I wasn't happy with the existing singlesided serial PCB; it didn't look like something
I could make using cheap hobbyist techniques like toner transfer. So I'm working on
a new version that keeps the same board size and "shield connector" placement, but
takes the typical trace width up to 0.8mm, increases pad sizes (restring 45%), avoids
running traces between pins (not that you COULD, with those design rules!) and takes
the minimum clearance up to 0.4mm. It has 9 jumpers (vs 6 in the original), all but one
of which are power/gnd. My initial version has a smt max233 on it, but other rs232
converters should fit. I think.
Is there any interest in this, or is everyone buying their boards?
(it has not been built or tested yet...)
Yes to both questions, but it may not happen very quickly. I'm willing to
provide the Eagle file early to someone else who can test, but not "publish"
them till after they're tested...
I like the idea of a dedicated rs232 transceiver - so much that I'm doing my own redesign to replace the transistor driver with a MAX232. I lost steam on that effort once I realized I'd made 4 or 5 standard single sided serial boards while playing with PCB etching.
The MAX233 has the advantage of requiring no external capacitors, but it's expensive - Digikey's single unit price is over US$10 each. The MAX232 needs 4 external caps, but the TI version at Digikey is less than US$1.
I don't "like" dedicated converter chips; they tend to be very overpriced for what they do. I was just getting frustrated trying to lay out the original circuit, and happened to have some 233s lying around that fit nicely. Pricewise, anybody but maxim has cheaper 5V drivers. I bought some national ds14c232 chips a while ago that are supposed to be pin compatible. But they're still BIG, and have more drivers than we need, and so on. there's the maxim ds275 in dip8/so8 with single drivers in each direction, but it's even MORE expensive. I found on the net a circuit with two resistors and two 2n7000 mosfets that looks promising (and it fits in the space left on the board...)
I found on the net a circuit with two resistors and two 2n7000 mosfets
I have a device that uses the 2n7000 for a 232 driver - I fried three of them in a short period of time; the 2n7000 is quite ESD sensitive. If you're going to use transistors, just stick with the original design, IMO.
I do like a dedicated tranceiver for the job. I've had problems before with devices that play the transistor game to steal voltages from the other end of the connection (besides the ESD death of multiple 2n7000s), but I've never once had trouble with a MAX232 or similar from Maxim, Motorola, Sipex, TI, etc. They're designed for the job, have ESD protection, and "just work".
If size is the issue, there are narrow SOIC versions of the MAX232. I tihnk the $0.90 TI version at Digikey is narrow.
What if a PCB (serial or USB) can be made and sold for USD 5 or less? Do you still need toner transfer method and a complete redesign?
What if you separate the "Development" portion of the board from the "Runtime" portion? A runtime PCB can be sold @ around USD 3.
The rationale is: you design and test one project at a time (using 1 Full Arduino board) and deploy your project using many many Arduino runtime boards.
I designed a runtime Arduino board. I will create a new post with the description of the Arduino runtime (including Eagle files).
What if a PCB (serial or USB) can be made and sold for USD 5 or less? Do you still need toner transfer method and a complete redesign?
Does the board meet my needs? It depends.
Maybe I want a specific set of peripherals attached to the ATmega. Maybe I have specific power needs. Maybe I do/don't need RS232. Maybe I do/don't want USB support. Maybe the board needs to be no larger than a particular size. Maybe I've got stuff lying around to make PCBs Maybe I just want to try to make my own PCB.
These are some reasons I made my own boards, but if I could have gotten a suitable PCB for US$3 - $5 I would have bought the PCB. Of course, I would probably be making Arduino shields instead of ATmega boards.
I don't want to disuade you - the first thing I looked at was a bare serial arduino board from PCB Europe. I don't remember what turned me off; maybe it was the shipping cost, or maybe the minimum order or something.
Just because I may not need them doesn't mean someone else might. And I'm always glad to see someone else's ideas - they give me ideas of my own.
What if a PCB (serial or USB) can be made and sold for USD 5 or less? Do you still need toner transfer method and a complete redesign?
Well, it's not supposed to be a complete redesign; it was supposed to be identical, only with design rules that would work with toner transfer and hand drilling. Otherwise (IMO) there's not a lot of point in having a singlesided design.
But this was at the root of my original question about whether everyone just bought their boards. The "bare-bones Arduino" from ModenDevices is $5; I think it's pretty tough to sell a PCB for $5 and make a realistic profit (enough to support a business, as opposed to enough to buy an occasional bottle of wine.) And an open-hardware effort ought not be dependent on generous people selling off their extras. Besides by publishing CAD files, you allow people not only to duplicate THAT board, but to modify it as appropriate to their actual needs (as someone pointed out.) And on the third hand, this sort of thing is good practice for use would-be PCB designers
Awesome. I think it would be great to have this design available to the community. As you said, even if you can buy a pre-made PCB for $5, there's still value in having the plans for something like this available. Not only does it let you create your own variants, but it provides a good option for people in places where it may be prohibitively expensive to order something from the U.S. Also, some people want to etch their own boards just for the educational value.
I tried designing my own serial single sided board - got too complicated. I found the "official" single sided board, but it doesn't maintain spacing between the female headers, and my efforts of designing something from scratch don't seem to me to be the most efficient way to get to the end result...
Are you interested in having me test your design from toner transfer through to running sketches?
I got a bit distracted doing the Freeduino boards. Professionally manufactured boards are SO pretty.
Yes, I would be interested in a beta-tester. However: what did you mean that the official board "does not maintain spacing" ?! I started with the official board and carefully left the connectors in the same place, so if they were wrong there they're wrong on my design too. (moving silkscreen images around, it looks like they're not QUITE exactly where the USB and serial arduino boards have them, however, looking at the boards I have unpopulated but drilled (CNC drill machine, so the holes ARE in the right place!), they seem to be "close enough" that the (Adafruit) shield I have would fit fine...
OK I take it all back. I swear I had previously printed the single sided board and found it didn't line up with my USB board. I just printed it out from Eagle and it seemed to work fine, so no idea what I did wrong.
Still interested in your version (assuming you are interested), I like your trace widths, I'd have more confidence making it myself.
I printed the .pdf of the single sided serial back before the CAD files were available. When it printed it didn't line up. I think my printer settings must have been wrong - I must have scaled the page or something.
