I’ve been playing around with a script in Tecan evoware for our old Evo, with the hope of adding some quality measures to our aliquot scripts.
There’s this feature called detect liquid (or detect volume), which puts a line into the script that detects the liquid level for the tips of your choice and puts the value in microlitres into a variable. I assume it uses a combination of the teaching (z-max?) and the labware dimensions to estimate the volume from the liquid level.
So- have anyone used this feature to get somewhat reliable volume readings in 96 well plates? I just need +/- 50 microliter accuracy for it to have some utility to me, I think. But more is better
Thanks for the answers! I’ll type up a bit of context at the bottom, in case you enjoy that kinda stuff Great idea to make a database. Then it wont really matter much if i cant make reality fit perfectly with the detected volume. 1,3mm’s seems like quite a lot, as far as I understand the EVO moves in 1/10 mm increments? If that’s the case, the resolution could be in the single digit µL’s (theoretical potential, in practice I assume its quite a bit poorer).
We’re mostly transferring blood serum from primary tubes to deepwell boxes and varying microtubes. Since we work with animal blood from animal health monitoring programs, we have all kinds of qualities and amounts in the primary samples (I recon goats doesn’t take too kindly to needles). Prominent clots for example, will sometimes lead to a lower-than-intended volume being aspirated without imparting a clot error. For the most part, as long as we can tell that there’s a fair amount (700-900µL) of serum in every tube, we’re ok. Usually i’d be confident in just eyeballing that, but our biobank tubes have dark sidebars which makes liquids invisible-ish from above! So a fairly inaccurate read is very helpful there.
We’re also going to do a few pooled samples, with 5x100µL’s pooled to each well. That would be more problematic to eyeball, and a fairly reliable volume detection would make me much more confident in proper aspirations. So I’m glad to hear that this function has some potential!
The accuracy is whole dependant on the well/tube manufacturers variance and the viscosity do the solution you are using. In the past I have used an EVO and been able to get the volume detection to +/- 6ul for a 20-500ul range in matrix tubes for workflow % recovery calculations. This was with buffered protein solution.
This error was mainly due to the meniscus of the solution changing at different volumes so a more viscous solution would be more linear so less meniscus changes and more accurate readings. So it is possible to get good accuracy.
The best way to test this is to set a quick program to dispense a volume range then volume detection to see linear range. This combines the accuracy of aspiration and dispense and volume detection but is a useful method to QC/QA your EVO before a run and give you regular run to run data variation. Plus spot if you EVO is not behaving itself before a major run.
What I’ve done in the past is if you really want to be as accurate as possible for liquid level detection is to tweak around with the labware definitions. I’ve been successful in even detecting 5uL on a V-bottom PCR plate
Thanks a lot for all the answers. Great idea to make a QC program based on volume detection - our EVO is very old, to the point where our local Tecan reseller didn’t want to do a yearly service program, but rather provide service on demand. Running such a QC program regularly will allow us to notice if accuracy starts declining, and book a service.
Since we’re only dispensing into two different tube-types for the foreseeable future, tweaking the labware shouldn’t be to big of a task. Thanks Daniel!
I realized yesterday that my tips are poorly aligned (x-axis), which seems to make them hit different parts of the meniscus in their respective microtube. Hopefully I can get that fixed, would be a bother to account for different profiles for every tip.
You would hope the difference between each tip is low after a PM service. Tecan do offer a tip align module but this is only if you need really high precision into 1536 plated or smaller wells.
The volume and detection QC runs I built were in a whole 96 plate, so you got values for each tip. We did not see much tip variation so tending to average across all tip for volume detection calculations. However this was a very well looked after system with users who were not afraid of telling people they crashed the tips. So we were always able to realign/bend the tip to a set reference position or call an engineer to check the arm and channel alignment.
Tecan IQOQ is very precise so you know the arm and channels are aligned and straight movement. Most of the alignment variations are in the metal tips or tip pickup not being clean alignment to the box.
Great idea to make a database. Then it wont really matter much if i cant make reality fit perfectly with the detected volume. 1,3mm’s seems like quite a lot, as far as I understand the EVO moves in 1/10 mm increments? If that’s the case, the resolution could be in the single digit µL’s (theoretical potential, in practice I assume its quite a bit poorer).