I have good experience using and programming a variety of schedulers and robotic arms as an automation engineer, but I haven’t used Thermo’s automation platform and software. I’m looking for user feedback from the group here.
How does the Spinnaker arm compare to the Preciseflex arms? Are they more “solid”? Does the vision teaching system work well, and actually automate much of the teach process?
How does Momentum compare to Cellario or GBG in terms of advanced features? Is there any way to access the code running a method the way that you can with Automata? If anyone has information on the software architecture within Momentum, please do share.
How is the general user experience with Thermo workcell systems compared to other workcells? Are there any missing features? How would users rate these systems for scheduling, dealing with errors, and running large, complex workflows?
How is the service and support for Thermo automation?
Teaching is quite easy with either vision system or with the manual system, by moving the arm and teaching block to the destination nest. Fine tune of the final coordinates as well the intermediate stop points along the path, is possible with their advance teaching tool. However it is not an automated process, if what you expect is to enter a list of nests and hit a button, so they are taught automatically.
Error recovery during the run is in general good. It allows to repeat or skip a step, skip an iteration or abort the work entirely. But you cannot write your own error handling logic, as far as I know.
The service support would depend on where you are based. I say that being based in Australia, where isolation can sometimes play a part in getting quick support onsite.
My recommendation overall is that you ask Thermo to demo the software beforehand and show you examples of workflows requiring a similar level of logic as what you need, and that you visit one or two labs with a Momentum implementation, where you can see a real workflow and ask any questions that come to mind as you learn how it works.
Thank you for the insight! I saw that there is a “healing” procedure that the arm can run to make small adjustments to the nest calibration after they are taught. If you have experience with this, or the general “robustness” of the arm, please let me know your thoughts.
For the scheduler, I would be interested in its integration capabilities - sending and fetching data from the Momentum database, as well as its driver capabilities particularly in regards to liquid handler operation.
Thanks for link to the repo and the recommendations - I’m working on doing exactly that.
We really don’t use that option, as once the teaching is done, if it was good enough it is very robust. The arm is quite robust. We have 3 systems that are used most days, and I can only recall 2 issues, one minor, in the last 4 years. I cannot compare it though to the other arms you mention.
As for advance integration I cannot comment. We don’t have them integrated on our LIMS.
Regarding drivers, it comes with a very large suite that includes most devices you would want to integrate. But to be sure, I suggest you check with Thermo if the specific devices you intend to integrate, have a driver already.
We have a Fluent integrated in one of the systems. Which liquid handler are you looking at integrating?
Good to know that for the arms - it certainly seems a bit more reliable than the preciseflex.
I managed to get access to the Thermo lab automation help resource site and was able to see how Momentum is set up. Unsure on liquid handler integration - likely an i-series.
I do have one specific question in case you would know. In Momentum, if you have a run in progress that is running a workflow utilizing half a resource pool, and you queue another workflow to use the other half of the resource pool, will the 2nd workflow start immediately?
Stated another way: if you have a running Momentum workflow that is using just 1/2 thermocyclers, and you queue another workflow to use the other thermocycler, will the 2nd workflow start immediately or wait until the first is done, even though the thermocycler is free?
I think the Spinnaker is a good robot for certain applications.
The “vision teaching” feature is mostly marketing hype—it works for basic needs but not for more complex scenarios, like teaching a nest on the carrier of a Hamilton STAR. In the example below, it was a real struggle: the grippers weren’t strong enough to hold the plate, so you had to squeeze them with one hand while guiding the robot with the other. This was made even more difficult by the dense arrangement of surrounding instruments.
One thing I do like is that the Spinnaker is sold as a bundle with an adapter plate, letting you add hotels and instruments without drilling into your table. This makes it easy to set up a reader–feeder workcell if you have basic automation skills, and you can repurpose the robot for other applications as your needs change.
For more complex systems, I’d recommend the Precise Flex—it’s significantly more robust. Thermo will sell it to you, but they won’t tell you that up front. The robot they put on a track has a very limited vertical reach, meaning the overall system ends up being much larger than one built with a Precise Flex arm.
Momentum also has a solid set of APIs—I just wrapped up a project using them.
Thanks, that’s great feedback! The preciseflex is so ubiquitous despite its flaws, I always thought that it was highly compact, has a very large reach, and is not overengineered, making it easy to interact with.
Would you happen to have insight on the Momentum scheduling question I posted above? I
For the thermocyler question, i believe you will be able to get the desired behavior you are looking for, i would have the sales/apps person demonstrate this for you in simulation.
Yes this can be done, I used Momentum for controlling a pool of thermocyclers for this exact application. The only complication is that you may have to split out your downstream steps into their own process, its been a while since I used Momentum but I’m pretty sure with the version I was on that if you have all of your steps in a single process the arm will wait in front of your thermocycler until the protocol is over before loading your second plate, not realizing that it won’t be done for another hour. That being said, it was a day of messing about and then it was working the way you would expect.
It should be able to use which ever device is free from the pool on the next iteration of your workflow, so you can stagger the starts for example based on other devices for which you can do only one plate at a time.
It is also very good handling lids, you simply tell if the plate is coming into a device with or without lid and if it has a lid in the storage position. Based on that the system de-lids / re-lids it as needed.
One thing you may want to compare with other systems you may be evaluating is the way it manages the devices api calls. In Momentum the arm waits until the other device confirms the plate handling has completed on their end, which sometimes make it slow. For example the arm places a plate in an incubator and waits until the plate is placed in the incubator shelf before moving to perform the next step. Compared with the Beckman Access system, this is much slower.