Difficulties with Tom Kole 4.3 Case/How to Fix


 

Paul Frere

TVWBB Member
Difficulties with Bryan's 4.3 Case/How to Fix

This post is intended to be informative for someone building an HM and planning to use the Kole 4.3 case with the four button holes. It is not intended to be critical. FWIW I worked my way through HS and college as an electronic tech before I got my MSEE. I also have a pretty complete shop, including a 3-axis DRO precision German mill, which was handy on this effort.

What I found is that the case is quite fussy about the soldered-together LCD and Button boards. There is a shallow cavity that locates the LCD and then four cavities that locate the switches. In addition there is a "fence" that locates the top of the LCD board in the case. I built my HM prior to buying the case so my LCD and button boards were already soldered together when I tried to insert them. To my eyeball, they were square, parallel, etc. Fitting this asembly to the case was by far the most difficult part of the HM build.

When the assembly was inserted, it would not come even close to sitting properly with the board against the fence, LCD in its cavity, and the switches in their cavities. First step was to file the top edge of the board at 45deg so it would slip farther under the fence and get the lcd properly aligned with its cavity. At that point with the LCD seated, the four switches were not centered in their openings. The buttons were toward the bottom of the opening and were shifted to the right as viewed from the front. I couldn't see inside, of course, but from the height of the buttons it was clear that they were not dropped into their cavities. Long story short, the switch board was immovably soldered to the LCD board, so when the LCD was properly in its cavity, the switches simply could not be in theirs. The consequence of this was that the whole HM assembly sat "high" in the case. The most obvious consequence of this was that the four probe jacks sat maybe 0.050" above the bottom of their cutouts.

After quite a bit of study, I ended up enlarging the switch cavities so that the switches would sit lower, then I milled 0.030" off of each of the main HM board pin headers, allowing the main board to sit closer to the button board and the probe jacks to sit closer to the bottom of their holes.

OK, finally, the point of the post: If I were to do this again, I would use the case as a jig for soldering the LCD to the button board. I'd make sure the LCD fit its cavity properly, filing the top board edge if necessary, then I would take it out, fit the button board, then reinsert the LCD board properly engaged with the button board pins. I would clamp the two board as necessary to make sure they were positioned correctly and stayed put. Only then would I solder the boards together.

Edit: Actually, as I think about it, I would probably also use the case as a locating jig for soldering the switches to the button board. IIRC there wasn't a lot of slop when doing this freehand but using a jig wouldn't hurt.

Hopefully this will help someone.
 
Last edited:
The case tries to be a compromise between creating something that fits like a commercial injection molded part and having enough tolerance built into it to support any possible soldering alignment. The latter isn't possible because of the distance between the attachment points and some solder points is far enough that it can create pretty significant differences between where things should ideally be and where they end up. Just one degree of angle at the LCD connector introduces 0.7mm of error at the other end of the LCD board. Allowing for 3 degrees of error would add over 2mm of tolerance which would allow other HeaterMeters to slosh around. There are some key areas to try to get as close as possible though:
  • Pinheader soldered in the LCD should be flush and perfectly vertical. Centering in the holes would be good but that's difficult to achieve so I don't put much effort into that.
  • Mating the LCD to the LCD/Button board, header should be flush with PCB. When looking from the side of the LCD, the two PCB should be parallel with each other. When sitting on the table with the LCD flat against the table facing up, the LCD/Button board should not touch the table. When parallel, the LCD/Button board isn't long enough to reach the table (via the riser pins that go down to the base PCB). Note that when inserted into the case, the buttons shouldn't go into the square button recesses, they hover just above them. (Better to have a case that can fit a slightly too large HeaterMeter than a case that bulges and compresses it)
  • The Raspberry Pi connector should be straight and somewhat centered. If you look from the bottom you'll see white silkscreen around the Pi connector. I solder them so I see a line of white along the top edge which is straight (not sloping off into disappearing).
  • The riser pin headers between the two boards should be vertical and perpendicular to the board edges. Having both headers skewed by a single degree shifts the top of the LCD by 0.85mm. There's tolerance for this but you can see how easy it is to get things very very out of of alignment with just a couple degrees here and there.

The LCD inset area was ~0.5mm too short up until April so inserting one of the newer white LCDs made things fit kinda weird. They are supposedly the same size as the older LCDs but seem to be just enough different that the old one fit mostly OK but the new one fit mostly not OK. Checking my calculations I found a ~0.5mm math error that corrected this for both versions. I use a jig to solder my LEDs into place because they can really wreck alignment if they aren't just right so I'd encourage folks to try to fit their case without the LEDs first and see if they can correct any misalignment before adding 3 more constraint points.

Here's some images I just took with the Yellow/Green LCD in my soldering jig. The LCD should fit under the "grab notch" but still have a little space before it contacts the thick bar at the top. The screen should be flush against the front. This old solder job isn't perfect so there's a bit of teeter-totter, and the buttons are a fraction of a millimeter higher than normal but it fits fine in the final case.



Finally just a side note, Tom Kole created the 4.2 case and I did the 4.3 case.
 
Last edited:
Thanks for the post, Bryan. I think it adds to the collective wisdom here. I have also corrected the thread title to give you credit for the case.

IMO your comments just underline the need for great care in combining these two boards. I did read and heed your instructions regarding the white line. I also did use the case as a jig for locating the LEDs in all three axes. What I did NOT understand was the need for beyond-eyeball precision in soldering the boards together. I think it would be good if you emphasized that more, even to the point of encouraging builders who plan to use the case to also use it as a soldering jig. Once all those header pins are soldered it is pretty much game over for adjustments. As I found.:(

For the record, my main problem was that the sensor jacks would not sit nicely in their cutouts. I assumed this was because the switches were not completely down in their cavities but from what you say that was probably not the cause. Anyway, after quite a bit of fussing and milling 0.030" off those two sockets, I got them close enough that "gently" forcing the case halves together pretty well solved the problem.

I guess you could use a piece of flex circuit in place of the hard soldered header but that would have its own problems with cost and for people who are not experienced techs. So IMO you are probably at a good point right now except for maybe providing a little stronger warning about alignment.

This is a very impressive project overall and, as I play with the web interface, I realize the work involved. Thanks again!
 
I appreciate you posting your specific problems especially with the exact modifications you needed to make. If you check the SCAD file in git, you'll see that I am still making occasional changes to the design based on my experiences and information I get from the community about how it fits for them.

It is strange that you had a problem with the probe jacks being too high to fit. That would mean that something from below was pressing it up, as all the LCD alignment stuff tends to force the board down and the mating length of the riser pinheaders should allow a little slack if the base board sits too low. Did it sit right if it were just the Pi and base board? Here's one I just popped together and snapped a photo of:


As you say it is near impossible to see inside to see what's not lining up properly to make the case close so it's really hard to tell what's going wrong if someone runs into a problem. I did all sorts of things when designing it, printing partial parts and cutaways and clear filaments and as it is now it seems to fit every HeaterMeter I solder. That just means I have built a part that works well with my process. Admittedly, there are a lot of variables involved on the soldering side and 3D printing side so hopefully it goes together more often than not.
 
Those jacks are quite nicely centered. After all my fiddling, mine are still not centered but the case does screw closed with about the same amount of gap between top and bottom as you have in your photo. One last thing I did, which did help to a small degree, is to flatten the bottom of the lcd cavity as it seemed like the LCD not going in all the way was helping to make the button board high.

If it were my design and I were fiddling with it, I would open up the LCD inset area and the four pushbutton insets enough that they could not interfere, leaving only the undercut tab ("grab notch"?) and the two screws to locate the board. The rebating those insets a little bit should not affect the stiffness the case face IMO. Then the USB jacks and thermistor jacks should finish the locating job when the case halves are assembled.
 
Undercut! So that's the term for that. I don't have any mechanical experience or time with a CNC or mill so I just sort of make up my own names for things. I appreciate your suggestions but I have a few questions to clarify.

Opening up the LCD inset area: you mean the size of the cutout in the back of the top of the ... wait let me get a picture. Extend this area toward the buttons? I'm not sure if going wider would work because the screw holes there are already barely able to be printed properly due to their clearance with the cutout.


Do you know if the case you printed was from the pre-April 05 version or after I made the change to move that LCD inset down?
 
Opening up the LCD inset area: you mean the size of the cutout in the back of the top of the ... wait let me get a picture. Extend this area toward the buttons?
Yes. I just reduced the height there because it didn't seem like the LCD was seating properly.

It's probably impertinent to make suggestions on such a carefully thought-out design but what this seems to me to be is what I call a "tight" design, where lots of things have to work almost exactly right or there are problems. The M3 screws and the undercut are the basic things that locate the button board and LCD assembly. In addition, though, you have cavities into which the four buttons are supposed to fit and a cavity where the LCD is supposed to fit. All these cavities are fine on their own, but if even one of these five items is a little out of position, nobody drops into his cavity. Since the plastic is opaque I can't see for sure, but this seems to be what is/was happening with my assembly. So my suggestion was that all the cavities be enlarged so none of the five items can possibly be so far out of position that the board assembly can't seat completely.

If that seating problem is solved, then the button board ought to drop right into place before soldering the LCD board and the thermister jacks should seat nicely in their half-holes. Similarly (before soldering) the LCD board ought to sit nicely under the undercut tab with the LCD bezel nicely flat against the back of the case front and the header pins protruding from their holes.

OK, if both those boards are seated properly with the header pins engaged, then the pins can be soldered properly because the case is functioning as a jig.

All my problems stemmed, I think, from the fact that the button board and the LCD board were eyeball-parallel, but apparently a degree or less off from exactly parallel. Hence, the original point of my posting here --- to suggest that folks take advantage of the case to use it as a jig. If my boards had been exactly parallel, the buttons and the LCD might still not have completely seated, but things would still have been better. Alternatively, if the buttons and LCD had properly seated then my sin in not getting things exactly parallel might not have caused as much difficulty. Hard to say without Superman's x-ray eyes to see through the plastic.

But regardless of all this fussy trivia, you have done a phenomenal multi-disciplinary job with the product and I continue to be grateful for it. I hope you are making some money.

Do you know if the case you printed was from the pre-April 05 version or after I made the change to move that LCD inset down?
No idea. Your invoice # was 2748, dated 5/29.
 

 

Back
Top