Thermal Compound with HBP - Consistent 0.1 Resolution Layer height and Consistent quick target HBP Temperatures!

[CKChavez]

I wanted to share some tips (I tried posting earlier but my post hung for some reason) that have helped me with my Replicator in printing, heat settings, and consistent 0.1 resolution prints. I've noticed a lot of people still having issues with leveling/re-leveling and with higher resolution layer heights.

When I first started using my replicator I used an IR thermometer to check the HBP platform temperature. I noticed that it varied greatly. When set at 110°C it measured anywhere from 85°C to 111°C. It was most consistent in the middle and front of the build platform but varied everywhere else on the platform. I also had sagging with the build plate when it was heated up for printing.

I didn’t give much thought to this but this simple solution made sense to me - I separated the aluminum build plate from the heated circuit board and applied thermal compound paste to the heated circuit board (like this - what you use between your CPU and its heat sink.) I reattached the aluminum build plate and the Circuit board and then tested the temperature again with the IR thermometer. The target temperature was now consistently within +/ -1°C anywhere on the build platform.

I then leveled the build platform using the standard method (not the dial gauge method) - doubel checked it. Note I have not had to the re-level the build platform at all and I’ve printed more than 75 items since then. The prints now stick consistently to the kapton tape and sometimes too well. I attribute this to the aluminum build plate expanding (and warping) in a uniform manner. I’m also able to print easily at 0.1 layer height. I now am able to set my build plate temperature lower from to 110°C to 100°C and the thermal compound helps it get to the target temperature faster. This helps me print my initial prints faster and with less time waiting for the build platform to reach the target temperature.

With the temperature of the build plate now consistent, I do not use rafts unless it’s absolutely critical for support material. Also note that I’ve enclosed my Replicator using (my poor man’s method picture for reference attached) clear laminating sheets. I didn’t insert any paper in the laminating sleeves but just ran them through the laminating machine blank. They are rigid enough for me to attach to the sides of replicator and for me to build a hood (I did put paper in the side and back hood sheets to make them a bit more rigid.) For the front of the Replicator I did use an acrylic piece.

I hope the thought of using thermal compound helps some of you who have found it necessary and required to use a raft or have had unsuccessful prints at resolutions 0.1 or less. I believe with the uniform heat across the aluminum plate the leveling remains consistent also. With this setup I’m using Sailfish successfully at extrusion rate of 200mm/s and travel rates of 200mm/s with excellent results!

[Dan Newman]

>
> I didn’t give much thought to this but this simple solution made sense to me - I separated the aluminum build plate from the heated circuit board and applied thermal compound paste to the heated circuit board (like this - what you use between your CPU and its heat sink.)

Except don't use that one: not only is it too expensive for the application,
it's also been implicated in shortings of the PCB to the aluminum build plate and
also in cases where it oozed to the underside of the PCB and shorted the SMD components
on the underside. While the compound is initially non-conductive, apparently it
can break down over time and become conductive. The manufacturer's own application
guide indicates that it should NOT be used in situations where electrical insulation
is necessary.

Note that MBI used to supply thermal compound with the Cupcake and ToM HBP kits.

Dan

[Adan Akerman]

This is a great idea. I have a tube right here, close at hand... I believe I'll be applying it first thing tomorrow! 

This is a 3.5g syringe of Arctic Silver 5. I believe it's identical to the product you linked to. It's a tiny tube. Is that adequate to cover the entire platform? Did you apply it and then smear it smooth, or just leave it in a zigzag pattern?

Oh, yikes, just saw your note, Dan. Do you recommend any particular alternative?

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.
To view this discussion on the web visit https://groups.google.com/d/msg/makerbot/-/L2NPS6a_cx8J.
To post to this group, send email to make...@googlegroups.com.
To unsubscribe from this group, send email to makerbot+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/makerbot?hl=en.

[Dan Newman]

On 1 Dec 2012 , at 3:26 PM, Adan Akerman wrote:

> This is a great idea. I have a tube right here, close at hand... I believe
> I'll be applying it first thing tomorrow!
>
> This is a 3.5g syringe of Arctic Silver 5. I believe it's identical to the
> product you linked to. It's a tiny tube. Is that adequate to cover the
> entire platform? Did you apply it and then smear it smooth, or just leave
> it in a zigzag pattern?
>
> Oh, yikes, just saw your note, Dan. Do you recommend any particular
> alternative?

Pretty much any of the mundane, non-conductive thermal pastes work.
The problem is that the stuff you see at Radio Shack or computer stores
tends to be the expensive stuff meant for a heatsink on your expensive PC's CPU.
That (1) doesn't need to be non-conductive [just don't get it onto the
PCB itself], and (2) people don't mind paying $5 - $15 for a tiny amount
because it's for their precious CPU. At a Radio Shack they may have
Artic Silver's Ceramique which is what AS recommends for situations
where electrical insulation is needed. However, it too is pricey for
the application.

I pretty much use the generic white stuff I can buy in a tub at the
electronics store. Just make sure that it's non-conductive. For example
McMaster-Carr's 3883K24.

Dan

[Adan Akerman]

Excellent, thanks.

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

[Chris Chavez]

Sorry, and yes thanks for pointing that out. I shouldn't have linked to the one in my post. I did use the nonconductive derivative. Good catch Dan!

On Dec 1, 2012 4:21 PM, "Mark Cohen" <markc...@gmail.com> wrote:

Use arctic silver arctic alumina
http://www.newegg.com/Product/Product.aspx?Item=N82E16835100015

On Dec 1, 2012 7:13 PM, "Mark Cohen" <markc...@gmail.com> wrote:

Ill post the name of the stuff when i get home

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

To post to this group, send email to make...@googlegroups.com.
To unsubscribe from this group, send email to makerbot+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/makerbot?hl=en.

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

[Chris Chavez]

Adam - I just smeared it a bit toward the corners to make sure it was somewhat uniform in as much of the board as possible. I've been really pleased with the result.

[RocketSled]

I'd already done this.  I used a material called Bergquist Gap Filler (because we use it at work).  The Thermal Pastes used for Heat Sinks aren't ideal for this application.  That kind of paste wants to be laid down thin, between two flat (ideally, machined) surfaces that aren't going to have a lot of "creep" under thermal expansion/contraction.  Bergquist is designed to fill larger irregular gaps and it turns in to a rubbery material when it cures.

It's important to use any TIM (Thermal Interface Material) sparingly.  It's actually an insulator.  You want to fill the gaps that would otherwise have nothing but air, since TIM conducts heat better than air.  But you don't want a thick layer that keeps the PCB from making direct contact where it can.  Direct contact conducts heat better than TIM.

The more uniformly the Aluminum plate heats, the flatter it stays as it expands. I've been thinking that insulating the back side of the HBP would help with this immensely, too, (and speed up heat-up time). But I haven't bothered yet...

[Gary Crowell]

I've used the Gap Filler at work too and it seems like a natural.  Since the heater pcb is just bolted to the aluminum plate around the edges, does the pcb really fit tight against the plate across the center?  I've never had mine apart - is there any electrical insulation between the board and plate, other than the soldermask?

Anyhow, I'm looking at this: http://www.digikey.com/scripts/DKSearch/dksus.dll?Detail&itemSeq=123238952&uq=634903049940885805

Twice as much as necessary (a 12x11" sheet) and not cheap, but it is adhesive both sides.

Gary

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

To view this discussion on the web visit https://groups.google.com/d/msg/makerbot/-/4pei-9aROVAJ.

To post to this group, send email to make...@googlegroups.com.
To unsubscribe from this group, send email to makerbot+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/makerbot?hl=en.

--
----------------------------------------------
Gary A. Crowell Sr., P.E., CID+

Linkedin  Elance  KE7FIZ

[hellphish]

I just use a layer of kapton on top of my PCB. It is an insulator but it has a rubbery thickness to it and seems to work ok. I don't have an IR probe but I have watched hairspray evaporate off the bed and it is pretty even and after a nice soak every part of the glass has enough heat for good adhesion.

[Andrew Russell]

This is another interesting thread, as I've suspected the HBP temp was not very uniform from various experiences.  (Also, when the HBP cable was failing, I noticed that prints would curl only on certain locations on the build platform, which I assumed was from non-uniform heat distribution)

Gary,  that link isn't working for me, what product is it?

RocketSled, is that Bergquist Gap Filler easy to remove? 

I can't tell from reading this link (http://www.bergquistcompany.com/thermal_materials/gap_filler/gap-filler-1000.htm) if it "solidifies" once cured, and if this would glue the ABP to the PCB.  I'm assuming the PCB will die one day due to the previous failing cable oxidizing parts of the male connector.  I'd like if I could replace it relatively easily w/o prying it off the aluminum plate.

I'm looking forward to trying out printing on glass, and I would assume a more uniform temperature of the aluminum build plate would help for this.

Y'all have offered up three suggestions and I'm not sure which to toss some money at! :) The bergquist gap filler seems like the most appropriate option for the job.

thanks!

-AndyR

 

[John D'Ausilio]

I'm guessing it's http://www.digikey.com/product-detail/en/BP100-0.005-00-1112/BER246-ND/1646693 .. looks interesting

[John D'Ausilio]

hmm .. but at only 5 thousandths, probably not thick enough to fill likely voids?

[Gary Crowell]

Right.

[Cymon]

I've been watching this with interest, but the problem is this doesn't jive with a desired upgrade I want to try one day. A buddy of mine made his plate removable using recessed holes for magnets instead of screws on the corners and I wanted to try that myself on my replicator. Having a sticky goo on the back isn't going to help with that. Is there something that will distribute the heat properly without being goopy or sticky?

[Jeremy Green]

Maybe something like this: http://www.itapestore.com/3mthermalpad5590h05mm-944x1foot.aspx

To view this discussion on the web visit https://groups.google.com/d/msg/makerbot/-/kXIMf4NfQJYJ.

[Adan Akerman]

Interesting... might be nice to have a layer between the aluminum and PCB, and another layer between the aluminum and glass. Starts to be a significant number of layers to be asking the poor board to heat, though.

Datasheet suggests our HBP temps are acceptable for this product: they age-tested it 2000hrs at 110°C without significant effect: http://multimedia.3m.com/mws/mediawebserver?mwsId=66666UF6EVsSyXTtNXfcN8T_EVtQEVs6EVs6EVs6E666666--&fn=02910.pdf

[Cymon]

Why not remove the aluminum and heat the glass directly? How thick is this glass, by the way? Does ABS like it or do you need Kapton?

[RocketSled]

RocketSled, is that Bergquist Gap Filler easy to remove? 

I can't tell from reading this link (http://www.bergquistcompany.com/thermal_materials/gap_filler/gap-filler-1000.htm) if it "solidifies" once cured, and if this would glue the ABP to the PCB.  I'm assuming the PCB will die one day due to the previous failing cable oxidizing parts of the male connector.  I'd like if I could replace it relatively easily w/o prying it off the aluminum plate.

We use some blue colored stuff that has a consistency like thick, cured silicon caulk (but it contains no Si, which is, ironically, bad for electronics).  It peels off like you'd expect.  We also have some pink stuff that stays soft like clay, which you need to smear off and then clean with a solvent.  We put this stuff inside our SSDs, to help them dissipate heat during operation (they're high performance).  We like the pink because it allows less force to be coupled to the PCBA through the drive housing, which makes the SSDs more rugged.  Blue is much stiffer but the thermal properties are about equal.  

For this application, the blue stuff is the right material to use.  The fact that the blue has some moderate adhesive properties is beneficial.  There is very little force at the center of the heater PCB to hold it in contact with the build plate.  I layer in some blue Berguist and clamped the PCB to the BP until the stuff cured.  I'm betting I get much more uniform heat transfer now.

You could get adhesive thermal heat sink paste, too.  But those things are usually epoxies, they cure *hard* and are not intended to be removed, which would make it difficult to service the HBP if you needed to disassemble it.  Blue Bergquist will just peel apart...

[RocketSled]

On Thursday, December 6, 2012 4:45:00 PM UTC-5, Cymon wrote:

Why not remove the aluminum and heat the glass directly? How thick is this glass, by the way? Does ABS like it or do you need Kapton?

That's my next project.  I'm just trying to decide how to do it.  I'm afraid to try and cut holes in the glass for screws, but was thinking I could glue the screws by the tops of their heads to the underside of the glass, and attach the heater and leveling screws basically the same as the Aluminum plate goes together.  There are some decent adhesives for gluing metal to glass.  It'd require opening up the holes in the heater PCB since they'd need to get past the underside of the tapered screw heads in order to contact the glass, or I'd have to layer in a lot of Bergquist (and the stuff isn't all that cheap, and like anything else it's efficacy goes down as its thickness goes up).  So I have to take my HBP apart to look at the PCB from the heater-element side in order to figure out if this would work, and I haven't had a chance to do it yet...

[Andy]

Great! Thanks for all the info.  Looks like the blue goo is what I'll try.  At least between the ABP and the PCB.  Between the glass and the ABP, I think I'll try one of those sheets posted previously, since one of my goals for using glass as a build platform is to be able to rapidly swap pieces and keep printing .  Maybe I can get a cheap temperature gun from harbor freight to check the results.

Do you happen to know the product name/number for the blue goo? :)

Thx,
-Andy

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.

To view this discussion on the web visit https://groups.google.com/d/msg/makerbot/-/uG9H1xOjNRsJ.

[Adan Akerman]

I've just placed a Digikey order for a 1m roll of the 3M acrylic thermal interface sheeting, as well as some TIM paste (and a bunch of Orion fans...). Cymon has a point, there may be value to heating the glass directly, but I am made more comfortable, somehow, by having the aluminum plate there. 

Here's hopin' it's the good stuff.

[CKChavez]

Awesome Adan. I'll be interested in how the 3M sheeting works out. I used a glass plate previously on top of the original aluminum plate before I had used the thermal paste with the original aluminum platform. Unfortunately, for me the heat didnt transfer well though the glass surface (I used glass cut from an old flatbed scanner I had laying around and binder clips to hold in place.)  With the heat not getting hot enough - The Temp on LCD read at target of 110c, but the glass surface was approx 15c less than target. At the time I was uncomfortable with setting the Target Heat setting much higher than 110c for fear of burning something up. Maybe I was too cautios then. Which is why I went back to the Aluminum plate with the thermal paste. I do have to say that I gave up quickly on the glass and didnt do much experimenting - maybe I'll give it another shot down the road. A "quick change" platform would be very nice. 

 

Luckily, I've had great success with the setup I have now.

[Andrew Russell]

Looks like the Bergquist product is: Gap Filler 3500S35.    Digikey has it only in minimum orders of 10, and its also pretty expensive ($45 for a single order)

They offered this as an alternative:

http://www.digikey.com/product-detail/en/832TC-450ML/473-1089-ND/655660

But from reading the application guide it looks like it will not be easy to remove.

I guess I'll be placing an order for the same items you listed Adan. :)

thanks everyone for the info and advice!

-Andy

[RocketSled]

Sorry I never even looked at the price, I just grabbed some left-over from the engineering lab at work so it didn't matter.  I guess it is kind of pricey (Berqguist, that is)...

The "gap pad" TIM is also stuff we've used.  This is a funny material.  It's actually kind of "non-Newtonian".  If you try and compress it quickly, it resists. But if you compress it slowly it offers very little resistance.  It makes it good at forming around parts on a PCBA without crushing them (you just need to press it down slowly).  Because of this, it does kind of "flow" and would probably be a decent alternative to something like Bergquist.  Not sure if the material you chose is the same stuff, Adan, but I expect it'll work anyway!  Be sure to give us an update once you've tried it.

[Gary Crowell]

Just for grins, I've been taking some fairly extensive temperature measurements of the HBP.  I want to try and duplicate them later today or tomorrow before I report what I see.

Gary

On Sat, Dec 15, 2012 at 12:09 PM, RocketSled <rscoh...@gmail.com> wrote:

Sorry I never even looked at the price, I just grabbed some left-over from the engineering lab at work so it didn't matter.  I guess it is kind of pricey (Berqguist, that is)...

The "gap pad" TIM is also stuff we've used.  This is a funny material.  It's actually kind of "non-Newtonian".  If you try and compress it quickly, it resists. But if you compress it slowly it offers very little resistance.  It makes it good at forming around parts on a PCBA without crushing them (you just need to press it down slowly).  Because of this, it does kind of "flow" and would probably be a decent alternative to something like Bergquist.  Not sure if the material you chose is the same stuff, Adan, but I expect it'll work anyway!  Be sure to give us an update once you've tried it.

--
 
 

[Gary Crowell]

Ok, I just finished an another hour taking HBP temperature measurements.  I'll take a few more later and while I'm writing this.  This session was done to confirm what I had measured on Friday night, and it did.

The setup:

• Fluke 80T-150 rigid temperature probe.  This was plugged into my no-name, but reliable and more importantly, consistent, digital multimeter.
• I printed a 10x10mm grid on the platform, measurements were taken in the center of the 'boxes'.  This helped enormously in finding the same location for taking repeat measurements.
• I made every effort to hold the probe at the same angle (i.e., vertical) and with the same pressure for each measurement.
• The HBP was cold, 22C at the start, which pretty much matched room temperature.
• The Replicator was set to 'preheat' with a target temp of 110C, and it reached that temperature in 8:20.
• The Fluke reads in 1/10 degree, and this was rounded to nearest degree for all measurements (exception for the cycles, see below).
• The Fluke was checked in a melted-icewater bath after the measurements (should have done it before too) and measured 0.2C.
• I have an Extech hand-held IR thermometer, but it was too difficult to hold it steady for the long periods needed for many measurements.  It was also very sensitive to angle at the close range necessary.  I did take some quick readings with it though.
• Just for grins, I also tried a Cen-Tech (Harbor Freight) handheld 'puck' IR thermometer.  See the attached .jpg.
• The Replicator has a full enclosure, except that the front panel was off.  There were no drafts in the room.  The enclosure interior temperature reached 30C, measured by a cheap thermometer I have installed in the enclosure.
• The Replicator (Sailfish) preheat times out after ~31 minutes.  It was rapidly restarted when it timed out.

Procedure: Of course the temperature is not going to be constant, as the power is cycled to the HBP to maintain the set temperature.  On the Replicator display, the temperature varies +/-1 degree from the set temp.  On the Fluke probe this appeared as a temperature cycle of about 2C degrees over a period of about 45-60 sec.  The 1/10 degree resolution of the Fluke was useful for this, as the cycles could readily be seen.  It would take the probe about 30 seconds to get to the temp of the point being measured, then by watching the cycle, it was very evident when the temp hit a peak of the cycle.  The measurement was recorded at the peak, rounded to the nearest degree.  The elapsed time from preheat turn-on was also recorded.

After running through this once on Friday, I had a good idea of the points of interest I wanted to measure.  I had determined that there was a smooth gradient between points (center-corners, center-sides) so this time I concentrated on taking measurements at the corners, sides and center.  One of the problems is the time it takes to make a measurement with the Fluke probe.  This made it impossible to take a 'snapshot' of all points at a particular time.  But having done it once before, it was possible to perform a measurement pattern that eked out the data I wanted. 

Findings:

• Shortly after reaching the set-point, the center of the HBP measured 90C.  Thirty minutes later it reached 92 degrees and maintained this temp for the remaining hour of measurements.
• At set-point, the left front corner measured 83C.  
• The other three corners were 87-88C.
• The temperature of the left front corner increased steadily over the next hour to 89C.
• The other corners also increased to 89C.
• In the interim times, the temperature of the left front corner always lagged the other corners.
• Measurements between the corners and the center showed a smooth gradient.
• Measurements at the center of the sides, were 1-2 degrees higher than the adjacent corners.
• Measurements with the Exetech IR meter, after ~90 minutes, were 103C in the center and 99C at the corners.
• Measurements with the Cen-Tech IR meter, after ~90 minutes, were 94C in the center and 87C at the corners. 
• The temperature variation due to cycling, at the center, was from 90.5 to 91.8C, over a period of about 45 seconds.  At another time it was 90.1 to 91.7C.

My conclusions (YMMV):

• The temperature reported by the Replicator thermocouple reads ~18 degrees higher than the actual platform center surface temperature.  (No wonder I couldn't get my tea hot enough.)
• The surface of the platform is not at it's final temperature when printing begins.  It requires an additional 20 minutes for the center to reach its final temperature.  After an hour, the corners are within a few degrees of the center.  The entire platform would probably require 2-3 hours to approach an isothermal state.
• My HBP has a 'cold' corner.  This is almost certainly due to an uneven thermal contact between the heater pcb and the aluminum plate (I have never removed my HBP).  For most of an hour after reaching the set-point, this corner is up to 7 degrees cooler than the center, and 5 degrees cooler than the other corners.  It is highly likely that my HBP will benefit from an application of thermal interface material.  How about yours?
• The temperature gradient consistently seen between the center and corners, center and sides, and corners to sides, show that a constant resistance pcb really isn't ideal for producing an isothermal platform.

A very long time ago, my actual first assignment as a shiny new engineer was to do some thermal modeling of power dissipation on a thermal plate (for the signal processor that went in the ADCAP-MK48 torpedo).  Though I haven't done thermal modeling in a long time I know quite certainly that a uniform set of heat sources, evenly spread across a flat plate, will produce a higher temperature at the center than at the edges.  Any thermal modeling software will show this.  From the CAD file for the heater pcb, it is evident that the trace that produces heat is a constant width throughout its length, thereby effectively producing uniform heat sources across the board.  That isn't what we want.  The traces at the sides and corners should be narrower (=higher resistance, more heat) than those in the center.  It would take some simple thermal modeling software (perhaps even just a spreadsheet) to determine those trace widths, but it wouldn't be inordinately difficult.

It is possible that my 'cold' corner is due to uneven trace etching or plating in the fabrication of the pcb.  This would depend upon the quality of the fabricator, but I don't think it too likely.  Even low-end fabricators should be able to hold that pretty close over a pcb of this size.

You might also wonder if it is really necessary to have an utterly even temperature across the plate.  I suspect not.  But even is probably better than uneven, especially when it doesn't cost anything more (recurring cost anyway, it might take an extra few prototypes to get it right).  That cold corner though, is probably somewhat important to know about.

Oh, and I know a guy, who I think has a thermal imaging camera.  I'll have to see what he's got.  If I asked around, I could probably find one somewhere at work, but hauling a Replicator in to do that might be a bit much. (Or I could just take in the plate and a variable power supply, hmmm.)

On Sat, Dec 15, 2012 at 12:44 PM, Gary Crowell <garyacr...@gmail.com> wrote:

Just for grins, I've been taking some fairly extensive temperature measurements of the HBP.  I want to try and duplicate them later today or tomorrow before I report what I see.

Gary

 

Gary

[RocketSled]

Gary,

Contact measurement with a thermocouple is going to read a couple of degrees low.  It was accurate in your glass of ice water because it was immersed. Completely surrounded and in direct contact everywhere.  Just touching the surface of the HBP, you get a point contact.  Heat transfer through that point is likely lower than the dissipation of the rest of the probe, the probe never really gets to the same temperature.  To get more accurate readings on the HBP you need some thermal compound to glop in to the center of each square, thick enough you can submerge the probe tip in to it.  

We usually use an epoxy to glue probes in place on ICs when we need accurate data.

Also, IR thermometers have trouble with shiny surfaces. Bare Aluminum, but also probably if covered in Kapton, the IR thermometers are likely to read low, too.

But excellent science, anyway!  You clearly have an overabundance of the very motivation I lack.  8*)

-R

[Gary Crowell]

Right about the point contact, of course.

Yeah, once, somewhere, I had access to an IR meter where you could adjust for surface.. emissivity(?).  But not on this one.

Wondering why I never get around to printing anything.

Gary

[Gary Crowell]

FWIW, I did go back recently and re-measured the center of the HBP with the contact probe, with a dab of Arctic Silver at the point.  Read right at 100C, where I was reading ~91C before.  

Gary

On Sun, Dec 16, 2012 at 9:21 PM, RocketSled <rscoh...@gmail.com> wrote:

-R

--
 
 

[RocketSled]

Wow.  More than I expected.