untitled
  • Hey Webmasters! New Photo Album Service Launched - Check it out!
Ghetto Reflow soldering

This web page documents my attempt to create a USB to RS232 converter board using the CP2101 chip from SILabs.

The problem is that my laptop doesn't have a serial port (without buying the $40 dock), but I really need one - for programming Microchip PICs mostly.

This chip requires almost no external components, a simple virtual com driver is provided free from the SILabs website, and it supports very high-speed communication - up to about 1 megabit per second, or 10 times faster than a 16F series PIC can handle. Furthermore, it appears to be the only chip of its kind easily available on the market.

There are USB -> RS232 converters commercially available, probably for pretty cheap too, but it's much more fun to build things yourself, so I didn't even look into buying one. I got two CP2101 chips from digikey for about $5 each and two 'USB Mini-B' connectors for about $2 each. I usually get at least two of anything just in case I screw up. All the other materials I needed I already had.

Materials
Total cost to me: $7, about 20 hours

Tools
*This is the 'secret weapon'. It is just a glorified hair dryer, the airflow is much lower so I suppose the output air is much hotter. I think its intended purpose is for stripping paint.

Procedure
Unfortunately, I took pictures after I was done so this is somewhat incomplete. I have done my best to 'recreate' the steps.
A lot of these images are scaled, in FF you can right click and go "view image" to see the whole thing.

The first thing I did was to get the datasheet for the chip and use it to design a board in Eagle. That is an entirely different article in itself and so, like the magic of TV - I present the finished product! *poof*



Note that my real name is actually on there (backwards! -oops). Thats because Ive decided to specifically claim copyright on this. Just saying. Note that this is not the original high-quality image I used to etch, and it will probably not work if you try to use it.

This image was then printed with an old laser printer onto photo paper. I unplugged it sneakily so that it didn't go through the fuser (heaven knows what would have happened to the photo paper?). I had to do this a couple times before i got a page that looked decently clean and defect-free. As you can see it smudges very easily and even flicking the paper makes some fly off.



This pattern was ironed onto my blank copper and etched in the usual way - sorry, no pictures. :(



Enter CP2102:

(link is VERY big)


I was getting kind of arty with the pictures at this point. Let us take a minute to appreciate the subtle lighting and how the penny compliments the copper board behind it. Or something.

The first thing that strikes most people is how darn small the thing is (Canadian pennies are the same size as American ones, btw). Most people would tell you you are crazy if you try to use a chip this small. Indeed, I wouldn't be writing up this project if it wasn't crazy. But it is so crazy that it just... might... work!

This is a 28-pin QFN chip, here is the back of it:




The problem with soldering this chip is that it has no leads.... everything is underneath the chip. This immediately eliminates traditional hobbyist "soldering iron" methods, and really the only choice left is reflow soldering (unless you have some kind of wave solder machine). My method of reflow is a little less than 'industry standard' however. First off, they usually use "solder paste" which is a gooey paste that magically turns into solder when you heat it up (its flux and solder dust, mostly). You are supposed to stencil it onto your board and place the chip precisely on it, then put it in a special oven that applies a very specific temperature profile to heat the solder hot enough so that it flows.

(This chip dosen't have a temp. profile in the datasheet anyways.)

I don't have any solder paste (or stencils), so instead I pre-tinned both the board and the chip so that there were small beads of solder on both sides of every connection. I then used one of those 'paper-clamps' to set the position of the chip accurately in one direction (lets call it the x-axis), like so (dramatization):



Then it was pretty easy to nudge it carefully until the pins on the 'y-axis' lined up. The first thing I tried to do was just heat it up with the heat gun where it was and hope that it would work.

And it worked... kinda. The chip was crooked, and when I tried to move it I guess I overmoved it a weenie bit... Long story short the pins were shorted all over the place underneath the chip. I simply baked it back off and started over. I did this a few times, it really only takes about 10 minutes to put it on, screw up, take it off and reset once you have the hang of it. I was having problems with the chip floating on too much solder in places, so on the last try, I moved the clamp onto the chip so that it was clamped down pretty hard against the board.

And beautifully, it worked! No shorts and everything looked nice and well connected. I proceeded to add the USB connector and caps so I could give it a plug & pray. I will summarize the next hour by saying that if you get water under a surface mount chip like this, it really matters. Use your heat gun to boil it away :P.

Well anyways, it worked, and passed the 'echo test' whereby i shorted all the signal pairs and used a serial terminal to send data to myself. this confirmed that both directions were operating properly. Then I cranked the speed up and watched the ASCII text fly at amazing speeds. Full Pass. :D. 

Finally, some more pictures to finish it off.

Some of the parts I used:



The board is finished:


Here is some nice close-up action of those solder joints:


And some more: (Link is VERY BIG)


The all-important LED shot:


Web Hosting · Blog · Guestbooks · Message Forums · Mailing Lists
Allwebco Web Templates · Build your own toolbar · Free Talking Character · Audio, Fonts, Clipart
powered by a free webtools company bravenet.com