Fan backwash causes warping?

First: Yes, the fan is installed the right way. I disassembled the fan, checked for the correct direction of air blowing into the fan housing/ heatsink duct.
My first prints were quite small, like the snowman.g or had no "big" mass like the orthogonal axis compensation test pieces ormaxis.g. I had no warping problems at all.

But when I first printed a longer, more "massive" part - an Arduino Mega2560 plate (from Thingiverse) with 12cm length on the y-axis and 7.5cm width on the x-axis,



I recognized warping on both edges on the part of the print which is to the left of the printer. The edges on the right were not affected at all.



This was with the supplied filament, which came with the printer. I checked all the variables like hot end temperature (205°C for PLA as recommended) and heat bed temperature (60°C) - the ambient temperature is around 20°C. The temperature of the heat bed was double checked with an infrared thermometer.
I also usually clean the kapton taped glass plate with isopropanol for propper adhesion of the PLA before every print.

While building the printer I took a thermal image of the heat distribution on the heatbed (with an Arduino Thermal Camera I built a while ago at about 80°C as far as I remember) because I was interested, if there were any irregularities. But it looked good, however it was without an active print which had to be heated. So under stress the heat distribution might look different.



It happened that I got new filament and tried the print again - but I still got that warping on each of the edges to the left. This time it even seemed to be worse. The quality of the rest of the print was almost identical to the one with the provided filament.
Looking at the print I saw a pattern on the backside which - after some thinking - led me to the assumption that the backwash of the fan might be involved (the hatched areas are more or less warped).



So I took a piece of baking paper and some kapton tape and isolated the fan backwash (and of course to guarantee inlet air) to the top of the fan. No warping at all!
(WARNING! Be careful to copy my cheap workaround because you get in trouble when the carriage assembly homes to X!)



Comparing both prints (left with workaround fan duct - no warping, right without - warping) which were done with the same settings, the question evolves whether I did some mistakes while building the hot end assembly, the heat duct print may be faulty or is there a real design problem with the fan position?



If it is not my mistake or a defect part and really a design fault, one may put the the Duet board in upright format. So a more durable workaround, like a printed fan inlet duct, could be done without crashing the X-axis.

And that's how it looks now:



Markus

[Edited for typos, Thingiverse URL, picture assembled]

Edited 5 time(s). Last edit at 12/22/2013 05:39AM by markbee.

---

XBee & electronics blog: [lookmanowire.blogspot.com]

Great work, Markus! I'll feed this back to our design engineer, to see what he comes up with.
Love the Arduino IR camera, will have to build one. Do you have any more information on that? It would be really useful to see how consistent heated bed temperatures are.

Ian
RepRapPro tech support

Thanks Markus for sharing, and good solution too. I wonder if its happened on every Ormerod due to design fault? any others Ormerod users facing the same problem? Nice useful IR thermal cam you have there !

Edited 1 time(s). Last edit at 12/21/2013 10:45PM by tru168.

Markus and any others viewing

I know I am going off topic here. Cruising around I could not decide on a better are to post in. I am very new to all this so please excuse any errors. I do not like the look of all the filaments draped between the various protrusions on your print. Is this inevitable? I have seen similar with other printers. Is there a setting which would control this?

As a general point it seems to me that this type of print might be better if the individual protrusions were printed seperately. A pre-requisite is that the spacing would need to be enough that the head etc would not foul protrusions already printed, the order of printing may be critical. Is there any way the existing software could be cotrolled to work this way? I think people who print small parts do several at once, presumably they create a large model by placing the various existing parts into a new model. I am thinking is it possible to print a raft, or base in this case, and then print various models onto it in various positions?

Rory

I haven't done any prints yet, but there is a huge amount of backwash from the fan on my Omerod, so much so that I thought I must have connected it the wrong way round. But I checked and it is rotating the right way to blow air through the heatsink. So I can quite believe that backwash from the fan is cooling the print too much.

@Rory: The filament threads between protrusions is usually a result of printing a little too hot, and fine tuning the retraction speed and distance. It can usually be tuned out. Filament can have quite widely vary properties; particularly from different manufacturers, and different colours. It's something you have to experiment with.

@dc42: We increased the size of the fan (from 40mm x 40mm x 10mm used on the Mendel/Huxley to 40mm x 40mm x 20mm) to increase the static pressure. The main function of the fan is to cool the top of the hot end, but we wanted to use air flow to help cool at the nozzle, because this improves print quality, without adding another fan. Because the fan output is ducted around the hot end, and out the nozzles, you need a bigger fan able to push air through it. A centrifugal blower fan would have been a better choice, but harder to package on the hot end.

The usual fix for backwash is to increase the length of the inlet tract, as Markbee has shown. Also, smoothing the laminar flow into and out of the fan helps greatly. That way, drawn-in air will counteract the effect of the backwash, which can't escape. Perhaps the nicest shape would be a duct that was swept upwards 45 degrees. I wrote an OpenSCAD script ages ago to provide nice ducting for these sort of fans; I'll dig it out and post it on thingiverse, and here.

Ian
RepRapPro tech support

Edited 2 time(s). Last edit at 12/22/2013 06:45AM by reprappro.

Great observation Markus- I printed two parts together yesterday, a small gear and a large one, the smaller gear was to the left of the larger one and lifted on the tenth layer, presumably because it was cooled down by the backwash while the larger gear was printing.

Perhaps a louvered fan cover would calm the flow while keeping the fan housing low profile, I'll try this and if it works will provide the stl file

Ray

Quote
reprappro
@dc42: We increased the size of the fan (from 40mm x 40mm x 10mm used on the Mendel/Huxley to 40mm x 40mm x 20mm) to increase the static pressure. The

..how about this idea - afaik the static pressure will increase if you join two fans running counter clockwise, my logic tells me (but sometime fail me) that it also would have to reduce the backwash, if this works it will be a very easy fix without any change to the form factor, just two 10mm fans on top of each other instead of one 20mm

btw, hallo all, I'm a total newbie to anything 3dprint, just finished putting ormerod nr. 168 together, ready for first power up, no smoke signals I hope..

168

Interesting idea but I forsee some prctical issue with achieving the counter roation. Apparently these type of fans only rotate one way, if you reverse the polarity no movement occurs. I do not know if two different 10mm fans could be obtained with opposite rotation. How about putting an 8500 rpm 10mm fan fed by a 5800rpm 10mm fan. The slower speed fan would presumably generate much less backwash. The static pressure of the 8500 is 2.4 compared to 2.7 cuft/min for the 20mm fan but I suspect th 0.9 of the slow speed fan would make up the difference. I agree counter rotation may be ideal but there seems to be a defacto standard of clockwise from the label and blowing towards the label. If anyone knows different please shout.

Rory

Edited 3 time(s). Last edit at 12/26/2013 02:24PM by Rory166.

Rory

Yes one would have to obtain two different 10mm fans with opposite rotation (if that exist) as the fan blades are made to curve for direction of rotation
I will see if I can find a pair in the new year - have to order a new one anyway, the one delivered is terrible out of balance and the sound is driving me (more) crazy :-)

Erik

Edited 1 time(s). Last edit at 12/26/2013 02:34PM by ormerod168.

I'm wondering (and I think I mentioned it in another post) whether a 10mm thick fan separated from the heatsink by 5mm or 10mm spacers would provide better cooling and less backwash. With the fan right against the heatsink, there will surely be no airflow near the centre or the corners of the heatsink.

DC

I dont think simple spacers will do as the fan is intended to be sealed more or less against the 5mm long duct leading to the heatsink. Ian has mentioned that they chose a 20mm fan because they require a high static head to force air past the heatsink. The air then travels through a side duct into the nozzles. The nozzles are 1.5x8mm making a total area for the three of 36sqmm, in addition there are 2x10 slots on the top and bottom of the heatsink total 80sqmm. Compare this with the fan area of 1100sqmm and we can see the fan is well constricted. Also less than 1/3 of the airflow is directed to the nozzles. I wonder if a baffle in front of the lower to heatsink outlets would be an improvement in that they may deflect the jets otherwise. The constriction will certainly cause backwash and I am not sure that extending the duct from 5 to 15mm will help apreciably. Where is Adrian Newey when you need him?

If the fan was spaced away from the heatsink it would need a duct to prevent all pressure being lost to the nozzles. The fan is not right against the heatsink but the 5mm duct is certainly short.

One feature of the design that puzzles me is the two apparently unused fan mounting holes which have correponding 3mm holes in the moulded parts, these will form two small jets pointing backwards at the top of the fan. I assume at some point a fixing was intended here but subsequently deleted? Am I missing something here not having completed the build.

Rory

Ah, I had missed the fact that there was already a 5mm long duct.

Quote
markbee
[...]
So I took a piece of baking paper and some kapton tape and isolated the fan backwash (and of course to guarantee inlet air) to the top of the fan. No warping at all!
(WARNING! Be careful to copy my cheap workaround because you get in trouble when the carriage assembly homes to X!)
[...]

Great detective work. As an alternative to making the circuit board horizontal, could the circuit board spacers be increased in length by maybe 30 mm. The spool spigot will need to be sorted out
but the spool might be mounted above the board.

Anyway I had better finish mine first!

regards
Andy