Tecan Infinite M200 Pro plate reader don't fully work

[isaacguerreir]

I have a Tecan Infinite M200 Pro and I was interested in use PyLabRobot to control it. On the beginning I was not even able to connect to the machine after pr.setup(). I used Zarig to transform the vendor locked driver into a generic one. After this change I was able to connect through setup.

I tried running basic commands like in/out plate and read absorbance, however only close door/in plate command actually worked, but the plate reader closed much slower if compared when I open using the vendor software.

I would you like to check if you guys had a similar experience, or if I something blatant wrong. Do you personally used this plate reader driver @rickwierenga? If so, only on a Linux distribution?

[rickwierenga]

we recently got a spark, the driver in plr was contributed in #798

I will try on mine and let you know how it goes

[rickwierenga]

what errors are you getting for reading absorbance / other things?

[isaacguerreir]

I tested this a bit more on my side, even making an Ubuntu VM so I could narrow it down better now.

I’m using:

• Tecan Infinite M200 Pro
• Ubuntu 26.04
• current main
• USB passthrough from a Proxmox VM

After fixing permissions for the raw USB device, I can consistently do this:

import asyncio

from pylabrobot.plate_reading import PlateReader
from pylabrobot.plate_reading.tecan import ExperimentalTecanInfinite200ProBackend
from pylabrobot.resources import Cor_96_wellplate_360ul_Fb


async def main():
    backend = ExperimentalTecanInfinite200ProBackend()
    pr = PlateReader(name="PR", size_x=0, size_y=0, size_z=0, backend=backend)
    plate = Cor_96_wellplate_360ul_Fb(name="plate")
    pr.assign_child_resource(plate)

    await pr.setup()
    print("setup ok")

    await pr.close()
    print("close ok")

    result = await pr.read_absorbance(
        wavelength=450,
        wells=plate.get_items(["A1", "A2", "B1", "B2"]),
        use_new_return_type=True,
    )
    print(result)

    await pr.stop()


asyncio.run(main())

What I see is:

• setup() works
• close() works
• read_absorbance() fails before returning data

So it looks like the problem is not the initial connection anymore. The backend is clearly able to talk to the reader at least enough to set up and close the drawer.

The failure seems to happen later, when the absorbance flow starts the actual plate movement / scan. From the traceback it looks like it times out around the first ABSOLUTE MTP,... / SCANX ... part of the sequence.

One more important detail: after this failure, the reader’s raw USB device can disappear completely from lsusb until I recover it physically. So on my side the sequence is basically:

• setup works
• drawer close works
• absorbance starts and then times out
• USB device may disappear after that

If it helps, I’m happy to test smaller sequences too, for example:

• just setup() + close()
• absorbance config without scan
• manual transport commands before SCANX

[rickwierenga]

@hazlamshamin does this ring any bells for you?

[hazlamshamin]

@hazlamshamin does this ring any bells for you?

@isaacguerreir i have tested the machine on both windows and mac before. on windows, would need setup like zadig as you did. meanwhile on mac, it is easier because no need to keep changing the driver. sometimes i also stumbled upon the timeout while reading the 3 modes, and since then, ive gone through the DLL files to RE it better. hence, there are several commits merged from the time you posted about this issue. it should now be able to handle better and work close to how the actual OEM backend do it. can you try with latest PLR version?

(additionally, what I notice would work better is, when plugging in USB to my computer, usually i power restart the microplate reader to avoid any unleft state, from previous computer it was connected from, especially if the previous one was connected with the official software)

[hazlamshamin]

@isaacguerreir additionally, i would recommend using the hello world notebook in the user guide folder to try the setup, since this has been tested to work :)