Author Topic: EEVblog #1126 - Mystery Teardown  (Read 1644 times)

0 Members and 2 Guests are viewing this topic.

Online EEVblog

  • Administrator
  • *****
  • Posts: 26824
  • Country: au
    • EEVblog
EEVblog #1126 - Mystery Teardown
« on: September 28, 2018, 09:01:33 am »
What's is this obscure 1970's bit of kit from a big name manufacturer?

AliExpress Affiliate: http://s.click.aliexpress.com/e/c2LRpe8g

 

Online TheSteve

  • Supporter
  • ****
  • Posts: 2542
  • Country: ca
  • GHz or bust
Re: EEVblog #1126 - Mystery Teardown
« Reply #1 on: September 28, 2018, 09:31:03 am »
Looks like it was used with a 5890 Gas Chromatograph.
VE7FM
 

Offline JonM

  • Supporter
  • ****
  • Posts: 134
  • Country: us
Re: EEVblog #1126 - Mystery Teardown
« Reply #2 on: September 28, 2018, 12:19:10 pm »
In the early 1990's I managed an analytical chemistry group that was responsible for several rooms full of these on both gas and liquid chromatography systems. They were connected to a HP-1000 minicomputer via a multiplexed serial bus.  The data connectors were known as "Grenoble Connectors", as I remember.

I was not hands-on for this system (thankfully, it was old even then) and I have reached out to one of the people who was hands-on at the time in case there are other interesting points.

 
 

Offline drussell

  • Frequent Contributor
  • **
  • Posts: 871
  • Country: ca
  • Hardcore Geek
Re: EEVblog #1126 - Mystery Teardown
« Reply #3 on: September 28, 2018, 12:23:22 pm »
Looks like it was used with a 5890 Gas Chromatograph.

I'm sure it was capable of being used with all sorts of lab equipment.

The 5890 wasn't introduced until 1984 so the 18652A must also have been used with other kit before that...  :)

Edit: The HP 7673B troubleshooting & service manual has a pin-out for a Y cable to the remote jack:



The HP 1050 Series Variable Wavelength Detector manual from 1992 talks about remote cable part numbers to connect an 18652A as well as analog output cables and a note that the output from the 1050 series can be configured to be 100 mV or 1 V maximum signal output.

The HP 3354B lab automation system also mentions the 18652A.  The 3500 series manual talks about the card-edge cable to go to the 18652A and says it is 1V input.
« Last Edit: September 28, 2018, 12:57:39 pm by drussell »
 

Offline ChunkyPastaSauce

  • Supporter
  • ****
  • Posts: 490
  • Country: us
Re: EEVblog #1126 - Mystery Teardown
« Reply #4 on: September 28, 2018, 12:24:58 pm »
Dunno what the insulation is but HP was using glass asbestos for some things in the 70s, so caution maybe be in order.
 

Offline drussell

  • Frequent Contributor
  • **
  • Posts: 871
  • Country: ca
  • Hardcore Geek
Re: EEVblog #1126 - Mystery Teardown
« Reply #5 on: September 28, 2018, 01:06:56 pm »
There is probably a bunch of information in specialized lab journals, etc, since you can find things like this on these systems, if anyone has access to old scientific journals:

https://cdn-pubs.acs.org/doi/pdf/10.1021/ac50037a723
 

Offline MrMobodies

  • Contributor
  • Posts: 21
  • Country: gb
Re: EEVblog #1126 - Mystery Teardown
« Reply #6 on: September 28, 2018, 01:29:31 pm »
I see it was used with another model HP 1090.
Listed as parts on this site:
https://se-source.com/hp-1090/

Also in the Agilent 1049A Electrochemical Detector Operators Hanbook.
https://www.agilent.com/cs/library/usermanuals/public/1049OperHandbook.pdf

I found this:
Journal of Chromatographic Science, Vol. 25, May 1987:
"The analog output from the HPLC detector is transmitted to the laboratory computer via the Hewlett-Packard Model 18652A A / D Converter Interface."

From here:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.859.7313&rep=rep1&type=pdf"

As I am looking at all this stuff The Tornados Telstar springs to mind.
« Last Edit: September 28, 2018, 01:55:05 pm by MrMobodies »
 

Offline JonM

  • Supporter
  • ****
  • Posts: 134
  • Country: us
Re: EEVblog #1126 - Mystery Teardown
« Reply #7 on: September 28, 2018, 02:23:37 pm »
I see it was used with another model HP 1090.
...

The HP 1090 was one of the most common liquid chromatographs, but the  18652A ADC box was used with many manufacturer's systems since it just got the analog signal from the detector. By the 1990's almost all pharmaceutical and agricultural analysis chemical separation was done with liquid chromatography (rather than the previously mentioned gas chromatography). It was common to mix and match the components (auto-sampler, pump, column, detector) with pieces from different vendors. There was nothing that tied the 18652A ADC to any specific HP product on the input side.

The HP-1000 based LAS (Laboratory Automation System) had the model number 3354B that presumably incorporated the HP-1000, interfaces, acquisition and analysis software.
 

Offline drussell

  • Frequent Contributor
  • **
  • Posts: 871
  • Country: ca
  • Hardcore Geek
Re: EEVblog #1126 - Mystery Teardown
« Reply #8 on: September 28, 2018, 02:45:12 pm »
It was common to mix and match the components (auto-sampler, pump, column, detector) with pieces from different vendors. There was nothing that tied the 18652A ADC to any specific HP product on the input side.

Yes, there are numerous references to the 18652A in other, non-HP manuals and papers citing various mixes of apparatus.  It was probably originally designed fairly soon after HP acquired F&M Scientific in 1965, just about when doing things like programmable temperature control was becoming a thing. 

Electronic data capture and analysis was the next logical step instead of chart recorders, etc. and I'm sure HP's expertise in those areas provided significant, rapid advances in the equipment of the era.  Next would be the automation of sample processing, etc.
 

Offline SilverSolder

  • Regular Contributor
  • *
  • Posts: 93
Re: EEVblog #1126 - Mystery Teardown
« Reply #9 on: September 29, 2018, 09:22:37 am »

Love this old stuff, not a CPU in sight...

Could the analog section be a precise voltage-to-frequency converter -  and the rest of the device essentially a counter that counts pulses when triggered for a new A/D conversion?  It would also send the result out as a serial data stream on the output port.

14 bits / 84dB range seems achievable with this method and would be considered very high performance in the 1970's.
 

Offline drussell

  • Frequent Contributor
  • **
  • Posts: 871
  • Country: ca
  • Hardcore Geek
Re: EEVblog #1126 - Mystery Teardown
« Reply #10 on: September 29, 2018, 10:57:46 am »
i would love to see Dave power this bad boy up and see if we can determine some basic specifications.  It shouldn't be that difficult to power it up, feed it a voltage from 0-1 V on the analog input, put it in RUN, and see what it spits out the other end.  :)

It should be relatively easy to probe with modern test gear and it probably still works.  :)
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 14899
  • Country: za
Re: EEVblog #1126 - Mystery Teardown
« Reply #11 on: September 29, 2018, 09:08:08 pm »
Was used in the HP1000 series, and I did use one ( and repaired mercury relay boards on it) that was part of a room size ATE system. Was used IIRC as one of a couple of ADC units that measured power supply voltages fed to the UUT, while the precision voltmeter ( it had a Nixie or Panaplex display, buried in the rack) did the precision readings, along with the synchro to digital converters, and the AC converters, to get the high resolution (14 bit) data.

The output is designed for transformer coupling both ends, and the input is fed to a transformer (Pulse one with a center tapped primary, and a 93R impedance secondary) that feeds a 7474 flip flop, biased so that the pulses set and reset the output, giving you the recovered data signal. IIRC 470R to Vcc, center tap to that and a 330R to GND, biasing the input nicely in the middle of "don't care" space, with the pulse amplitude alternately driving the input stage into conduction on the negative edges.

Replaced a lot of 5474 IC's in that position, they did not like being driven above Vcc on the edges, and would eventually have enough threshold shift to stop recovering the data reliably. Remove the IC, place in a test jig and it works perfectly, but you find the high threshold is now 1V or so, and the gate input current is around -5mA. We had tubes of them around, but is is a standard.
 
The following users thanked this post: thm_w

Offline JonM

  • Supporter
  • ****
  • Posts: 134
  • Country: us
Re: EEVblog #1126 - Mystery Teardown
« Reply #12 on: October 02, 2018, 11:28:09 am »
I see it was used with another model HP 1090.
...

The HP 1090 was one of the most common liquid chromatographs, but the  18652A ADC box was used with many manufacturer's systems since it just got the analog signal from the detector. By the 1990's almost all pharmaceutical and agricultural analysis chemical separation was done with liquid chromatography (rather than the previously mentioned gas chromatography). It was common to mix and match the components (auto-sampler, pump, column, detector) with pieces from different vendors. There was nothing that tied the 18652A ADC to any specific HP product on the input side.

The HP-1000 based LAS (Laboratory Automation System) had the model number 3354B that presumably incorporated the HP-1000, interfaces, acquisition and analysis software.

I have been corrected, 3354B was actually a model of the HP-1000. I think that it had arrived shortly before me in 1983 and I remember the guy who managed it calling it "a Ferrari", compared to the previous system. We moved to a Model A990 HP-1000 in 1992.

From my friend PVR:

"The 3354B had a whopping 2MB of main memory and would boot from almost anything that HP had on the HPIB bus.  The computer didn't even have a main board, it was three boards that all worked together.  HP built this thing for their own test purposes (the Real Time Executive operating system was hardware-interrupt driven).

It supported four current loop cards that with 15  18652A devices each (last address was reserved for broadcast) so it would handle up to sixty chromatographic detectors.  I think we were close to that number.

User response sucked though, until we got the A990.  That was a righteous RTE system.  That handled five loop cards and I wrote the spooling routines that finally freed up a lot of the user sessions.

I swear I could have emptied the magazine of a .45 into one of these things and it would still take data."

We did modify the 18652A boxes so that new samples would not be injected if the data system was unreachable.

Back in those days data were stored in proprietary format binary files and people were just starting to worry about data portability into the future. So, we developed software to convert the HP binary files into a proposed industry standard using NetCDF. I'm giving a portion of a talk on NetCDF in December where I will show that I can read and plot one of those data sets from 1995. The rest of the talk will be by someone else on how NetCDF is used at NASA and NOAA. Apparently the chromatography community now has some XML based common file standard.
 
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf