On Fri, 2011-01-21 at 22:24 +1100, Erik Christiansen wrote: ... snip > Kirk, are you thinking of using something like an A4 xmega, in a 44 pin > TQFP, for ease of soldering, or going for more I/O? (Eagle has 44 and > 100 pin TQFP macros.) Ah, I see the xmega100 breakout board. It adds $19 > to the chip price. We can have a prototype PCB done for that, if it fits > within 6.3 x 4 inches. Still, I've only hand soldered TQFP44, using a > hollow-tipped iron. That was so easy. TQFP100 is said to be manageable > as well. > > One option is to lay out the PCB for the breakout board _and_ an xmega100 > within that. Then it can be built either way.
I was trying to use an ATmega because they are 5V chips, but I kept running into a problem of having the different functions I wanted being on the same pins. The Xmegas seem to have less pin function sharing. I am thinking the first batch of boards should have just the parallel port interface, power supplies and other essentials to get the uC to run, then break out the rest of the pins, so peripheral development could continue. The more pins available would allow for more flexibility in how the uC could be used. For the soldering issue, I am thinking that there are people on the list that already have the skill and tooling, and might be willing to run a batch or two for a nominal fee. "Many hands make light work" so if (big IF) a batch of boards gets made, there may be an inquiry made for soldering them. > I see that "Absolute Maximum ratings" won't allow feeding in 5v signals, > but TTL levels would just about be OK if Vcc is near 3.6v. Some > resistive dividing could add a safety margin, but clamping to Vcc (e.g. > 3.3v) via schottky diodes, with resistors on the inputs, would provide > more consistent input voltages with a greater variety of PP > implementations. > > For xmega to 5v conversion, a common emitter transistor with (say) 4k7 > on the base, and 1k on the collector, would suffice. If there are no > more than 5 xmega -> PP signals, then a small transistor array like the > LM3046 would save fiddling with individual transistors. (An SMD LM3046 > is convenient, individual SOT23 transistors are a pain, I find.) I would like the parallel port interface to handle 5V or ~3V I/O from the PC. The peripheral signal voltages could vary widely, but I tend to think, in order to combat noise, 12 or maybe 24V DC or AC would be more appropriate in a CNC environment, but would need to be addressed according to the application. On the other hand, my vision of the board's purpose is to handle only fast signals, such as steps, PWM and encoder input, so the higher voltages may not be appropriate. Slow (higher voltage) signals could be handled by a standard parallel port and buffer/isolator board. I would also like to follow best or better design practices. So I will need those that know better, to provide guidance. To that end, I also tend to copy other designs that are common for an application. So far, I have only looked at this chip: http://focus.ti.com/lit/ds/symlink/sn74lvc4245a.pdf but currently, I see that understanding the EPP software as my next goal. I have an ATmega setup that I can use to get an EPP component sorted out, then I can move this to the Xmega. > > If there is enough interest, it might be worth while to have a batch of > > PC boards made and/or maintain a parts list or kits. > > It costs so little to have even one made, that I'd do schematic capture, > have several of us check the design, do the layout, send the eagle file > off, and have a cheap first prototype back inside a week. If a track or > two need to be cut, and a wire run, then that's normal. > > Since an ATxmega128A1 is about $11, the whole interface PCB plus CPU can > be had for the same money as the xmega100 breakout board. (If out > eyesight is up to it. :-) > > Just some initial thoughts. > > Erik A few years back (maybe more) I had Sparkfun make some boards for me. The price was very good, but I had to wait a few weeks for them to fill a panel, then send the job to China, have the board house find a space in their normal production schedule, then ship the panel to SF. SF would then break the panel out and ship. It seemed to be the best deal going at the time, but I guess hobby fab has come a long way since then? By the way, I tend to prefer GEDA for schematic and board layout. I had a hell of a time making new components in Eagle, which I find easy in GEDA. On the other hand, GEDA has very few pre-made components so one ends up making a lot more of them. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA ------------------------------------------------------------------------------ Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)! Finally, a world-class log management solution at an even better price-free! Download using promo code Free_Logger_4_Dev2Dev. Offer expires February 28th, so secure your free ArcSight Logger TODAY! http://p.sf.net/sfu/arcsight-sfd2d _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users