DIY gas chromatograph: Design troubles of an improved detector

[ChristofferB]

Hia all!

A month or so ago i got my gas chromatograph working, and made a big thread about that,
and now i'm constructing an improved detector: A 4-element thermal conductivity detector! (see pics)

Idea is simple: 2 fitting tees in series times 2: 1 set for pure carrier gas, 1 set for column output detection. the 4 elements (either hot filaments or NCT's) form a bridge circuit. All the Tee fittings are bolted onto a central block of aluminum to ensure they're all at the same temp.

I've hit a few snags, though.

1. I have no idea why a 4-element wheatstone bridge is better than a 2-element (+ 2 resistors) and i've seen both.

2. thermal conductivity is related to pressure(or flow?) of the gas stream. if i have one side pressurized by my carrier gas and the other open to air that side won't
    hold much pressure at all. how this is resolved in commercial systems, i don't know.

3. What is the best approach for a heating element in this setup? a pad heating element sandwhiched by the two plates? a couple of power resistors bolted on
    symmetrically on both sides?


#2 is my biggest concern at the moment. I could just put a needle valve on the output, but then i would be simultaneously regulating column pressure/flow upstream and downstream.


The detector will consist of: - A small balancing potentiometer to zero the bridge.
                                        - An instrumentation amplifier with variable gain
                                        - maybe a +/- analog meter movement to zero in the instrument after, maybe a pos/neg switch
                                        -output to ADC for datalogging. (so far just an arduino punching in values in excel).

What do you think? Thanks for the interest!


--Chris


Edit: I guess #2 really comes down to a fluid problem like this:

        ____________          ______________
                              \____/
const. p                   ____                                 atm. press
        ____________/       \______________
                 A                                B

will the flow through A and B be the same? I feel like it should, since whatever change in flow the narrower tube produces, the opposite should happen when the tume expands again, but my engineering physics aren't good enough to actually math it out.

[LaserSteve]

Most  of your problems will be resolved by modifying your flow path to thst used by modern machines and driving your four element detector with a constant current source. When i get home in an hour I'll draw this out for you.

Steve

[ChristofferB]

Thanks for your reply!

I'm not completely sure what you mean with 'modern' flow path, as the TCD element cavity layout is fairly similar to they're done today - as well as how they've always done them, since it's probably the oldest and least favored detector today.

I did plan on supplying my bridge circuit with a constant current source, possibly with multiple ranges, since I plan on both a set of bead thermistor elements and hot tungsten filaments.

--Chris

[babysitter]

Hi Chris,

about everything you need to read out balance of wheatstone bridges is found in this document:
https://www.analog.com/media/en/technical-documentation/application-notes/an43f.pdf
It just doesn't care for heating the elements the bridge consists of.

i think the most sensitive detector for cooling is the lowest-mass sensor/heater (resistor) you can get, attached by wiring that provides the lowest coupling to the wall (thin enameled cooper wire with lööps) and a stem that isolates well (glass? aerogel?)

You can attack the pressure dependent temperature by modeling (what is the difference between what my carrier gas should do and what actually happens?) or attaching just the right equipment: pressure sensors in front and back.

(Just considered to recommend a MS5607 to measure the temperature and pressure of gas at off-heater locations.)

[ace1903]

In one project I used flow sensor based on thermal conductivity. guess here goal is to measure conductivity under fixed flow.
It was  IST INNOVATIVE SENSOR TECHNOLOGY model FS1L.0.1L.195.
My interest in analytical chemistry is only driven by curiosity and i have no formal training in it.
Would recommend short look at it's datasheet and related application notes.

I would like to see more info about existing components of the system. I saw some pictures of in related thread but have no idea what is inside  those
closed boxes. Would like to see how valves look like and how flow regulation is achieved.
Maybe then I can give some useful suggestion.

 

[ChristofferB]

Ohh that detector looks quite neat! it's even small enough in width to fit in the side arm of an 1/8'' OD swagelok tee. I'll look into them, but they seem to be for quite high flow velocities. Having multiple sets of sensor elements that are interchangeable isn't a bad idea, so I might just try them and see.


Here are some detailed pictures of the flow path. Helium cylinder feeds directly into a Brooks 8601 regulator, and i have a pressure gauge on the output side, leading directly to the inlet: a fitting Tee with the column at one port, and a rubber seal/septum under the third to inject samples through.

The column then outputs into the detector and the output port of the detector is just open to air.


The new detector will be installed in the same place as the old, and the other "half" of it will be put inbetween the pressure reg. and the inlet for reference.

[ChristofferB]

I realised i've never provided a schematic overview of the 'plumbing' of the instrument, so here it is! I don't recommend using expressch for it though 

Sample is injected through a septum, pushed by helium through the column to the detector.

This is what i'm GOING to implement. right now i have a simpler setup as per old threads.

Enjoy! And as always thanks for the interest!

[ChristofferB]

I'm sorry that I keep posting in my own thread but I was too pleased with the mechanical looks of the new detector/inlet arrangement not to share.

It'll probably be a while before I can pressure test it all, but all leaks i've encountered with these fittings are stopped by a quarter turn more on the compression nuts.

Man, I love these tiny process tube and fitting systems! it's like process engineering lego!


A point I may have overlooked regarding NTC thermistor elements for the bridge - having a rapid thermistor response time is probably key, so should I go with glass encapsulated thermistors or SMD ones (which might be uncoated) with legs soldered on?