Installed Carl's Tecto blocks on Replicator 2X, now need to adjust PID tuning - how?

[Gian Pablo]

I installed a "100% Carl" setup on my Replicator 2X - Alu Dual block and the Tecto hot blocks.  Everything went well (though I would welcome some tips on cable routing) but I do notice that the heating performance is quite different from before.

While everything heats up faster, there is quite a bit of oscillation around temperature - with a target temperature of 235 the actual temperature bounces between 229 and 240. That range is big enough that it produces noticeable artifacts in the print.

I imagine this has everything to do with the PID parameters - any suggestions on where to go with this? It looks this: http://makerbot.wikidot.com/pid-controller-tuning outlines a methodology for doing the tuning, has anyone done this successfully?

Current parameters are:

Extruder 0:

P = 7.05

I = 0.45 <= I changed this one up since it was a little slower to heat

D = 36.28

Extruder 1:

P = 7.08

I = 0.35

D = 36.42

[Gian Pablo]

Changing the D parameter to 72 as per manual tuning instructions now gives me an oscillation between 236 and 233 with most of the time spent at 235. That's probably good enough to start with.

Carl's blocks and thermal barriers are amazing. With the extruders at 235C for 15 minutes the cooling bar is cool to the touch. My extreme cooling solution is not really necessary any more.

[Ryan Carlyle]

You want to do a "full manual tuning" or just ballpark it? Dropping P and I and raising D will tend to decrease oscillation. Can't drop I much on Makerbots/Sailfish though or your temp will sag below the set point a few degrees. (Depending on insulation quality and heater power.)

[Gian Pablo]

I'm mostly ballparking it for now. I'll try reducing P a bit too.

I notice that my right extruder takes longer to heat, probably variation in the heater core?

[Ryan Carlyle]

Probably. The two heater blocks do get different amounts of fan wash, but iirc the left block gets more airflow, so if the right is slower then that is probably heater variation. Sailfish printers heat at 100% power until within 10C of the setpoint so there's no PID stuff to worry about until you get close. 

[Gian Pablo]

This is working much better now. Carl suggested blocking off the gap between the front of the cooling block and carriage, which I did by putting Kapton tape on the bottom of the heat sink. This allowed me to lower the P and I parameters, and raise D without compromising the preheat time.

On Tuesday, February 23, 2016 at 8:52:09 AM UTC-8, Gian Pablo wrote:

[Gian Pablo]

FWIW my PID settings now are:

P = 6

I = 0.35

D = 72

This limits the oscillation to about 1C.

[Gian Pablo]

I think a bad crimp on the heater cartridge connector was affecting performance. I'm fixing that and doing another set of tests.

In general I'm still getting worse quality on the Rep 2X than on my Rep 1, and I think the PID setting is affecting it. Surface finish on planes is a bit rough, and I think it is due to the constant temperature variation. On the Rep1 the temperature does not budge at all.

[Jason Gross]

I found this thread to be very helpful so I'm adding in my experience.

I think I have the the same setup, here's what I just installed on my otherwise stock makerbot 2x.

• Alu-Dual Cooling Block 
  
• 12V / 24V Heater Cartridge × 2
  
• Tecto Heater Block × 2
  
• AluCarriage Dual × 1
  
• Insulation Material × 1
  

I also had issues with the temperature fluctuations (it would swing +/- 10 deg C and rarely stabilize when just preheating). Modifying the PID settings to what you posted and blocking off some of the airflow with tape did help and reduced the fluctuations. For me they tend to keep within +-2 most of the time, but will swing +/- 5C from time to time during the print. My surfaces were a bit rough at first and the infill was a little sparse, so I also reduced the "Filament Diameter" setting for both extruders to 1.73. Temp is at 230 printing ABS (one spool is Makerbot branded and the other from MakerGeeks which has been decent).  After after a couple of prints and fine tuning my plate leveling the prints are consistent with what I was getting prior to mods. I tend to print with .15mm layer height and reduce the speed a little bit. 

This is the result of a lot of trial and error and so not all settings may be relevant to any success. The outlines, first layer and surface fill speeds seems to be the most relevant and slowing them down seems to have positive effect. 

First layer  (20mm/sec with raft. 10mm/sec with no raft)

Outlines 40mm/sec 

Surface fills (floor and roof) 45mm/sec

I keep the max speed at 80mm/sec and left the default speeds for everything else.

Also, I leave the top off the printer, leaving it on with a dual extrusion ended up with a failed print (and it was quite warm in the chamber). Leaving off the printer top has not resulted in any failed prints so far, but I've only done three. 

Haven't tried anything other then ABS, but I'm looking forward to trying out my aging spools of PLA sitting on the shelf. :)

[Gian Pablo]

Thank you for posting, glad to hear that someone is having similar issues!

Like you, I now need to leave the top off in order to successfully complete prints, and run my externally controlled cooling fans at a higher speed. I think heat creep has become more of an issue. I suspect that the thicker Alu-Dual cooling block leaves much less of a gap between that and the Tecto blocks, so it may be getting some radiant heat transfer. I've insulated the heater block using Carl's recommended aluminum tape, and it seems to have helped a little.

More things I'm going to try:

• Swap in an alternative heater cartridge
• Re-do the wiring of the heater cartridge connectors
• More insulation on the blocks
• Use thermal barriers from P3D - I'm using these on my Rep1 which is pretty much flawless, perhaps the internal geometry is different
• One more round of PID tuning using a guide that Carl posted (attached)

While I'm getting quality that is close to what I had before the upgrade, it is not as good as my Rep1. Layers look more uneven, especially noticeable around thin corners and thin parts.  (However, when using the same tweaked acceleration settings, the R2X shows LESS ringing than the R1, which is to be expected). I'm pretty certain that my print issues are due to the temperature oscillation, which is affecting the flow rate.

There is a small possibility that it might be due to problems with the motor cables, so I'm looking into that too.

[Carl]

Would be helpful if you could post photos of your installations - one focusing on the heater blocks and a second showing the rest of the carriage setup. Hopefully it will help with diagnosing the problems you are seeing! :-)

[Carl]

On the hot end photos... I am particularly interested in seeing how you have connected the thermal barrier to the Alu-Dual cooling block! ;-)

[Ryan Carlyle]

I think the aluminum fiberglass tape is less effective as an insulator than the usual ceramic cloth and kapton. Which is fine, because it's way more convenient to apply -- maybe two layers would be merited. 

[Jason Gross]

To accurately show that I think I have to pull some things apart. I'll attempt to describe it, and if necessary I can dismantle it to take some more pictures...

The filament tube is threaded into the Alu-Dual and has about 1.5/2 turns above the top. Afixed is the supplied nut and tightened, the second nut is tightened against the Alu-Dual The heater block is threaded on the tube and has probably 1-2 mm of clearance to the nut from the insulation. The nozzles are tightened against the tube in the heater block as per the instructions for the T shaped blocks. There's a very small gap between the nozzle (flange?) and the heater block. I left that so I knew they mated but tried to minimize this gap. 

First Q. 

From the pictures there is only one nut in the T-shaped block, but two for the Tecto... I think that it should stay stable with one nut, should I remove the lower nut, and can/should I use a little loctite on it? 

Second Q.

As Ryan suggested, how should I better insulate the heater block. I was looking at some options on Amazon and it seems there's a lot of products for racing/exhaust systems. Any thoughts on those? Or should I just order some more of the same foil fiberglass stuff? (Do you have a link / model / manufacturer, so I could source it in the US?) This shipping from Europe was pretty speedy, but Amazon is still quicker. :)

A couple more data points:

My first attempt to assemble it all I made the mistake of having the heater block rest against the lower nut, the heat transfer was such that I could never get the block up to temp but heated up the whole carrier nicely... 

And last night just for a test I did a factory reset on the makerbot settings. The standing temperature was stable within +/-2, but once I ran a print it went +/- 15. Reapplied the PID settings from a above and things were reasonable.

-Jason

[Gian Pablo]

Here are some pictures. Much like Jason, I threaded the thermal barrier such that only 1mm or so protruded above the Alu Dual block, then put a nut there. Locked it in with a nut on the bottom of the Alu Dual block. Both nuts are tightened against each other, pressing against the block. The hot end block is then threaded onto the thermal barrier, such that there is a small gap between it and the lower nut. There is a small gap between the bottom of the nozzle and the hot end block. (Of course the threaded part of the nozzle is tightened hard against the thermal barrier.)

In terms of actual performance, with default PID settings I get large temperature swings, a range of +/- 7C at least. With my modified PID settings, I never get more than +/- 1C, exactly like Jason reports.

My setup differs from Jason's in that I have higher CFM, externally controlled fans installed. Running these at a relatively high setting consistently lets me finish prints. At a lower setting my prints fail, almost certainly due to thermal creep.

Side view of the top nut as installed:

Side view of the lower nut as installed:

Top view of the upper nut as installed:

I'm beginning to think that more insulation is critical, both on the top and bottom of the heater block. Also might turn the heater block 1 turn lower on the thermal barrier. This will make it closer to the nozzle, but will leave more of an air gap on the top.

On Tuesday, March 29, 2016 at 5:27:12 PM UTC-7, Carl wrote:

[adam paul]

I thought the general consensus these days was to use a single thin brass nut on the bottom?  My CTC has the original style dual cooling bar with the extra step on it, I use no nut on the top, and the thermal barrier is pretty close to flush.  

[Carl]

Thank you for the photos! :-)

In all my tests using the Tecto across several machines - Including a Replicator 2 that I run as both a single and a dual... I typically only see a temperature variance of 1deg C higher or lower than the target temperature. 

I typically run my heater blocks without insulation... and NEVER need to perform PID tuning on any machine running a Mightyboard based controller...

I am curious if either of you have airflow directed straight towards the heater blocks from any source? This would be the most likely explanation for the variances you are seeing.

I am working towards an installation guide for the Tecto and hope to have it available fairly soon... but as Adam posted... I recommend installing only one of the thermal barrier nuts of the two that are supplied. This nut can be installed either above or below the Alu-Dual cooling block depending on your feeder and the material you are printing.

The photos below show the orientation that I recommend... the example shows the thermal barrier nut mounted below the Alu-Dual cooling block.

[Gian Pablo]

Thank you, that is super helpful!

Good to know that these should work without PID tuning or insulation. I'm struck that Jason and I have had such similar results.

I have been careful to avoid having air blowing straight onto the hot block, by filling gaps in the carriage as per your suggestion.

If you only use 1 nut and make the top of the thermal barrier flush with the cooling block, that adds 2mm air space between the hot end and the cool block, might be enough to make a difference.

Also interesting to see how you run the wires. I found that there was not enough space to run 2x heater wires and 2x thermocouple wires between the stepper motors, so I ran the heaters up front between the fans and the thermocouples up the back. I offset the thermocouples so that the tip doesn't stick out much - as per your pictures.

I will try this and report back.

[Carl]

It is fairly tight... but there is plenty of space available for the wires... 

The majority of the assembly - including installing the steppers... feeders... and just about anything else that mounts to the Alu-Dual cooling block - can be done off of the machine and then dropped into position on the AluCarriage Dual when ready! ;-)

[Ryan Carlyle]

I recommend one brass nut beneath the bar. That maximizes contact pressure on the lower end of the barrier tube where it's most critical. Steel works ok but not quite as well.

Don't put nuts on both sides, there is measured data proving that can "float" the threads inside the cooling bar and greatly reduce heat-shedding.

Don't leave any threads visible below the cooling bar nut. The most critical place in the entire hot end is the first few threads above the heat break.

[Gian Pablo]

Very helpful! Good info about the floating threads and threads below the cooling bar nut.

Still curious as to why the Rep1 works so much better.

[Jake]

I'm about to embark on a Tecto upgrade of my 2X, so I appreciate this thread.  Thanks to everyone for the information they've contributed.

Its been a while since I've rebuilt a hot end, but one thing I wonder about is.... anti-seize.  Any need for it anymore?  Even though its not exactly thermal compound, at the very least it might serve to facilitate heat transfer between threaded components.  My recollection is that anti-seize was only called for between the nozzle and the heating block.

Looking forward to receiving my Tecto parts and doing this upgrade.

[Ryan Carlyle]

Since we have all dissimilar metals in the hot block, there's no galling risk. (An aluminum nozzle could gall in the hot block though.) The main thing that will stick the parts together is molten plastic cooking in the threads. If you tighten the nozzle against the heatbreak properly, that shouldn't happen.

Most anti-sieze compounds will have trouble with the temps we're using here. And if you get any on the contact area between the nozzle and thermal barrier, that might actually contribute to leaks. I don't recommend using any kind of goop on the hot threads. Thermal compound may be a good idea on the cold end threads. 

One thing I'm wondering about the Tecto thermocouple is whether the time constant of the sensor response is much higher than the brass thermowell style. A slower sensor will absolutely cause oscillation. (On the other hand, more thermal mass with a bigger aluminum block counter-acts that effect.) Everything else being equal, for slower sensors, you have to decrease the PID gains to compensate. Normally, that would be just fine, but MBI's PID code hard-caps the accumulated error on the integral term, so if the I gain is too low, it will never reach the target temp. What I recommend to compensate is to drop the P gain (which will make it "hunt" less aggressively) and raise the D gain (which will resist rapid changes). It's hard to say exactly how much though -- maybe cut P in half and double D from the default tuning and see how that works. 

I don't know if there's legitimately any sensing speed difference. Unfortunately, my workshop is packed up, so I can't measure the thermal properties any time soon. 

[Gian Pablo]

That's interesting, I am in fact running P = 3 and D = 72, which is about half the regular P and double the D, as you recommend, and that has dramatically improved temperature stability. Dropping P to 2 didn't make much of a difference.

I also wonder about the placement of the thermocouple. The hole lets you put the thermocouple all the way through so that it pokes out the other side, but perhaps it is best to mount it so it doesn't stick out at all. 

[Jake]

On Wednesday, 30 March 2016 20:25:02 UTC-4, Ryan Carlyle wrote:
Since we have all dissimilar metals in the hot block, there's no galling risk. (An aluminum nozzle could gall in the hot block though.) The main thing that will stick the parts together is molten plastic cooking in the threads. If you tighten the nozzle against the heatbreak properly, that shouldn't happen.

Fair enough.  Thanks for the recommendation.  

For historical reference, MBI used to include some high temp anti-seize with their stepstruders, a very long time ago.  

"If you ever plan on disassembling your hot end, it's imperative to apply the included anti-seize compound to the threads. This stuff will protect against rust, corrosion, seizing, and galling that can happen at high temperatures. Simply smear a bit on your finger or a swab and apply it to the external threads. It should be applied to the nozzle threads and the thermal barrier tube threads."

I believe it was "Saf-T-Eze", a nickel based product and was rated to 2600F.  I have a tube of it left over from ages ago.  I believe this is the data sheet: http://www.saftlok.com/safteze/tds/TDS-Nickel_Premium_Grade_Anti-Seize.pdf

I'm guessing, like many things, MBI didn't quite have the engineering quite figured out when they wrote these instructions, and perhaps were over concerned about the nozzles seizing.

I'll take your advice, though and leave it out.

[Carl]

Although not on a Replicator... this video (https://youtu.be/fclR702JM8o) shows the Tecto installed and printing on a modified WanHao i3 with the CloneBoard Mini installed - running Sailfish. 

Absolutely no PID tuning was done on this installation... the heater block has no insulation material... and the system is using a 24V/40W heater cartridge - so pretty much the same setup I would install on a single extruder replicator. 

The temperatures remain within 1deg C above/below the target temperature throughout the duration of the 2 hour print shown in the video! :-)

[Jake]

I'm beginning to think that more insulation is critical, both on the top and bottom of the heater block. Also might turn the heater block 1 turn lower on the thermal barrier. This will make it closer to the nozzle, but will leave more of an air gap on the top.

I've started assembling my setup and I'm using P3-D barrier tubes, which are a little shorter than Carl's.  With the tube recessed just below the top of the cooling block (no nut on top) and a nut on the bottom, this leave the heater blocks between 4.75-5mm from the bottom of the cooling block before insulation.  This is closer than the ~8.6mm gap between my stock Rep2X heater blocks and the cooling block.

Is this a concern?  What's the prevailing thinking about the gap between the heating block and the cooling block?  Should I go back to "OEM Carl" tubes to get a larger gap?

Update:  I should clarify that Carl's block has an extra 3mm feature that's not on the Rep2X cooling block, so my 4.75-5mm is really ~8mm if you measure from the cooling block mounting surface rather than the feature right over the heater block.

P.S.  My mild OCD doesn't like that there's a 0.25mm bigger gap for one heater block, but that's just the way the threading works in order to keep my block orientation. Nozzles are bottomed against the tubes, and measure as level with a caliper.

[Carl]

OCD seems to be a common problem on the Google Groups these days! :-)

If installed correctly... you should be seeing a 7mm gap between the top of the uninsulated Tecto's and the thermal barriers' point of entrance into the Alu-Dual cooling block... and 10mm to the mounting surface - on both of the heaters. Anything less... and you will need to lower the thermal barriers in order to increase the gap.

[Jake]

On Tuesday, 5 April 2016 18:45:09 UTC-4, Carl wrote:

Anything less... and you will need to lower the thermal barriers in order to increase the gap.

 

Thanks Carl.  Your barrier tubes are about 3-4mm longer than the "high lubricity" tubes from P3-D.  With the shorter tubes, I can't push the heater blocks as far down as that.  Do you think this is unworkable, or should I abandon "high lubricity" in favor of moving the heater books down to get that 7mm gap?

[Carl]

It is certainly worth giving a try! ;-)

[Gian Pablo]

Ok - quick question for you. You say that I don't want any threads visible below the cooling bar nut. If I do that, then the gap between the top of the Tecto block and the protruding feature on the cooling bar is about 5mm. Carl recommends lowering the cooling bar so as to get a 7mm distance, but then there will be exposed threads.

What is most important? Since I'm using insulation, I'm thinking that following Ryan's recommendation is probably the best bet, and sticking with a 5mm gap.

[Carl]

My recommended installation instructions can now be found in this post on the Clone Google Group. This set of instructions is what I advise following! :-)

[Gian Pablo]

Fantastic! Thank you!

I really like the suggestion of using the gino pad as compressible material in order to facilitate leveling. I've been making a Kapton shim, but that is a tedious process...

[Gian Pablo]

Printing now, with all the various changes. Temperature is holding rock solid at 230C. 

I'm hoping this is the long-sought change that will make my Rep 2X print as well as my Rep 1!

[Jake]

I'm seeing similar behavior on my setup.   As you probably gathered from previous posts, I'm using P3-D barrier tubes that put the heater blocks closer to the cooling block. I'm currently insulating with (old school) ceramic cloth and kapton.  I went with a different orientation because I wanted to insure that there was adequate separation from the moving belt.

I ran the following test with the cooling bar and extruders sitting just as you see them in this photo, on the bench.  No airflow, heat sinks, fans, or anything else.  I see a similar variance in temperatures.  After about ~6.5minutes it settles down to a +/- 1deg C variance.  The PID settings were stock, untouched (P=7.02745104, I=0.32549021, D=36.14117813 as per ReplicatorG)

I think I'll have some PID tuning work to do in my future as well.  Maybe I'll hook the stock hot-ends back up and run the same test for comparison if I have some free time.

 

I'd value any thoughts from the experts.

[Ryan Carlyle]

Jake, what are you using to export temp data from RepG?

[Dan Newman]

On 07/04/2016 10:19 AM, Ryan Carlyle wrote:
> Jake, what are you using to export temp data from RepG?

You can enable logging in RepG via the preferences (advanced). RepG will then
write to ~/skeinforge.log. You can then use a variety of tools to extract the temp
data from the resulting log file. I don't recall offhand what level/granularity of
logging you need to enable to get the temp data logged.

Dan

[Jake]

I'm using a crude Python script that queries the hot end temperature once per second over serial. I can post it to github if there's interest.

[Carl]

That would be awesome! :-)

[Gian Pablo]

Keeping my fingers crossed, but it looks like success. I printed one of Ryan's caliblocks in ABS, and it is as good as anything I've gotten from the 2X. I'm printing one in nGen now too. This has already gotten farther than other nGen attempts on the 2x.

My hypothesis as to what was happening: by using two nuts, I was in fact floating the thermal barrier, and preventing it from shedding heat into the cooling bar. This accounts for the cooling bar feeling cool, and yet having extrusion issues due to thermal creep. I note that the thermocouple on the second extruder also stayed quite cool.

Now, the cooling bar gets noticeably warmer, and the extruder cooling fans are quite obviously blowing warm air. The TC on the second extruder is giving a high temperature. More importantly, prints are finishing!

[Ryan Carlyle]

On Thursday, April 7, 2016 at 10:54:50 AM UTC-5, Carl wrote:

That would be awesome! :-)

 

I'm using a crude Python script that queries the hot end temperature once per second over serial.  I can post it to github if there's interest.

Yep, I'd definitely have some uses for this. 

[Jake]

Here is a comparison between the stock Rep2X hot end and a Tecto hot end: (Stock PID settings)

Any thoughts?

You can find my very crude temperature logger here: https://gist.github.com/jake-b/eba2ac042ff9a0ddbf747d7b93ba7bbd

It logs tool 0, but could be made to log Tool 1 with some changes to the "Request" packets on lines 84/90 (the tool ID byte, and the checksum byte)

This is an adaptation of a crude tool I used to log the heat up of my Bottleworks HBP.  You can find that code here: https://gist.github.com/jake-b/2d6745a928da627090cef0f07681dcca

You'll likely need to edit the serial connection string if you use something other than "/dev/tty.usbmodemfd121"  and there are probably some package dependencies that I installed that you'll need to install as well.

[Ryan Carlyle]

Are the two charts synchronized on the start of heat-up without any offsets? Looks like more dead time or a longer time constant on the system. Which could be attributable to slower sensor response, larger hot block heat capacity, or some combination thereof. 

[Gian Pablo]

That is consistent with what I have seen using the default PID settings. Using my modified settings (P=3, I=0.35, D=72) and with correct installation, my Tecto setup gets to a stable temperature and does not oscillate. I have tried printing at 230C (ABS) and 240C (nGen).

[Jake]

On Thursday, 7 April 2016 14:30:19 UTC-4, Ryan Carlyle wrote:

Are the two charts synchronized on the start of heat-up without any offsets? Looks like more dead time or a longer time constant on the system. Which could be attributable to slower sensor response, larger hot block heat capacity, or some combination thereof. 

Yes, they're synchronized within a second, given I'm only polling at 1 second intervals.

[Jake]

On Thursday, 7 April 2016 15:18:12 UTC-4, Gian Pablo wrote:

That is consistent with what I have seen using the default PID settings. Using my modified settings (P=3, I=0.35, D=72) and with correct installation, my Tecto setup gets to a stable temperature and does not oscillate. I have tried printing at 230C (ABS) and 240C (nGen).

I ran the test with these PID settings just for comparison.  Your setup could behave different than my bench-mounted setup, but it still doesn't look critically damped to me. I get more oscillation with these settings than the stock MBI settings.

[Jake]

So another test.  I imagine that the thermocouple junction in in tip of the stainless thermowell.  Most of the builds (even Carls) show the tip sticking out a bit.  I tried centering the tip within the block, rather than sticking it out.  With stock PID settings, this position seemed to oscillate less.

Interesting if you look back at the stock Rep2X extruders with stock PID settings on some of the other tests, it seems to dip just before target temperature.  (almost like it is over damped, before settling down into a normal oscillation and dampening.  This is behavior is interesting to me because most x3g programs start printing immediately when they hit the target temp.  With default PID settings as they are, it takes a few minutes for even the stock setup to stop ringing.  This means that you're likely starting the print over temp, with a few oscillations.

This also doesn't account for the fact that you're feeding a "coolant" in the form of cold plastic into the extruder.

[Ryan Carlyle]

The PID code has some quirks that explain a lot of the behavior you see. For example, it bypasses PID and runs bang-bang when the temp is more than 10C from the set point. Your curve for the stock hotend shows the power output dip when the initial 100% power mode switches over to PID and the integral feedback hasn't adequately kicked in yet. So you're just getting P and D action, with the D trying to reduce heater power because the recent history is rapid heating. Basically at 10C below the set point it switches from 100% to 0%, then the D action dies down ~3 seconds after the temp stops climbing and you start seeing the PID code do what it's supposed to do.

Downside to the >+/-10C bypass is that it will lead to overshoot if the amount that the temp keeps rising after shutting off the power is more than 10C. You'll see that with high power beaters and slower temp sensors.

Additional sidenote, for this kind of application you want to be way overdamped, unless you are trying to do very high speed printing.

[Jake]

On Thursday, 7 April 2016 20:35:55 UTC-4, Ryan Carlyle wrote:

Downside to the >+/-10C bypass is that it will lead to overshoot if the amount that the temp keeps rising after shutting off the power is more than 10C. You'll see that with high power beaters and slower temp sensors.

I think this is whats happening with the Tecto setup, you get ~5deg overshoot on the first "bounce".

Based on some subsequent tests, my recommendation is to position the tecto thermocouple's tip so that it doesn't stick out of the heater block (flush) if you can.  You get a little less overshoot (~1deg less) and it behaves about the same if you try to center the tip in the block, so there's no benefit to "burying the tip" deeper in the heater block.

[Carl]

The analysis and feedback you are providing is awesome Jake! :-)

I would love to see a comparison using the recommended setup... an apples vs. apples comparison would be very interesting to see... and would help me to tweak the recommended setup if needed! ;-) 

[Jake]

I would love to see a comparison using the recommended setup... an apples vs. apples comparison would be very interesting to see... and would help me to tweak the recommended setup if needed! ;-) 

Here is a comparison of standard PID settings with the thermocouple in 3 positions: "Tip sticking out" is what I'd call her "recommended per instructions" configuration.  The "tip centered" configuration is where the tip is buried to align approximately with the nozzle.  "Tip flush" is with the tip not sticking out the other side of the block, but aligned to be flush with the far end of the block.

The "tip centered" and "tip flush" configurations trace nearly an identical curve, but the "tip sticking out" one has a little more overshoot and more oscillations.

[Jake]

It would be awesome to see the results using my thermal barrier... a brass nozzle... and if possible... a comparison showing the Tecto heater block without insulation... with the aluminium foil insulation... and with your current kapton and ceramic insulation! ;-)

I already started re-assembling, unfortunately.  I want to see this puppy print again!  The reason I went with ceramic/kapton though is because I suspect I'll be disassembling this again and I didn't want to waste my one shot at the aluminum insulation.  If there are opportunities to take more data in the future I will.

I'd be surprised if the insulation has a major effect on the bench.  I suspect it'd be more prevalent when there are fans and air currents from actual printing.

  

[Jake]

Okay, I took things back apart to run some more tests.

This one is ceramic/kapton insulated vs. uninsulated, default PID settings:

Starting temp was a little off (I guess because I was handling the block/thermocouple to remove the insulation.  I think it shows that it settles down to about the same as the insulated version, but you get a little more overshoot on the first "bounce"

[Jake]

More charts for you.  This compares insulation types with non-insulated.  I've aligned the charts at 27-degrees instead of the point where the heaters were turned on.

I'm really not sure which one is "best" at this point.  Any thoughts?

[Ryan Carlyle]

No meaningful difference between those, I think.

[Gian Pablo]

I'm here as a private citizen, no longer part of Autodesk. I'm at Studio Fathom now.  However, the team is aware of your gear, and I've certainly recommended it!

On Friday, April 8, 2016 at 4:57:46 AM UTC-7, Carl wrote:

Happy to hear you are up and running Gian! :-) 

Since I know that a large number of regular contributors to the various 3D printing related Google Groups are now under an NDA on the AutoDesk group that they and I were invited to join by you... I am curious if you are posting as a regular user or if you are posting as an official AutoDesk representative... and if this somehow ties into your developments? 

Would be awesome to know that AutoDesk is testing my products! :-)

[Gian Pablo]

Results so far: printing ABS works for the left & right extruders. Quality is very good. nGen works on the left extruder but not the right. Problem is filament gets hung up somehow and the extruder starts clicking. Looking at the filament after extraction, looks like a bulging of the filament right before it enters the thermal barrier tube, suggests an issue with thermal creep. I can try cranking the fans higher. Might also be an assembly issue on my part, I will take it apart and take a look at whats happening.

In the picture below, white is ABS, orange is nGen, black is ABS printed on the Rep 1 for comparison. I've lit them from the side specifically to highlight variations in the layers.

[Dan Newman]

> I'm really not sure which one is "best" at this point. Any thoughts?

Between insulated types, the one which hasn't started lifting too much at
the edges in 6 months might, by that measure, be deemed superior. Within
a couple of months, the ends of the kapton tape I use to hold down the ceramic
material has significantly lifted.

Dan

[Gian Pablo]

Yeah, I ended up with the "tip flush" config pretty much through luck. Sticking it out was definitely no good. 

[Jake]

It would be awesome to see the results using my thermal barrier... a brass nozzle... and if possible... a comparison showing the Tecto heater block without insulation... with the aluminium foil insulation... and with your current kapton and ceramic insulation! ;-)

The last test you requested. This compares the P3-D nozzle and barrier with Carl's barrier and brass nozzle.  I tried 2 configurations Carl's parts -- "near" which has the heater block closer to the cooling block  and "far" which moves it further away.  Not possible with the P3-D tube, which is shorter, and so its the "near" configuration.  Also the P3-D tube as tighter treads, perhaps due to the thickness of the coating, I'm not sure if that affects heat transfer or not.

I am not sure what to attribute the differences, but it seems like Carl's tubes perform somewhat similarly no matter the distance the heater blocks are mounted.  The P3-D tube seems to have less overshoot and oscillation.  I don't know why, but that doesn't necessarily make it better, just different.

I think my final build will be the P3-D nozzle and tube, with Carl's insulation (since ceramic/kapton) is not appreciably better, but I may add a layer of caption to the bottom of Carl's insulation to protect from caked on filament and whatnot.  At the same time, I'm going to add one small bit of ceramic insulation to insulate the "exit end" of the heater and thermocouple because it can't hurt.

[Ryan Carlyle]

I would expect tubes and insulation to change the heat-up rate more than the amount of oscillation. Very interesting.

Are you using the same heaters for all these tests? There's a lot of variation in cartridge resistance and that can make a big difference in behavior.

Question for Carl... Is there any possibility of the TC tip being positioned at different depths in different Tecto thermocouples? Some manufacturing variation could explain why people are getting different results.

[Jake]

Are you using the same heaters for all these tests? There's a lot of variation in cartridge resistance and that can make a big difference in behavior.

Question for Carl... Is there any possibility of the TC tip being positioned at different depths in different Tecto thermocouples? Some manufacturing variation could explain why people are getting different results.

Same heater, same thermocouple. 

[Gian Pablo]

And here is some carbon-filled ABS, probably the best print so far. The little dots just brush off. No ringing, no warping.

[Ryan Carlyle]

Nice-looking Caliblocks.

[TobyCWood]

Are these with acceleration slowed down?

[Gian Pablo]

Thank you! It is a great model, thank you for sharing it. Much more fun and useful than a bunch of 20mm cubes!

On Friday, April 8, 2016 at 6:48:54 PM UTC-7, Ryan Carlyle wrote:

Nice-looking Caliblocks.

[Gian Pablo]

Yes. Tweaked the acceleration settings. I think I can get rid of the remaining artifacts by adjusting the JKN settings.

[Jake]

By the way, if you haven't wrapped your heater blocks in Carl's Alu/Fiberglass insulation, and then again in Kapton, then you're missing out.

I want to wrap my car in this stuff.  Looks like it fell off a NASA mars rover.

[TobyCWood]

I don't know man... with those going back and forth while I watch... before i know it I'll get hypnotized!

[Andrew Aurigema]

Ummmmm been buzzy with printing and missed this thread..... what is that perty lookin insulation again ???  Would I bust off the ceramic insulation and put that directly on the hot section or put it over the ceramic things that the MBI factory installed  on my R2X ???

And I know we have covered this a bunch of times but my fans are dying and need to be replaced. I can push on the blade assy and the fan spins faster for a few minutes then slows down again.   I think they are 24 volt but want to be sure so I dont blow the driver on the board.    A good part number to get replacements would be appreciated too.  

Drew in sunny FLA....... 2600 hrs on my Replicator 2X and it just keeps a printin.