NetStamp project

XCore Project reviews, ideas, videos and proposals.
User avatar
skoe
Experienced Member
Posts: 94
Joined: Tue Apr 27, 2010 10:55 pm
Contact:

Post by skoe »

Right, that's why I didn't write anything here for a while :)


User avatar
lilltroll
XCore Expert
Posts: 956
Joined: Fri Dec 11, 2009 3:53 am
Location: Sweden, Eskilstuna

Post by lilltroll »

What is the latest status with the stamp :?:
Probably not the most confused programmer anymore on the XCORE forum.
User avatar
Folknology
XCore Legend
Posts: 1274
Joined: Thu Dec 10, 2009 10:20 pm
Contact:

Post by Folknology »

I am making some final small tweaks this week primarily on the paste and silk screen layers, I would then like to get some feedback from the Xmos engineers before making the first prototype. I am worried about L2 placement and am looking for some help in assembling the first one, any suggestions here from the experienced folks would help (particularly in the Surrey/Hants/London area). I have most of the components ready, although I am having issues sourcing the Ethernet RJ45s because the only supplier has a bulk policy, but that will be resolved even if I have to buy a boatload its basically a negotiation.

Also this is a good time for any further feedback from folks on this thread. Anything You noticed that I need to change, any recommendations, any subtle mistakes please let me know.

Regards
Al
User avatar
lilltroll
XCore Expert
Posts: 956
Joined: Fri Dec 11, 2009 3:53 am
Location: Sweden, Eskilstuna

Post by lilltroll »

Lets soon make a new render then - so we can help others get the overview.

Last week I tried "auto-placement" with solderpaste, a small single-row QNF and hot air.

I was curious if I really could misalign the circuit and that it would float pack to the perfect position by it self (as seen on youtube) - but it did every time and it was like magic.
I didn't use any stencil - I used solderpaste on the centralpad and very little on the other pads. That way I could put some extra solder by the iron on the small pads on the side.
But the "autoplacement" worked great.

I believe that it's almost impossible to fit the L2 without a stencil - if you do not apply the trick with manual "re-balling" of the L2 with help of low temp airheat an raising the paste with a small tool.

But as a summary - maybe you do not need to worry about the alignment at all, it's not as then using the iron.
Probably not the most confused programmer anymore on the XCORE forum.
User avatar
Folknology
XCore Legend
Posts: 1274
Joined: Thu Dec 10, 2009 10:20 pm
Contact:

Post by Folknology »

Thanks Mika

I will definitely be using a stencil and oven, I would not even attempt hot air stations and manual soldering for the L2 as I wish to maximise chances of success! Thus I am looking for someone with such facilities in the area (1st preference) or a commercial company that can do so at reasonable prices.

regards
Al
User avatar
lilltroll
XCore Expert
Posts: 956
Joined: Fri Dec 11, 2009 3:53 am
Location: Sweden, Eskilstuna

Post by lilltroll »

Leon had some good contacts for that i UK i believe - somewhere here he posted a link to a UK company that was specialized in fitting components on prototype boards.
Probably not the most confused programmer anymore on the XCORE forum.
User avatar
leon_heller
XCore Expert
Posts: 546
Joined: Thu Dec 10, 2009 10:41 pm
Location: St. Leonards-on-Sea, E. Sussex, UK.
Contact:

Post by leon_heller »

I think it was ASK:

http://www.asktechnology.co.uk/About%20ASK.htm

I've used them a couple of times.
User avatar
Folknology
XCore Legend
Posts: 1274
Joined: Thu Dec 10, 2009 10:20 pm
Contact:

Post by Folknology »

Thanks Leon I will check them out, they are just up the road from me infact.
kster59
XCore Addict
Posts: 162
Joined: Thu Dec 31, 2009 8:51 am

Post by kster59 »

This is a very interesting project and I just came across it.

I have several module designs of my own that I am working with and plan an L2 version.

What is your estimated selling price for an assembled unit?

I am surprised that XMOS doesn't provide a module type board.

First comment:
Obviously you will not be able to make everyone happy but my suggestion/suggestion is try to make it use standard PDIP 40 (.100" pins with .600" spacing). I think if you professionally assemble it you can get everything into that format with components on both sides.

An easier way to do is to move the 40 pins interior from the edge. Then the board will be bigger than the pins underneath There is no reason that the 40 pin headers need to be on the edge but a very good reason to making them interior is so that you can use low cost 40 pin sockets or 40 pin Zif sockets.

The 40pin propstick USB is incredibly useful but I have moved away from propeller in favor XMOS.

Another option which I use on my current L1-128 design is 40 pin .100" separation. This is the same as 40 pin IDE cables which makes very inexpensive cabling and connection. 40 pin sockets with .100" headers is also inexpensive. It is also very easy to mount as a module with standard breakaway headers and female sockets.

Second comment:
For a general low cost board, I really encourage you to redesign to use the FT2232D chip. I would hookup the FT2232D in FT245 mode for 12mbps data. There are two big reasons 1) onboard JTAG doesn't require an XTAG2 2) FTDI drivers are available for multiple OS for USB. Have you successfully written any programs that make use of the XMOS ULPI USB for high speed communication between a PC and the XMOS device? I'm not sure if it has changed but last time I checked XMOS didn't even have the USB to UART drivers available. USB to UART probably won't be that fast (FTDI drivers in UART mode are limited to about 1.2mbps) which means someone would need to write PC side drivers. I would be willing to bet good money nobody is going to write these drivers and thus the XMOS ULPI would not be utilized. Even if you do write the PC drivers updating the drivers is a nightmare for multiple OS (mac/linux/Windows XP/Windows 7/Windows 7 x64, etc) and you will likely need to write new drivers for most OS updates. This will severely limit the commercial potential of your board.

I am thinking of developing my own PDIP 40 or possibly dual PDIP 40 board based on a L2 design but I will probably just wait and buy yours if you are almost finished! The Xstamp and others never went commercial available and just kind of died so I hope you have more success!

These are just my comments and of course it is your project so feel free to ignore them.

Sign me up if you need another beta tester.
User avatar
Folknology
XCore Legend
Posts: 1274
Joined: Thu Dec 10, 2009 10:20 pm
Contact:

Post by Folknology »

Hi kster59
Obviously you will not be able to make everyone happy but my suggestion/suggestion is try to make it use standard PDIP 40 (.100" pins with .600" spacing). I think if you professionally assemble it you can get everything into that format with components on both sides.
Thanks for the feedback, the dimensions are unlikely to change at this point as I am too close to prototype. I would love to get the dimensions down to a 0.600 spacing but doing so is a real challenge. Yes it would require components on both sides but even with that it would be tough, it would also need more layers and it might still be near impossible without removal of certain components. This design if actually a hardware reference design for Amino and as such needs certain features, if those dimensions were to be met it would almost definitely be for a lower requirement. If you look at the routing for the L2 particularly including Ethernet phy it is very tricky indeed, but may be possible. However I wouldn't rule out a slimmed down design in future perhaps meeting reduced design requirements, but some development works needs completing first to make that feasible.
For a general low cost board, I really encourage you to redesign to use the FT2232D chip. I would hookup the FT2232D in FT245 mode for 12mbps data. There are two big reasons 1) onboard JTAG doesn't require an XTAG2 2) FTDI drivers are available for multiple OS for USB. Have you successfully written any programs that make use of the XMOS ULPI USB for high speed communication between a PC and the XMOS device? I'm not sure if it has changed but last time I checked XMOS didn't even have the USB to UART drivers available. USB to UART probably won't be that fast (FTDI drivers in UART mode are limited to about 1.2mbps) which means someone would need to write PC side drivers. I would be willing to bet good money nobody is going to write these drivers and thus the XMOS ULPI would not be utilized. Even if you do write the PC drivers updating the drivers is a nightmare for multiple OS (mac/linux/Windows XP/Windows 7/Windows 7 x64, etc) and you will likely need to write new drivers for most OS updates. This will severely limit the commercial potential of your board.
I have looked at FTDI options for Xmos designs before, they actually recommend the Xtag2 route for new designs, but I understand the advantages of supported multi-platform driver support for things like virtual UART over USB having used such things before, I continue to stress that this really should be supported by Xmos's Xtag2 solution as many developers benefit from this sort of functionality. I am also hoping to use a different angle to solve the kinds of problems that those features help with but you will have to bear with me on those until I get some real hardware.

The reason I include the ULPI is two fold, first in the hope that drivers may be developed for all sorts of USB applications and the second is to support audio applications for which drivers already exists. As you are probably aware audio applications are an Xmos sweet spot and I didn't want to exclude the audio developments from NetStamp's applications.


As for cost I am still talking to suppliers and assemblers at this point so I cannot be certain. Clearly I want to make it as affordable and competitive as possible to reach the largest audience, obviously costs are also a function of volume and so its success will also determine its eventual purchase price.

I do welcome your feedback

regards
Al
Post Reply