[tesseract]

Hello All,

First post here. I'm developing a small biodiesel manufacturing plant. ASTM D-6751, sub part D-6584, requires that a GC be used to determine free and total glycerine though there are many other methods of performing this test. At any rate, rather than sink a minimum of $8k into a new GC or that much into a used but of questionable vintage GC, I want to take the real low road and buy a tired old 5880A and, somehow, incorporate an A/D converter board so it can talk to an inexpensive (err... are there any?) peak matching program. So.... I'm just wondering if anyone else had this bright idea before I did and what their results were. Any comments, except the kind like "don't be such a cheap s.o.b. and buy a new GC" will be welcome! That said, if I really am better off buying, say, an SRI Instruments 8610 then I will listen to reason. Thanks!

Jeff Jenkins
Principal Engineer
Tesseract Mobile Electronics
(branching out into the totally unrelated field of biodiesel)

[eugenedakin]

Hi Jeff,

Hmmm.... an interesting idea.  My background has experience with biodiesel and HP GC's (5880A, and 5890), but I do not know much about SRI 8610's.  I do know that the resolution on a 5880A is not very good (this is dependant on the column, software, and sampling rate to name a few parameters).

I know that you can hook up a HPIB board (is this an analogue to digial board?) to the 5890's, and you should be able to write almost any computer program from there.  I am not sure if the 5880 series has the technological capability for an HPIB board .. (is it too old?  I'm not sure)

Negative aspects of using either of these two are: It is very difficult to find existing parts for these machines.  I believe HP stopped manufacturing parts for the newer (5890) machine about 4 years ago.  The 5890 machines are rugged, durable, and a very good machine.  If there is an 'aftermarket' company which creates parts for the 5890, this GC would almost last a lifetime.

I do not claim to be an electronics expert, but I have learned enough to try and keep the existing equipment running.

I hope this helps.

Sincerely,

Eugene

[Dude]

I don't have much to offer either.  Expanding on what Eugene had stated, there is an aftermarket company called Alpha Omega Technologies (they bought up a lot of the HP equipment and resell it now).  I don't know anything about the 5880A, the 5890 was the older generation GC equipment that I was familiar with.  The technical service department at Agilent was pretty good with those type of questions (or I suppose Alpha Omega, although I never talked to them directly).

It's been about two years since I've touched a GC, however, the A-D converter generally came from the company that sold the software (neither was cheap).  I suppose the cheapest way would be to send the signal directly to an HP printer/ chart recorder.  Some chart recorders calculated areas and recorded retention times at the end of the run.

If you are going to do the GC analysis, I would buy new equipment.  A GC will be a significant investment anyway you look at it.  You will have to buy and replace tanks of He, air and hydrodgen at a minimum along with tubing and line dryers along with a lot of expendable vials and syringes, septa, etc.

Another approach would be to see if you can get a good deal with an analytical lab that runs the ASTM analysis.  If your process seems OK, it might be cheaper in the long run to add batch runs together and test a smaller # of samples at an outside lab.

http://www.alphaomegatech.com/

[tesseract]

Thanks for the comments, eugenedakin and dude, even if they were almost depressing enough to make me want to slit my wrists...

Well, not really. So far I am just as out in left field as I was to begin with, but that's okay as I knew this was going to be a long shot in the first place (how's that for a mixed metaphor).

What I was envisioning was ignoring whatever woefully superannuated HPIB junk that might still be out there and instead "revinventing the wheel". More specifically, amplifying the mV signal from the FID - perhaps logarithmically, if necessary - then sampling it with an off-the-shelf A/D converter chip then presenting the digital output (time + magnitude) to, eg., a USB port. The hardware should be very much doable depending on what the actual signal output from an FID looks. Unfortunately, I haven't been able to find that sort of info out. Furthermore, as you pointed out, Dude, the software is inextricably linked to the detector to the a/d board, and while making the hardware would be trivial for me, writing the software would definitely not. Hey, I learned my lesson back when OS/2 Warp came out. I was so fired up about it but, damn, there wasn't hardly any good software available for it so I sat down to write my own e-mail client in REXX. Well, six months later I still hadn't even figured out the API hooks so I cut my losses and hung up my programming (dunce) cap right then and there. Not about to repeat that mistake again today.

Anyway, until I, a) find some hard data on FID output signal parameters; b) find a piece of 3rd party peak interpretation software that will (theoretically) work with whatever I concoct to interface it to the FID; c) or find someone else who had this same dim-bulb idea and met with success (or failure), I suppose I will have to shelve this one... is that the consensus thus far?

At any rate, eugenedakin, here's a link to the GC I was contemplating: http://www.srigc.com/2005catalog/cat33.htm.

For a little more background on what I'm up to (and which might also help justify my selection of the "chemical engineering" forum), I want to test the ability of several ion exchange resins to perform acid-catalyzed esterification of the free fatty acids (typically oleic, linoleic and linolenic acids) in semi-refined vegetable oils. I have heard that Nafion might possibly be up to the task which would be a real hoot to me since Nafion is the darling polymer of the fuel cell crowd these days (I did *lots* of research in fuel cells a couple of years ago and concluded that they are way too ahead of their time from a lack of infrastructure, to say nothing of the cost of Pt or the polymers themselves).

Of course, the GC will also be used to certify the biodiesel meeting ASTM D-6584 (the specific subpart of D-6751 that calls for a GC), but as Dude pointed out, certification testing would be best left to a specialized analytical lab in the beginning, at least.

But anyway, back to the homebrewed GC to analyze homebrewed biodiesel... euegendakin, you mentioned that one of the handicaps of the 5880A was its low sample rate. I assume that this is the fault of the "Integrator" that interfaces with the unit and not with the FID itself. Please correct me if I am wrong, otherwise would it be safe to say that a higher sample rate of the FID output will result in substantially better accuracy?

-Jeff

[eugenedakin]

Hi Jeff,

At any rate, eugenedakin, here's a link to the GC I was contemplating: http://www.srigc.com/2005catalog/cat33.htm.

Good Searching.  That is a good 'starting' GC.  It may be challenging to measure small concentrations of esterification, but it should work well for large conversions (it will be easy to see the larger peaks).

For a little more background on what I'm up to (and which might also help justify my selection of the "chemical engineering" forum), I want to test the ability of several ion exchange resins to perform acid-catalyzed esterification of the free fatty acids (typically oleic, linoleic and linolenic acids) in semi-refined vegetable oils. I have heard that Nafion might possibly be up to the task which would be a real hoot to me since Nafion is the darling polymer of the fuel cell crowd these days (I did *lots* of research in fuel cells a couple of years ago and concluded that they are way too ahead of their time from a lack of infrastructure, to say nothing of the cost of Pt or the polymers themselves).

Wow.  I hope that it works for you.

Of course, the GC will also be used to certify the biodiesel meeting ASTM D-6584 (the specific subpart of D-6751 that calls for a GC), but as Dude pointed out, certification testing would be best left to a specialized analytical lab in the beginning, at least.

Yes.  I agree with Dude also.

But anyway, back to the homebrewed GC to analyze homebrewed biodiesel... euegendakin, you mentioned that one of the handicaps of the 5880A was its low sample rate. I assume that this is the fault of the "Integrator" that interfaces with the unit and not with the FID itself. Please correct me if I am wrong, otherwise would it be safe to say that a higher sample rate of the FID output will result in substantially better accuracy?

Yes, I believe your right about the integrator.  It sounds like you are really handy with electronics, and increasing the sampling rate (through your electrical savyness <-- is that a real word?) would increase the resolution and the detection limits of the GC run.

Chuckle ... Does this sound more 'upbeat', while being honest?

Take care.

Eugene

[tesseract]

Thanks again for the helpful comments! The SRI GC is certainly attractive, but I know right off the bat I'd have to ditch the 1m long silica gel column for a 30m 5% phenyl{insert alkyl group here}silicone one... Geez, what can only 1m of silica gel separate accurately? Oil and water, maybe? You should get the understatement of the year award for this one:

It may be challenging to measure small concentrations of esterification...

Still, at less than $6k for the total package - including electronic pressure control and software?! - it seems rather churlish to complain about spending $300 or so on a proper column... 

Okay, back to the geriatric HP 5880A (or 5890, should I be so lucky). Ditch the "Integrator" and this thing has a fighting chance of getting back into fighting trim. There are plenty of COTS* a/d boards so I don't even have to reinvent the wheel in that department, I just need to find out the exact response characteristics of an FID. Time to go to the local university's library and start sifting the stacks.

Thanks again!