Another DIY Reflow Controller

[MatCat]

Well as it goes I needed a reflow controller, so I made one!

This will be driving your fairly common 1500W walmart special.

All is tested and working, now to finish the firmware!  I did do the board to be open source, so I will post the files up for it soon, also I will post up the firmware once I have it completed.

Specs:

MCU               : ATMEGA644P
Thermocouple: MAX31855 / K Type
USB                : FTDI
SSR                : NTE 75A
Indicator        : RGB LED

Power In and Out

Power In

USB and Thermocouple connection

ISP Header

And the main guts! 

[IanJ]

Hi,

Nice one!

Just interested, did you use PID or a simple table, and are you using PWM (slow) for the heaters or just simple on/off.

My own home made oven I found great for small boards but for larger ones (euro card sized etc) I'm having to fit a small fan (running pretty slow) inside to assist in circulating the heat inside the oven.

PS. love the well-used tools etc lying on your workbench......a real sign you often knuckle down to some hard work.......

Ian.

[Psi]

As awesome as it is to build things and as much as i am all for hobby projects..
ya really don't need to worry about reflow profiles for hobby toaster ovens.
A variac on the cable and a one-off temp sensor to figure out the ideal variac position is all you need.

Sorry, i dont want too come across as negative.

[JuKu]

As awesome as it is to build things and as much as i am all for hobby projects..
ya really don't need to worry about reflow profiles for hobby toaster ovens.
A variac on the cable and a one-off temp sensor to figure out the ideal variac position is all you need.

Sorry, i dont want too come across as negative.

Not my experience. You might do without proper profile if you are doing mostly resistors and not too demanding components and are willing to do some manual rework. But even a simplest thing like a 317 with a PCB heatsink: Good luck soldering the regulator properly without tombstoning a close by 0603 or 0402 resistor unless you have a proper pre-heating phase. With slow enough heatup and long enough reflow phase you might, but then you kill your 100+ pin ICs. MatCat is doing it right.

[Kremmen]

+1 for a nice project

@PSi: While i also acknowledge that people manage to do what is recognizably SMD soldering using the most primitive of methods, starting from frying pans and kitchen stoves, it doesn't necessarily mean that those methods are good, reliable or desirable. What you suggest will probably work nicely in the majority of cases and i don't want to bash it outright. That said, i did what the OP has done and extensively modified a toaster oven, rearranging the heating and air circulation as well as employing a commercial prigrammable, sequencing PID controller. I didn't do it because i had to, but because i wanted to so there are cheaper alternatives. Still i am happy to say that the performance of that oven has been 100.0%. Not a single failed joint in any of the various boards i have fed it.
I won't argue that similar results can't be achieved with simpler means but this is so damn convenient! No guesswork, no fiddling and watching over the process, no nothing. Just a nice result time after time.

P.S. i moved all heating elements to the top side. This allows me to do double sided loads without fear of the bottom side parts dropping away because that side won't reach liquidus during the normal cycle. I recommend it to others who modify their ovens.

[Psi]

Not my experience. You might do without proper profile if you are doing mostly resistors and not too demanding components and are willing to do some manual rework. But even a simplest thing like a 317 with a PCB heatsink: Good luck soldering the regulator properly without tombstoning a close by 0603 or 0402 resistor unless you have a proper pre-heating phase. With slow enough heatup and long enough reflow phase you might, but then you kill your 100+ pin ICs. MatCat is doing it right.

Mike uses the variac + toaster oven method as well

Reflow bit starts at 14min

[Kremmen]

Fair enough but he seems to have a wetware PID controller in the loop anyway .

He makes an important point towards the end of the video - avoid blasting away with the heater elements right from the start because the IR from the glowing heaters will fry the composites on the board. It is exactly the behavior one sees with a PID controller and a soaking ramp: the elements are just starting to show deep cherry / oily look which means they are somewhere around 500-550 C. The temp ramps up with nearly constant heating power that is a fraction of the max. That does the trick for presoak and then smoothly increase power to ramp up to liquidus and then down after the 10-20 seconds of liquid solder.

All of that can be done with a variac and eyeball. However unless you already have the variac, a PID controller will probably be the cheaper option. Of course a variac has other good uses so why not.

[IanJ]

All,

May I speculate:-

Unattended, reliable reflows across varying sizes of pcb's & smd technologies etc require a reasonable setup by way of reflow profile, control hardware/heaters etc.

Anything else then you either have to monitor it closely, take lady luck in your arms or have had run your setup sooooo many times your experience has won out.

Ian.

[MatCat]

I am using a PID loop in this setup, I am actually going to be using this setup to do small production runs so I do need it to be a descent profile.  The oven I am using is convection so has the fan, and I am going to be insulating it too to minimize temperature variant as much as possible.

It might be considered a bit overkill for the general hobby use, but I do a lot of prototyping and my own production runs (sometimes up to 5k units), so yeah I think it's worth it.

I plan to have the firmware workable today, so if I do I will do a video of the first test reflow which is to test the PID loop and temperature reaction times so I can fine tune if need be.

[Psi]

Be wary of the fan, they can cause intermittent problems with components moving during reflow even when you think you have the air volume low enough.

Depending on the position and orientation of objects inside the oven, and on your PCB, you can get funneling effects in various locations which concentrate the air.

[Kremmen]

Since i mentioned the fan, let me elaborate that my fan is not an original pizza oven gizmo. It is an adaptation from a microwave oven fan and sits in a separated compartment with a diffusor grating in front of the fan prop. The airflow is relatively modest and definitely turbulent. Just enough to move the air in a cycle but not to create any kind of blast or streamlines.

[MatCat]

I pretty much have the firmware in a workable state to do a test run... the only hardware hurdle left is isolating the thermocouple since the entire oven including the rack is tied to ground, my plan is to use a little bit of insulator to isolate it from the rack and have the thermocouple attached to the bottom of the rack. 

I also still need to whip up the computer program so that I can log the reflow profile, as well as control it from computer.  I will probably do this in java so that it can be cross platform easily.

The system will have the ability to hold 10 reflow profiles, with 20 points each, the time and temperature can be anything within the 20 points, and of course programmable from PC.

[calexanian]

I want to make a cam/variac operated one. As the cam turns it wiggles around the variac setting the heat. It worked for the staging control in the Saturn rockets. All of those stages and separations... Motor and cam controlled.

[MatCat]

Murphy always gets you somewhere!  First I found out I actually traced the RX/TX lines to the FTDI wrong on the PCB!    Since The MCU has plenty of horses I just implemented software serial to make it work, then Murphy came in even worse, causing ground inductance issues with the Thermocouple, stick that thing in the oven and it just goes haywire!   So I am probably going to have to rethink my nice compact design for better shielding from the AC.

[MatCat]

It turns out it was the thermocouple I was using, I tried a different thermocouple and blam things are working as expected!  Here is a chart of the first dry run to see how close reality would get to my profile.  Obviously a bit of tweaking to be done, but not too bad!  Odd saw shape at the end is because I kept opening and closing the door and seeing how the pid would react hehe.

[MatCat]

After a little tweaking on the profile here is a workable profile, I did reflow a single test board that I picked because it has a 5A inductor and DPAK5 with large heatsink plane so it was a good test to reflow, and that went ok, it could of used about an extra 20 seconds in reflow so I will probably lengthen the reflow profile slightly.

Here is the new chart

[MatCat]

Got a little video put together of a reflow, figures the one on camera has to have it's cap go sideways ROFL.  All is well though I think maybe it's in reflow slightly too long?  There was a slight yellowing to the silkscreen which I did not expect at all, otherwise went quite well though.

[Kremmen]

Nice work 
One suggestion: you can see how the inductor reflows later than the smaller components when you ramp up the heat. At the same time you also make the notice that the reflow phase should maybe be a little shorter. I have tried to avoid over 20 sec in liquidus. One way to get that is to prolong the soak phase at ~150 C. I have set it at 2 minutes and in practice it is about 90 seconds due to the lag of actual temp vs setpoint. Anyway, that gives the larger components some time to warm up to the soak temperature and from there they reach liquidus a bit quicker making the relative delay shorter.

[MatCat]

Nice work 
One suggestion: you can see how the inductor reflows later than the smaller components when you ramp up the heat. At the same time you also make the notice that the reflow phase should maybe be a little shorter. I have tried to avoid over 20 sec in liquidus. One way to get that is to prolong the soak phase at ~150 C. I have set it at 2 minutes and in practice it is about 90 seconds due to the lag of actual temp vs setpoint. Anyway, that gives the larger components some time to warm up to the soak temperature and from there they reach liquidus a bit quicker making the relative delay shorter.

Yeah I planned to shorten that phase a little bit, plus when I did that run I had 6 boards in the oven in various locations to look for hot zones, really the only boards that lagged where the ones towards the front by the window, which makes sense.   Overall I am pretty happy!  Now it's time to put some polish in the firmware, and whip up a good companion PC software for it too.

[MatCat]

Today I did 100 boards so it was a good test of the system, I did tweak the reflow time so that liquid stage was about roughly 20 seconds, and that worked perfectly, though I think I will slightly lower the peak temperature by about 20F.  I also aluminum-foiled the door (except small window).  First reflow was 63 boards since that was all I could fit on the rack, only had 2 fails do to parts going cockeyed but that was mostly because of an uneven spot on the rack, and I was happy to see pretty much all of the boards reflowed at the same time, showing that there is little in hotspots, the oven is getting warmed up pretty evenly.  I do have a video which I am currently editing of that.

[MatCat]

[Milan]

Hi, nice project. 

I also made one controller for my workshop.

MCU: PIC18f4520
Sensor: PT100
Oven heaters power: 600W

This project are in test phase.

On the last picture is the new PCB design.
If testing goes well, I will post schematic and code.

Milan

[MatCat]

I will post full schematics for my design shortly, I need to polish it up a little and fix the UART RX/TX issue.  I will post the full Altium Designer project, as well as gerbers.  I still have some work to do on the firmware to enable full PC functionality, plus PC code to go with, so once I have all that working I will also publish all the code.   

[jesuscf]

Today I did 100 boards so it was a good test of the system, I did tweak the reflow time so that liquid stage was about roughly 20 seconds, and that worked perfectly, though I think I will slightly lower the peak temperature by about 20F.  I also aluminum-foiled the door (except small window).  First reflow was 63 boards since that was all I could fit on the rack, only had 2 fails do to parts going cockeyed but that was mostly because of an uneven spot on the rack, and I was happy to see pretty much all of the boards reflowed at the same time, showing that there is little in hotspots, the oven is getting warmed up pretty evenly.  I do have a video which I am currently editing of that.

Nicely done!  In my experience if the reflow should be less than 60 seconds or the silkscreen changes color.  if the reflow temperature is over 230C the solder masks may get darker.

[scopeman]

Wow MatCat,

Really Sweeeeet!

Now I wonder if I can steal that new Black and Decker that Ali just bought at Kohls for 25 bucks.....

Somehow I think she would notice.....

Doe's Altium offer an Export option that can save the files in P-CAD 2006 format?

Sam
W3OHM

[Rick~K]

Really nice controller, even a oven full of boards reflowed pretty even.

Rick
K8OHM

[scopeman]

I will post full schematics for my design shortly, I need to polish it up a little and fix the UART RX/TX issue.  I will post the full Altium Designer project, as well as gerbers.  I still have some work to do on the firmware to enable full PC functionality, plus PC code to go with, so once I have all that working I will also publish all the code.

Hi MatCat,

Did you post this project somewhere?

Sam
W3OHM