Logxen

"Ornitottero" ... a mega sized Papilio

8 posts in this topic

  Greetings! I am a contributor to the Smoothieboard project (http://smoothieware.org) and am interested in developing a Papilio-compatible board with higher pincount for cnc control. The basic idea involves bumping the fpga up to a bga package, much like the Pipistrello, but then breaking out 112 pins instead of the standard 48 pins. Here is a first draft concept layout:

 

Ornitottero-brd-pre0.png

 

  Clearly, this is built off of the Papilio Pro as a base, and is meant to be as compatible as possible with existing hardware and software designs. To this end, the lower three 16-bit wing ports A, B, and C match a standard Papilio and should accept existing Megawings. The next port, D, is left free for system concepts such as the built-in LED. The upper 4 ports are assigned ports E, F, G, and H. This neatly fills out the 128-pin potential of Zpuino's standard gpio block. The upper 4 ports are also spaced to allow two sets of inward facing wings without blocking a Megawing in the lower slots.

 

  64MBit flash seems more than sufficient for basic programming even on to lx25 chips (especially with the addition of an sdcard slot) so I figure that can stay an easy to assemble soic. I still haven't really decided if it's really worth bumping the ram up from 64MBit soic to a more expandable bga package...

 

  Other than that I intend to follow the Papilio Pro designs closely, including the upgraded power supply systems. I will probably prototype this to some degree for my own purposes over the next few months... let me know what you all think and whether this is worth turning into a real product! If enough interest is shown I will act on it sooner and maybe even go past prototype. :)

Share this post


Link to post
Share on other sites

Assuming the need to go to multi-layer for the BGA part, that looks like it will be a very expensive board. You might look at a part such as the  XC3S500E-4PQG208I which is in a 208 pin PQFP package. The FPGA is a bit more expensive but the PCB will be a fraction the cost if you can keep it to 2 layers. Also at least when it comes to Seeed, the cost of the board increases rapidly with size. 5cm * 5cm max is around $1/bd, 10cm * 10cm max is $2.50/bd, while the next size up, 15cm * 15cm max is close to $9/bd. If you can manage to keep it within 10cm square then the bare boards are ridiculously cheap.

1 person likes this

Share this post


Link to post
Share on other sites

Yeah, I'm not sure I'd want to prototype bga through iTead (who I'd assume those quotes came from)... Oshpark just upgraded their 4-layer fab to 5/5 traces and 10/4 vias which is actually good enough to 0.8mm pitch bga. The $10/in2 for three boards certainly adds up, but I'm not too worried about prototype cost. If enough people are interested and I do a production run the pcb cost becomes negligible next to the component costs.

 

Edit: Oh I missed that you mentioned Seeed. I have to admit I've not ordered from them before. At any rate I'm happy with the quality of the boards where I currently source them.

Edited by Logxen

Share this post


Link to post
Share on other sites

Here's an example of that exact chip on a pcb prototyped through Oshpark:

azonenbergspartan6bga.jpg

(photo courtesy of azonenberg on irc.freenode.net/#oshpark)

 

And the gentleman that did that board isn't alone... several others in the Oshpark community have been successfully doing bga chips for a while now. With their guidance (and maybe buying myself a reballing kit off eBay) I expect to be able to prototype the board I described without much difficulty. As to the production run my assembly house is fully qualified to assemble bga. I recently e-mailed them and they assured me that their assembly machines support bga chips of this pitch.

Share this post


Link to post
Share on other sites

Oh BGA is certainly possible, I've successfully soldered a couple of them with a hotplate and plenty of flux. My point was just that they usually require a multilayer PCB, and those tend to be far more expensive than regular double sided boards. I haven't looked into quantity pricing though so perhaps it will drop to something reasonable if you build enough of them.

Share this post


Link to post
Share on other sites

Yeah, the multilayer PCB boards become pretty reasonable in quantity. Its the assembly costs on BGA parts that get expensive.

 

Jack.

Share this post


Link to post
Share on other sites

Ok, so I tried decreasing the board size to see what would happen:

 

Ornitottero-brd-pre1.png

 

  On one hand I added the WING_D at Jack's request bringing it up to a neat 128 pins. I did a pincount and that would have used pretty much exactly ALL the pins on the BGA256 chip I was originally targeting (and using all the pins on a BGA256 is a bit iffy on a 4-layer board.) So I went ahead and swapped up to the CSG324 which is going to make routing a lot easier anyway. Other than that and the *squish* this is the same board as above.

 

  I think adding WING_D is a good plan... for one thing it (in theory anyway) would allow one to install a MegaWing facing either direction. I have the two MegaCocoons separated my 0.8" to make sure that MegaWings can fit without blocking access to the other 64 pins. However, I don't think it would fit two of any current MegaWings even with the *squish* design. So, I think it may be best to go back to my original board outline that allows a single MegaWing facing either direction on the lower MegaCocoon, and full use of 8bit and 16bit Wings on the upper MegaCocoon.

Share this post


Link to post
Share on other sites

Posted (edited)

This was spam. Please delete this post

Edited by Manufer
spam

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now