I wasn't happy with the performance of the side mounted fan duct. Especially on small parts the air did not come close enough to the nozzle and because of the sideways acting force provided by the airflow the parts could deform during printing. Additionally the cooling effect was different on all sides of the model.

There are some alternative designs like https://www.thingiverse.com/thing:2808152 which looks very promising but i decided to design my own.

Design goals:

• airflow as close as possible at the nozzle
• equal cooling on all sides
• simple mounting, if possible single part design
• include 40mm hotend fan adapter for silent operation

After a lot of tinkering and testing i came up with a very unique (?) solution.
It's a single part print without support which attaches magnetically to the screws/nuts on the extra mounting holes on the smart effecor. So these can still be used for mounting i.e. a zesty nimble (like on my printer) and the assembly can be taken off within seconds.
For part cooling i'm using a blower fan because high pressure is needed which axial fans do lack, i used a 50mm one but 40mm might work, too (couldn't find one locally which was specified at 24V)

(sorry for the bad image quality, it was too dark and the images hat to be reduced in quality quite a lot before uploading)

Thickness below the heatblock is 2.2mm, so there is still more than 1mm of clearance to the printed part, weight is 19g.
The prototype was printed in PLA which obviously isn't a great choice when mounted that close to the heat block It's ok when the part cooling fan is on but will deform quickly if its off.
I will probably use nylon when i'm happy with the design.

I'm just 1 hour into the first print so i can't say much about the viability of the design yet.
The design is not yet published, but will be soon and will be open source.
The smart effector model which i used to design the cooling solution is already available on thingiverse: https://www.thingiverse.com/thing:3240616

I would be happy if you could provide some feedback

This looks similar to my problem a few weeks ago which turned out to be a bad insulated temperature sensor:
https://forum.duet3d.com/topic/5785/how-to-debug-heater-faults/4

Most of the time it didn't trigger a heater fault and some of the graphs looked exactly like yours.

I suspect if the air can be directed very close to the nozzle tip it isn't necessary to provide a 360 degree cooling.

My current plans for the next iteration:

• 270 degrees cooling, so the nozzle tip can be seen again during printing
• remove the 'fingers' and prohibit the air from blowing directly at the nozzle so the air stays cooler
• use a 40mm blower instead of 50mm to reduce footprint and weight
• maybe use a 40mm blower for the e3d heatsink as well? The fan duct uses a lot of space currently

Any idea where i can get a high quality 40mm 24V blower fan?
The 50mm one i'm currently using has pretty bad vibrations

Maybe the old style silicon sock would be better suited as it shielded the nozzle better.
I suspect the radiated heat from the nozzle is a problem both for the print and the fan duct.

Where the screw that tightens the heater cartridge radiates heat, the fan duct fails first. It burns a hole in the PLA the size of the screw head.

The design is now up at thingiverse:
https://www.thingiverse.com/thing:3366268

Sorry for the delay, i had to setup a new pc and was on vacation for a few weeks.
I incorporated some small improvements but noting major changed.

It's still a work in progress and im not happy about some design choices:

• the 'fingers' that point to the nozzle can deform easily and get destroyed
• the air blows at the nozzle and is reflected down, this cools the nozzle and heats up the air
• i can no longer see the nozzle while printing, maybe a 270 degrees cooling would be good enough?
• the 50mm blower is probably oversized, i may try a 40mm blower

I did print one using Taulman Alloy 910 (Nylon), but havn't tested it yet.

@samlogan87
Which inner/outer diameter silicone tube did you use?

@Danal
No i'm still iterating on the design, but i uploaded the stl here so you can have a look:
0_1543426343933_Fan Duct.stl

I designed it for printing with my 0.25mm nozzle, so wall thickness is usually 0.6mm. Should be printable with a 0.4mm nozzle as single walls however. Layer height: 0.15mm
I added a sacrificial bridge where the nozzle will go.
Magnets used are: dia 5mm, length 8mm but you can use smaller ones and glue them in.
Blower: dia 50mm, width 15mm, hotend fan: 40x10mm
Weight is down to 12.9 grams in the current iteration.

@RCarlyle
Only one vertical 'post' is directly touching the the sock and that's only because the heat block is rotated the way it is. The two other posts have about 8-10mm clearance to the sock, just rotating the heatblock 45 degrees would give more than enough clearance on all vertical posts.
However i think contact to the silicon sock is not a problem as the plastic is a very poor at transfering heat, so only a part of the post is getting warm (and the silicon itself limits the amount of heat transfered, too). My hotend cooling fan printed from PLA dropped directly touching the heatblock and only melted at the contact surface.

@samlogan87
I already considered the berd air and i may use it if my design fails.
However the berd air has a lot of downsides:

• noisy
• additional electronics needed
• additional mounting parts needed
• not easy to assemble
• still requires separate hotend cooling fan
• no international shipping from the United States to Europe and not available locally

In fact the berd air was one of the inspirations for my design as i wanted to design something that provides similar performance but without some of the downsides. Of course my design has other downsides like the amount of space needed and additional weight at the effector...

I wasn't happy with the performance of the side mounted fan duct. Especially on small parts the air did not come close enough to the nozzle and because of the sideways acting force provided by the airflow the parts could deform during printing. Additionally the cooling effect was different on all sides of the model.

There are some alternative designs like https://www.thingiverse.com/thing:2808152 which looks very promising but i decided to design my own.

Design goals:

• airflow as close as possible at the nozzle
• equal cooling on all sides
• simple mounting, if possible single part design
• include 40mm hotend fan adapter for silent operation

After a lot of tinkering and testing i came up with a very unique (?) solution.
It's a single part print without support which attaches magnetically to the screws/nuts on the extra mounting holes on the smart effecor. So these can still be used for mounting i.e. a zesty nimble (like on my printer) and the assembly can be taken off within seconds.
For part cooling i'm using a blower fan because high pressure is needed which axial fans do lack, i used a 50mm one but 40mm might work, too (couldn't find one locally which was specified at 24V)

(sorry for the bad image quality, it was too dark and the images hat to be reduced in quality quite a lot before uploading)

Thickness below the heatblock is 2.2mm, so there is still more than 1mm of clearance to the printed part, weight is 19g.
The prototype was printed in PLA which obviously isn't a great choice when mounted that close to the heat block It's ok when the part cooling fan is on but will deform quickly if its off.
I will probably use nylon when i'm happy with the design.

I'm just 1 hour into the first print so i can't say much about the viability of the design yet.
The design is not yet published, but will be soon and will be open source.
The smart effector model which i used to design the cooling solution is already available on thingiverse: https://www.thingiverse.com/thing:3240616

I would be happy if you could provide some feedback

This looks similar to my problem a few weeks ago which turned out to be a bad insulated temperature sensor:
https://forum.duet3d.com/topic/5785/how-to-debug-heater-faults/4

Most of the time it didn't trigger a heater fault and some of the graphs looked exactly like yours.

I disasembled the printer and checked the wiring, connections, board, heated bed and so on. Sadly no issues whatsoever.

But this time i noticed my multimeter picking up something right at the limit of the 20 MOhm range when measuring heatblock against wiring of the thermistor. I switched to another multimeter with a wider input range and sure enough there is a current path from heatblock to thermistor wiring just around 20 MOhm.

Interestingly the resistance drops over time to around 15 MOhm and there is no current path the other way around (input range is 60 MOhm). I'm not familiar with the construction of these thermistors so i'm not sure if this diode effect is normal.
After removing the thermistor i measured directly against its housing and its just 4 MOhm.

I checked my old thermistor and there is no current path, so i replaced the thermistor with the old one.
As the error is very rare i cannot be sure this fixed it.

It would be pretty unlikely to have additionally a sporadic leaking heater cartridge, so i wonder if static electricity could be to blame? It could be provided by the gearing of the zesty nimble, the plastic flowing through the nozzle or the nozzle rubbing on the previously printed layer.

Just a quick update: (i hadn't time to disasemble the printer yet)
I was printing a vase and didn't touch the printer.
Just after an hour of printing without problems, the hotend temperature started to fluctuate. It was commanded to 255 and jumped between 240 and about 265. On the print it was very obvious the temperature was much too high, as the PCTPE filament began to cook.
I manually reduced the temperature a few degrees and the problem was gone and temperature stopped fluctuating.

However the PanelDue touchscreen was no longer working while otherwise the display was working (temperatures did update)
Webinterface worked, but only a few seconds at a time so i don't have a graph of the temperature fluctuation.
Reset did not resolve this, a powercycle was necessary to get the PanelDue touchscreen responding again.
Could this problem be correlated to the temperature issues?

I couldn't find the cause yet, probably will have to disable it and inspect the wiring and connectors in detail. I measured the heater cartridge and thermistor against the heatblock and resistance was >20MOhm, so no short.

I printed another 20 hours without any problems and then got again small jitters at the heated bed:

(blue line, near end of graph)
This time it was during heatup, so no movement on the printer.
I monitored it at the PanelDue and it jumper 2-3 degrees up and down, then stabilized by itself and print started shortly after. No problems during printing.
After this printed again 6 hours without issues.

I'm really puzzled about the cause of this, especially as it sometimes affects the hotend and sometimes the heated bed and never both at at the same time.

Better version of my drawing:

Bad wiring or parasitic voltages from leaking cartridges can explain the jitter in my other images, but this looks like the heater continued to heat up while temperature reading stayed bang on at 195.
Looks like the temperature reading freezes while the heater is running away?