Drip Governor - A most valuble first hack

[cephdon]

What about capacitance sensing? It seems like the proper application of conductors along the side of the printed tank should have a capacitance that changes with the increase in water. If they could be arranged in such a way as to produce some predictable outcome (change in frequency from LC circuit?) then we could determine height of the water. Maybe the frequency produced could be returned through the mic input instead of drip counting as a way to represent position.

BTW, I don't expect anything to come from my ramblings. When I have my peachy printer, I will be making the stepper mods myself and testing other ideas to see what is possible. :-) However, in the sprit of participation, I figured I would throw out a few ideas to see what others thought of them. I love the approach for this printer and I just can't wait to get my hands on one.

[Slatye]

A capacitive sensor would be sort of cool. It'd definitely have to be an end-user addon (since the Peachy doesn't come with a container and the capacitive sensor is definitely going to depend on the container) but at the same time a simple aluminium foil sensor can offer pretty decent accuracy.

The easiest way to get it into a computer is probably the same as before: replace the capacitor in the standard astable 555 timer circuit with the capacitive sensor, and use the microphone input to measure the frequency of the resulting wave.

Slayte..... now I see where you're coming from !!!! But for my part, I think Rylans valve set to large drips (unalterable) and the 555 light switch will give us all we want (fine Z, coarse Z and combinations of both) all under software control. The components and PCB and valve together, probably under $10. No changes whatsoever to the main Peachy. I would be very happy to receive a Peachy with that config.

I'm torn between the two options. Having a pure open-loop system would be neat, and being able to adjust the speed is nice. However, I can't get away from the feeling that feedback of the actual water level would really help for Z-axis repeatability in large jobs. Especially for something like the canoe, I'm having trouble accepting that an open-loop system will be able to keep the height accurate enough.

[rylangrayston]

Well I haven't got an inexpensive valve working quite yet, I think more than one of the designs we have explored will work well when we have the time to do a higher quality job.

For now iv made an expensive one, using a servo. here are a few pics and some arduino code.

20140318_205947.jpg 20140318_210001.jpg

/*


with a servo shaft connected to a ball valve 
use 2 pots one to set the open position and the other to set the closed position. 


*/






#include <Servo.h>


Servo valve_servo;  // create servo object to control a servo 


int potPinOpen = 0;  // analog pin used to connect the potentiometer that sets the open postion of the valve
int potPinClosed = 1; // analog pin used to connect the potentiometer that sets the closed postion of the valve
int val;    // variable to read the value from the analog pin 

int openTime = 300;  
int closedTime = 300;




void setup() 
{ 
  valve_servo.attach(7);  // attaches the servo on pin 7 to the servo object 
} 






void loop() 
{ 
  
servo_valve_on();


servo_valve_off(); 


} 






void servo_valve_on()
{
  for (int loopCount = 0; loopCount <= openTime; loopCount += 1){   
  val = analogRead(potPinOpen);            // reads the value of the potentiometer (value between 0 and 1023) 
  val = map(val, 0, 1023, 0, 179);     // scale it to use it with the servo (value between 0 and 180) 
  valve_servo.write(val);                  // sets the servo position according to the scaled value 
  delay(10);
}
}


void servo_valve_off()
{
  for (int loopCount = 0; loopCount <= closedTime; loopCount += 1){   
  val = analogRead(potPinClosed);            // reads the value of the potentiometer (value between 0 and 1023) 
  val = map(val, 0, 1023, 0, 179);     // scale it to use it with the servo (value between 0 and 180) 
  valve_servo.write(val);                  // sets the servo position according to the scaled value 
  delay(10);
}
}

PS

after some work on the capacitive feed back on the peachy printer pro, i think that capacitance feed back is a real possibility.
I think if one was to make a plastic laminated strip of aluminum foil placed in the top and bottom containers you could actually have two capacitors one rising and one falling in capacitance over time.

Ive noticed that there are some very inexpensive digital calipers on the market that use capacitive feed back. It would be fun to hack into one and replace the variable cap with our own. ..
it might be an informative little experiment. The digital calipers are available for 10 dollars retail and have .005 mm resolution!

Here look at what r2k-in-the-vortex has to say at http://electronics.stackexchange.com...c-caliper-work

Just had some fun trying to scope the signals, something really funky is going on there.
"Here is a good web page" <- that page? wrong! not what is happening there at all, there is only one input signal, not sin and cos
"The key is using unevenly patterned conductors in proximity of two capacitors." <-- wrong again
If you ever find a webpage where someone has actually built a copy of one of these then i'll believe what they are saying.
Anyway this is what i measured, cant find any of that info from google
The vertical strips that are grouped by 8, these are connected to digital outputs of the chip on blob, they are driven by PWM signals - approximating sinewave. 8phases, sinewave period 1800us(YMMV), pulse period ~5.6us. Each phase shifted by 1800us/8 = 225us
The receive plate gets the summa summarum that comes through stator by capacitive coupling. Now the receive signal is bunch of garbage mostly, but the signal peaks that correspond with output pulse rising edges do form a sinusoid. Phase of that sinusoid depends on position of the stator. Im guessing rx measurements must be timed with output pulses, and then there is some funky signal proccessing to get the phase shift, im not 100% sure on how to do the rx side of this.
As stator pattern and pattern of tx plates repeats every 5mm that means the final value is summa of coarse and fine measurements. Coarse measurement is the count of 5mm repetitions, counted and remembered just like regular encoder values, you can mess this count up is you move the scanning head on the caliper too fast, caliper loses its 0 point. Fine measurement is the phase shift measurement of the output sinusoid. These are summed and displayed on the LCD.
Here is an illustration: http://no.life.ee/rainer/pics/stator01.jpg
Why is this even important?
a) If anyone has managed to copy it to a diy project, then atleast i cant find it on google. Im sure someone has done it just doesnt seem like they published their project. Meaning that for such a common item the how-to information is simply not out there.
b) Ability to make dirt cheap diy linear encoders counts for a lot, for example you know how prone to failiure all the diy 3D printers are? Thats because they are open loop control systems, little jam or slippage and control system doesnt know where the robot is anymore. Now for an industrial robot you buy a linear encoder, one for each axis. Heidenhein and 100 other companies will gladly sell you one for ~1k€. Basement hobbyists unfortunately dont kave that kind of budgets. But they would gladly buy(or make, manufacturing is simple enough) capacitive linear encoder like ones used in digital calipers. If the how to information was out there somewhere.

stator01.jpg

[Slatye]

Well, just a minor update. I've ordered one of the ultra-cheap eBay diaphragm pumps just to see what I can do with it. Mostly just seeing if the one-way valves are in any way good enough to provide reliable drip rates - if 1000 cycles gives me 200mL once, does it give me 200mL the next time too? Or do I get 250mL the next time?

[Pete]

I've ordered one of the ultra-cheap eBay diaphragm pumps

Sorry think this might have been my suggestion earlier in the thread, and I forgot to post when I got it.....the wrong thing arrived, something second hand (wish I could work out what it's from) but didn't have an in and an out port as pictured, I think it was for sitting in a reservoir but I was kind of hoping not to cut holes in the bottom of my containers, I figure dangling pipes in the top is much easier and safer. I asked for a replacement as described and just got a refund :-( . I've now got a cheap peristaltic dosing pump on the way. Also now looking at water pumps for coffee machines...I'll let you know how I get on with this.

Thought for everyone, although Rylan's laser cutter design is good for the peachy team to manufacture (Respect Rylan, I've actually got an idea for a pump from laser cut slices) since we'll all have a stock printer can we not invent a drip governor that can be mostly printed and maybe hand wound!?

[Pete]

Topically, the peristaltic pump I ordered arrived today and I think £6.99 well spent.....http://www.ebay.co.uk/itm/1910373982...84.m1439.l2649

First impressions are very good, I set up a quick test with a PSU and it ran with no load between 3.5 and 6V and 7-14Hz ish came out (there's 3 rollers).

Set up a second dirty test (see picture) and hooked it up to a USB lead so was running from ~5V and I got 50ml in 46 seconds.....and I repeated and got the same numbers three times even with my crude setup, it even gave me the same number once in reverse.
IMG_20140402_173707.jpg
and a horrible video if anybody wants it (I need to clean my kitchen!)



A couple of small cautions if anybody else tries this
-The pipe supplied is short short, I found a slightly longer silicon pipe in my old nitro RC car, and it was easy to replace...I think it's 5mm with 1mm walls but I don't have a measuring stick here
-The pipe seems to work its way around the pump slowly (this seemed better under load but that's not scientific). I think this can be easily fixed by cutting the pipe shorter and putting those fish tank pipe connector things in and a couple of cable ties butting up to the housing

A few things I noticed
-this appears to drip rather than flow, I'm not sure if this is a drip per trapped compartment or more, I'll try to work this out, it's easy to see at low speeds that the motor slows down as another roller hits the pipe so this might just be differing flow rates over the cycle.
-when you stop the pump the liquid remains exactly where it is...this is great as starting and stopping the pump does not appear to need priming to get flow again
-you can hold the output shut with air going in and actually see the pipe inflate slightly (and go phut when you let go)

My next steps are
-get some longer pipe
-add opto feedback so I can detect the rollers going past

Hopefully once I've got the feedback mechanism I'll be able to find out if the volume of water per roller is constant regardless of RPM...I think this is a bit hopeful but I'll setup to plot frequency vs flow and we shall see.

[Pete]

I've got some initial testing with my peristaltic pump so I thought I'd share...seems to good to be true for the moment

So i added some feedback, I found some IR photo interrupters from an old line following robot school project and made a tiny board which handily fits in a space inside the pump housing out of the way of the rotors
IMG_20140404_154719.jpg
I can now read back with my scope every time a rotor goes past.

I setup some kitchen scales (this is my biggest inaccuracy, it shows whole grams ad I doubt it's particularly good but hopefully repeatable) and PSU with this feedback counting and then tried shifting some water at different rotor RPMs.
At 5V I was getting about 8.9Hz (that's 180RPM, 3 rotors per revolution) and 14Hz at 6V

One shot 100ml at 6V I got counts of
833
819

and at 5V single 100ml shots
812
816

then I tried shifting 100ml in several on off on cycles
at 6V I got
835

and at 5V
840
833

so the RPM doesn't seem to make any difference (I need some more precise scales!). I'll try to see this afternoon if I can count flow both ways so that you can potentially put more in and take out again to overcome surface tension, I think there's another thread about this somewhere.

[Pete]

I just did a quick in and out test, I had to move my reservoirs, previously I had the header above but I tried to run the pump the other way and it wasn't so good at pumping uphill a couple of feet.

With them level I pumped 150ml in and 50ml out and got 1282 counts in an 430 out so 852 compared to the above is pretty good (or compared to the 150/1.5 value, 855), I suppose the slight extra can be attributed to my measurement error or the lower pressure this time on the input side.

Regardless this looks to be a pretty good mechanism for getting in and out a known quantity of liquid. I'll set up a better test fixture that counts in and out a set amount and mark on the resevoirs the high and low, then I can leave it running for a longer time and see if there's an accumulated bias on one side.

[rylangrayston]

Wow!!!! I only just saw this now!

Thanks Pete!!! This is realllllly valuable work your doing.
Id say your, once again, leading the way for the pro! First with the micro controller and now with the pump!
Its just so cool to see people like you and everyone else here on this forum leading development ahead of even those of us that work at peachy!
Seriously a big Thankyou from the hole team.

Since you have had success with this pump im ordering some too!

PS I think all this wonderful pump stuff could use a thread of its own.. it seems obvious to me that both the pump method and the drip method will be widely used. Any mods reading this? I suggest we move the last few posts to there own thread called Pete's pump's

@Slatye
Im supper curious to know how diaphragm pumps preform, did yours arrive yet?
I have a bunch of gear pumps but there are 2 problems .. they lots of tiny bubbles and when the pump shuts off
it still lets water flow.

[Anuvin]

I have tried 3 times now to write a post that is both engaging and reflective of the effort you put into your posts, but you set the bar really high. Great work Pete, I really enjoy reading your findings, and I can't wait to see what else you come up with.