No sparks or mechanical fatigue. 1500Watts can be handled by a Q4025L5 triac driven by a MOC3011 triac driver - all for less than $4 from Jameco. No snubber is needed because the load is not inductive. See http://www.fairchildsemi.com/an/AN/AN-3003.pdf
-Randy
Keep in mind that a triac driving a 1500W 120V load would itself dissipate about 15 watts. This would require a decent heatsink, whereas a mechanical or solid state relay runs fairly cool. That level of dissipation may be hard to get out of an insulated tab pkg too, so the heatsink would either need an insulator pad or the heatsink would be electrically hot.
The main benefit behind a triac might be its ability to reduce the power without resorting to harsh on/off cycles. Has anyone seen a need for this?
Oznog:
The main benefit behind a triac might be its ability to reduce the power without resorting to harsh on/off cycles. Has anyone seen a need for this?
I believe that using a PID control algorithm, you can maintain a much tighter control on the temperature. Harder to do this with a relay and I suspect a much shorter operating life (chattering contacts don’t last as long). By the way, phase control is not really needed for this application when using a triac.
Its best to line up the possible solutions and their pros/cons plus pricing and go from there.
relay - need to handle at least 20A and I’d go for a 25 or 30A spec:
cheap, easy to drive, good isolation
mechanical, subject to wear (spec at least 1M rated cycles), noisy (audible and electrical)
triac. same current needs as a relay.
cheap, easy to drive, good isolation (via opto driver), faster switching, phase control
heat dissipation, electrically noisy (unless using zero crossing logic)
SSR
very easy to drive
expensive
one should also add IGBTs for completeness sake but I don’t know enough about them.
From surfing digikey, the relay and triac solutions could be had for under $5. SSR, is way out there in prioce. Based on the above, I’d opt for the triac approach. I’d also, as I said before, move all the AC stuff off the controller board.
Check the new tutorial Nate’s posted- it’s a hoot. The final outcome is that hotplates are cooler (read better) than toaster ovens. It was truly impressive to watch the hotplate do its thing.
He’s also been talking about offering a stencil service for a while now. You might see it in the near future.
I’ve been lurking here for a while, but looking over the existing reflow controller, and reading the tutorials, I wonder if it might be better to drive the upper elements at a different drive level than the lower elements in the reflow oven.
Given the propensity of the upper elements to heat so much faster, I would think that it would be amazingly slick to do both topside and bottom side sensing, and adjust so that the board heated evenly on both sides. SSRs are available for cheap on the surplus market (I have a pair of them I grabbed for $10 each, 20A/250V, 5V trugger), and I think a slow PWM signal could keep things under control. It might be as simple as characterizing the heating rate of the upper and lower elements, and adding a calibration constant, but I think there’s real power to be had with a dual-element control. Thoughts?
And yes, I know, the first version was only just released today. I figure if I’m going to do this, I’m going to do it right, and only once
I have been using a toaster oven for SMD reflow quite successfully for about 4 years now. I’ve reflowed thousands (literally) of small dual sided boards. Toaster oven was mom’s old one (read: free) and is much smaller than the one in the tutorial. Rememer that the bigger the oven, the more thermal inertia it will have, making it harder to control.
The first time I used it, I burned a batch of boards. I found that since the two bottom elements were placed one above the other in the center, there was a hot spot. This was easily fixed by putting a 4x6" piece of aluminum above them. Since aluminum transfers heat very well, it tends to diffuse the heat acoss it. It also acts like a buffer to keep the oven from changing temperature too fast.
I made a little stand for my boards. They are double sided (and I do both sides at the same time) so they need to be suspended. I placed a thermocouple directly under the board. BTW, all my boards are hand pasted using an electronic dispensor, a syringe and a lot of patience. But after seeing the cheap stencills here, I might be converting!
Using the ‘toast’ mode, I manually modulate the the toast switch to follow the reflow curve. Initially, the oven will heat up at just about the right ‘preheat’ rate. Once I get close to 140, I turn it off. The temp will continue to rise but will start to level off just like the profile. Once 160 is reached, I will again turn it on and spike it up to about 200. I’ll turn the oven off, but the temp will continue to rise. I’ll leave the door closed until 215-220 and then open it enough to allow the temp to start to drop at a gradual rate. Waite 5 minutes before trying to move the board. (I dropped a board that was still ‘wet’, thinking it was cool enough and all the parts came flying off the board.)
After a few runs, you learn how not to overshoot your target temp and using my scopemeter trend-plot feature, I can match the reflow profile exactly. I have thought of making something to do this automatically, but its so easy to do it myself it wasn’t worth it.
A couple of comments to the admin. From what I read on your site, you were having problems with the USB connectors not reflowing or burning other connectors… If you are ramping the temp up like the profile you have here
this could be your problem. The ‘soak’ period is important to get all the components up to just under reflow temp. You profile doesnt have a soak period. Try to let them soak at 160 for 30-60 seconds and see if it helps.
I also think that solid state would be the way to go here.
Not saying that t wouldn’t be easier to use a solid-state relay or triac or whatever, but I think a relay would work perfectly well here. You could PWM the heaters with a period of, say, 5 seconds and the contacts would last forever. Electrical noise won’t be significant. (It’d certainly be less than from a phase-controlled triac!) It’ll make a click but I guess I don’t see that as being a problem.
Anyway, just wanted to defend the old electromechanical technology a little.
Oh no - the curve was just from a dry run from On to Off. No control, just data. And we’ve never used the toaster to reflow boards. Instead, the $2300 industrial oven has all sorts of problems - all of which preset themselves as poorly reflowed and melted connectors. The hot-skillet doesn’t have any of those problems.
Congrats getting your toaster going! My brother recommended a ceramic brick (similar to your aluminum plate) but I was afraid of adding anything the would retain heat as it would dampen/smooth the rate at which the oven could be ramped/cooled.
We are waiting for the new PCBs to come in with the corrected relay footprint and a ground on the thermocouple (I gotta read the datasheet closer next time).
My limited experience in oven control tells me that the issue with the toaster over may be the top elements. They not only create a wicked thermal gradient inside the chamber, but they will tend to cook things like your prescious connectors cajun style because of the direct heat. Welcome to the joy of IR ovens.
The thing I noticed most “real” reflow ovens (read: convection ovens with inert gas atmosspheres) is that the heat is indirect. I think Mike’s aluminium plate is a good start. I think that top plate with sufficient airflow above it (to make sure the upper elements don’t melt down) would go a long way toward solving your cooked connector issue.
I also think that recirculating the air in the oven would help. It looks to me like the fan on the side of the toaster oven (mislabeled as an AC transformer in the photo takeapart) only pushes air through the oven, which would lead to the rather tepid slope of the DTemp/Dt curve you posted (looks like 0.8dC/sec?). Four 350W elements should be able to heat the oven cavity at something close to 1.5dC/sec without much trouble.
The final thing that comes to mind is something reflective for the door of the oven, like tinfoil, shiny side out. The oven works on IR, and 5 sides of the box are reflective, but the 6th lets it out, mica window or not (likely not, I think the windows are all pyrex derivatives these days). You’d need to keep a close eye on the top and bottom side temperatures, but with dual thermocouples that wouldn’t be hard. I’m debating building a pair of controllers to run the top and bottom elements independently with the same curve. In theory, if I make them dependent on each other, they’ll lock together nicely, and all will go swimmingly.
All this fancy talk is nice (and Great Fun), but I’ve been reflowing in a $19 Target toaster oven for months. 225 for 4 minutes, 350 for 2 and 450 for 1. Granted, no plastic connectors, but it works. Not a single ‘toasted’ board.
I’d like to have a computer controlled oven, but is it really necessary? I doubt it. Do I want one? YOU BET! But let’s keep things in perspective. If your starting with a toaster oven…well…there you have it :roll:
And right now, using a toaster is the last thing on my mind. We are so slammed with assembly, I don’t have the time to develop/modify a toaster oven. We use what works. And that’s the skillet - for us at least.
Well, I tried the “skillet method” last night on a fairly complicated prototype board and it worked fine.
A few problems I ran into:
My skillet bottom isn’t exactly flat so the heating wasn’t as even on the “low” spots. I measured the surface temperature on a dry run before I reflowed my board and was surprised that the heating inside the circle formed by heating element was actually pretty consistent.
I didn’t do such a great job hand pasting the board so I had some touch up work to do. I need to find a cheap and easy way of getting stencils made. I may try the mylar stencils that Sparky and his guys at Sparkfun use. I’ve got inexpensive sources for all the other parts of the protyping chain now – except stencils.
I also was digging through my toaster oven stuff this morning and found that it came with metal “deflectors” that mount over the heating elements. I may have to install them and try again with them installed. Since I have a convection oven, I’ll bet they will help even out the heating inside the oven.
I do extensive surface mount work. The biggest problems are the new microwave chipl like the JEDEC MO-220. These have surface mount connections which are entirely under the chip, o that hand soldering with an iron is impossible.
I built a small hot-plate from a 2X2X.1 sheet of aluminium with 2ea 25 watt 30 Ohm power resistors in parallel attached to the bottom of the plate. This is then supported by 4 stand offs on a wood base. I apply solder paste with an acupuncture needle, put on the parts, and watch the process under a microscope while I apply 40 VDC. (40 X 40 / 15 = approx 100 watts.) This heats to soldering temp in about 2 minutes, and the MO-220 IC’s self center and bond very nicely.
Clearly, there are some dangers, like forgetting to turn the hot plate off, but otherwise it works well.
MGP:
I also was digging through my toaster oven stuff this morning and found that it came with metal “deflectors” that mount over the heating elements. I may have to install them and try again with them installed. Since I have a convection oven, I’ll bet they will help even out the heating inside the oven.
Nice work! I’d love to see some pictures of a plate like that!
We were contimplating building a profile controller for the skillet with a PIC, relay, LCD, thermocouple, etc. The PIC would PWM the skillet and monitor the temperature of the plate, etc.
A little background. We have two toaster oven reflow ovens running. They are made from Oster toaster ovens with all the built in electrical controls bypassed.
120VAC Connection:
For those of you worried about that lethal 120VAC, here is how I solved the problem for my toaster oven controllers.
I bought a surplus SSR from All Electronics. I made a heavy, grounded extension cord terminating in a rounded/square outlet box. I bolted the SSR in the bottom of the box on one side and mounted a regular 2 outlet gang on the other side.
The SSR is hooked up so that it controls just 1 of the outlets. The other outlet is always hot. The always hot outlet is used for the wall wart that powers the controller. The “switched” outlet is used for the toaster oven. You have to have an outlet gang that allows the two outlets to be separated to do it this way. Most of them have a connecting tab you can cut.
Take the digital inputs to the SSR and run them to a plug on the box. I used a 3.5mm audio jack mounted in the cover.
I had to make an outlet cover. I bought one that had a place for a switch on one side and dual outlets on the other. I filled in the switch with some epoxy and then drilled my hole for the audio jack to mount.
On the controller side all you have do do is have a matching jack with a cable to run to the outlet box. Any kind of cable and jacks you have handy will do. You could hardwire it, but being able to disconnect it easily is a plus.
When you’re done you’ll just need to use 1 wall outlet to power the whole thing and the 120VAC is in a grounded box where it belongs.
Burning Down the House:
One of our controllers lost it’s mind one day and just kept heating. It turned out that it was due to a connection that had gone bad where the thermocouple connects to the PC board. The PIC read it as a low temperature and just kept the oven running.
It’s bad enough that it turned 4 boards, with components, into black, crunchy, crispy critters but the real problem is that it could have started a fire. It took half a day to air the place out.
If you build one of these oven controllers, make sure you implement a watchdog timer. Set the trip point for a bit longer than the longest heating state you have. If the controller doesn’t get to the next state within that prescribed time, the watchdog time isn’t touched before the trip time. If the watchdog timer trips the SSR is shut off.
I have a plan for a very simple thermal cutout, with an old-school thermal breaker right in the oven between the line and the coils. An opto will sense when it gets tripped and activate a fail-safe routine to keep things like this from happening. I’ve had thermocouples go bad a few times, and the results can indeed be disasterous.
I still haven’t figured out a foolproof way to PWM the elements however. I’m working on an industrial light dimmer here at work that might be able to donate some topologies, but the dimming board (fets, driver, filtering, etc) will need some uprating and modification to handle the load. If I ever get the hardware built (I just bought a house so things are moving slowly on th ehobby front) I’ll be posting details.
I should have considered the failure modes a bit more completely right up front and had the watchdog timer in there from day one. At least I have it set so that the reset state of the PIC has the bit controlling the SSR turning the relay off.
The thermal breaker, right in the oven, is an excellent idea; even if it isn’t monitored by the controller. In my case that breaker would have to be able to handle 120VAC since that’s all there is at the oven itself.
My plan is to put it on the outside of one of the ‘inside’ oven walls, and have it inline with the AC line feeding the relay. That way, if the oven overheats, it not only turns off the coils, but the power to the relay, effectively de-energizing all of the 120VAC parts of the circuit at once. Also, it won’t be in the direct heat of the coils. If things get too warm, there’s a distinct chance plastics could soften, and I prefer to take the power away completely if that’s even a remote possibility.
I’m also thinking about a Robin Hood/Friar Tuck scheme, where a small, cheap micro would piggyback onto the system and serve as an active check for the main microcontroller, in both watchdog and monitor capacitites. That will no doubt add a lot of complexity to the project, and is thus at the very bottom of my list, behind the differential heater drive and the active fan control, but it will definitely make the project close to bulletproof.
