Ronny,
how do you place a two sided board in the oven? I know someone mentioned the solder paste will hold the components in place on the bottom but do you just place the board directly on the metal grill or hoist it up on buffers along the edges or mounting holes? (if so, what material do you use for the buffers?)
kaiki:
Ronny,how do you place a two sided board in the oven? I know someone mentioned the solder paste will hold the components in place on the bottom but do you just place the board directly on the metal grill or hoist it up on buffers along the edges or mounting holes? (if so, what material do you use for the buffers?)
A lot will depend on the size of the board(s) in question. The last one I did had most of the components top side. The few bypass caps on the bottom actually fit between the grill wires if I positioned it just right. If that doesn’t work, or is just too cheesy, get the wide spring clips used for holding papers together. 2" wide works well for me. clip them to two opposing edges, and lay them perpendicular to the grill wires. You can put in a number of boards this way, and it lifts them about 1/4" above the grill. No plastic please…
Yeah - you just have to support the previously reflowed side off of the grill (or whatever it is sitting on) so the components are not under stress and move when they reflow. This metod really relies on convection to reflow, so it’s best left to the oven guys and not the hotplate guys 
I personally use FR4 material to support the boards - you can gat blocks and channels from various sources.
Good luck!
Steve
Are there any problems if the bottom side of the board is mostly copper/ground plane either delaminating or preventing enough heat to transfer to the top side and melt the solder?
My board is 2.5" x 4", double sided. All the components are topside
and I’m using leaded solder.
sd_proto:
Are there any problems if the bottom side of the board is mostly copper/ground plane either delaminating or preventing enough heat to transfer to the top side and melt the solder?My board is 2.5" x 4", double sided. All the components are topside
and I’m using leaded solder.
Nearly all of my boards are mostly copper on the bottom and I’ve never had a problem.
The preheat phase is the key here. By raising the overall board temperature to a temperature that will not kill the components short term, you will shorten the time that the board needs to stay at the potentially fatal reflow temperature in order to thermally saturate the board and components for proper solder adhesion. The cool down phase is important too. Too slow and failure may occur, but too fast and you can get some thermal cracking and separations internally on some components. Stay as close to the manufacturer’s recommendations, and all will be well.
Ron
Thanks for answering my question and for the valuable advice.
Ronny, I just wanted to be clear about your comment of following the manufacturer specs. Do you mean for the type of solder or for the components?
I’d expect it has to be a happy medium of everything, but there can be so many potentially different component with various specs.
How are you measuring the temp? I hadn’t planned to get a thermometer if it wasn’t necessary, but if it will help, I will.
Do any of you have any general profile that you follow, or do you just get a hang for it after a few times?
A couple of hotplates are locally (The Netherlands) are available, but most of them go up to 215 degrees Celsius.
Would that be hot enough to reach the reflow temperature?
JD
Edit: Found this tutorial http://www.sparkfun.com/commerce/presen … %20Skillet that said that at 225 degree Celsius the USB component was reflowed completely…
I think this answered my question. A temp of 215 degrees C should be enough for most components…
Most of the sloder reflows at around 185C, so, you should be ok.
While cooking off a SMD board is really a good way to get started, a more advanced method is to use a small toaster oven. The 550W oven I chose has two quartz heater elements, one above and one below. The goal is to be able to do double-sided SMD reflow in one shot, or maybe two-pass.
The current design uses a PIC18F252 as the brains. An AD595 is used to interface the thermocouple, which outputs a nice 10mV/Celsius signal. The board also uses a hacked Nokia LPH7366 LCD panel 84x48, and supports an RS-232 port, and a 5 button UI.
The controller uses a 20A solid-state relay with zero-cross switching. A later version might use two of these relays to control the upper and lower heaters independently while monitoring two thermocouples.
I hope to get my first spin of the controller board back in a couple of weeks, so we’ll see.
sounds pretty good. 2 questions:
Would you describe your UI? especially how you used the buttons.
did you take a look at the max6675 instead of the AD595? I like what I see in it though it requires wires.
Philba:
sounds pretty good. 2 questions:
Would you describe your UI? especially how you used the buttons.
did you take a look at the max6675 instead of the AD595? I like what I see in it though it requires wires.
Not sure what you mean about the AD595. It requires no external components for 10mV/C operation, save power decoupling and obviously a K-type thermocouple. To wire it up is simplicity itself… power ground, thermocouple and the vout… I also implemented the alarm via the built in OC driver, useful for signaling to the micro that the thermocouple has failed.
The UI is multi function, and modal, via the center button(enter/mode)… the U/D/L/R cluster is use for selection and parameter modification. Feedback on UI is given on the LCD. Such minimalist interfaces are easy to implement.
The goal is to allow the user to select from a few default profiles in Program ROM, or a few custom ones stored in the PIC’s parameter FLASH.
Or create a new profile in a simple ‘Profile Slice Editor’ by allowing the user to draw the profile curve, and set “hold-points” and “key-slices” on the LCD, using the arrow keys. A later version might also include support for injecting external air via a small fan to speed up -∆T ramps.
The core thermal tracking process is a simple P+D controller. I am leaving out the ‘I’ component (P+I+D) since it can easily cause overshoot and low frequency ‘ringing’ in thermal applications. ‘I’ is only needed if you need very fast ‘seeking’ to the target parameters, and ‘Integral Windup’ is acceptable. I may also implement a simple ‘sequenced hysteresis’ controller which won’t need as much tuning as the P+D controller, and is rather independent of the oven’s specific thermal capabilities.
Additionally the RS-232 interface will support high-level and low level direct access to the controller, as well as profile management, performance logging, and code updates via a simplistic boot-loader.
Hope that answers your questions.
BTW: a screenshot of the PCB pattern is posted here:
metaforest:
Not sure what you mean about the AD595. It requires no external components for 10mV/C operation, save power decoupling and obviously a K-type thermocouple. To wire it up is simplicity itself… power ground, thermocouple and the vout… I also implemented the alarm via the built in OC driver, useful for signaling to the micro that the thermocouple has failed.
I assume from your comments that you didn’t look at it. I typo’d my comment and meant to say “requires 3 wires”. The Max6675 has the ADC on chip and gives 12 bits of resolution vs 10 on the PIC. Plus it means that the analog signal doesn’t spill onto the PCB where it can pickup noise. It supports detecting open thermocouple as well. Comes in a SOIC8 package. The SPI interface is very easy to use but it does take up 3 pins on the PIC (vs 1 for the 595).
By the way, there’s a microchip app note that uses opamps and the PIC’s voltage ref to do cold junction compenstation. It costs about 1/4 the the 595 or 6675 but seems fairly complex to use. It’s not clear how accurate it is, either.
Philba:
I assume from your comments that you didn’t look at it.
Actually I looked at it a long time ago… when I first spec’d this controller design. It might ease the analog issues, but it complicates the SPI driver in my system due to sharing the SPI bus with the LCD. Also after looking at it again I see that the ADC in that chip is pretty noisy. 2LSB of noise at ambient and 3LSB noise at 85C… that sucks… and my board does have to operate in the oven’s chassis… temps are likely to be around 60 - 70C… so there goes the 12 bit advantage…
The PIC18F252’s ADC has a noise spec that shows 0.5LSB from -40C - 125C, and a 10mV/C signal is not hard to keep clean…
I based the design on what I have on hand, and wanted to avoid SMD parts. This design is a boot-strap. Might revisit the design after I have baked a few goodies with the current design.
Thanks for the input it is very much appreciated. 
I understand your dislike of the 6675’s ADC spec but the PIC noise spec doesn’t include the fact that the trace from the 595 to the PIC will pick up a pile of noise. I’ve always found that even with really good design, the results on a PIC were no where near the spec. Typically, I’ll see 1.5 to 2 LSBs of noise. A reflow oven is a fairly noisy environment.
Also, you have to keep in mind the total number of bits involved. 2 LSB on a 12 bit ADC beats .5 LSB on a 10 bit ADC. Assuming all else is equal, I believe that the 6675 will be more accurate than the 595+PIC. Not that it’s going to matter THAT much as a few degrees error won’t make much of a difference and you can filter out a fair chunk of noise.
Philba:
I understand your dislike of the 6675’s ADC spec but the PIC noise spec doesn’t include the fact that the trace from the 595 to the PIC will pick up a pile of noise. I’ve always found that even with really good design, the results on a PIC were no where near the spec. Typically, I’ll see 1.5 to 2 LSBs of noise. A reflow oven is a fairly noisy environment.Also, you have to keep in mind the total number of bits involved. 2 LSB on a 12 bit ADC beats .5 LSB on a 10 bit ADC. Assuming all else is equal, I believe that the 6675 will be more accurate than the 595+PIC. Not that it’s going to matter THAT much as a few degrees error won’t make much of a difference and you can filter out a fair chunk of noise.
:lol: Are you a MAXIM Rep.? (jk)
I only need 5 samples/sec. So if the noise becomes an issue I’ll just over sample the heck out of the input, apply a steep IIRF to the ADC output and 64:4 interpolator to get mah bits back, and then some. I have plenty of headroom so I don’t really need to worry much about noise, and the App doesn’t require better than 5C accuracy anyway. :lol: