[Automated-sequencing] Re: DNA SEQ: 3730 Tray on Deck errors

Phillip San Miguel pmiguel at purdue.edu
Wed Sep 6 09:58:07 EST 2006

 > At 12:24 AM 6/09/2006, Phillip San Miguel wrote:
 >>     [Just to widen the pool of 3730 users, this is a cross post to
 >> ABRF forum and bionet.genome.autosequencing]
 >>     There has been some prior discussion of a particular error that
 >> has bedeviled our facility more and more of late:
 >>     The error is "Tray on deck does not match tray type in run
 >> setup". For us, this generally occurs between the A1 and B1 quads
 >> of a 384 sequencing reaction plate. Generally, once it occurs, we
 >> are shut down until and engineer fixes it. We have two 3730XL's and
 >> have actually seen this on both instruments during the last month.
 >> The fix is generally for the ABI Field Service Engineer to replace
 >> all the sensors in the gripper assembly.
 >>     Usually this seems to do the trick--for a time.
 >>     The FSE will almost universally warn us that sensors can be
 >> damaged by liquids being spilled on them. But then just as
 >> universally note that there was no evidence that this had occurred
 >> with our sensors. (The sensors are dry and have no buffer residue on 
 >>     The situation is somewhat more complex than I'm describing
 >> here--frequently this error seems to presage the failure of some
 >> major system of a 3730. But not always.
 >>     Has anyone else been seeing sensor issues in their 3730's?
 >>Phillip SanMiguel
 >>Purdue Genomics Core Facility
> Date: Wed, 06 Sep 2006 10:24:32 +0800
> From: Frances Brigg <F.Brigg at murdoch.edu.au>
> To: ABRF Discussion List <ABRF at list.abrf.org>
> Subject: Re: DNA SEQ: 3730 Tray on Deck errors
> We have had a 3730 since 2003 and have had to have most or all of the 
> gripper sensors replaced on 3 occasions since installation. The AB 
> engineers are always surprised and say that they don't see this kind 
> of thing at other sites. The sensors are always dry with no buffer 
> residue, and the replacement sensors are wrapped in plumbers' tape to 
> keep moisture out. After the most recent replacement the engineer 
> suggested that it could be due to power failures or brown-outs that 
> affect the sensors first, and this would probably explain why your 
> sensor failures sometimes precede major systems failures. What we 
> tend to see is intermittent failures with the sensors, so that they 
> will run fine for a while and then start to fail with increasing 
> frequency. If the sequencer is shut down for a while, it will again 
> start off OK but then fail once it has been running for a while (ie 
> after 1 or 2 runs). The engineer said that this occurs because the 
> sensors run better when cold and are more likely to fail as they heat 
> up, which could explain why they tend to go at a particular quad on a 
> 384 well plate. It is a good idea to run the sequencer beforehand if 
> an engineer is coming out to look at the sensors to make them fail on 
> cue, as it often takes an hour after switching on with active runs 
> for them to start failing.
> The error message will depend on which particular sensor goes and at 
> which point in the process it fails, so you could also see messages 
> such as "autosampler cannot hold tray" from the same problem, and 
> find the autosampler holding two plates at the same time. I have on 
> some occasions left the sequencer on and gone back to find the 
> capillaries sitting in air as the periodic automatic re-homing of the 
> autosampler has started and a sensor has failed in the process. It is 
> possible using the 3730 service tools/service/diagnostic/autosampler 
> test to do a dry run (ie it does all of the plate movements in a run 
> without the polymer fill and electrophoresis) to work out which 
> sensor is failing when not all of them are affected, but if they all 
> go there is not much you can do, and it only really helps if you can 
> persuade your FSE to leave you a spare sensor or 2 for emergencies.
> The engineer also said he had seen errors at other sites where the 
> sensors were fine but the bottom of the plate holders were filthy and 
> so encrusted with buffer that it formed a wall where there should 
> have been a cut-out and made the plate appear to be a different type 
> than it was. (It is a good idea to examine the base of the trays to 
> see if there is anything unusual about them, particularly in the area 
> that the sensor is having trouble with if it is the same plate holder 
> and sector each time).
> Finally, I have also seen software problems cause apparent sensor 
> errors, but only in a very unstable system. (Our FSE did the upgrade 
> from V2.0 to V3.0 for Windows 2000 and couldn't get it to work, we 
> opted to wait until the XP version came out rather than trying to get 
> the Windows version reinstalled properly from scratch, and hobbled 
> along with a partially restored V2.0 but with bits and pieces of V3.0 
> installed. The overnight runs were fine, but during the day we kept 
> getting sensor errors. After a while we noticed that they occurred 
> whenever the in-stack door was opened during any part of a run, 
> apparently triggered by the StackerMsgBean "%% an instrument door is 
> open" error message, and whatever stage the run was at it would fail 
> the next time the autosampler tried to pick up a plate. Normally it 
> isn't a problem to add to the in-stack with runs in progress, 
> although the %% error message still appears. We would find that the 
> error would also keep recurring if we tried to repeat the run using 
> the same plate record, but if a new plate record was made up with a 
> different name and the door was left alone it would run perfectly 
> well. This took a long time to diagnose, because there was sometimes 
> quite a long gap between opening the door and the error that it 
> triggered, eg while the oven was getting up to temperature during a 
> run, but once discovered the error could be generated on cue every time.)
> But sensor damage due to power fluctuations is the most likely cause 
> of ongoing problems.

Thanks for the detailed response, Frances. I'm also starting to hear 
that the reoccurring problems I'm having are likely the result of power 
fluctuations or spikes. (With suggestions that I should purchase a 
UPS/power filter.) Of course it is very difficult to rule this 
possibility out, or it would be. I think I *can* rule this out because 
one of my 3730XLs *is* on a UPS/power filter. Actually it is the one I 
have had the most trouble with sensor-wise.

Anyway, the Field Service Engineers no doubt have a hard job here. The 
3730 is a complex instrument. But I have to admit that after a while I 
get a little tired of hearing the standard list of culprits for 
instrument failures:

(1)	I'm sloshing buffer or other liquids on the plate type sensors.
(2)	Someone/something is accessing my console computer through the 
internet in a way that causes mysterious failures.
(3)	The power to the instrument is bad.


More information about the Autoseq mailing list