Varian 3800 Split Injection

[benhutcherson]

I have a recently acquired 3800-300MS/MS system, and fortunately now have the MS working well(aside from gaps in my knowledge of how to use it).

I'm still learning my way around the 3800 GC, which seems a very capable instrument, but in a lot of ways somewhat more "clunky"(albeit a lot more configurable/flexible) than the 5890/6890/7890s I'm use to.

The one I'm working with only has a single 1177 injector and no built in detectors(only a hole in the side for a transfer line).

So far, it's been working fine in splitless mode.

I tried today to run some samples in split mode, and that's where it all went to heck. I'm running MS Workstation 6.5.3, and selected "split" on under the injector menu then set the ratio(I tried several, from 1:1 to 20:1). It accepted and activated the method fine.

Once split has been toggled on, though, I can't get it to come up to the set pressure, and the instrument reports no flow through the column(not sure if I believe that, as there is some head pressure and the MS source pressure is consistent with changing column flow)

I'm at a bit of a loss on this, and don't know the plumbing well enough to even hunt for potential issues. There's a small plastic tube hanging out the side which I suspected might be the split vent, but I didn't see any flow out of it. Just to be sure, I disconnected it completely even though it blows clear with air pressure.

Setting it back to splitless restores normal operation.

Any thoughts on what could be going on in it?

[tdgcuser]

Ben,
do you have the manuals for the 3800 and the specific inlets on your instrument? The two that we have both have the SSL inlets which have split vent and septum purge lines on the left side of the instrument. You’ll need two low flow meters to balance the flows using the valve on the front panel on the 3800. Its tricky because it probably doesn’t identify which line is which.

The inlets and EPC on the 3800 are primitive compared to the Agilent 6890 EPC. You will be disappointed if comparing them. The 3800/300 is still a great teaching tool, because the software gives you a huge amount of information and control over the QqQ.

That being said, we use ours as follows, one with liquids and spme for determining QqQ parameters and the other as thermal desorption GC-QqQ. The TD system buys us factor of 10-1000 increase in sensitivity due to the elimination of the noise floor common to thermal desorption.

[benhutcherson]

Thanks-I do have a manual for the 3800 that covers the 1177 injector and the type 1 EFC. I've tried to make heads or tails of them, and I'm not convinced that everything is working correctly.

Fortunately, I was able to locate the purge vent line-it's a teflon tube right behind the valve mounted at the top of the oven. I did set the flow rate I wanted and adjusted the purge to 2mL/min per my electronic flow meter.

I'm still not getting any flow out of the split vent, though, and I'm wondering if the valve is defective. I've gone over that area of the instrument and made sure all the connectors were seated. I was running it at one time with the plastic tube completely disconnected. It turned out that this was PROBABLY a good move, as somehow or another it had pinched in the instrument side panel. I got that straightened out, though, and the line blows clear when I put compressed air through it, so I don't think it's the issue.

When I toggle split mode(with any split ratio-I've tried a range of settings from 1:1 to 400:1 and a bunch of more typical ones in between like 10:1 and 20:1) and I get no flow out of the split vent. I initially measured none with the flow meter, but hooked up a good old fashioned soap bubble meter and then even tried the expedient of the tube in a beaker of water.

I agree with your general assessment of how primitive on the whole this instrument is. The flow controller technology seems about on par with my EFC 5890, and definitely a step behind the 6890. On paper the 3800 does seem like a great GC, but I have a hard time imagining needing 3 inlets and 3 detectors. I'm honestly wondering if I should try the 3900 GC I have here that went with a long-dead Saturn, although I've never seen any documentation that discussed the pairing of that particular GC with a 1200 or 300(although the Saturns could be specced with either a full blown 3800 or the stripper 3900).

I do agree with you also on the software. Now that I've been using it for a month or so, I've gotten very comfortable with it. With the MS in particular, it's a lot less "turn key" than ChemStation and you have to have some idea of what's actually going on in the MS. I'm glad to actually get that hands-on experience, though. The example I've been using is that when I do SIM on an Agilent, I'll often write a custom tune file that widens the peaks up a bit and also increases the EM voltage to maximize the SIM signal. With EDR on the Varian, the latter is unnecessary. For the former, it prompts you(both for various MS-MS modes and also SIM) to select both Q1 and Q3 resolution, so the "back door" I have to use to do that on Agilents is instead right in my face on the Varian.

It's actually making its teaching debut hopefully this week. It will be in a 200 level(first semester science major) chemistry lab with me doing a demo. Given the target audience, I'm not going to talk about MS-MS. For the time being, I have a grad student dominating the 5975 so getting that is hard, and I love that on the 300, it's easy to point out all the major parts of it without breaking vacuum. I'm hoping it gets more teaching use in the coming year.

[benhutcherson]

I decided I was going to dig into this today and trace/check lines or whatever else I needed to do to figure out where the problem was.

To start things off, I turned on split at a 10:1 ratio from the front panel, and was immediately rewarded with flow out the split vent. I tried it with a range of split ratios-from 1:1 to 1:100, and it continued to function fine.

So, from there, I updated it in a method, activated the method, and once again everything worked perfectly.

I'm not sure why things have suddenly decided to resolve itself, but they have.

I was about at the point of trying to see if it would interface with a 3900 GC(it's not documented, but don't see why it wouldn't) but it looks like the 3800 got a reprieve