Current pump-tubing attachment assembly is based on flexible tubing and insertion, with a compression housing. This is a bit fiddly.

Idea:

• LPM pumps are designed with an O-ring (of the same material as the internal surface, so no additional material compatibilities) to facilitate incorporating into a manifold. See [https://www.theleeco.com/uploads/2021/11/PDS-129-2018-06-LPMSeriesFixedVolumeSolenoidDispensePump.pdf ]
• Design a manifold (fabricate with SLA) where the pumps screw into the manifold (with O-ring providing a seal) and then exits to a 10-32 female port
• 1/16" tubing is connected with a flangeless 10-32 PEEK Fittings (e.g., [https://www.idex-hs.com/store/fluidics/fluidic-connections/flangeless-fitting-peek-10-32-flat-bottom-for-1-16-od.html] )

Requires:
[] Redesign of the Pump assembly

Eliminates:
[] Adapter clamp
[] 1/8" tubing
[] screws/bolts associated with Adapter clamp

The M5 ecosystem is really neat... and maybe there are opportunities to do a more plug-and-play approach to the Sidekick control electronics that would minimize soldering. For example:

• StepMotor Driver ($22) combines both an ESP32 CPU, power supply input, and 3x stepper motor drivers for kinematics. It also has 4 limit switch inputs...
• And about those limit switches; those could be on Grove connectors too ($3.50)...
• Use a grove connector to add a 4-relay unit ($12) and we've got pump control
• Serial input is by R485, so add a USB to 485 adapter ($5)

@rodolfokeesey — I had thought that the 3D_Assets contained the Fusion360 file, but this does not appear to be present. Can you upload this? We have an engineering student coming on, who is going to work on modifying the design.

Getting a tight, leak-free connection between the 1/8" OD to 1/16" OD tubing (pp. 37-39) is tricky with the current build.

Consider using LuerTight fittings or chromatography fittings from Sigma Aldrich which would convert each fitting into a Luer connector to facilitate the build process; would eliminate the need for the adapter clamp assembly.

These are made of PEEK (like the pump wetted surface), so we would not be introducing additional material incompatibilities. Disadvantage is that these are pricey ($18 ea! for the LuerTight, $43/10 from Sigma) which raises the price. Would need 2x of these for each pump, or a total of 8x.

It might be possible to print these by FDM in polypropylene (PP), as described in Price et al. 2021

WebUSB/WebSerial is a feature in Chrome and Edge (Safari & Firefox have stated they will not support it) for interacting with serial devices from web-pages.

https://developer.chrome.com/en/articles/usb/
https://developer.chrome.com/articles/serial/

This might allow us to create a web-interface, rather than having users use a serial terminal program, simplifying the setup.

The new Raspberry Pi Pico W was announced earlier this week, which adds WiFi support (and only costs $6 USD as opposed to $4).

There's a quick rundown of the status online, suggesting that the MicroPython support is pretty good. [https://www.hackster.io/news/low-cost-connectivity-for-the-iot-hands-on-with-the-raspberry-pi-pico-w-696b9f42011e],

How might the Sidekick take advantage of WiFi support in a scientific setting?

Some initial ideas:

• Would allow for placing a Sidekick inside a glovebox or anaerobic chamber without the need to run a USB cable inside to control it.
• Facilitate driving multiple sidekicks from a single computer

Design challenges:

• We would need to add a 3.3V regulator to power the Pico off the existing 12VDC supply, but this would be pretty trivial. We might squeeze by with a cheap linear voltage regulator like the LD1117 would be fine (but calculate current, taking into account the voltage...maybe a beefy heat sink would be needed; or use an MPM3610 Buck converter for more efficiency

What types of scientific use cases would WiFi enable that would justify building this out?

I've been observing students have trouble with the soldering for making custom wires. Admittedly this is a pain. Some ways to eliminate this:

1. Use stepper motors 1 2 with JST or DuPont connectors (rather than bare wires) already attached to simplify the process. This will necessitate adding the proper header (if not DuPont) on to the PCB
2. Use crimpable connectors to JST or find limit switches with JST connector cables attached to eliminate the need to solder these on (or at least make the attachment easier). Need to add JST headers to PCB
3. Can also use momentary pushbuttons with crimpable contacts1 2 and use the crimpable->JST connectors above , and add PCB boards
4. Take advantage of the castellated Raspberry Pi Pico pcb to avoid having to solder header pins on to pico and cut pins

(not a concern with the pumps)

[HARDWARE]: Where can i buy the LPM pump? The only way i can found is texting with company supporter. Are there anyway to buy its?

This looks like a great project! I don’t see any license listed, though. Are there certain restrictions on who is allowed to build it or needing to pay for a license? I’d like to try the build myself.

Naruki Yoshikawa from Tokyo Medical and Dental University asks:

I leaned about your Sidekick at the Acceleration Conference in Toronto last year, and I am trying to build one.

I tried to purchase the pump mentioned in the paper (LPMA1250110L), but I was told it is not available anymore.
The provider company suggested using closest alternative LPLA1250625L, but I am not sure if it fits Sidekick.
The spec is written here: https://www.theleeco.com/uploads/2021/04/PDS-116-2018-06-LPL-Series-Fixed-Volume-Solenoid-Dispense-Pump.pdf

If you can provide me your opinion about the alternative pump, that is very helpful.

Problem
The design uses a supply of 12V, which runs through a Darlington array (ULN2803A). That array is expected to have a voltage drop of ~1.1V at the LPMA1251110L rated current. This means the pump is being operated at 10.9V if there are no other losses. That’s 9.2% below nominal which is a bit outside of the manufacturer's ±5% spec.

Proposed solution
Replace the ULN2803 with a Toshiba TBD62083A (or its cousins in that series)—it is effectively a DMOS version for the (bipolar) ULN2803, and thus should have less voltage drop, but is otherwise pin compatible. (The output response is slightly slower, but still much faster than the pump cycling frequency for our application so that shouldn't matter.)