35mm film driven repstrap

ok i think i maybe found it, in the extruder.h tab of the extruder firmware (i was looking in the 5d extruder.h file)

i tried changing line 26, so:

#define TEMP_PID_INTEGRAL_DRIVE_MAX 110
#define TEMP_PID_PGAIN 5.0
#define TEMP_PID_IGAIN 0.1
#define TEMP_PID_DGAIN 100.0

became:

#define TEMP_PID_INTEGRAL_DRIVE_MAX 110
#define TEMP_PID_PGAIN 0
#define TEMP_PID_IGAIN 0.1
#define TEMP_PID_DGAIN 100.0

With this, the temperature took 20 minutes to get to 40 degrees. it would go up a few deg, then go back down.

so i tried #define TEMP_PID_PGAIN 2, the temp rose much faster but (going for 240) just floated between 220-230.

is this even the right bit of code to be changing? i'm not sure.

... i think your IGAIN should be higher and the PGAIN not zero.

I'm actually adjusting a PI-temperature controller and when using only "P", the controller stops before reaching the target temperature, so you need "I" to get it to the target temp.

My actual hack uses only a high "P=5000" with "I=0" until reaching (target-temp - 0.5 centigrades), then i set "I=2" and the temp starts oszillating some minutes until it's +/- 0.001 centigrades around the target temp.

With "I=0.1" i had no oszillation and the controller needed some hours to reach the target temp with 0.001 centigrades accuracy.

But this is different for every algorhytm, so you need other values of "P" and "I" ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

ok thanks i'll have a play around.

My aluminium block is from a heat sink. I'm pretty sure it was an alloy, beacuse the aluminium was harder to cut through than normal. I hope this doesn't work against me.. i guess the heat sinks are designed to dissipate heat but i'm hoping through its design rather than the alloy...

can't get it to hit target temp, best i can get is fluctuating between 220 and 230 (target 240). have to call it a night.

Ok the best i can get is hitting 240, dropping to 230, REPEAT. So i'm assuming it's not the PID, but my aluminum block either:

-has too much surface area
-is an alloy designed to dissipate heat... (it's taken from a heat sink).

but these are still just assumptions.

brass and aluminum sheets not as easy to come by as i thought, so surprised i didn't have any hoarded.

there are a couple of algorithms for tuning PID on wikipedia, basically they involve finding the P value where the system oscillates then taking fractions of that value for the various elements..

I just played with the various parameters until I gained an intuitive feel for what each of them did.. the P factor is the overall coarse adjustment, the I factor is for fine tracking after that, and the D factor can either accelerate changes or make them slower.

I limit is important- too much, and your loop oscillates as the I term alternately overshoots and undershoots, too little and your I term maxes out without attaining target temperature. When this is set correctly, I factor basically only changes the speed at which the I term settles down.

I found that slowing changes was very important since my max6675 only gives 5 readings per second and the thermocouple's own time constant caused additional lag, so I use a fairly large negative D term to cripple output during fast heating stages and avoid massive overshoot. D also lowers I term catchup effects after large set-point changes.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

... our PI-algorhythm is like this:

// block heater
fTemp=GetTemp(); // fetch block temperature
fDeltaTemp=fSetTemp-fTemp; // SetTemp = 60.0
// PI controller, takes about ???? ms
if(temp_control)
{
fProportional=fDeltaTemp*fKProportional; // the proportional value
fIntegral+=fDeltaTemp*fKIntegral; // the integral value
if(fIntegral>4095.0) fIntegral=4095.0;
if(fIntegral<-10.0) fIntegral=-10.0;
fDAC=fProportional+fIntegral;
if(fDAC>4095.0) fDAC=4095.0;
if(fDAC<0.0) fDAC=0.0;
UpdateDAC(); // set heater
}

The heater is activated by a DAC-value >0, where 0=heater_OFF and 4095=heater_MAX ... in thermal 'equilibrium' the DAC-value is something around 1200 ...

I have optimized the temp-control with a P-value of 6000 and an I-value of 20 - when heating from roomtemp, the DAC is 4095 at full power, when the temp is near SetTemp, the sum is lower 4095 and the controlling starts.

I can heat the device from 20deg (roomtemp) to 60deg (SetTemp) in 10 minutes with maybe 20Watts thermal energy, then the controller 'overshoots' with maybe 0.4deg and starts a damped oszillation with 2 to 3 every smaller overshoots with a poeriod of maybe 2 minutes ... then the oszillation is stable by +/- 0.001 centigrades ...

For faster finding the correct values you can use this procedere:
- set I=0
- set P really big (i tested wit 20000), until the temp starts to oszillate around the SetTemp.
- reduce P until the oszillation stops ... then reduce the value by 10%
- then start to increase the I-value, until you get a fast enough setting on the target temperature ...

Edited 1 time(s). Last edit at 10/20/2010 09:01AM by VDX.

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

thanks heaps for your help guys.

but after another couple of lengthy tuning sessions, i think i have realised my problem.. because there is no PEEK or similar material block readily available, and because i'm using an older version of the brass nozzle (shorter), I just went with some aluminum for support:



i thought i should get away with it because the aluminum is butting up against the PTFE and not at all touching the brass. But i felt the aluminum when the nozzle was around 240 and it was also quite hot, so i think that's where my nozzle is drastically losing heat.

so i've ordered the proper stuff and will wait...

i'm wondering how easily resistor+block is going to fit into future updates of the firmware. I think it should definitely be the default heater method, but we also have to work out a way for the firmware to work without everyone having to tune their own PID. obviously my setup was going to require tuning, but i'm wondering if people go with Adrian's specs on the heater block, use brass, and buy the exact same resistor, will adrians firmware just work instantly as it did with cutting 6ohms of nicrome?

also i'm confused as to why we are tuning a PID formula to emulate bang bang, shouldn't it be a different formula all together?

yep the alu was definitely the problem:



if i can think of an easy way to cut up the heatsink it could make about 10 heater blocks

eh couldn't get PID happening for now so back to nichrome for now...

Is it possible to revert to the default java preferences somehow??

So filmstrap has been treating me pretty well, printing out parts for my mendel:



here is a timelapse of it printing a vertex:

i should note, i previously posted photos of cast mendel parts i recieved from fotonlabs, i said they were fine but upon trying to drill them out they just started to snap and turned out they were way too brittle, made from polyester... the guy said he is now casting in resin but i just went for a refund. i'm sure they would be fine in polyurethane resin but just make sure that's what you're buying



annywho my only problem (why i'm wanting to revert to default preferences in reprap) is that it's started printing layers in two different spots:



i don't think it's a mechanical fault because when i go back and print old gcode from before i tweaked settings it prints the same as it did. I just can't work out what setting i changed to make it do this and i think i've changed most back to default using my laptop as a guide. seems to be only the outline that is occasionally off center, or prints from 0,0 where as the main part is a few mm into the build zone...

Hi,

I came upon your blog while searching for 35mm film sprockets for a project. They're very difficult to find and cost $$ because they're replacement parts for projectors and cameras (not all cameras have sprockets, they have a slot and guide). I am interested in either making some and/or buying some of yours. In one of your posts you provided a link to the file for making them and stated that, if anyone is interested in making them, you would improve the 'pins'. If you're still offering, I would like a copy of the file for the improved version. As well, what would you charge for the sprockets? I live in Toronto, Ontario, Canada. You can correspond with me via email. Thanks.

ok my first design didn't print with a hole for some reason (from shapeways)
second design i talked about i beefed up the teeth TOO much, got it from shapeways the hole is there but the teeth are too big and the film gets caught,

so at the moment i'm using film gears and a pulley system that have been appropriated from a projection room in an analogue manner (touching them up with clay, adding other parts then making a mold and casting in resin). it's not the easiest mold to make though if you have no experience

so now my printer works, although i am out of filament. i have to remodel the gear and i have drawn on paper a pulley system that i also need to model. i will try them out as soon as filament arrives (hopefully today)

i think the new ones will be the way to go, and i can sell you them with a pulley system. if not i can also sell you copies of the ones i'm using now.

i know i keep saying this but i will update the wiki with heaps of photos real soon.

I just saw the questions on PID as I was trying to figure out how you got away with using PTFE and no support. Normally the tips push out as the threads soften or it bulges and you get a leak between the PTFE and brass.

Anyhow... you will waste some time performing needless calculations, but otherwise it ends up being exactly the same. After calculating the integral and derivative terms, the firmware uses this to set the PID output:
output = (int)(error*pGain + integral*iGain + derivative*dGain);

By setting P to something high (255 should be high enough since it gets constrained and I think decimal values are dropped or rounded off from the temp reading), I to 0, and D to 0, you end up with the same control logic as in bang bang. You waste some time (probably not a significant amount) calculating the integral and derivative because the I and D value will ultimately cause these terms to zero out anyhow.

My firmware may be a newer version or maybe the slave firmware doesn't have this, but the version I run on my RAMPS kit actually has provisions for both modes. You control which mode is used by setting E_TEMP_PID_BAND. Setting it to 0 causes it to run in bang-bang only. Setting it to 500 or so causes it to run in PID only. Lower values like 20 or 80 will make it use a mixture of both modes. It actually changes the control logic so that it doesn't run needless calculations rather than using bang-bang emulation. Older versions may have required emulation just because it was simpler than trying to re-write the control logic.

mannn i wrote a long response to this but it's gone.

short answer: nozzle not falling out cause i'm using camiels V2 nozzle which it arrives pre glued from mendel-parts.com which i highly recommend but for my mendel i am going to use adrians latest version where the ptfe screws into the brass which i have already bought from reifsnyder, untested but it looks promising. (also a peek block for support)

also i have been using resistor in a brass block also from reifsnyder on my v2 nozzle and it has been working really well for weeks even though the best i could get the PID to control temp is floating between 240-250C, so thanks for the tips i will certainly give it another try.

can anyone tell me where the PID for the heated bed is (using 5d firmware)

at the moment my bed is just going into a 12v power supply... i need control.

Look in configuration.h.

You should find something like this:

#define B_TEMP_PID_PGAIN 2.0
#define B_TEMP_PID_IGAIN 0.07
#define B_TEMP_PID_DGAIN 1.0
#define B_TEMP_PID_BAND 0

These are the PID settings I have. There are a number of motherboard and slave configurations, I don't think all features are supported on all platforms. You have to check the "#if MOTHERBOARD" lines to make sure you are working with settings that are relevant to your hardware.

awesome thanks.

ok camiels V2 hot end just spat the brass end out, lasted for about a 1 1/4lb of ABS
\

I assume it was held in with PTFE? To get any reasonable life out of it, you need to constrain it with band clamps or something similar (IE a tight fitting tube). This will keep it from bulging and forming a leak. Then you need to trap it between your extruder block and some kind of support structure. I thread a nut on the barrel after the PTFE is installed. Then I put on a washer with two extra holes drilled in it. The washer has to be at least 1" in diameter. Any smaller and you won't be able to put the bolts in. Two M3 bolts go through the holes and clamp everything together.

Here are a couple of views of it. Side view
Bottom View

luckily i have adrians new design with peek block sitting at home waiting, not much time to do anything at the moment as i'm working 7 days a week at the moment because two projectionists are on holidays ---killing my life!! can't wait to try new hot end design!!

with my extruder out i had a chance to wack on my USB microscope and do a stopmotion pan

this is over blue scotch tape:

i'm sure people know this but i just realised after talking with my dad that you can open up GCODE in excel,

when you open you must select 'all documents', then open and choose 'delimited' and hit next, then choose 'space' in the delimiters options and press finish.

this way all the actions for each axis come up separately in columns:

and you can do the same in open office by opening a spreadsheet, then in the insert menu click on 'sheet from file', open your file then select 'seperated by' and click 'space', then hit ok and ok again.

here is an idea for the extruder smaller gear drive that i could make it in reality but not in digital reality

for my wades extruder kits i sell the smaller gear now including dual tension screws, by embedding two nuts on either side.



my last resin gear didn't fail me but it did creep to the point where i had to retighten so i thought why not double up.

This is an awesome idea!

Some properties of film strips which I believe weren't mentioned in this thread,

Regarding coding of the strip, it's actually already available. I don't know about cine film, but the vast majority of 35mm stills stock has DX code light-printed onto it. On the film itself, half of it is static data (film brand and ASA rating) and half of it is location data (frame number). This data is meant to be read optically, and to be read faster than reprap moves. It even has human-readable frame numbers in-between.
36-frame films are about 1.6 meters long, if clipping out DX-less areas it should still be about 1.4 meters long.

The problem is that it's not designed for high-resolution positioning. The code changes every half-frame, which is about 19mm distance. But it could be possible to add to the DX positioning perforation positioning. For every half-frame you have four perforations. Each perforation hole is about 2mm, with gaps being about 2mm also. This allows higher resolution positioning using almost the exact same method.
--

Regarding film strength, it's indeed quite well as mentioned before but it certainly has its problems. In practice a film strip is resilient to tearing down its width, however it is susceptible to tearing down its length. Since it's fairly narrow, tearing down the length shouldn't happen too much but it's still easier than favorable. The perforations themselves tend to hold strong most of the time, but in the case of resistance during pulling of the film they do get damaged easily. If for example the toothed spool will attempt to pull the film while the film is being stopped down the road, the perforation will break.

Different brands produce film of different material properties. Generally speaking highly-praised films tend to be of thinner film-stock which is great when you want optical resolution, but not so great when you want mechanical prowess. Film can also break to form creases, because it is flexible but not elastic, and with creases it is less flexible and certainly less predictable.
On the other hand it can have its material properties modified using conventional chemistry. Improper processing can lead to overly hard or overly soft films, such errors can be applied intentionally to film used as drive for altered properties.
--

Another problem with film is its tendency to get static build-ups because it's from polyester. It gets very quickly, and dirty drives are a problem. Should be better when coated with anti-static.
This however leads to some pro to cine films - unprocessed cine films are coated on their back with some carbon layer called rem-jet, probably used for preventing static build-ups. But then you miss on all the optical positioning fun.

i have got loads of junk film from ads and trailers which is generally thick (made mostly by the same multinational corporation that owns the cinemas)

i considered layering two lengths of film together to double the width for my x axis (because it is holding the y axis) but it doesn't seem to need it. it has worked perfectly so far except when you treat it badly. Like a few times i have knocked my opto flag out of position and didn't notice, then go to print and the bed kept on homing up till the bed hits the nema 23 motor gear and then it will tear or stretch. on my y axis with the same problem the nema 17 doesn't really stretch it though. (i'm sure it could with more power) but if you scratch its back it scratches yours and keeps its shape.

another problem if you have those accidents and you have a lot of tension is that the film can wear away at the plastic gears teeth much faster.

the same multinational corporation that owns my soul is this year beginning its 2d/3d digital revolution of every cinema, replacing all 6 of our 35mm projectors with a big digital system that will control everything and i mean everything so eventually they can close down their other subsidiary who churn out most of the expensive film for all the cinemas in australia. so i'm hoarding the best i can.

I don't have my mendel kit yet, so I don't know about lengths and stuff. I just assume that the 1.6 meters long stills films would be enough, and there's no need for cine films specifically.
Also I know Australia is a stupidly huge place, but apparently there's some old Aussie who designed and built a machine that produces 35mm film. If you find the guy, he may be of some help. And if you find him, please send him my regards and a sincere offer to buy a copy of the machine.