Cheap peristaltic / doser pumps investigation

IMG_6129.JPG


Got an Adafruit Motor Shield V2 just to simplify the initial development.

Using microstepping (PWMing the position between steps), you can get a butery smooth (well, except for the detents, which is actually a firmware driver problem since the waveform enters a delay and hold between steps) very slow 2rpm or lower movement speed. The default Adafruit libraries have a hard time with microstepping and higher than 4-5RPM, and going to single or double stepping is significantly louder. I'll have to hack on a driver for these.

The motor on these units produces plenty of toque at a mere 3V (where PWM stepping between two poles draws about 1A). There is little reason to run at a higher voltage.

I'm trying to locate some Norprene tubing which will fit into the pump head as opposed to the very thin wall silicone tubing it ships with.
 
Two more samples: the "6V" variants, and a "12V" variant which was nearly identical to the first one, but shipped with much larger ID (and OD, but similar wall thickness) silicone tubing.

The 6V motor variants are actually driven by a gear on the motor shaft - no reduction gears though, so torque is limited (its impossible to run thick wall norprene tubing - not enough torque).

p6.jpg


p7.jpg


p8.jpg


Fresh from the envelope, it was super squeaky as there was no lubrication on the inner slip joints of the rollers (yet there was ample lubrication on the outside). Luckily the roller assembly is friction fit, so just pops apart.

Since this pump is gear driven, and has no inherent shaft slip (though can still slip on the tubing) it runs at the highest flow rate of any of the samples. Motor torque requirements means its very hard to run at a low speed. It also has no position feedback.

I'm working up a controller for the stepper motor variants, and probably will use that with a new doser setup. I still want to experiment with a position sensor for the cheap DC motor variants.

The 12V variant has been in constant operation for 1.5 weeks now. Still running with no tubing breaks, though there are signs of wear on the rollers from the shaft (blackening mostly).
 
Messing around with the $35 pump this weekend, I've discovered that the l298n driver cannot regulate current (the heatsink gets HOT), and is extremely hard to use. The pump quality seems good, but I'm waiting on a better driver to arrive before I continue testing it.

I wouldn't dose anything with the cheap $10 pumps over the long term, but they are stupidly easy to work with using PWM and a MOSFET.
 
Ok, I'm going forward with building a doser around the stepper motor variants. I think they give me the best control and most dependable rate.

The 12V doser under continuous duty basically turned useless (motor spinning, no dosing) after about 4 weeks. Derp. Obviously continuous duty is extreme, especially since most people would run them a few percent of the day depending on needs.

So, for the doser control, I want to emulate what I did with the LEDBrick units (sorta) and make it programmable over Bluetooth LE, so I can use my phone to adjust settings and store programs. I'm going to use the BMD-300 module, which contains the nRF52832 MCU. Operating is totally stand-alone.

The stepper motors are actually 3.3(3)V units in NEMA17 sizes, so the normal stepper driver boards are ill-suited to correctly operate them (the common Toshiba driver is really a 8+V driver). I'm going to instead use a DRV8836 H-bridge chip from TI.

Since most wall-warts are 5V units, I need a high current 5->3.3V supply. Looking at using the ST1S32 4A Buck converter. Its a high frequency unit (1.5MHz) so the inductor size remains small .

Current rough outline of what the board is looking like (2x3in)

b1.png
 
4 weeks isn't too bad, depending on how it failed - if it continued at a high %age performance level until near the end, I'd consider that worth the money. If it was continuous degradation (more likely I'd think?) then is becomes less useful :(

Tim
 
Some of the pumps in here actually look pretty decent.

I got some about 2 years ago for a project. I go the $15 ones that are just drums spinning on a shaft, no gears like the very first ones pictured.

The actual dosing rate was about half of what it was rated for, and just using a pot I was able to control the flow from about 50% to 100%.

I got 6 to dose 3 part, do water changes and dose vodka. Over the first year they ran ok, the flow rates were not the same across the pumps but seemed to maintain the flow rate decently. the tubing they come with breaks often, and needs to be replaced about every 3-6 months. If they do break and leak, the motor will rust out FAST. After 2 years I had replaced every single one of them at least once. I ended up trashing the unit and getting BSR pumps.
 
Sadly, I didn't monitor the before and after rate of the failed unit, bad experiment practice to see how much it started slipping before outright failure :).

And thanks for the information on the tubing EnderG60. The stuff that ships with these pumps is very very thin wall silicone stuff (it makes airline tubing look robust). I grabbed a bunch of sizes of norprene tubing awhile back but most of the pumps here weren't sized (and with the exception of the stepper motor model) had nowhere near enough torque to handle anything thicker. I assume the tubing cracked or split in the pump head?

I should revisit the tubing problem for the stepper motor head, I think it should be able to handle a thicker wall silicone tube. It had problems with the smallest norprene tube I could find, but that stuff was like 95% wall 5% tube.
 
Also, I'm swapping to the DRV8834 motor driver. Its a fully integrated stepper controller with 32 microsteps capability (so you get 6400 positions per rotation, mainly for smooth driving than actual dosing accuracy), current decay and ramp, indexer, and more. This will simplify writing firmware to keep everything nice and smooth and quiet. Its a mere extra 50c per channel.

http://www.ti.com/product/DRV8834

(I'm trying to keep the low quantity build cost at <= $5/channel).

Also, any interest in an ESP8266 enabled version? WeMos or on board? Digital control lines (now much easier due to not needing real time motor control) RaspberryPi / Arduino? Something with the electric imp boards (bluefish etc)? At this stage its pretty easy to at least add some kind of pinout or pads if someone wanted to substitute another controller.
 
karimwassef said:
So how accurate do you think you can control the flow? Can you get it to one drop?
Click to expand...



The actual step size is absolutely tiny, but for repeatability I'd say the smallest volume would be the volume contained in the tubing between two of the roller wheels (gravity and surface tension variables could draw this much out of the tube eventually, and surface tension would gate minimum drop size). That's pretty small since there are about 5 rollers in this pump head.
 
You must log in or register to reply here.

Similar threads

R
Reefer 250, Trigger roller mat sump, varios 6 pump. cheap!
Replies
1
Views
108
tgurske
T
B
JBJ Arctica 1/10 hp Chiller Modification for External Control
Replies
9
Views
763
griss
griss
Reef Kinetics
How ReefBot Lab Elevates Aquarium Testing to the Next Level
Replies
0
Views
305
Reef Kinetics
Reef Kinetics
B
Open source robot for testing and supplementation
Replies
10
Views
818
Benoit T
B
P
Jebao Doser 3.1: Convert to TUYA app
Replies
0
Views
1K
papibi
P
Back
Top