Dual sided reflow assembly of Eggfimder Quantum boards

[caveduck]

Something I've been working on incrementally this year amongst all the machining projects - assembling Eggtimer board kits efficiently and repeatably.

This is a 3D printed fixture with an insert that has registration pins that fit the Quantum board mounting holes, and is set up so that the top of the board is flush with the top of the main red fixture. It's used for paste solder application with OSH stainless steel templates. The top and bottom side templates are tape-hinged on either side. Most of the printed blue insert is cut out so that boards that already have one side populated can be flipped over to have paste applied to the other side. The screws that hold the insert to the red table are countersunk flush. I'm planning making another insert to hold Quantum boards, and another whole fixture for the much larger Protons and TRS's. I'll get this up on Thingiverse after the AirFest crunch.

I'm using two kinds of solder with different melting points to do the dual-sided reflow, which is done with a mechanical-controls toaster oven and an Arduino based reflow controller I got off of Kickstarter a few years back. On the bottom side I'm using regular SnPb (217C) and on the top side it's SnBi (137C). The bottom gets done first with the higher temp SnPb. Then the top side can be done without affecting the bottom side since the SnBi cycle never reaches the melting point of the SnPb. The pre-soldered chip on the other side will not fall off, it's not heavy enough to defeat the surface tension holding it on. Yes, the WiFi module is getting reflowed onto the board - I just run a line of solder down each row of pads and put a couple of resistor leads in diagonal corners. After the board tests good they get epoxied down since the module has some mass.

[Edit: forgot to mention I'm using some 1.0 mil kapton tape under the WiFi module to protect the vias under it. Much thinner than the suggested masking tape (less likely to make the pads not bond) and more compatible with reflow temperatures.]

For batch assembling I be sure to put a Brother label on the WiFi chip with the access key before assembly. It survives the reflow run just fine. Through-hole parts get hand-soldered afterward.

 

[rharshberger]

Very nice work.

 

[caveduck]

Four Quantums at once going into the oven, which is a Cusinart unit from Target modified so that that the convection fan and both the top and bottom heating elements all run at the same time for maximum heat output and uniformity. Pre-modification it would only run one of the heating elements when in "convection" mode. You probably don't really have to do that though. Opening and reassembling the oven was a PITA, you need a foot-long screwdriver to screw the front control panel on *from the back*.

Here's the reflow controller - a Reflowster from about 5 years back. No longer available but still one of the nicer fully-integrated units I've seen.

Yesterday's output is shown below. I solved the problem of holding the thru-hole parts in place for batch hand soldering by dropping them in from the top and tacking down the edges with some 5-minute epoxy. Beats the pants off of trying to tape them in place.

The only other gremlin I hit was that a couple of the wifi modules tilted enough even on the 1mil kapton tape to have marginal connections on a few pads. I'm going to look carefully at whether I can omit the tape or just grind off a couple of the unused pads on the left end of the carrier, which are the only ones at possible risk of shorting to the vias underneath the module.

 

[caveduck]

Links to things used in this project

Low temperature SnBi paste solder: https://www.amazon.com/Solder-Bi57-6-No-Clean-Lead-Free-Temperature/dp/B0195V1QEI

Regular SnPb paste solder: https://www.amazon.com/SMD291AX-REWORK-SOLDER-PASTE-CLEAN/dp/B0186IM0P0

Liquid flux in syringe (this stuff is magic if you have to rework): https://www.mouser.com/ProductDetail/Chip-Quik/SMD291?qs=8BX3xQzFIvmwkympZqnNNA==

Solder paste stencils - both kapton and stainless steel available: https://www.oshstencils.com/

Reflow controller ("Controleo3"), oven build kits and fully build ovens: https://www.whizoo.com/ This is not what I currently use but there is a ton of good info on their site about building a more accurate reflow oven. When I finally upgrade I'll probably start here.

 

[Dan Griffing]

I'm using two kinds of solder with different melting points to do the dual-sided reflow, which is done with a mechanical-controls toaster oven and an Arduino based reflow controller I got off of Kickstarter a few years back. On the bottom side I'm using regular SnPb (217C) and on the top side it's SnBi (137C). The bottom gets done first with the higher temp SnPb.

I thought about doing this too but someone, I'm not sure who, advised me that two different kinds of solder paste weren't necessary so I went with only the 137C SnBi paste.

I constructed a $250 reflow oven using the Whizoo.com Controleo3 kit and a brand new $30 Black and Decker toaster oven. Since the Controleo3 has its own language for programming and SD card storage of the reflow profiles based on the optimal specifications for the paste being used, I spent a couple of days making and testing a profile for the 137C paste, and tested the profile on 2-sided practice surface mount solder boards. I also learned how to dispense the paste onto the pads with a dispensing needle, and constructed a vacuum pick-n-place tool from an aquarium pump with one part reversed, all based on online videos.

So far I've successfully assembled half a dozen Eggtimer kits, and have more yet to do. The whole 137C reflow profile takes less than 30 minutes to complete. This has made assembling Eggtimer kits as fun as building the rockets they're used for.

The reason why two different temperature solder pastes aren't required is that the flux paste with tiny solder particles seems to act differently than the solder once it's melted and the surface tension has stuck the chips to the board. If your reflow profile just goes over the specified MP for a short period of time, it melts the solder in the fresh paste and the liquid solder surface tension centers the chips onto their pads and sticks them there quite well.

But if you leave some pads bare on your practice board for a subsequent reflow iteration, you can add fresh paste and chips to these pads and go through the reflow process again, and the previously soldered chips do not budge at all and the solder does not appear to have been melted again. The same thing appears to happen with chips on the underside of the board.

Now bear in mind that the Controleo3 reflow oven I used had an aluminum tray, and I put several layers of paper towel as an insulator between the underside of the board and the tray, so the underside might have bee a few degrees cooler than the topside, and the measured oven temperature only went to 155 of 160C for about a minute. So its likely that the soldered chips on the underside of the board never got to the 137C MP to melt the solder.

I also used the 137C paste because I wanted to use the reflow process for some of the boards that had sensitive chips already soldered, and I wanted to make sure that the solder temperatures didn't come anywhere close to damaging the existing chips.

 

[cerving]

I gotta get me one of those Controleo controllers... I use the B&D oven, but I manually control the temperature. Leaded solder paste is MUCH more tolerant of temperature inconsistencies than non-leaded solder. Considering that I've mounted baro sensors on several thousand boards with a near-zero failure rate, it's "good enough".

 

[Dan Griffing]

I gotta get me one of those Controleo controllers... I use the B&D oven, but I manually control the temperature. Leaded solder paste is MUCH more tolerant of temperature inconsistencies than non-leaded solder. Considering that I've mounted baro sensors on several thousand boards with a near-zero failure rate, it's "good enough".

I chose the 137C solder paste because I didn’t know how sensitive the barometric sensor was to reheating.

The Controleo3 is a pretty amazing overall product for the price. As a retired firmware and control engineer I couldn’t avoid diving into the firmware source code. It’s not really PID control and the design is a mess but the product works overall so I didn’t touch it, even though I could have.

Having a simple “programming language” for reflow recipes, a touch screen interface, SD card recipe storage, Arduino Zero compatible electronics, AND a rock-solid electrical and mechanical design — all for $250 — makes it quite a winner! The instructions are somewhat amateurish with the writer not keeping his personality in the background, but they work if you print them out, pay particular interest to the pictures, make notes, and use the instructions as a checklist.

Pay attention to the instructions’ warning about the sharpness of the internal surfaces of toaster when you open it to seal and add insulation. Wear rubber gloves to protect yourself. And wear protective gear when you do the final step of the final ceramic wool insulation because it’s like asbestos to inhale or get on yourself. But this is only 15 minutes and can be done outdoors.

Once you get the recipe dialed in for the board size, the oven works flawlessly. Daubing solder paste onto the pads and using a diy vacuum pick-n-place to stick the chips onto the pads is amazingly more fast than hand soldering. Once you do it, you’ll never go back.

The Controleo3 is the same kind of winner that Eggtimer products are, in general.

Contact me via email or call for special details.

 

[Dan Griffing]

Four Quantums at once going into the oven, which is a Cusinart unit from Target modified so that that the convection fan and both the top and bottom heating elements all run at the same time for maximum heat output and uniformity. Pre-modification it would only run one of the heating elements when in "convection" mode. You probably don't really have to do that though. Opening and reassembling the oven was a PITA, you need a foot-long screwdriver to screw the front control panel on *from the back*.
View attachment 429441
Here's the reflow controller - a Reflowster from about 5 years back. No longer available but still one of the nicer fully-integrated units I've seen.
View attachment 429442
Yesterday's output is shown below. I solved the problem of holding the thru-hole parts in place for batch hand soldering by dropping them in from the top and tacking down the edges with some 5-minute epoxy. Beats the pants off of trying to tape them in place.

The only other gremlin I hit was that a couple of the wifi modules tilted enough even on the 1mil kapton tape to have marginal connections on a few pads. I'm going to look carefully at whether I can omit the tape or just grind off a couple of the unused pads on the left end of the carrier, which are the only ones at possible risk of shorting to the vias underneath the module.
View attachment 429443

I did just one board at a time with the Controleo3, I used the aluminum tray for the Controleo3 instead of the rack, and I did all of the smd parts one side and a time, all with the 137C lead free low temperature Sn Bi solder paste. I hand-soldered the through-hole parts and large components instead of using reflow on them. I put the board on several layers of paper towel to insulate the already soldered components on the backside, and kapton-taped aluminum foil over altimeter chips, WiFi daughter board, and anything else that I thought needed temperature protection.

The optimized 137C reflow profile only had the temperature above the MP and up to 160C for a few seconds so everything worked out.

 

[caveduck]

Yeah I'm thinking of getting a Controleo3, at least the controller part. Things have worked so well thus far with the hacked convection oven that I don't feel like further oven mods are going to help me very much. I was really interested in a vac picker for a while, but I've improved enough with the tweezers that it doesn't feel urgent anymore. Both of my cycles run in < 10 min...the most time-consuming thing now is soldering the thru-hole parts and connectorizing, though the epoxy-tacking has helped a lot.

 

[Dan Griffing]

Yeah I'm thinking of getting a Controleo3, at least the controller part. Things have worked so well thus far with the hacked convection oven that I don't feel like further oven mods are going to help me very much. I was really interested in a vac picker for a while, but I've improved enough with the tweezers that it doesn't feel urgent anymore. Both of my cycles run in < 10 min...the most time-consuming thing now is soldering the thru-hole parts and connectorizing, though the epoxy-tacking has helped a lot.

The key part in making it work with the toaster oven is sealing all of the cracks and insulating, so that the heating elements can bring it up to the required temperature and with uniformity. I’d recommend getting the whole kit and buying a new $30 B&D toaster oven. Otherwise you’re likely to be trying to figure why its not working right instead of making optimized reflow recipes for assembling boards.

Even with all of the insulation the ramp rate is quite dependent on the thermal load from the size of board. With heat leaks and poor insulation it simply won’t work.

 

[caveduck]

...yet mine *does* work and the temperature ramps quite nicely, and I've not had to do much fiddling at all with profiles...

 

[Dan Griffing]

...yet mine *does* work and the temperature ramps quite nicely, and I've not had to do much fiddling at all with profiles...

Each reflow solder paste comes with a data sheet that contains a profile for that paste.

The activation of the flux in the paste takes place at a lower temperature than the melting point of the solder, and it optimally whets and etches the pad and chip surfaces at this temperature before they’re ready for melting the solder.

Just heating the stuff up might work but the action of the flux might not be sufficient, or too much heat might be applied to the chips that could damage them. The recommended reflow profiles are done by professionals in the field and they exist for a reason.

Just like the other areas of high power rocketry, it’s based on solid science and engineering.

 

[caveduck]

and my controller implements said profiles pretty decently, and actually uses a PID algorithm, and they are pretty well aligned with the mfrs profiles. BTW I am a degreed scientist.

 

[HHaase]

One thing I don't like about Tin Bismuth solders is they are quite brittle. For passives like resistors and ceramic capacitors it's no big deal, but for electrolytic caps or other parts that are large sized with small terminations I wouldn't recommend it. For this application I think it's pefectly fine..... just something to be aware of. But having rails or conveyors to make double sided SMT work smoothly with a single paste isn't really something easy to implement at home. You'd be surprised though how big of a component the surface tension will hold in place without issue.

The Controleo oven system is pretty nice. I have a Controleo2 that I use for my own boards, I've probably run near two thousand cycles on the thing for my own home business. All my stuff is lead-free since I have too many international customers. I think the whole front of the thing is orange now from flux fumes. Proper convection is just so superior to an IR system, particularly if you're using plastic body IC's.

I do this stuff for a living, and as a hobby. It's a curse.

 

[Dan Griffing]

and my controller implements said profiles pretty decently, and actually uses a PID algorithm, and they are pretty well aligned with the mfrs profiles. BTW I am a degreed scientist.

I appreciate that your controller implements the profiles pretty decently and actually uses a PID algorithm.

I looked at the control source code for the Controleo3, and as someone who has implemented many varieties of PID algorithms in firmware throughout my career, there’s no discernible PID algorithm there. But overall the Controleo3 works sufficiently as a product so I resisted the temptation most firmware engineers have to make any changes. I also appreciated the simple profile recipe programming language of the Controleo3 that permits people to write executable recipes to closely follow the spec solder paste profile for any solder paste they use.

As I learned about reflow soldering, if I had wanted to design the firmware for a diy reflow tool myself, it would have taken many months and then resembled this one but not been any better. Instead, I spent $250 and several days to assemble this one, and it works in a professional manner. It’s not perfect but it’s hard to ask for a better overall value.

I’m 5 years retired but was a degreed chemical engineer with process control systems and firmware design as my specialties.

 

[HHaase]

Truth be told, I wouldn't go too crazy worrying about the profile anyway unless you're measuring actual on-board temperatures with a thermal profiler. Typically the PCB is around 10-15 degrees below the convection temp on a lightly populated board like what we deal with in rocketry. As long as your ramp rates are within reason, and you control the cooling rate, you should be fine as long as your peak time/temp is good. Even then, the main critical area to look at is the peak temp and time above liquidous. If your board is hitting 215C for 30-45 seconds on a leaded profile, you should have some great solder joints.

The paste print is much more critical to a good SMT board than the reflow process either way.

-Hans

 

[Steve Shannon]

...yet mine *does* work and the temperature ramps quite nicely, and I've not had to do much fiddling at all with profiles...

It’s difficult to argue against success, isn’t it?
Both of you have done really interesting work. The extra work you did enabling all the heating elements in the Cuisinart probably compensates nicely for the heat loss and I just don’t believe that a $30 Black and Decker starts out as a higher capacity oven in the first place. Nice job documenting what you’ve done. I don’t often subscribe, but I did as soon as I read your first post.

 

[HHaase]

Yeah, the cheap B&D's take a lot more work. I think I stuffed every available volume on mine inside the housings with mineral wool for insulation. Another surprising area of heat loss was the glass window, enough that I couldn't initially get mine to do a complete cycle without an error. I ended up wrapping the whole door in tin foil, which was a huge help.

If I were to do it again I'd look harder for something that is double-walled all the way around, and with some kind of gasketting/seals around the door.

-Hans

 

[Tobor]

Hey guys,
Many thanks to each of you for posting to this thread. Great info and discussion. A reflow oven has long been on my list of DIY projects.

 

[KilroySmith]

You guys are too particular.

I got my toaster oven from Goodwill for $10. I measured the temperature ramp rate of just cranking the temp dial up to max, and found that it actually reasonably matched the specified ramp rate of the Kester Sn63Pb37 solder paste I had. So my reflow controller is a piece of paper that says:
1. Turn temp to 250C
2. When temp gets to 170C, adjust knob to keep between 170-180 for 60 s
3. Turn temp to max
4. On reaching 220C, turn off oven and open door.
Heck of a lot cheaper than buying/building a controller, and works great for the 1-2 times a year I use it. Perhaps the profile isn't perfect enough to run a million boards through a factory at acceptable quality levels; but that's OK.

 

[Dan Griffing]

It’s difficult to argue against success, isn’t it?
Both of you have done really interesting work. The extra work you did enabling all the heating elements in the Cuisinart probably compensates nicely for the heat loss and I just don’t believe that a $30 Black and Decker starts out as a higher capacity oven in the first place. Nice job documenting what you’ve done. I don’t often subscribe, but I did as soon as I read your first post.

Whenever I come across something that I think is really cool, my 55-year old Catholic upbringing still leaves me with the tendency to proselytize about it, and I often fail to consider that others who have equally cool ways of doing the same thing have been there before, and might consider my exhubrance a challenge to their equally valid way of doing it. Right now I’m still in the afterglow of discovering high power rocketry in my retirement.

 

[Dan Griffing]

Yeah, the cheap B&D's take a lot more work. I think I stuffed every available volume on mine inside the housings with mineral wool for insulation. Another surprising area of heat loss was the glass window, enough that I couldn't initially get mine to do a complete cycle without an error. I ended up wrapping the whole door in tin foil, which was a huge help.

If I were to do it again I'd look harder for something that is double-walled all the way around, and with some kind of gasketting/seals around the door.

-Hans

The Controleo3 has gold reflector tape for the window, except for a small viewing port. My estimate is that half of the value of the Controleo3 is the meticulous design of the insulation of the toaster oven. You don’t need to make up for heat that is not lost.

 

[curtisheisey]

Great tips. Thanks.

I bought a zellus controller. It was a project by a student, successful kickstarter item, and he sold it only for a few years, but looks very similar to the ControleoX. Not available now. He did seem to master a good PID control loop.

Do you solder the ESP32 right to the PCB and not use the wire standoffs?

 

[HHaase]

You guys are too particular.

If you ever drive through a traffic light at an intersection, you'll be glad that I am superbly particular about my solder quality.

The Controleo3 has gold reflector tape for the window, except for a small viewing port. My estimate is that half of the value of the Controleo3 is the meticulous design of the insulation of the toaster oven. You don’t need to make up for heat that is not lost.

I used too big of an oven, and ran out of the gold tape that came in the kit. That stuff is expensive! I didn't have the money for more at the time so I used what I had.

 

[Dan Griffing]

Great tips. Thanks.

I bought a zellus controller. It was a project by a student, successful kickstarter item, and he sold it only for a few years, but looks very similar to the ControleoX. Not available now. He did seem to master a good PID control loop.

Do you solder the ESP32 right to the PCB and not use the wire standoffs?

I’m away from home now so I can’t look at the board but I only reflow soldered the smd parts and used stand-offs for hand soldering as the instructions called for. I also heat-shielded the gps and Wifi daughter boards with aluminum foil, insulating paper, and Kapton tape.

 

[caveduck]

I'm reflowing the ESP32s but they are a little prone to some slight tombstoning, because having to put even the thinnest tape under them creates a small gap, and solder doesn't like to jump airgaps. Putting some flux and more solder paste across the vias on the back of the board post-reflow has fixed a couple of problem children. Having to put in 16 wires in to make up for having to tape over one via in an inconvenient place is a major loss of time so I'm going to try to engineer that out of the build. Next batch I am going to try out removing the plating from the no-connect pin on the ESP32 that sits over the via to abolish the tape. Too busy in the AirFest departure crunch to do any more on this right now (or even post much).

One more note, when you send your files to OSH you can modify the gerber files to take care of things like making a cutout for the pre-soldered chip, enlarging the too-tiny holes for the 3-legged chip, adding holes for the ESP32 pads, etc. There are lots of gerber tools out there, and OSH has a simple editor right in the web interface. I ended up having to cut my stainless templates, which was a considerable PITA.

Oh yeah, firmware programming on the Mac - you need a couple of things

• Good serial drivers for the Prolific devices used on the eggtimer boards. https://mac-usb-serial.com/. Costs a few Euros (Paypal for the win), worth every nickel. You want the PL-2303 driver. The Prolific drivers have been buggy as #$%! for years.
• Mac version of command-line esptool can be found here: https://github.com/igrr/esptool-ck/releases
• Alternately you could try the Python esptool via 'brew install esptool'. More work to set up.

Here's the command line for the command-line esptool, assuming you have the mac-usb-serial.com drivers:
./esptool -vv -cd ck -cb 115200 -cp /dev/cu.Repleo-PL2303-00203405 -cf Quantum_1_07A.bin

Be sure to check your serial device name with 'ls /dev/cu.*' and adjust accordingly. The mac-usb-serial drivers will all start with Repleo-PL2303, YMMV if you try to use the Prolific factory drivers.