Author Topic: Hints about the next generation 3458A  (Read 1640 times)

0 Members and 1 Guest are viewing this topic.

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 94
  • Country: us
Hints about the next generation 3458A
« on: October 21, 2017, 03:29:03 PM »
Dave did a 5 part video series interviewing John Kenny. There is a lot of good information about the test and measurement industry.


The exciting information is in part 3 at 8 minutes. John talks about how they started working on the replacement for the 3458A but the Agilent management did not want to proceed with the project. I got the impression the Keysight management will probably be more receptive to the idea.

To help the design team, what features do you feel are most important to keep or improve on in the new 3458A?

 
The following users thanked this post: Andreas, zhtoor

Online zhtoor

  • Regular Contributor
  • *
  • Posts: 207
  • Country: pk
Re: Hints about the next generation 3458A
« Reply #1 on: October 21, 2017, 06:37:23 PM »
is it me or anyone else has the impression that the golden era of hp 3458a is basically gone forever?  :-//
maybe over to fluke.

regards.
 

Online lukier

  • Supporter
  • ****
  • Posts: 495
  • Country: gb
    • Homepage
Re: Hints about the next generation 3458A
« Reply #2 on: October 21, 2017, 07:40:22 PM »
Maybe it would be easier for them if they simplified the ADC, removing the fast mode pathways and focusing on the slow (precision) mode - for metrology and autocal purposes and use a discrete 1 MSPS ADC for fast mode and digital-RMS (AFAIR Keithley DMM7510 does something similar, but their slow ADC is not that linear, 1ppm+1ppm only unfortunately).
 

Online Echo88

  • Regular Contributor
  • *
  • Posts: 218
  • Country: de
Re: Hints about the next generation 3458A
« Reply #3 on: October 21, 2017, 08:16:45 PM »
Keysight knows exactly to whom they sell: a big amount goes to semiconductor-fabs and a smaller amount to metrology-labs as far as i understood MisterDiodes.

The analog designers already know what can be improved in the 3458A, the only question is: will the management allow them a big enough budget and time to design it. That depends on the market and the achievable price for such an improved instrument.
 

Offline Awesome14

  • Regular Contributor
  • *
  • Posts: 186
  • Country: us
Re: Hints about the next generation 3458A
« Reply #4 on: October 21, 2017, 08:31:15 PM »
A LCD display would be nice.
War: how Americans learn geography
 

Online zhtoor

  • Regular Contributor
  • *
  • Posts: 207
  • Country: pk
Re: Hints about the next generation 3458A
« Reply #5 on: October 21, 2017, 08:34:56 PM »
adding an lcd and the associated electronics would mean polluting the electromagnetic environment
around *very* sensitive circuitry is not without its hazards, what needs to be done is to probably
further isolate the digital electronics part from the analog section.

regards.
 

Offline dr.diesel

  • Super Contributor
  • ***
  • Posts: 1980
  • Country: us
  • Cramming the magic smoke back in...
Re: Hints about the next generation 3458A
« Reply #6 on: October 21, 2017, 10:41:03 PM »
I suspect most customers are happy with the 3458A just the way it is.  However KS can't be caught off guard if say Keithley made a surprise release of the DMM8510, with comparable performance.

I have a renewed respect for KS after John Kenny's interview, if a 3458A revamp is necessary, KS is listening and bring what's needed.

Of course we'd all enjoy higher performance, but i'd be pretty happy with just the addition of ethernet.

Online HighVoltage

  • Super Contributor
  • ***
  • Posts: 3357
  • Country: de
Re: Hints about the next generation 3458A
« Reply #7 on: October 22, 2017, 01:04:17 AM »
Of course we'd all enjoy higher performance, but i'd be pretty happy with just the addition of ethernet.
... and SCIPI and BenchVue compatibility
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline Edwin G. Pettis

  • Frequent Contributor
  • **
  • Posts: 288
  • Country: us
  • The plural of anecdote is not data.
Re: Hints about the next generation 3458A
« Reply #8 on: October 22, 2017, 02:30:24 AM »
There is a grave error by Kenny when he talks about the3458A and the LTZ1000, HP did NOT design the LTZ1000, Linear Technology designed it, HP's contribution was the idea for the thermal isolation on the LTZ1000A.  Kenny misspoke about a number of things in the interview but they are mostly of little importance.  Frankly, I think the main reason Keysight can't produce any significant improvements on the 3458A is because the engineering department hasn't got the same set of skills and knowledge that the original design team had, all four of those guys are gone from Keysight.
« Last Edit: October 22, 2017, 02:33:18 AM by Edwin G. Pettis »
 
The following users thanked this post: quarks, TiN, Monittosan, lukier, 2N3055, fonograph

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1535
  • Country: ca
Re: Hints about the next generation 3458A
« Reply #9 on: October 22, 2017, 02:40:15 AM »
Hi,

One point that he did make, which shouldn't be under estimated, is that a lot of the 3458A are used in automated test systems where speed is really important:



This is an important attribute that goes beyond the cal lab and precision measurements.

Jay_Diddy_B
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 3439
  • Country: de
Re: Hints about the next generation 3458A
« Reply #10 on: October 22, 2017, 03:13:58 AM »
The high speed and high accuracy are kind of different properties of the 3458. So there might be more than one meter to replace the 3458 - one for higher accuracy and a different one for higher speed.

Especially the higher speed part should not be that difficult today. For a true high speed measurement, especially of low level signals the 3458 like other meters with traditional auto zero also give away about half the time for the AZ phase. So there would be even rather obvious room for improvement.  AFAIK meters like the DMM7510 and 34470 already are not that bad when it comes to fast readings.

Especially for the AC part there are quite a few possible extra function to a DMM, like predefined bands and this way lower ranges. With digital sampling it is more than just a single RMS values to calculate from it.
 
The following users thanked this post: splin

Online e61_phil

  • Frequent Contributor
  • **
  • Posts: 432
  • Country: de
Re: Hints about the next generation 3458A
« Reply #11 on: October 22, 2017, 03:16:17 AM »
Of course we'd all enjoy higher performance, but i'd be pretty happy with just the addition of ethernet.
... and SCIPI and BenchVue compatibility

I would say all these things could easily handled by a small box with ethernet on one side (with LXI and standard SCPI commands) and a GPIB connector on the other side.

Most of the time I use a 3458A it is connected to a computer. Therefore, a LCD or something like that wouldn't be a real benefit. I think it is the same case in test and metrology applications. User experience on a 8.5 digit multimeter isn't very important, I think. Most of the time an computer is talking to the instrument.
A 8.5 digit DMM isn't an electronics engineers workhorse which should have auto hold and stuff like that (like 3446x).
 
The following users thanked this post: splin

Offline Flinstone

  • Regular Contributor
  • *
  • Posts: 157
  • Country: be
Re: Hints about the next generation 3458A
« Reply #12 on: October 22, 2017, 03:53:28 AM »
I also liked this open interview with John Kelly - and little bit confused who commissioned the design of the LTZ1000 and who the designers were (Bob Dobkin/Carl Nelson ... ) and if it was HP why they made the part available for the open market as companies like Keithley, Datron, Fluke (in its early days), a.m.o. ... took advantage - but a lot of early parts in the 3458A had special HP part numbers ...

Last years I have been in contact with KS local service center many times - and it is gorgeous to see how the old HP spirit revives - when I am asking for spare parts / boards - they put one of the 3458A veterans (JD) from Loveland in the loop ... to confirm technical info is correct - one regains respect for this company Keysight that used to be an unrivaled leader in T&M.

cheers
Fred Flintstone

 
« Last Edit: October 22, 2017, 04:14:46 AM by Flinstone »
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 636
  • Country: us
Re: Hints about the next generation 3458A
« Reply #13 on: October 22, 2017, 04:51:49 AM »
I could see a two meter release with different firmware and modified thermal control. Datron/Wavetek had the 1271/1281 DMM depending on the customers's needs. Speed vs accuracy. A metrology based meter with no external fan and more stable internal temps would be great. Keep the fan and ACAL for the wider operating temperatures.

A big improvement would be to increase the maximum value per range. The 10V range could measure to 19.99xxxxx. That would put it back in competition with the 8508A where the upper range readings are needed at the same resolution.
Give it a 1 ohm range and an electronic front / rear switch.

All digital side improvements will probably come from their newer products.
 

Online Moon Winx

  • Regular Contributor
  • *
  • Posts: 55
  • Country: us
Re: Hints about the next generation 3458A
« Reply #14 on: October 22, 2017, 05:58:04 AM »
I guess when you already make the best meter there is not much incentive to spend $$$ on the development of a replacement. If, say, the 8508A was an order of magnitude more accurate, we'd probably see an improved meter coming from keysight. Keysight is making some crazy good bitstream generators now, maybe that is what they are focusing on.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 94
  • Country: us
Re: Hints about the next generation 3458A
« Reply #15 on: October 22, 2017, 08:21:56 AM »
Four techniques to increase time between when AutoCals are needed or performed.

1. I believe someone mentioned that the 3458A could benefit from a variable speed fan. The fan would work to keep the critical internal parts at a constant temperature of say 40 deg C. This would take a little thermal engineering but you may be able to keep the critical internal parts within 0.5 deg C as the external temperature changes 5 deg C. The software would post a warning if the fan was at max speed and the critical internal temperature is now out of regulation. The control software or the operator would then do an Autocal as required.

2. Calibration memory is cheap, why not use more of it? When Autocal is performed at a new temperature close to 200 calibration parameters are updated (I don't remember the exact number). Why not store those parameters as data points for that temperature. Then when Autocal is performed at a new temperature you will have another set of 200 values at a new temperature. With two or ten of these Autocals in memory the main processor can interpolate the cal that applies at intermediate temperatures. Now the meter can determine the correct calibration constants every time the meter changes temperature by 0.1 deg C.

Valid Autocal parameters would only be stored if the internal temperature was stable (+-0.2 deg C?) for xx? minutes.

Autocal also corrects for component drift with time or humidity but that is different from temp co.

Maybe you can program in a Super Autocal All mode. By using the variable speed fan the meter can do it's own internal temperature sweep. This Super Autocal All mode will operate over the weekend. The fan starts at a slow speed for 2 hours and the meter automatically does an Autocal All. Then the fan speeds up 20% (reducing internal temperature 2 deg C), the meter waits 2 hours for temperatures to settle and it performs an Autocal All again. This continues till the fan is at 100% and the internal temperature is lowest. Now the calibration data in memory can be interpolated for the next week.

3. Why can't Autocal All be performed at 5 am before anyone gets to work? Then the meter is calibrated and ready to go for the day. This may not be useful in an ATE environment but it would be helpful for bench or metrology use. Of course it would cut down on excuses for coffee breaks...

4. If the meter is doing high speed 6 1/2 digit work all day, how often is Autocal needed? Is it when the meter internal temperature changes 10 deg C? It would probably help the high speed ATE community if there was guidance in the manual about this.

Just some thoughts. Maybe they will spark a usable idea.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 3439
  • Country: de
Re: Hints about the next generation 3458A
« Reply #16 on: October 22, 2017, 08:50:49 AM »
Having something like a variable fan can be a good idea, but it can also be tricky. Such a board does not have a single temperature, but many different heat source on the board and thus hotter and colder areas. If you increase the air flow over the board one would not just lower the temperature of the hole board, but have a usually larger effect on the hot parts, thus also changing gradients. This is likely more of a headache. However it can work if the critical sections are inside metal boxes so that you only have a temperature for each box - the thermal design would than still need to couple those boxes or make them move the same way. A fully shielded box around some parts may be a good idea anyway as there is more and more RF around from phones and WLAN.

For the auto-cal part, there might be more granularity. Just for doing voltage readings you don't need the adjustment of the Ohms and amps. Especially some of the volts part only need a rather short adjustment, that does not take long. Other parts, like AC amps might take a rather long internal adjustment, maybe no repeating everything just because of the temperature change. Correcting for temperature drift is already done in some meters (e.g. the DMM7510). However this does not really work with humidity, as this usually is a delayed effect, like the average RH over the last 1 days to 10 days depending on the part and temperature. Also sensors are way more complicated.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 2893
  • Country: tw
  • Volt-bolt
    • xDevs.com
Re: Hints about the next generation 3458A
« Reply #17 on: October 22, 2017, 09:54:35 AM »
Or set of hybrid assemblies just like 57XX MFCs, where you have critical circuitry blocks hosted on own ceramic substrate module with film heater resistor and oven control loop. That stuff not cheap though.
There is no much point to keep stable temperature over control logic which occupy nearly half of the current boards :)

What I'd love to see is NFP-version with battery option, so it could be used for floating applications, like null-metering or high-voltage applications, where common voltage to ground can be large.
This would need 10mV/1mV ranges on DCV though.
xDevs.com | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 94
  • Country: us
Re: Hints about the next generation 3458A
« Reply #18 on: October 22, 2017, 11:48:14 AM »
On the first video, at 19 minutes, John talks about not needing to do Analog ASICs anymore. The resistors and switches available today can do the job. Maybe, no more A3-U180 Issues!
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 1610
  • Country: de
Re: Hints about the next generation 3458A
« Reply #19 on: October 22, 2017, 03:50:33 PM »
1. I believe someone mentioned that the 3458A could benefit from a variable speed fan.

Hello,

I think that a variable speed fan makes the temperature gradients within the device unpredictable.
I´d prefer a fan-less device together with a "low power" design. (no filter cleaning anymore).

At least for the 10V-range a factory calilbrated T.C. correction together with the internal temperature sensor would be useful to increase stability.
http://www.eevblog.com/forum/metrology/project-pimp-a-keithley-2000/msg1106839/#msg1106839

The auto-cals performed could be used as refinement by a learning function.

With best regards

Andreas





 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 3439
  • Country: de
Re: Hints about the next generation 3458A
« Reply #20 on: October 22, 2017, 10:32:13 PM »
For a rack mount system an really high performance and thus some power consumption there might be a need for a fan. In a stable 23 C lab, one might get away without a fan, but difficult in a dense rack at 40-50C.

A way to allow for a variable speed fan would be to have the critical analog part inside a close (preferable metal) box, so that the fan would only cool the box as a whole, but not the individual parts. Not problem with direct cooling the digital part or power supply. With the critical parts protected from the fan, there might not be a need for a fine filter anymore. At least cleaning the filter would not be that important.

The DMM7510 pictures already shows some boxes around the reference and something that I think is the input amplifier. The ADC itself might not be that sensitive at that level.
 

Offline Flinstone

  • Regular Contributor
  • *
  • Posts: 157
  • Country: be
Re: Hints about the next generation 3458A
« Reply #21 on: October 26, 2017, 05:05:01 AM »
Hello VoltNuts,
                       I have posted a mail from John Pickering (co-Founder DATRON) reviewing the video blog with John Kenny about the LTZ1000(A) - for historical reasons and correctness - we decided to publish that part of our conversation:

http://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg1332678/#msg1332678

Enjoy reading !

Best regards
Flinstone

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf