Chromatography Forum

Hello!

The short question:
Does anyone have, or can point me to a site with HP's (Agilent's) "RS232 Programming Manual", P/N 3396-90335, for the 3396 integrator? We have acquired one of these integrators, and can capture the .RAW or .BNC files sent out the serial port to a PC. However, the manuals we have on hand only recommend reading the above referenced manual for details using the serial port.

The longer question:
Does anyone have any references on the file formats of the .RAW, .BNC or formatting information on the serial output? We can capture the serial output, and can see it is ASCII encoded hex. Using a hex editor on what few .RAW and .BNC files we have (which were from a copy of the PEAKS-96 software downloaded from the computer museum), we can see there is a distinct header containing ASCII text followed by binary information. However, we can not determine if the header is fixed length, or variable using 'flags', and the format of the binary data block(s). Other references indicate it uses a 32 bit floating point format (IEEE-754), however, attempts to apply that format to the data block has been unsuccessful.
Thank You!
Jay

Hello,
I can not answer either of your questions, but I dug up some emails I had with some people, who used it with their 3396 integrator, hopefully they may give you some ideas.

My previous work stopped using the 3396a, and converted the gc to use the GPIB cable, with chemstation 1701 software ($$$$).

Alex

1) ( from Gorazd Bercic)
Thanks very much for your efforts with explanation of communication problem. I hope that we will be able to solve this problem. What we are doing now ? We capture BNC (or BCN) file which integrator sends to PC. The captured output is saved to disk and after that we import this file into the Peak96 software – where we can see results. From the Peak96 we can also export data file to ASCII format; such files could be used for presentation of chromatograms when needed. I now that this is not optimal path however we can at least save our results in useful form for future analysis. I will inform you if we succeed to optimize this path.

2)(from Justin M Ferrentino)
What we did discover was that the cable we were using (which was the recommended HP cable) had a different pin configuration than the one at the working GC/Peak96 setup. Currently we are building a custom RS232 cable that will duplicate this pin-out and will hopefully let Peak96 work with the integrator. Do you have any interest in using the 3396A again? I can send you the pin configuration when I go to my lab tomorrow,

Peak 96 Data Transfer Cable Configuration
.....DB 15 Connector..........DB 9 Connector
............1................................... 1
............2....................................3
NC........3....................................
SH........4....................................4.....SH
SH........5....................................7.....SH
NC........6....................................8.....SH
NC........7....................................
NC........8....................................
............9...................................5
NC........10..................................
NC........11..................................6.....NC
NC........12 ................................9.....NC
NC........13..................................
............14.................................2
NC........15.................................

NC = No Connection
SH = Pins shorted together

{p.s I gave this to another person, who mention the problem of only recording the first 40 bytes was still present, so when I contacted Justin, he wrote}

I'm not sure what else that could be. I guess I should note that I was running Peak 96 on an old (Windows 98) PC and I ran Peak in DOS mode only. With a newer computer there may be issue with the wait time for the integrator handshake. Not really sure what to tell you- all I can say is that I was able to get Peak to work with this cable configuration and the factory directy HP cable didn't work.

3) (from Luis Costa)
Thank you very much for your concern and for the information. After our first contact, I gave up on the HP3396 and started to consider other options. Fortunately, I was able to bypass the integrator and move directly to acquiring the GC signal with an external data acquisition circuit board, a self-made cable, a laptop and software. It has been working quite well so I'm not interested in changing it.

Hello Mr/Ms. Chemstation,

Thank you for your reply, it's good to know that someone recognizes this old unit. Unfortunately, your older messages were not any help, as we have the RS-232 lines set up properly and can capture the text files sent by the 3396. We have another program which converts the text file (which as noted is simply 2-byte hex encoded data) back into a .BNC file which (sometimes) we can import into Peak96, which then (sometimes) will export out to a plain text .ASC (ASCII) file. I was hoping to find information on header blocks information and number encoding in the .BNC and .RAW files, so I could skip the import/export into Peak96 and convert it to a Galactic format file directly.

Once again, thanks for your help. If I find anything elsewhere I'll post it here as well.

PolyChem

Hi,

I was looking in internet for a manual of the Agilent integrator 3396a and end up in this forum and in your post. Would you send me the manual you have (in case you have it in pdf?) We got our integrator a couple of weeks ago and I have to test it. Unfortunatelly I have never used one of those before and I do not know how to start

Thanks in advance
Marina

my email: msafont@sionpower.com

Hi
goto http://www.hpmuseum.net/display_item.php?hw=668

click on the product documentation link, enter the verify code, and download the PDFs,

otherwise the Agilent website has the pdfs for series III,

suggest typing in Google --> filetype:PDF 3396

Alex

Good morning
I contact you all regarding this HP integrator 3396 and its (impossible) connection to a pc.
We tried the original cable. We built several connection cables, following manual instructions, following your tips and our experience.
Unfortunately these cables do not work!!!!!!!!!!!!!!!!
We tried several pc configurations, with different MS-DOS version.
The best one seemed to be MS-DOS 3.3 (the one suggested by hppeak96 manual).
In detail, the integrator and the pc commucate, very well for almost all the operations.
When we try to send a file to the integrator it works properly, BUT when we try to send a file from the integrator to the pc, the procedure freezes!
An empty file, or a file of 40 bytes, arrives....

Done that the integrator has the writing head corrupted, we are not able to see our chromatograms.

Can you help me??

In case, can the signal output form the GC be acquired by anything else?????? (we tried an ADC but despite a classic chromatogram, with triangles, we obtained several squares!!!!!!!!

Hopefully
Susanna

PolyChem wrote:
Hello!
The longer question:
Does anyone have any references on the file formats of the .RAW, .BNC or formatting information on the serial output? We can capture the serial output, and can see it is ASCII encoded hex. Using a hex editor on what few .RAW and .BNC files we have (which were from a copy of the PEAKS-96 software downloaded from the computer museum), we can see there is a distinct header containing ASCII text followed by binary information. However, we can not determine if the header is fixed length, or variable using 'flags', and the format of the binary data block(s). Other references indicate it uses a 32 bit floating point format (IEEE-754), however, attempts to apply that format to the data block has been unsuccessful.
Thank You!
Jay

Just for fun - can you send an example of captured HEX output to me - unichrom at unichrom dot com.
Btw if you plan to develop acquisition driver - have a look at http://www.unichrom.com/chrom/ucddke.shtml

Hello everyone.

I am currently building a similar setup (HP 5890A GC HP 3396 Series II integrator hooked up to PC). I believe, I have some answers to the addressed issues already, just have to test the system a few times.

Well, I believe I got it.

My system consists of the following parts.

1. Instrument: HP 5890A GC unit equipped with a standard 19242-60010 INET interface
(also two detectors, TCD and FID, if that is important)
2. Data collection: HP 3396B Series II integrator, connected to the unit by a standard HP-IL cable pair. The printing part on it was removed (as it was defective and prevented normal load and POST of the integrator). The setting of 10 DIP switches is in factory mode (switch 1 down, i.e., OFF; the rest up, i.e., ON; that meaning 9600 baud speed, no hardware handshake and no handshake delay, and 15 sec timeout)
3. Connection to PC was made by a standard HP cable, Part No. 03396-60530 (pinout can be found on p. 8-14 of Agilent manual 03395-90530). The serial port used was COM1.
4. PC workstation consists of a MSI MS-6555 motherboard (a special model with one ISA slot) with a Celeron D 1200 MHz processor and 1 GByte of RAM onboard. Operating system is Windows XP Service Pack 3 2002 version. Disk file system is FAT32.

The software (HP Peak 96, in my case version A.02.01) must be installed to C:\PEAK default folder and it can be ran in two modes.

I. A pure DOS mode. Port Address in BIOS has been set as 3F8/IRQ4 and IRQ 4 has been assigned to Legacy ISA, Reload G
Global Time Events function has been enabled for everything except IDE secondary and IRQ.
What is the most important thing is that the HP 5890A device MUST be in a READY condition. Otherwise the software behavior would be quite erratic.
With such a setup runs/sequences are initiated smoothly and files are written directly to PC (actually removed immediately after recording according to the integrator logs), with no errors observed.

II. From Windows XP (the conditions of BIOS functions and device status are the same). In this case a DOS emulator must be utilized. I have employed a well-know freeware DOSBox v. 0.74. The only thing to be made is to change the serial port status in the [serial] section in the dosbox-0.74.conf file. Instead of
serial1=dummy
the following:
serial1=directserial realport:COM1
should be typed in in order to enable serial port communication through COM1.
The parameters of the COM1 port in Windows were default, i.e. the following:
9600 baud, 8 data bits, Parity - none, stop bit - 1, Flow control - none, FIFO buffers set as default.

After that DOSBox can be launched, the disk C can be mounted as virtual C:\ (command: mount c c:\) and Peak96 can be ran. It works perfectly.

When launched directly in Windows XP, even in compatibility modes, it communicates with integrator, but on starting the run sequence and/or requesting integrator status the program crashes with a following error:

Fatal error evm0009 EMM Error: 8B; function 5001

None of compatibility modes help.

Of course, on a much older PC (Entrada Desktop computer PD6-TLX from CompuLink Research; PI-166 ММХ, 32 МВ RAM, DOS 6.22) the software also works perfectly, the READY condition of integrator even not being prerequisite.