XML1

[rp181]

XMOS Minimalist Board - L1

As most of you know, I have been working on the board. At this point, everything works - but the flash (an easy fix, just a wrong package). The board features:
- L1 Single Core Processor - 500 MHz
- 1x 8 bit port
- 8x 1 bit ports
- On board Flash
- XTAG Compatibility (via breakout)
- Castellated holes
- Very small size

The main processing board (not final, a few minor modifications have been made):
http://api.ning.com/files/WOcd0cy41LJTN ... llated.jpg
The board measures 1.6" x .7625" (Smaller than the XTAG) and has castellated holes (maybe non-castellated version), allowing for the soldering onto parent boards (containing periphrials).

This board can be soldered to a breakout board:

measuring 2" x 1.1" (Breadboard compatible), providing IO and JTAG for the XTAG2.

Right now, I am unsure how to have a temporary board simply for programming - maybe just holding in the headers?

As of right now, for 100 boards, pricing stands at:
~40$ for the main processing board
~45-50$ for the processing and the breakout

Thoughts?

EDIT: For a sense of size, this is the first prototype (which is a teensy bit bigger)
http://api.ning.com/files/jMp9-Jstzmuld ... /ravi5.jpg
http://api.ning.com/files/q6qtrPNPGGOha ... /ravi1.jpg

[Interactive_Matter]

What I like:

The minimal size - it is easy to integrate it into existing circuit - that is great

What I don't like:

Would a XTAG port on the base board itself make sense?
This would make it more versatile.
Pogo Pins or a small connector (JST style or something like this) would be great.

[TonyD]

rp181, the jpegs are not working. png works

[rp181]

The jpegs are just links, they seem to be working fine for me. Anyone else having an issue?

I was thinking, for the XTAG, to try an stagger the holes a bit, so a header could easily, and temporarily, pressed in.

EDIT: I pushed all the components closer together, so I will be adding an SMD header to the top of the board, that will connect to the JTAG. This way it can be programmed in system.

And from a price standpoint, $37 for a baseboard and $7 for a XTAG breakout and Breadboard compatible is looking feasible (or $43 for a complete set).

[Corin]

Hi rp181,

I've just had a look at your design - nice work, but there are a few issues that I've noticed:
- The power supplies are not sequenced. Please see: https://www.xmos.com/published/xs1-l-ha ... -checklist
- R11 / R12 are not required.
- MODE[1:0] should be tied high to 3V3 for your oscillator configuration rather than left floating.
- The 1V0 core power decouplers need distributing evenly around the L1.
- The 3V3 rail around the chip could do with a couple more 100nF decouplers.
- The ground plane on the board is sub-optimal - particularly under the L1 device - all the return current goes through a very small path back to the ground pin which can cause quite a few issues on the board. e.g. high PDN impedance, chip instability, EMC and cross-talk issues.

I hope this helps.

Kind Regards,
Corin

[jonathan]

Corin - thanks on behalf of the community, this kind of feedback is very important and I am sure it is very valued.

[rp181]

Thank you for that list!


-The power supplies are not sequenced. Please see: https://www.xmos.com/published/xs1-l-ha ... -checklist
They are actually sequenced. The 1V reg has an RC circuit on the enable pin that gives a excellent delay. Subsequently, reset triggers are supplied. Signals look great on a scope!
- R11 / R12 are not required.
I was doubtful of this, thanks for the confirmation!
- MODE[1:0] should be tied high to 3V3 for your oscillator configuration rather than left floating.
Both MODE1 and MODE0 are tied to 3v3. MODE 2,3 are tied to TRST_N.
- The 1V0 core power decouplers need distributing evenly around the L1.
I will work on this. I had placed them hastily just trying to get them of the bottom.
- The 3V3 rail around the chip could do with a couple more 100nF decouplers.
Will work on this too!
- The ground plane on the board is sub-optimal - particularly under the L1 device - all the return current goes through a very small path back to the ground pin which can cause quite a few issues on the board. e.g. high PDN impedance, chip instability, EMC and cross-talk issues.
I was under the impression I wanted a small return path. I can remove quite a bit of via's to lengthen the return path.

EDIT: I see what you mean by a small path, I confused it with a short path. That has been difficult. Will board houses solder to a pad with a via in it?

EDIT2: Here are the eagle files. Having trouble putting decoupling capacitors on the "north" corners.
Note that the FLASH in this is wrong at the moment.

[Corin]

Ah - I think I had the old version

For the ground plane - ideally the whole of the bottom of the PCB should be a solid ground plane with a few a slots (caused by tracks) as possible.

Kind Regards,
Corin

[Bianco]

Hi rp181,

- MODE[1:0] should be tied high to 3V3 for your oscillator configuration rather than left floating.

I hope this helps.

Kind Regards,
Corin

The mode pins have internal pull-ups that are strong enough to be left floating?

[Corin]

Bianco,

This is correct, but we still recommend that they are tied, rather than just left floating.

Kind Regards,
Corin