Products > Test Equipment
Some old school instruments showing how it's done (HP 3325A and Fluke 8506a)
<< < (61/91) > >>
joeqsmith:

--- Quote from: garrettm on February 17, 2021, 08:01:30 pm ---
--- Quote from: joeqsmith on February 17, 2021, 04:39:34 pm ---There may be software already for the Fluke meter.   I would check that first.   

If you have never used LabView,  the first thing I would suggest is you spend a day going through the getting started manual.  Work all the examples and get a basic idea how it works.  It's not magic and like any language takes time to learn.   You may also want to join their forums.   

The 5-byte data is what the LED displays.  No need to correct it like the 3-byte.  Also, no 2's comp.  So a little easier.   

I've not seen any problems using the technique I mentioned to sync up with the data.   Then again, that serial data is very slow.  But the PCs are so fast... I guess you just need to try it.

--- End quote ---

I'm sure there is software used by the military, but I haven't seen any open source software that supports the Fluke 8505/6A. I'll try looking on NI's site in case someone has uploaded something for one of the paid LabVIEW editions.
--- End quote ---

First hit on Google "labview fluke 8506a":
http://sine.ni.com/apps/utf8/niid_web_display.download_page?p_id_guid=E3B19B3E921B659CE034080020E74861



--- Quote from: garrettm on February 17, 2021, 08:01:30 pm ---All fixed-point binary data (except the exponent byte) require two's complementing as an entire word before conversion to floating point. This is done to make the value positive before merging the fraction bits into the floating point mantissa and letting the processor figure out the bit fields for the integer portion (using FP addition). Otherwise, all single byte binary data is in signed char format (i.e., "two's complement"). So no work to do there.

That said, there are free fixed-point libraries for C++ available which would eliminate the need for conversion, but I haven't tried using them yet.


--- End quote ---

5-byte is stored in 2's comp where 3-byte requires 2's comp.   I implemented both in Labview but obviously am only using the 5-byte. 



--- Quote from: garrettm on February 17, 2021, 08:01:30 pm ---I'm hoping to get at least 360 readings/s using the HS mode. I can do it with the external trigger input but I've yet to get it to work using bus triggering with my current setup. I plan on making an AR488 Arduino GPIB interface here soon which should solve my inability to directly control the talk/listen operation of the DMM.
..
If I can't figure out LabVIEW I may write a program in Visual Studio 2017 or Matlab to control the instrument, generate CSV files and plot the data--as I have a little more experience with them. Though, from what I can tell, LabVIEW is supposed to make doing this pretty simple. There is also a program on the forum for reading from multiple instruments but it appears to only work with SCPI instruments. I need to look through the documentation and see if I could make it work with these older meters.

--- End quote ---

I've not ran into any problems controlling various test equipment with LabView, but I've been using it for many years.   It's not something I picked up overnight.   
garrettm:

--- Quote from: joeqsmith on February 17, 2021, 09:01:19 pm ---5-byte is stored in 2's comp where 3-byte requires 2's comp.   I implemented both in Labview but obviously am only using the 5-byte.

--- End quote ---

All three binary formats use a fixed-point type where only the number of integer bits and fractional bits about the radix point is changed. Hence, the conversion is the same for each format when ignoring the error bit and implied base 10 exponent for 3-byte binary or explicit exponent byte for 5 and 6-byte binary (which is always 10^1 for voltage measurements in any binary format).

Thus, all negative valued binary data require two's complementing to convert to floating point. This is to change the signed value to unsigned, for which FP uses a sign bit rather than two's complement.

LabVIEW must have native support for fixed-point data types, otherwise negative valued readings would be incorrect. I also assume you drop the exponent byte, as it is not a part of the fixed-point word and would break the conversion if not handled separately.
joeqsmith:
I wrote both from scratch, handling the sign, exponent and fractional parts separately.   Obviously I have not tested the 3-byte out side of the example in the manual.   For the 3-byte, I perform the 2-comp before doing anything else as the manual states.  After 2s-comp, I split it up pretty much like shown.   For the 5-byte, it is already presented as 2's comp so I just split it up.  I'm sure we are talking in circles.  Obviously I have been posting data that has been properly converted so no worries on my end.

Some conversions, like these, are not built into labview or at least not that I am aware of.  Coding them from the standard libraries has not been a problem.         
garrettm:
Here's the output of my meter on the 10V range after power up using each binary type. The decimal representation given below the hex values uses my code for the conversion.

|  Vs |  3-byte  |     5-byte     |       6-byte      |
-------------------------------------------------------
|+10V | 10 00 00 | 00 FF FF 99 01 | 00 FF FF C2 00 01 |
|-10V | F0 00 00 | FF 00 00 89 01 | FF 00 00 84 00 01 |
-------------------------------------------------------
|+10V |+10.00000 |    +9.99994    |     +9.999963      |
|-10V |-10.00000 |    -9.99992    |     -9.999921      |

I would be curious to see the output of your code with what's shown here.

The above data was collected using my 8506A and Advantest R6144 set to 10V. With average on, I saw +9.999962V and -9.999927V on the DMM’s display. With the voltage source set to +0V I got +0.00011V and for -0, -0.00009V, so it has some offset and the DMM wasn't zero corrected or warmed up. But the readings look correct to me.

All binary types returned by the DMM use two's complement format. However, only negative valued FXPs need to be two's complemented during conversion to change their bit pattern to an unsigned value of the same magnitude. Positive valued two's complement types have the same bit pattern as unsigned types and consequently can be used as is.
dietert1:
Meanwhile i performed another test on our two Fluke 8502As. I used a 650 uF film capacitor and a 100 MOhm resistor to sweep both instruments and found both of them bad.
In the "yellow" instrument the ADC is misaligned and produces steps of about 50 ppm at 10 V input and about 25 ppm every 2.5 V. The other "blue" instrument does not work with negative input voltages (random results). These instruments should really have some builtin ACAL procedure with automatic fine-tuning of the ADC ladders. Will have to setup a calibrator for them. And fix that negative voltage fault.

Appended some code i have been using to log those 3-byte messages from the instruments. Nothing nice, yet seems to work.
Meanwhile i received the Fluke Application Bulletin 25 from ArtekMedia and it includes example code showing the "!" mode and detailed timing specs (Trigger delay, ADC conversion time, data transmission time etc.). I think with a fast GPIB interface the instruments should perform up to the specs.

Regards, Dieter
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod