 8508A.
2. If one range will run out of spec, you have to adjust the whole meter and you lose all your history.
3. ..

[maxwell3e10]

It would be best if the meter could spit out all calibration constants (offsets/gains) for all the ranges in a user-friendly format (better than a calibration memory dump). That would allow one to keep track or potentially go back to a known previous state.

[Magnificent Bastard]

What behavior of the 8508A do you mean?
I am using ACAL on my 3458A once every day and it has shown to be working perfectly, based on resistance measurements.

I meant the stability of the 8508A. The 8508A is stable without daily ACALs. The fact that the CAL? constants move quite a bit and the reading is stable on a 3458A (if you use ACAL) means that the internals of the 3458A drift by the same amount as the "constants" drift.
If you don't use ACAL anymore, the 3458A will drift like the shown constants (which are constant without running ACAL). My point was only that a 3458A isn't nearly as stable as a 8508A. The 3458A (at least ours) really need ACAL to compensate for drifting components.

There are two approaches to engineering DMMs--- the first is to use low (or medium) quality components in the critical areas; with a high-spec voltage and resistance standard; then use auto-cal to compensate for time and temperature drift.  The second method is to use ultra-high-spec components in all of the critical areas; and these keep time and temperature drift to a minimum.

There are two different intended use cases for these DMMs.  The 3458A was meant to be a high-spec systems meter-- used on the production floor with only mild control of ambient temperature and humidity.  The 8508A/8558A/8588A DMMs are all meant to be mollycoddled in a nice cozy calibration lab, where the temperature and humidity does not change much at all.

95% of the market for the 3458A has been (and still is) as a systems meter for semiconductor companies-- out on the production floor.  The 3458A *CAN* be used for metrology-- but that is a very small segment of the market, and I don't think HPAK ever cared about it very much at all.  This is why they designed in artifact calibration-- to keep the end-user's production cost down.  A big semiconductor company can have over a thousand of these meters-- can you imagine what the cost might be to adjust each one of these against primary standards for each range in each function?  Wow!

The 8508A/8558A/8588A DMMs are an entirely different matter-- and they are specifically designed to support metrology labs--- so, they must remain stable (in the calibration lab) and have calibration traceability to primary standards.  In other words, each range in each function must be measured against primary standards before, during, and after adjustment-- hence there is absolutely no reason to build in artifact calibration, since the full complement of primary standards are right there (and handy) to make the adjustments anyway.

Apples and oranges people.

I would love to see Keysight come out with an 8.5 digit meter that is specifically designed for metrology use (and then maybe another similar model for systems use).  I would still like to have the artifact calibration myself-- but real calibration labs would probably rather use individual primary standards for adjustment.  In the new meter, the artifact calibration should create a "traceability report"-- showing the uncertainties of the standards used, and the uncertainties of each range after adjustment.  Probably, this could be shown on the screen (with realtime updates) as you are performing a measurement, because the final measurement uncertainty can be dependent on what is being measured.

[e61_phil]

@Magnificent Bastard: I don't see much contradiction to what I said. I'm arguing only from the metrology side of view. But, at least here in germany, ACAL isn't enough if you need traceability. There is no way around a proper calibration if you are an accredited company in germany.
And the Fluke 8558A is designed as a 3458A replacement. Hence the name. And that is not only for the metrology applications, also for the applications which needs very fast DMMs. -> System-DMM

use low (or medium) quality components in the critical areas; with a high-spec voltage and resistance standard; then use auto-cal to compensate for time and temperature drift.

That is exactly what I said: You cannot use a 3458A without ACAL.
And what can be wrong with a meter which is stable without ACAL? That gives only more confidence when you make range transfers, because that really shows up something odd. If you meter is build with medium quality parts you will always assume that some little shifts are normal and that can cover problems.

hence there is absolutely no reason to build in artifact calibration, since the full complement of primary standards are right there (and handy) to make the adjustments anyway.

Nothing to add


I think we are around the circle with the discussion. Main points are on the table. We all would like to have some Cal Check function which reports drift. And if you have something like this it is just software to use the latest findings to correct for the actual value or switch the function off and display the values with the CAL constants found during the last calibration.

[MiDi]

SN18A states 0.43ppm/day average for A3 in spec -> 90d ~40ppm, 1y ~150ppm 

[Kleinstein]

[quote author=e61_phil link=topic=174129.msg2289000#msg2289000 date=1553279985
....
I think we are around the circle with the discussion. Main points are on the table. We all would like to have some Cal Check function which reports drift. And if you have something like this it is just software to use the latest findings to correct for the actual value or switch the function off and display the values with the CAL constants found during the last calibration.
[/quote]

An internal cal check would be really nice for a metrology grade meter. It may be there as part of the self test. However it may need a little extra hardware to provide the internal test signal - not much, so one would hardly notice this at the price point of a 8 digit meter. As it helps to have the same internal reference, an external source may not be a full replacement.

Depending on the degree of test, it might also need things like a second voltage reference and thus possibly a major addition. The Datron 1281 had 3 voltage refs (2x LTZ, 1 x LM399) and from the HW side the possibility to read them separately. It is still not really possible to show the actual drift, more the indication of something to be wrong - like having two clocks and still not knowing the exact time. It would be only for amplifier gain that one could show inconsistency with the ADC  - it could still be the ADC aging, though not very likely.

The Fluke 8588/58 could also profit from a kind of internal cal: they seem to use different ADCs for the slow (> 100 µs aperture) and fast conversions and the digitize mode. The specs for the fast conversions are not that great, and at least the longer term drift or TC could really gain something from an internal comparison of fast and slow conversions.  The fast modes are likely rarely used for metrology applications and may not have the strict traceability needs.

It is a little odd to see 2 years specs for the reference meter - for real cal lab use those meters are more like on a less than 1 year cal cycle. I would see more use in specs for a longer cal cycle for the lower resolution meters. Also the longer the time, the more attractive ACAL gets - resistors / dividers tend to drift over time, while the ADC's INL is expected to be rather stable over time.

[TiN]

The Datron 1281 had 3 voltage refs (2x LTZ, 1 x LM399)

Sorry, only 2 x LTZ in 1281/4950. That was 1271 replaced "check" LTZ with LM399 instead.
However 8508A cheaped down and used only single LTZ1000A (LTFLU in later revisions).

And the Fluke 8558A is designed as a 3458A replacement.

If early 8558A specs are close to the truth, then 8558A is not that good competitor to working 3458A (DCV/Resistance), unless cheaper than 10K$ to fill tiny market spot above 7.5 benchtop meters but below metrology/reference meters.

[e61_phil]

The 8558A will cost ~10-11k€. Therefore, almost the same price but RoHS compliant.

I haven't checked everything in detail, but it seems they gave the 8558A the same specs as the 3458A opt 2

[Magnificent Bastard]

In the "things that I would like to see" department, my wish to Santa Clause would be an 8.5-digit SMU.  This would be able to generate as well as measure Ohms, DCV, ACV, DCI, and ACI.  Since it would already have the sources, then doing a "Cal Check", or "A-Cal" or even an "Artifact-Cal" would be possible.  This could be used to calibrate "lesser" DMMs-- 6.5 or maybe even 7.5 digit meters-- as well as measure all kinds of things.  It would be the ultimate machine for a volt-nut.  Unfortunately, the market is so small for something like that, we will never see it happen.

[TiN]

my wish to Santa Clause would be an 8.5-digit SMU.

Like we engineers say - When you want something...you build something.
I would laugh together with Santa too...if I didn't build 5720A last year... 

[e61_phil]

CalPlus send me Product Specifications today (I think they are online now?).
Unfortunately, the footnote 15 (transfer within 10%) is still there, I will ask...

[TiN]

Yes, it's online public now.
Documents published are byte to byte exact to ones posted few weeks ago, that we already discussed nicely.
I cringed a little on advertisement video:

6.5-digit 5522A workhorse calibrator with Pomona banana jack + Belden cable, while voice behind the scenes says "designed by metrologists for metrologists". I'm used to make same cables for my K2002's. Aww...

Promo PDFs show proper 5730A's , 5790B's and Fluke Everett JJ system and "Fluke Volt" to as reabilitation tho.



Keysight, now eyes on you. Let's see 9-digit meter, pretty please. 3458A after all already gives 9.5 digits of noise over GPIB.
Year 2019 => SI Volt 2019 => 9-digit, get it ?     Just don't use LM399 , and we are golden.

[e61_phil]

Hopefully your picture should show how a 5522A is calibrated 

[mzzj]

30 amp range on 8588 is quite exceptional on dvm.

And I feel sorry for that poor bastard who has to enter all the data from the calibration certificate to stability tracking file 
675 calibration points in the certificate..  Fluke rep wasn't sure if the calibration certificate is available as file but agreed that it would be a good idea

[ScoobyDoo]

Hello Voltnuts folks,

                              The Fluke 8588A demo video is up and running again:



Further info:
https://us.flukecal.com/fluke-calibrations-8588a?lcid=778c7b5f-7d67-e911-80cd-00155d027460&utm_source=YouTube&utm_medium=Social-Media&utm_campaign=Video

Herzliche Grüße/Meilleures salutations/Best regards

ScoobyDoo

[srnec]

some pics from inside of beast

[bsdphk]

I must admit that I have a hard time recognizing the picture you two guys paint of HP3458's ACAL.

While you two focus on the "top line" specs for DCV and OHM stability, in the 3458 ACAL firmware those are almost a detail.

The real meat of the ACAL code, and what makes the 3458A special because of it, is things like calibrated zero offsets for all the relay combinations, all the AC* gains and so on.

Likewise to claim that the new Fluke does not do ACAL flies in the face of the laws of nature:  There is no way they can deliver the stated specs without doing some kind of ACAL on a lot of the parasitics.

I suspect the main reason for 3458As has the manual ACAL is that it takes a long time (mainly because of the ADC and averaging), because it puts wear on the reed relays and because you really should run it with nothing connected to the input terminals.

With modern computing power, a 10+ times faster ADC and solid state range/function switches I expect ACAL to be "hidden" behind function changes, and to deal with long term drift by periodically "stealing" single conversions to get zero/ref measurements as input for a real-time mathematical model of the measurement hardware.

In the 3458 that would have required at least a 68K20+FPU and caused constant clicking of relays.

What I'm most looking forward to in these new meters (not that I can afford one) is if they have for instance three ADCs running pipelined, each with its own voltage reference, to get the noise down by sqrt(3).

[Magnificent Bastard]

An 8.5-digit meter is more of a RATIO measuring device, and less of an absolute standard.  Even with these new Fluke meters, the absolute specs are +/- a few digits in the SIXTH decimal place.  The final 2 digits are useful for ratios.  These meters are NOT a "calibration lab in a box"-- please stop thinking of them that way.  Achieving uncertainty less than a few PPM requires different, more precise, (and mathematically provable) techniques than just whacking a resistor or voltage standard onto a "golden" DMM, and calling it "good".  That's why NMIs and Tier-1/Tier-2 calibration labs still use old-school techniques like bridges, null meters, and self-calibrating dividers.

[TiN]

srnec
Thank you for insight. Have any more? 

Resistor network datecoded 2017, so that gives an idea when Fluke had prototypes (if this is photo of prototype, not the production unit?) of this meter.
Bunch of LTC2057HV's spread around the board. Shunt near the relay looks like VPG BMF type.

bsdphk
To make ACAL useful ADC must be extremely linear. That often eliminates everything else other than SD or multislope integrating type.
FS Noise limitation also comes from DC reference, as it is hard to get noise lower than 1-2 uV, available from best of the best like LTFLU or LTZ1000 references.
So I doubt full ACAL can fast enough to be able run in background, even with solid-state switching. This class meters run autozero in background, and even just that already cuts speed more than in half.

[e61_phil]

This class meters run autozero in background, and even just that already cuts speed more than in half.

I fully agree to your post, but the Fluke meters (at least the 8508A) will not do auto-zero. That prevents the current spike, which comes from recharging the input.

[Kleinstein]

For using ACAL / internal cal measurements there is no simple yes or no. There are different steps where numerical corrections or internal adjustments are used. The simplest one is the auto zero - this could include the input amplifier (like with most HP meters) or just the ADC (e.g. Keithley 2000, Datron 1281,...).  Another point is the ADC gain - this is also quite fast and older meters like the Keithley 19x did that in real time (measuring signal , zero and reference - thus 3 conversion for a single reading). Another point are possible numerical corrections, e.g. for reference or gain drift - the Keithley DMM7510 seem to use this to some degree. These may be based on separate ADC for the temperature.

The ACAL step for the ranges like used in the 3458 is also not limited to non or all: One might use it for the high voltage input divider as these tend to be not that longtime stable, while measuring the high current range shunt's may be less desirable. Linking the ohms current sources to the shunts may be a good idea too.
Adjustments / calibration for the AC ranges are another point. This may not need a high accuracy ADC, but more like special hardware for the tests. The AC part may not have much in common with the rest up to the point of having a separate ADC. 

For all those extra measurements one could choose between in the background or as a special ACAL /zero step called by the user. Even if not used for ACAL essentially the same measurements could / should be used as a kind of self test in a really good meter.

I don't think a full ACAL in the background is feasible. Through thermal effects there can be some kind of slow settling and thus an effect on the next few measurements. Normally this would be a kind of slow settling effect, so it needs a few seconds to get better than ppm after a large jump.

As the new 8588 seems to be quite fast/ low noise I don't think they will use much adjustment in the background - maybe / likely AZ for just the ADC. For the old 8508 that was really slow, I would not be surprised to have the ADC gain adjustment in the background, as this can also help with linearity, as it can compensate some thermal effects.

[srnec]

TIN, you are right, this one is preproduction unit. I am more curious about real units and difference between this one and "final" version.

[srnec]

An 8.5-digit meter is more of a RATIO measuring device, and less of an absolute standard.  Even with these new Fluke meters, the absolute specs are +/- a few digits in the SIXTH decimal place.  The final 2 digits are useful for ratios.  These meters are NOT a "calibration lab in a box"-- please stop thinking of them that way.  Achieving uncertainty less than a few PPM requires different, more precise, (and mathematically provable) techniques than just whacking a resistor or voltage standard onto a "golden" DMM, and calling it "good".  That's why NMIs and Tier-1/Tier-2 calibration labs still use old-school techniques like bridges, null meters, and self-calibrating dividers.

This depends also on what kind of precision you are looking for. For commercial cal. lab is it precise enough. If you want to go higher you should look for something like 732C or Josephson.

[iMo]

After 30 years the community is waiting on something better than the 3458A in all parameters.
At the indicated price (15kE) the expectation is pretty high.

TIN, you are right, this one is preproduction unit. I am more curious about real units and difference between this one and "final" version.

It is my understanding you have/had access to the preproduction unit. Did you play with it?

[srnec]

Yes, I had chance to play with it. What I had chance to try, it is promising piece of hardware. Specially with high speed sampling is Fluke going to for them new territory. To use all new functions there will be necessary to talk with the instrument over network or GPIB.

It is my understanding you have/had access to the preproduction unit. Did you play with it?

[Chipguy]

The asking price for the Fluke 8588A ist 15.400 EUR from a Noth German reseller.

[dr.diesel]

talk with the instrument over network.

This feature alone will win many customers.

How long did it take to boot up?  Pretty much instant on or is it running an OS?

[e61_phil]

How long did it take to boot up?  Pretty much instant on or is it running an OS?

Does the boot time matter for an instrument which takes severeal hours to be within spec?

[dr.diesel]

Does the boot time matter for an instrument which takes severeal hours to be within spec?

Simply curious, given the display and network it's a safe assumption.  Which also likely means years worth of firmware updates till it's stable, like the 7510.

[srnec]

In system menu  "Licenses" is lot of about Linux, so I expecting similar system as in Fluke 400 thermal imager. It is booting faster than their 289 handheld DMM

[e61_phil]

The Fluke 5730A seems also to be running on Linux. The boot time is ok.

[srnec]

Someone who needs this kind of instrument may not be interested in absolute accuracy. This is reason why the 8558A exists. You may need only resolution and cold start few times per day.
 

How long did it take to boot up?  Pretty much instant on or is it running an OS?

Does the boot time matter for an instrument which takes severeal hours to be within spec?

[dr.diesel]

I don't personally care how long it takes to boot, I only asked because watching/how an instrument boots does shed further clues to how it's built, design choices etc.

[srnec]

Don't worry, my reaction was more about e61_phil's comment.

I am heavy using Ethernet communication also on 8846A (gives me another digit) and other instruments, for many purposes it is much more effective than GPIB.

I don't personally care how long it takes to boot, I only asked because watching/how an instrument boots does shed further clues to how it's built, design choices etc.

[guenthert]

[..]
I am heavy using Ethernet communication also on 8846A (gives me another digit) and other instruments, for many purposes it is much more effective than GPIB.

Not to mention, that Ethernet connectors provide galvanic separation via 'magnetics'.

[srnec]

Yes, burn only one board/IC when light strike hit it. :)

This one says a lot:

Not to mention, that Ethernet connectors provide galvanic separation via 'magnetics'.

[iMo]

What is on the sdcard there?
I can see an ADSP-BF527 Blackfin on the inguard pcb..
More pictures, plz 

[jeremy]

what is the purpose of these "fish scale" patterns all over this board?

https://electronics.stackexchange.com/questions/85633/what-is-copper-thieving-and-why-use-it

[TiN]

Yup, that board is impedance controlled, so copper area need to be uniform to maintain trace thickness during etch.

srnec, can you show us analog boards pics please? I'm particulary interested in resistors for resistance and current ranges and voltage reference. I'd expect it's like later 8508As, LTFLU + network?
Do you have access to calibrator like 5700/5720? Curious minds want to see linearity. Fluke, as always, keep this important specification ommited and replaced by marketing woowoo instead.

I don't get the excitement over network connectivity. If somebody fork 10+K for meter performance like this, spending 500$ for GPIB/network bridge/etc is mandatory and standard already anyway. 

[dr.diesel]

I don't get the excitement over network connectivity. If somebody fork 10+K for meter performance like this, spending 500$ for GPIB/network bridge/etc is mandatory and standard already anyway. 

In production facilities these things end up in every cell, or station, which can number in the hundreds.  At this point the GPIB bridge is a 1:1 unnecessary accessory, plus cable, plus clutter at the rack etc.

One instrument per IP is also easier to manage, with less equipment in the middle.

[TiN]

Matter of replacing GPIB bridge with LAN router and more worry about security holes. It's relative, I'd expect. 

[srnec]

I hate fatty GPIB cables and connectors :)
Ethernet has many advantages and only few drawbacks compare to GPIB. Galvanic separation and long distance connection is for me only cherry on top.

[Mr. Scram]

Interesting times.

[srnec]

This one? Everything except current shunt and digital is on one huge board.
It is covered by shielding, when I will get final product, I will do measurement against 5730 and 732C. Today I had discussion also about 8846 linearity, one guy tested id due interlaboratory comparison measurement. He told me that it was pretty linear.

srnec, can you show us analog boards pics please? I'm particulary interested in resistors for resistance and current ranges and voltage reference. I'd expect it's like later 8508As, LTFLU + network? :)
Do you have access to calibrator like 5700/5720? Curious minds want to see linearity. Fluke, as always, keep this important specification ommited and replaced by marketing woowoo instead.

[iMo]

@Srnec: Plz, be so kind and do make pictures of everything there in the box, if possible.
People do make and do like teardowns of DMMs on the eevblog.

[TiN]

Yees. So as I expected, it is based on LTFLU reference. But what's going on with all those LM324's on right side of the board 
Also interesting box between WIMA caps, perhaps LTC1043 based capacitor divider/multiplier? 

[e61_phil]

This one? Everything except current shunt and digital is on one huge board.

Are there two AD ADUM ICs for galvanic isolation of the analogue board?

Do they use a standard transformer with extra windings for the analogue part or some special DC/DC converters?

[Kleinstein]

Yees. So as I expected, it is based on LTFLU reference. But what's going on with all those LM324's on right side of the board 
Also interesting box between WIMA caps, perhaps LTC1043 based capacitor divider/multiplier? 

A capacitive voltage multiplier (or divider) would absolutely make sense. The Datron 1281 also uses this, though not LTC1043 based.

The lots of LM324 (though hard to read) are really odd, as I see essentially no passives a round them. LM339 would make more sense with the chain of 4095.

[dr.diesel]

it is based on LTFLU reference

Aside from their familiarity of/with the LTFLU, this would not have been a cost issue IMO.  Q&A with their engineers would be sure be fun!     

[srnec]

The SD Card is internal memory for measurements storing.

What is on the sdcard there?

[dr.diesel]

Keysight Germany has commissioned the design of the successor of the 3458A to a German engineering house - three years ago -
their new unit will be launched soon - in term of weeks but no longer ... several sources inside Keysight told me such.

 

[pansku]

Looks like the 3458A follower launch is getting closer. Today I got an application note from Keysight titled "Calculating Uncertainty using Digital Multimeter Ratio Measurement Techniques" and look what was lurking there 

[Mr. Scram]

I guess the successor of the 3458A is the 3458A. I have to say I don't mind the VCR look.

[EEVblog]

They finally changed to the black labeling huh. Worth the upgrade 

[David Hess]

They finally changed to the black labeling huh. Worth the upgrade 

If Darth Vader had a multimeter, that would be it.

[kj7e]

They finally changed to the black labeling huh. Worth the upgrade 

If Darth Vader had a multimeter, that would be it.

I was thinking BatMeter.

[splin]

What chance of a glut of dirt cheap 3458As as cal labs worldwide rush to update to the latest and greatest black-meters? Or at least half so they can't be accused of colour discrimination? 

[maginnovision]

What chance of a glut of dirt cheap 3458As as cal labs worldwide rush to update to the latest and greatest black-meters? Or at least half so they can't be accused of colour discrimination? 

I'm guessing 0.

[splin]

What chance of a glut of dirt cheap 3458As as cal labs worldwide rush to update to the latest and greatest black-meters? Or at least half so they can't be accused of colour discrimination? 

I'm guessing 0.

Ahh, you're no fun; cold hard reality is an unwelcome intrusion into a pleasant, if futile, daydream... 

[TiN]

Yup, zero. Nobody sane would replace meter with a history for a new untested black box.
Enough to say, Fluke managed to get key transfer specifications worse on every generation (Datron 1281 -> 8508 -> 8588A)

[splin]

Yup, zero. Nobody sane would replace meter with a history for a new untested black box.
Enough to say, Fluke managed to get key transfer specifications worse on every generation (Datron 1281 -> 8508 -> 8588A)

Worse?  Or more realistic perhaps? Maybe they are simply being more conservative with their specifications?

[TiN]

Well these are specifications from manufacturer. Datron 1281 had no known problems to meet it's specification, so calling it's amazing performance unrealistic is a long shot.  We can all hope and expect that meters perform better than specifications, but it is hard to put just hopes into the calibration report. I'd love to be proven wrong if somebody actually have calibrated standards (732B/742A/SR104 etc) and 8588A to collect the data, can supply with Python scripts to do so. I have access only to 1281s, 3458s and 8508.

[Kleinstein]

Checking if a meter meets the transfer specs does not need calibrated standards. It is about short time stability and linearity.  Anyway a check on a few meters does no say much on how conservative the specs are. These specs need tests on many meters and probably some estimates that can be more or less conservative. There even is a chance they may have to (or can) be revised when more units are in service an tested.

[e61_phil]

I was searching for Datron 1281 transfer specs, but could only find stability specs.
Has anyone a link for a document with transfer specs?

[maxwell3e10]

From the manual https://www.elso.sk/media/download/pdf/fluke/man/Datron-1281-user-manual.pdf
the transfer accuracy for DCV is 0.2 + 0.05 ppm on all ranges.
In contrast, both HP3458 and Fluke 8588 have 0.05 + 0.05 ppm transfer accuracy on DC 10 V range. But for some of the other ranges it is indeed worse than 1281.

[e61_phil]

From the manual https://www.elso.sk/media/download/pdf/fluke/man/Datron-1281-user-manual.pdf
the transfer accuracy for DCV is 0.2 + 0.05 ppm on all ranges.

Do you have a page number for me? I searched the document for "transfer" but couldn't find that.

I'm especially interested in the 10k Ohm transfer, but voltage would be a good starting point.

[maxwell3e10]

Its called TEN MINUTE STABILITY SPECIFICATIONS. 10 kOhm is 0.2 + 0.1 ppm for 1281, while for Fluke 8588 it is 0.2 + 0.5 ppm, so indeed significantly worse.

[e61_phil]

Its called TEN MINUTE STABILITY SPECIFICATIONS. 10 kOhm is 0.2 + 0.1 ppm for 1281, while for Fluke 8588 it is 0.2 + 0.5 ppm, so indeed significantly worse.

stability isn't transfer!

stability describes noise only, while transfer includes other errors. And you're comparing 95% specs against 99% specs

[The Soulman]

Its called TEN MINUTE STABILITY SPECIFICATIONS. 10 kOhm is 0.2 + 0.1 ppm for 1281, while for Fluke 8588 it is 0.2 + 0.5 ppm, so indeed significantly worse.

stability isn't transfer!

stability describes noise only, while transfer includes other errors. And you're comparing 95% specs against 99% specs

Is it?
What other errors will be included in the transfer specification?

[e61_phil]

Its called TEN MINUTE STABILITY SPECIFICATIONS. 10 kOhm is 0.2 + 0.1 ppm for 1281, while for Fluke 8588 it is 0.2 + 0.5 ppm, so indeed significantly worse.

stability isn't transfer!

stability describes noise only, while transfer includes other errors. And you're comparing 95% specs against 99% specs

Is it?
What other errors will be included in the transfer specification?

Linearity for example (not only of the ADC, also current source)

[The Soulman]

Its called TEN MINUTE STABILITY SPECIFICATIONS. 10 kOhm is 0.2 + 0.1 ppm for 1281, while for Fluke 8588 it is 0.2 + 0.5 ppm, so indeed significantly worse.

stability isn't transfer!

stability describes noise only, while transfer includes other errors. And you're comparing 95% specs against 99% specs

Is it?
What other errors will be included in the transfer specification?

Linearity for example (not only of the ADC, also current source)

Good point, that doesn't really matter when comparing two resistors of the approximately same value, but
obviously does when comparing a 1K to a 10K resistor for example.
I would assume the "general" linearity specifications apply, I would also assume the linearity doesn't change significantly over time,
so short term specs wouldn't be much better if they were given.

[e61_phil]

That might be true for two resistors equal in value, but that isn't specified by Fluke. They specify a range. So one have to be careful what specifications are compared. I would assume the Fluke would behave also much better for equal values.

I also ran into this trap as I tried to transfer 1k to 10k with my 7081 (sold now), which has also transfer specs and behaved very bad on this 1:10 transfer. We discussed that on another thread here.
I would say the 8508A and 8588A TRANSFER specs aren't comparable to any STABILITY specification.

btw: Solartron called it also "transfer" Spec, which seems to be only valid for equal values.

To get an idea of stability (of one or very equal resistors) it should be ok to make some measurements and calculate the standard error of the mean for a given time.

[maxwell3e10]

If you read Fluke 8588 specification notes, it says: "Transfer specification for DCV, DCI, and Ohms applies to measurement made between 10 % and 120 % of range for deviations of up to 10 % of the initial measurement made using the same configuration for range, filter, aperture, delay etc. Specification accounts for linearity and noise but excludes temperature coefficient which should be calculated from the data provided according to the environment in which the instrument is used."
So it only allows values different by 10%, which is not much of a test of linearity.

[e61_phil]

yes, and there is an ongoing disussion with Fluke about that. But even 10% is more than the same value.
And that 10% isn't specified for the 8508A.

[niner_007]

Seems incredible - the 8508A adds 0.3ppm and the 3458A adds 2ppm for NIST traceability.

Is that because of the uncertainty of the equipment used for calibrating the 3458A?

[EEVblog]

Fluke Australia have just offered a short loan of the new 8588A Reference Multimeter, I can only have it for a day or two. Presumably only one in the country and it has to go around the country doing a dog and pony show.
Obviously an actual full review is out of the question (and pointless), as I have no other bleeding edge voltnuttery gear to measure it against.
What would you want to me to do with it?

[TheSteve]

Fluke Australia have just offered a short loan of the new 8588A Reference Multimeter, I can only have it for a day or two. Presumably only one in the country and it has to go around the country doing a dog and pony show.
Obviously an actual full review is out of the question (and pointless), as I have no other bleeding edge voltnuttery gear to measure it against.
What would you want to me to do with it?

Will they let you open it and take detailed shots of the inside?

[TiN]

Take it apaaaaart :-D. But I'd expect that is out of question for a demo unit either. 

Like you said, anything else is kinda pointless, it would take a week only to warm-up and settle for any meaningful measurements.
I'm working on making review comparing old classy Datron 1281 vs newer Fluke 8508A and touching a bit on 8588A (which I don't have access for).
Sadly some of the metrology key parameters such as transfer specifications for resistance and DCV are getting worse and worse with every new meter.

[EEVblog]

Fluke Australia have just offered a short loan of the new 8588A Reference Multimeter, I can only have it for a day or two. Presumably only one in the country and it has to go around the country doing a dog and pony show.
Obviously an actual full review is out of the question (and pointless), as I have no other bleeding edge voltnuttery gear to measure it against.
What would you want to me to do with it?

Will they let you open it and take detailed shots of the inside?

Don't know yet, I've asked.

[amspire]

Fluke Australia have just offered a short loan of the new 8588A Reference Multimeter, I can only have it for a day or two. Presumably only one in the country and it has to go around the country doing a dog and pony show.
What would you want to me to do with it?

I guess you will be calibrating some of your references with it. It is interesting to see how accurate some of your references and meters actually are.

You could show how the references change with temperature.

Also, perhaps you could demonstrate the very low frequency noise of zeners and compare the noise ordinary zeners to the buried zeners found in the good references.

Richard

[e61_phil]

Fluke Australia have just offered a short loan of the new 8588A Reference Multimeter, I can only have it for a day or two. Presumably only one in the country and it has to go around the country doing a dog and pony show.
Obviously an actual full review is out of the question (and pointless), as I have no other bleeding edge voltnuttery gear to measure it against.
What would you want to me to do with it?

Hi Dave,

it is a very nice meter. I had it for 1.5 weeks and now I ordered one. It would be nice if you could compare the sampling against your 7.5 digit keysight meter. The fast sampling will not suffer from short warming up.

Sadly some of the metrology key parameters such as transfer specifications for resistance and DCV are getting worse and worse with every new meter.

TiN, you write that again and again, but I think that isn't true. The Datron 1281 hasn't any transfer specs at all and the 8588A has a limited range for compararison (that is really a pity). Which means you have to apply 24h specs if you want to make 10:1 transfers.

The 1281 is the winner in the 100mV range. The other ranges are divided by the 8508A and the 8588A (that's why I keep the 8508A). The 3458A is still unbeaten on paper up to 100V.

I also think that Fluke has very carefull specicifications on the meters. I did some linearity/transfer tests and there is no visible difference in transfer accuracy between the HP 3458A and the Fluke 8588A loaner.

[TiN]

TiN, you write that again and again, but I think that isn't true. The Datron 1281 hasn't any transfer specs at all and the 8588A has a limited range for compararison (that is really a pity). Which means you have to apply 24h specs if you want to make 10:1 transfers.

Maybe so. We already discussed this before  . I tend to be more interested in design capability and actual performance, rather than careful specmanship from Fluke, which is so careful, that even critical specification like linearity are nowhere to be found. Again, I don't say that 8588A is bad meter, I just disappointed in lack of transparency from metrology targeted product. So I've used 10 minute stability for D1281 and same 10 minute stability for 3458A/HFL specifications to compare with + INL specification. And unlike 3458A, Datron and Fluke is easier to apply for transfers because internal GPIB-controllable switching for inputs. I love using 1281 to compare three references together (with 8508/001 or 8588A can do only two references comparison).

[e61_phil]

From my side of view the most interesting "INL spec" is the transfer spec. Saying 0.1ppm INL on the first page of the 3458A datasheet is only marketing and you can't use that number for any calculation. But, you are right we already discussed that over and over

What kind of specification are you missing? Perhaps one can ask to publish that. My "feeling" is, that Fluke Calibration is really knowing what they do and you can get support from people with metrology know how. My experience with Keysight is the complete opposite.


How do you connect a third reference to the 1281? Sounds interesting

You can compare Ref A against the internal Ref and Ref B against the internal. This way you can also compare"three", but I guess you want to do something else
For this purpose, I build a small box with low thermal relays to compare references with my 3458A at home.

[TiN]

That is secret knowledge of xDevs IRC team 



I just use both rear inputs on 1281 and scan each channel in sequence.

[e61_phil]

Very nice!
Do you know what kind of relays they use? The best ones I could find (thanks to Johannes!) are specified with 300nV typical. It seems that my box is better, but perhaps there are better ones..

[maxwell3e10]

I did some linearity/transfer tests and there is no visible difference in transfer accuracy between the HP 3458A and the Fluke 8588A loaner.

Could you explain what is going on in the attached plots and table. I am not quite following what all the numbers refer to.

[niner_007]

Saying 0.1ppm INL on the first page of the 3458A datasheet is only marketing and you can't use that number for any calculation.

My "feeling" is, that Fluke Calibration is really knowing what they do and you can get support from people with metrology know how. My experience with Keysight is the complete opposite.

The unpublished INL spec is suspect, my feeling is that the Flukes have more marketing on their back that anything else, it's 2019 after all. My feeling is that the ADC in the 3458A is still unbeaten and Fluke is still playing catch up.

[niner_007]

Very nice!
Do you know what kind of relays they use? The best ones I could find (thanks to Johannes!) are specified with 300nV typical. It seems that my box is better, but perhaps there are better ones..

Which relay are these? The COTO 3500 are < 500nV, that's the only ones I'm aware of with such low emf spec.

[e61_phil]

Saying 0.1ppm INL on the first page of the 3458A datasheet is only marketing and you can't use that number for any calculation.

My "feeling" is, that Fluke Calibration is really knowing what they do and you can get support from people with metrology know how. My experience with Keysight is the complete opposite.

The unpublished INL spec is suspect, my feeling is that the Flukes have more marketing on their back that anything else, it's 2019 after all. My feeling is that the ADC in the 3458A is still unbeaten and Fluke is still playing catch up.

If the INL number (I wouldn't call ist spec) is that important could you tell me how I can derive the transfer specification of 0.05ppm + 0.05ppm (10V range -> direct ADC) from that "0.1 ppm dcV linearity"? Or any other way of calculation, if I want to compare two voltages?
How do you define the INL number given in the marketing section of the datasheet?

[TiN]

INL (of the ADC or meter, it's an important difference) is the total error of the range, while transfer specification is much more limited parameter, that applies only for specific conditions.

By means of verification ADC's INL (which comes from ADC design and hard to improve) and verification of DMM front-end path INL and error factors (which we can work to improve/workaround weak spots), one can determine and get characterized actual transfer specification.

E.g. ADC can have INL 5 ppm over its input range, that looks like a smooth curve, and with doing the smart math and digital equilibristics you can get 0.5ppm transfer spec out of it. But you cannot do the opposite.

Relays in 1281 (and 8508 too) that handle inputs are S4-L's. We did repairs and teardowns of 1281 multiple times before. There is no much need is super-low emf, since errors from thermal EMFs are not changing faster then the measurement cycle.

Could you explain what is going on in the attached plots and table. I am not quite following what all the numbers refer to.

Not sure what is difficult here? This is same Datron 1281 DMM measuring three different 10V voltages.

[e61_phil]

Nice information with the relays! Thanks.

The datasheet 0.1ppm is meter INL for sure, nothing else makes sense.

How do you define "total error"? My question is simple, I guess: What is the "0.1ppm dcV linearity" useful for? I don't know how I should use that to calculate the error of a 3V to 6V transfer for example.

And I guess the definition is not everywhere the same. The Fluke calibrators also have a linearity spec in ppm, but it seems that these values are simply added to the error from the rest.

[TiN]

You need to know all error contributors to arrive at actual transfer error of 3V to 6V. INL alone is not enough, otherwise you wouldn't have all those limitations in transfer specifications.
The whole reason why limitations are specified - so other error contributors are minimized. It is that for high end DMM or MFC, short-term stability often limited only by signal path noise and INL.
Noise we can attempt to average/filter out. INL you cannot filter out or predict, so that's why I think it's one of the key design parameters that can be helpful to know in uncertainty calculations.

Of course, most of the calibration applications don't go down this rabbit hole and direct you just using manufacturer's specification like 24-hour stability or 15 minute transfer specification, so manufacturer already did the calculation and estimation of the worst-case error for you. Knowing INL and actual errors from DC amplifier, switching circuits, thermal stability, reference noise, ADC resolution and so on is important if you want to verify manufacturer's specifications instead of just trusting datasheet's numbers like 0.05 ppm Umeas + 0.05 ppm Urng.

Fluke 57xx MFC already have included INL errors in the specifications, you don't have to add anything. You can review ACAL function verification paper for some more details, how transfers are actually covering various functions in MFC.

[e61_phil]

TiN please make a point. It is again only "it is important"-bla bla (sorry). If you want to verify the manufactures data you should start from scratch and then you don't need a ADC INL spec (which is also not given for the 3458A. The 0.1ppm are for the meter). If you want to make transfers between standards with least possible (specified) uncertainties you need transfer specs not marketing 0.1ppm INL nonsense. There are measurements shown by Fluke for the 8508A INL. http://support.fluke.com/Calibration-Sales/download/asset/2114953_a_w.pdf
I don't know what specification is missing here.


The Fluke 57XX and also the old 5440B have linearity specs. You can use that if you want to verify linearity of a meter for example. The 5730A specifies linearity with 0.3ppm + 2µV on the 11V range. That is a spec you can simply add to your measurements. It is not the span of an error.
Here is an example with applied calibrator linearity uncertainties: https://www.yumpu.com/en/document/read/11827423/a-wheatstone-bridge-for-the-computer-age-les-huntley-fluke

[Kleinstein]

The INL error can be a relevant part of the transfer specs and also limit the accuracy of the ACAL method. However if the type of INL is known the transfer specs can be tighter than just the INL number would suggest, as linearity has more aspects than just the upper limit of the INL curve.
For the INL specs 0.1 ppm of a 10 V full scale would mean a possible 1 µV error from nonlinearity. For the example 3 to 6 V transfer this would be some 0.3 ppm contribution to uncertainty.

INL specs are tricky, as it is rather difficult to really test INL and not all units may be the same. One can do a rather time consuming test at some 1000 point's with a very accurate source (JJA), but this still is only a partial (though usually sufficient) test and not at all practical for every unit or even a larger sample. So I can understand that Fluke is reluctant to give INL specs, as they may still have to measure more units.
With an ADC like in the 3458 there is a chance that a drifting resistor array may also effect the INL - so it may change over time. Not so much in the 3458 but more like the Datron1281 and ADTV6581, there could also be aging (ingress of humidity) of the integration cap that can make INL worse over time.

A special test point for linearity is the turn over error (comparing positive an negative reading). However a good values for the turn over error does not guarantee good INL: a good example are integrated SD ADCs with differential inputs that often by design have a very low turn over error, much better than the full INL.  Also some thermal effects are not caught by the turn over specs.

There are measurements shown by Fluke for the 8508A INL. http://support.fluke.com/Calibration-Sales/download/asset/2114953_a_w.pdf
I don't know what specification is missing here.

The INL error curve is just an example - other units may and in some aspects will behave different.

There is no absolute need for INL specs. Transfer specs for enough cases would be sufficient. However these are usually only give for a few conditions and not even the same for all meters. Not sure if this is more like cherry picking or just manufacturer specific preferences.

[TiN]

My point is very simple. I repair and test 8.5d meters and 7.5d calibrators, so as part of troubleshooting I always have to do INL test on DUT, to ensure functionality built on top of INL performance are correct. Using full transfer spec as a limiter in such tests is not adequate. You should be able to understand why already. Testing INL shows dodgy out of spec 3458 ADC very easily, even though meter still meets transfer specifications, thanks to ACAL.

There would be no reason in testing any calibration issues or ACAL or transfer features if core ADC or DAC cannot deliver the required sub-ppm INL performance. Since my tests are also automated, I use two or more verified 3458A's as "composite 0.1ppm 10V INL standard" in such tests, with two DCV meters confirmed lack of the temperature dependency. This is why INL number in datasheet is important for me. Don't have my own JVS yet. Other alternative is manual labor-intensive testing with 720A, SR10*0 and alike. As your Fluke linked doc outlines, INL is also not a problem for transfer between the same nominal value standards, typically used in metrology.

and not all units may be the same.

Precisely! I'd say 2 meters are never same, actually.

Manufacturer specifications apply for most of the units out there. But each particular unit still behaves little differently. More listed design parameters in documentation help me to separate expectations, as I obviously can also perform measurements inside the DMM to find out each domain error and compare that to expected manufacturer values.

With an ADC like in the 3458 there is a chance that a drifting resistor array may also effect the INL

.
My measurement results (from >8 different ADCs, good and bad) show that all drifty/bad ADCs cannot meet 0.1ppm INL in -11 to +11V sweep. Good ADCs have no problem with this.

There is no absolute need for INL specs. Transfer specs for enough cases would be sufficient. However these are usually only give for a few conditions and not even the same for all meters. Not sure if this is more like cherry picking or just manufacturer specific preferences.

It is more of a desire item, then actual metrology field need. We are deep down the rabbit hole for this one here.
Maybe I'm expecting too much, thinking about a customer who paid $20k+ for "reference DMM" and not having that little bit of optional information? Kind of same goes to traceabillity in calibration reports. Desired wish - to see data of all upstream standards chain to SI, so one can calculate own numbers for U. Like open-source calibrations, sorta. Actual metrology field tho provide you only final measurement result with assigned uncertainty and expecting you to trust the lab in their measurement. Lab is audited by accredited body to ensure their measurements are in order, but there is no way to be 100% sure your particular calibration report is correct. Maybe the calibration tech had bad mood during your DMM calibration So it's all marketing and specmanship, one would say, no any different to Vishay's "typical 0.0 ppm/C TCR" resistors? Truth is hiding somewhere in between... .

[e61_phil]

I never said it doesn't make sense to measure linearity of a meter. This is done in every proper calibration. (And that is why you need linearity specs of the calibrator)

I don't get your point why it is so important to write in the marketing section of a datasheet 0.1ppm linearity (they don't tell it INL). Both (Fluke and HP) showed "typical" values of their meters against a JJA.

You pretend that Fluke wants to hide something. That isn't the case in my opinion. I think it is more the other way around. Why aren't there absolute specs on the 3458A datasheet? There is only a footnote that you should add 2ppm to be traceable.

[Kleinstein]

....
Why aren't there absolute specs on the 3458A datasheet? There is only a footnote that you should add 2ppm to be traceable.

At the high end the absolute specs are less important as the difference to the relative specs only reflects the cal lab quality. So it's only valid for the 1st year or so.  After that it depends on where and how the meter is calibrated. The extra 2 ppm  for tractability are only valid for the initial calibration at HP (this may have even changed over time).

[e61_phil]

....
Why aren't there absolute specs on the 3458A datasheet? There is only a footnote that you should add 2ppm to be traceable.

At the high end the absolute specs are less important as the difference to the relative specs only reflects the cal lab quality. So it's only valid for the 1st year or so.  After that it depends on where and how the meter is calibrated. The extra 2 ppm  for tractability are only valid for the initial calibration at HP (this may have even changed over time).

That might be clear to the people here in this forum section, but one can see many comparisions between Fluke 8508A and HP 3458A were people compare the specs from the 3458A datasheet against the absolute specs of the 8508A. And it is comfortable to have absolute specs which can directly be applied, that is what I expect from a "reference" meter. That doesn't mean they don't give relative specs.

[splin]

Very nice!
Do you know what kind of relays they use? The best ones I could find (thanks to Johannes!) are specified with 300nV typical. It seems that my box is better, but perhaps there are better ones..

Which relay are these? The COTO 3500 are < 500nV, that's the only ones I'm aware of with such low emf spec.

Panasonic TX-S

[EDIT] Take a look at this thread:

https://www.eevblog.com/forum/metrology/relay-choice-for-scanner/

[e61_phil]

Panasonic TX-S

Yes, I used Panasonic TX-S. They seem to work very well.

[EEVblog]

I've been told no on taking the lid off. They need the cal intact in order to send it to various customers over the coming weeks.
So I can have it tomorrow for the weekend, but can't do a teardown.
It might be available in December for that though.
Should I bother?

[Andreas]

What I do with every Instrument is a tilting test when having connected to a stable reference
E.g. here with a K2000 which is surprisingly bad (mostly due to the LM399 reference):

https://www.eevblog.com/forum/metrology/project-pimp-a-keithley-2000/msg1106829/#msg1106829

with best regards

Andreas

[daqq]

I've been told no on taking the lid off.

So, you're saying X-ray the crap out of it? 

[Kleinstein]

A longer run with a shorted input is definitely an interesting point. Kind of getting independent noise data that also include longer times (e.g. Alan variation curve).  There is little need to do this with many parameters. The main interest would be longer integration times, though very long times are very likely internal averaging. So something like 10 PLC would  be enough. For the voltage ranges, the 0.1 V range would sense mainly the input amplifier, the 10(20) V range would mainly sense the ADC part and the 100 V range and 1000 V range should not be very different, both having possibly quite some effect of the divider (unless they actually use different dividers.

For the noise I hope that Fluke did not include the same odd 0.1 Hz noise bump like many Keithley meters.

[laichh]

Good reading material about the Fluke 8588A:

https://arxiv.org/pdf/2205.11321.pdf

https://us.flukecal.com/literature/articles-and-education/electrical-calibration/application-notes/inside-newest-85-digit-re