From: Gordon K. <go...@gu...> - 2005-02-22 00:48:14
|
OK- Here is what I'm working on for Robostix. Boards are basically ready-to-go other than your commentary. For images see: http://www.gumstix.org/tikiwiki/tiki-index.php?page=+Robostix Pertaining to past feedback: 1) You'll see that we are thinking of using SPI in a way that would allow a software download over SPI (NSSP on the gumstix side) if you take advantage of the Power-on-reset autodownload over SPI. On this board we have a LDO between the V_BATT (Power Jack) and the supply to the ATmega128. So: dropping L_DD14 will drop the AT power enable, and dropping L_DD13 will drop nPROG_EN. If the programming_enable line is low when the chip powers back up it will download boot code over the SPI. 2) Errata pertaining to the schematic passed: the high-side supply (5V) the voltage-transceiver will be pwered directly by V_BATT and not Vcc5 as noted; ATM_PROGEN will not be tied-high because of an internal tie-high. 3) What will be most helpful is commentary on the selection of connectors. All that is shown are the pads and at this point, tin-posts are contemplated (e.g. Molex 53047-- http://dkc3.digikey.com/PDF/EU042/0037.pdf ) 4) I have transceibvers for 2 more lines to signal from pxa to at128 and 1 from at128 to pxa255. Preferences? 5) Sadly-- no stackability on this go-round. Frankly, I'm interested in how the I2C support goes before deciding between the number of choices for board-addressing amongst daughtercards. 6) Choices made: a) No rs232 level from the atmega to the world, and b) No uart connection between the atmega and pxa255, and c) The form factor is the same as the Waysmall boards. Thanks, Gordon |
From: Dave H. <dhy...@gm...> - 2005-02-22 03:40:44
|
Hi Gordon, Looks pretty good. Comments below. On Mon, 21 Feb 2005 16:47:40 -0800, Gordon Kruberg <go...@gu...> wrote: > OK- Here is what I'm working on for Robostix. Boards are basically > ready-to-go other than your commentary. > > For images see: > http://www.gumstix.org/tikiwiki/tiki-index.php?page=+Robostix > > Pertaining to past feedback: > > 1) You'll see that we are thinking of using SPI in a way that would > allow a software download over SPI (NSSP on the gumstix side) if you > take advantage of the Power-on-reset autodownload over SPI. On this > board we have a LDO between the V_BATT (Power Jack) and the supply to > the ATmega128. So: dropping L_DD14 will drop the AT power enable, and > dropping L_DD13 will drop nPROG_EN. If the programming_enable line is > low when the chip powers back up it will download boot code over the SPI. Sounds reasonable. The schematic shows an ATMega103. I'm assuming that this is just because they have the same pinout as the ATMega128? > 2) Errata pertaining to the schematic passed: the high-side supply (5V) > the voltage-transceiver will be pwered directly by V_BATT and not Vcc5 > as noted; ATM_PROGEN will not be tied-high because of an internal tie-high. I'm confused. Are you saying that the i2c transceiver WILL be powered by V_BATT? If so, doesn't that make it pretty much useless? Because V_Batt will have to be somewhat higher than 5v in order to power the regulator, and then the i2c bus won't work with devices that don't like voltages above 5v. Also, for part U53, what is it? I'm guessing it's a resistor array? If so, why are pins 5 and 10 unconnected? It seems like it would be a voltage divider the way it's shown (maybe there's a hidden connection to something). > 3) What will be most helpful is commentary on the selection of > connectors. All that is shown are the pads and at this point, > tin-posts are contemplated (e.g. Molex 53047-- > http://dkc3.digikey.com/PDF/EU042/0037.pdf ) It would be really useful to have a row of Vcc alongside the ADC connectors. Many analog sensors need this. I would be good if an external connector could be made available to supply this. I really like being able to plug in sensors without having to make adapter boards, just so I can get the wiring right. Possibility 1 - use the same spacing of connectors that you have now, but have an extra row of Vcc. Then 3 pin headers could be used. Possibility 2 - Bring out as many individual 3 pin connectors (like the molex ones) that you can fit, with one analog signal per 3 pin connector. I'm assuming that the servo connectors could to use 0.1" headers that would go out the bottom of the board? It would be useful to have ground/Vcc available for the ATM_PCx and ATM_INTs too (I figure I might as well ask!) It would be good to have a 4 pin connector for the UARTs so that an external level shifter could be used (like http://www.acroname.com/robotics/parts/S13-SERIAL-INT-CONN.html) I noticed that PA0-7 aren't connected anywhere. Would it be possible to have some solder pads on the bottom or something to make these easy to get at? I'd rather not have to solder directly to the pins :) Looking at the schematic (and board layout), it appears that there aren't any pullups on the gumstix side of the i2c connection. I believe that these need to be pulled up 3.3v. > > 4) I have transceibvers for 2 more lines to signal from pxa to at128 and > 1 from at128 to pxa255. Preferences? I see that there is also one unused pin going in the other direction. PE2, PD5, and PB4 would be likely candidates. Perhaps having PB4 going from atmega to pxa. > > 5) Sadly-- no stackability on this go-round. Frankly, I'm interested in > how the I2C support goes before deciding between the number of choices > for board-addressing amongst daughtercards. > > 6) Choices made: > a) No rs232 level from the atmega to the world, and > b) No uart connection between the atmega and pxa255, and > c) The form factor is the same as the Waysmall boards. That's all I can think of for now. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: mark g. <mar...@th...> - 2005-02-22 04:30:00
|
On Monday 21 February 2005 16:47, Gordon Kruberg wrote: > OK- Here is what I'm working on for Robostix. Boards are basically > ready-to-go other than your commentary. > > For images see: > http://www.gumstix.org/tikiwiki/tiki-index.php?page=+Robostix > > > Pertaining to past feedback: > > 1) You'll see that we are thinking of using SPI in a way that would > allow a software download over SPI (NSSP on the gumstix side) if you > take advantage of the Power-on-reset autodownload over SPI. On this > board we have a LDO between the V_BATT (Power Jack) and the supply to > the ATmega128. So: dropping L_DD14 will drop the AT power enable, and > dropping L_DD13 will drop nPROG_EN. If the programming_enable line is > low when the chip powers back up it will download boot code over the SPI. I've never used SPI, is it hard to work with? > > 2) Errata pertaining to the schematic passed: the high-side supply (5V) > the voltage-transceiver will be pwered directly by V_BATT and not Vcc5 > as noted; ATM_PROGEN will not be tied-high because of an internal tie-high. > > 3) What will be most helpful is commentary on the selection of > connectors. All that is shown are the pads and at this point, > tin-posts are contemplated (e.g. Molex 53047-- > http://dkc3.digikey.com/PDF/EU042/0037.pdf ) > > 4) I have transceibvers for 2 more lines to signal from pxa to at128 and > 1 from at128 to pxa255. Preferences? > Hook them up to some bit bangable GPIO's. Preferably ones that can generate interrupts on either pxa or atmega. > 5) Sadly-- no stackability on this go-round. Frankly, I'm interested in > how the I2C support goes before deciding between the number of choices > for board-addressing amongst daughtercards. > > 6) Choices made: > a) No rs232 level from the atmega to the world, and > b) No uart connection between the atmega and pxa255, and Without a serial communications between the atmega and the PXA you are making the software development for the atmega program MUCH harder. I urge you to reconsider this. ttyS1 should talk directly to the TX/RX UART on the atmega. > c) The form factor is the same as the Waysmall boards. This is too small! If marketing is telling you to keep it that small tell them the user need access to most all the signals coming from the atmega part to interface with the robotics gear. Giving up on those just to keep it in the form factor of the Waysmall board is a crime. > > Thanks, > > Gordon > > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real users. > Discover which products truly live up to the hype. Start reading now. > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Dave H. <dhy...@gm...> - 2005-02-22 06:03:03
|
Hi Mark, > > 1) You'll see that we are thinking of using SPI in a way that would > > allow a software download over SPI (NSSP on the gumstix side) if you > > take advantage of the Power-on-reset autodownload over SPI. On this > > board we have a LDO between the V_BATT (Power Jack) and the supply to > > the ATmega128. So: dropping L_DD14 will drop the AT power enable, and > > dropping L_DD13 will drop nPROG_EN. If the programming_enable line is > > low when the chip powers back up it will download boot code over the SPI. > > I've never used SPI, is it hard to work with? Not particularly. It's basically just another bidirectional communication mechanism. It's actually simpler to deal with than i2c. Here's an app note from Atmel: http://www.atmel.com/dyn/resources/prod_documents/doc2585.pdf This app note talks about doing in system programming over SPI: http://www.atmel.com/dyn/resources/prod_documents/DOC0943.PDF From my quick review of the pxa datasheet, any of the GPIO pins can be configured to generate an interrupt. So, you should be able to do anything that can be done over a serial port using the SPI port. > Without a serial communications between the atmega and the PXA you are making > the software development for the atmega program MUCH harder. I urge you to > reconsider this. ttyS1 should talk directly to the TX/RX UART on the atmega. Why do you think that the serial port is so important? I'm just trying to understand the issue. > > c) The form factor is the same as the Waysmall boards. > > This is too small! If marketing is telling you to keep it that small tell > them the user need access to most all the signals coming from the atmega part > to interface with the robotics gear. Giving up on those just to keep it in > the form factor of the Waysmall board is a crime. I'd certainly agree that it would be worth having a bigger board, especially if the pins could all be brought out in a power/ground/.signal pin arrangement using 3 rows of 0.1" headers. This would make interfacing for the hobbyist MUCH easier. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Craig H. <cr...@gu...> - 2005-02-22 04:37:01
|
On Feb 21, 2005, at 8:29 PM, mark gross wrote: >> c) The form factor is the same as the Waysmall boards. > > This is too small! If marketing is telling you to keep it that small > tell > them the user need access to most all the signals coming from the > atmega part > to interface with the robotics gear. Giving up on those just to keep > it in > the form factor of the Waysmall board is a crime. Gordon and I will take this to Marketing and beat them up with it -- thanks C |
From: mgross <mar...@th...> - 2005-02-22 06:21:34
|
On Monday 21 February 2005 22:02, Dave Hylands wrote: > Hi Mark, > > > > 1) You'll see that we are thinking of using SPI in a way that would > > > allow a software download over SPI (NSSP on the gumstix side) if you > > > take advantage of the Power-on-reset autodownload over SPI. On this > > > board we have a LDO between the V_BATT (Power Jack) and the supply to > > > the ATmega128. So: dropping L_DD14 will drop the AT power enable, and > > > dropping L_DD13 will drop nPROG_EN. If the programming_enable line is > > > low when the chip powers back up it will download boot code over the > > > SPI. > > > > I've never used SPI, is it hard to work with? > > Not particularly. It's basically just another bidirectional > communication mechanism. It's actually simpler to deal with than i2c. > > Here's an app note from Atmel: > http://www.atmel.com/dyn/resources/prod_documents/doc2585.pdf > > This app note talks about doing in system programming over SPI: > http://www.atmel.com/dyn/resources/prod_documents/DOC0943.PDF > > >From my quick review of the pxa datasheet, any of the GPIO pins can be > > configured to generate an interrupt. > > So, you should be able to do anything that can be done over a serial > port using the SPI port. > > > Without a serial communications between the atmega and the PXA you are > > making the software development for the atmega program MUCH harder. I > > urge you to reconsider this. ttyS1 should talk directly to the TX/RX > > UART on the atmega. > > Why do you think that the serial port is so important? I'm just trying > to understand the issue. > Code reuse. All the software I've used for my robot work has been serial port centric. For instance consider boot loaders. Most existing ones use the serial port. Without serial port communications between the atmel and pxa a new boot loader will need to be written. (This may be a poor example, as a boot loader specific to this system would be needed) Debug print's from the atmel side will be harder to implement, and will need to use non-standard libraries. On both the PXA side as well as the atmel side. I would like to re-use all the linux host based tools for atmel that are built around the assumption of the serial port being the communication to the atmel. > > > c) The form factor is the same as the Waysmall boards. > > > > This is too small! If marketing is telling you to keep it that small > > tell them the user need access to most all the signals coming from the > > atmega part to interface with the robotics gear. Giving up on those just > > to keep it in the form factor of the Waysmall board is a crime. > > I'd certainly agree that it would be worth having a bigger board, > especially if the pins could all be brought out in a > power/ground/.signal pin arrangement using 3 rows of 0.1" headers. > This would make interfacing for the hobbyist MUCH easier. |
From: Dave H. <dhy...@gm...> - 2005-02-22 06:58:32
|
Hi Mark, > > Why do you think that the serial port is so important? I'm just trying > > to understand the issue. > > > Code reuse. > > All the software I've used for my robot work has been serial port centric. > > For instance consider boot loaders. Most existing ones use the serial port. > Without serial port communications between the atmel and pxa a new boot > loader will need to be written. (This may be a poor example, as a boot > loader specific to this system would be needed) With the setup Gordon is proposing, you don't need a bootloader, since the gumstix will be able to program the AVR without a boot loader, using the same technique that the AVRISP. > Debug print's from the atmel side will be harder to implement, and will need > to use non-standard libraries. On both the PXA side as well as the atmel > side. > > I would like to re-use all the linux host based tools for atmel that are built > around the assumption of the serial port being the communication to the > atmel. So, I agree with you a bit. But I also think it's a waste to tie up two serial ports, one on the ATMega and one on the gumstix. An option that I think would be good would be to take one of the serial ports and level shift it to 5v and put it right next to the ATMega one so that they could be jumpered together (to do wat you want) or be used separately. It also just occurred to me that with the proposed board, none of the serial ports on the gumstix are available, since you can't use the Robotstix alongside the Waysmall. So, I now think that the extra voltage shifters should be used to convert the Tx/Rx from one of the gumstix serial ports from 3.3 to 5v. Then at least it could be connected to one on the ATMega. I'd also like to see the other two serial ports brought out to headers along with some more GPIO pins from the gumstix. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Nate W <del...@gm...> - 2005-02-22 08:53:27
|
Some best-of-both-worlds ideas: Provide jumpers to tie an Atmel serial port to the gumstix serial port - with jumpers removed the serial ports are available to use separately. Are there any readily available connectors with two rows of pins that can be jumpered when the connector is removed? I.e, to bind the two serial ports, remove the female connector and jumper the pins; to use the serial ports separately just attached a female connector to the male header. I'm pretty sure this would be straightforward with .1" pin spacing, but it might eat up a lot of real estate. Instead of enlarging the board to expose more I/O, how about just using readily available connectors with cabling (perhaps ribbon cable) to allow most of the extra bulk of the robostix-robot interface to be located away from the robostix itself? How about two versions of the board, one waysmall-sized with higher-density headers and one prototyping-friendly-sized with .1 headers for everything? Or would that end up costing substantially more? In any case, I think you guys have a great product line on your hands! On Mon, 21 Feb 2005 22:58:24 -0800, Dave Hylands <dhy...@gm...> wrote: > Hi Mark, > > > > Why do you think that the serial port is so important? I'm just trying > > > to understand the issue. > > > > > Code reuse. > > > > All the software I've used for my robot work has been serial port centric. > > > > For instance consider boot loaders. Most existing ones use the serial port. > > Without serial port communications between the atmel and pxa a new boot > > loader will need to be written. (This may be a poor example, as a boot > > loader specific to this system would be needed) > > With the setup Gordon is proposing, you don't need a bootloader, since > the gumstix will be able to program the AVR without a boot loader, > using the same technique that the AVRISP. > > > Debug print's from the atmel side will be harder to implement, and will need > > to use non-standard libraries. On both the PXA side as well as the atmel > > side. > > > > I would like to re-use all the linux host based tools for atmel that are built > > around the assumption of the serial port being the communication to the > > atmel. > > So, I agree with you a bit. But I also think it's a waste to tie up > two serial ports, one on the ATMega and one on the gumstix. > > An option that I think would be good would be to take one of the > serial ports and level shift it to 5v and put it right next to the > ATMega one so that they could be jumpered together (to do wat you > want) or be used separately. > > It also just occurred to me that with the proposed board, none of the > serial ports on the gumstix are available, since you can't use the > Robotstix alongside the Waysmall. > > So, I now think that the extra voltage shifters should be used to > convert the Tx/Rx from one of the gumstix serial ports from 3.3 to 5v. > Then at least it could be connected to one on the ATMega. > > I'd also like to see the other two serial ports brought out to headers > along with some more GPIO pins from the gumstix. > > -- > Dave Hylands > Vancouver, BC, Canada > http://www.DaveHylands.com/ > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real users. > Discover which products truly live up to the hype. Start reading now. > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users > -- Nate Waddoups Redmond WA USA http://www.natew.com/ <== for nerds http://www.featherforum.com/ <== for birds |
From: mgross <mar...@th...> - 2005-02-22 15:16:06
|
On Monday 21 February 2005 22:58, Dave Hylands wrote: > Hi Mark, > > > > Why do you think that the serial port is so important? I'm just trying > > > to understand the issue. > > > > Code reuse. > > > > All the software I've used for my robot work has been serial port > > centric. > > > > For instance consider boot loaders. Most existing ones use the serial > > port. Without serial port communications between the atmel and pxa a new > > boot loader will need to be written. (This may be a poor example, as a > > boot loader specific to this system would be needed) > > With the setup Gordon is proposing, you don't need a bootloader, since > the gumstix will be able to program the AVR without a boot loader, > using the same technique that the AVRISP. > Oh, Thats cool then. The AVR is new to me, so I'll trust you. :) > > Debug print's from the atmel side will be harder to implement, and will > > need to use non-standard libraries. On both the PXA side as well as the > > atmel side. > > > > I would like to re-use all the linux host based tools for atmel that are > > built around the assumption of the serial port being the communication to > > the atmel. > > So, I agree with you a bit. But I also think it's a waste to tie up > two serial ports, one on the ATMega and one on the gumstix. > Why? every robot board I've played with (albeit PIC boards) tie up a serial line for the software development system. I'm just worried about being able to reuse existing devleopment ulilities for debug and development. It looks like the SPI boot load will be cool. But we still need some type of debug printf capability that works with current software tools for the AVR. If I'm making too big of a deal about this let me know, I'll trust you Dave. > An option that I think would be good would be to take one of the > serial ports and level shift it to 5v and put it right next to the > ATMega one so that they could be jumpered together (to do wat you > want) or be used separately. > > It also just occurred to me that with the proposed board, none of the > serial ports on the gumstix are available, since you can't use the > Robotstix alongside the Waysmall. ???? You need serial or USB at a minimum to talk to the gumstix from the host development system. > > So, I now think that the extra voltage shifters should be used to > convert the Tx/Rx from one of the gumstix serial ports from 3.3 to 5v. > Then at least it could be connected to one on the ATMega. > > I'd also like to see the other two serial ports brought out to headers > along with some more GPIO pins from the gumstix. YES! --mgross |
From: Andreas T. <li...@da...> - 2005-02-22 13:24:05
|
Hi there, Just my 2 cents... > 1) You'll see that we are thinking of using SPI in a way that would > allow a software download over SPI (NSSP on the gumstix side) if you > take advantage of the Power-on-reset autodownload over SPI. On this > board we have a LDO between the V_BATT (Power Jack) and the supply to > the ATmega128. So: dropping L_DD14 will drop the AT power enable, and > dropping L_DD13 will drop nPROG_EN. If the programming_enable line is > low when the chip powers back up it will download boot code > over the SPI. > Which LDO are you planning to use? The MIC5255 datasheet i found, didn't show up a version with 5V output. 150mA output power is also way to small for hooking up a lot of sensors. The ATmega128 don't uses default SPI lines for ISP programming. You have to connect PE0/PE1 to the gumstix if you want to program the AVR that way. Otherwise you will need a bootloader. PEN is only needed to enter ISP mode if you cannot be sure that SCK stays low during power-on reset. Usually only ISP_MOSI (PE0), ISP_MISO (PE1), ISP_SCK (PB1) and RESET are needed. Have a look at the chapter about 'Serial Downloading' in the datasheet. > 2) Errata pertaining to the schematic passed: the high-side > supply (5V) > the voltage-transceiver will be pwered directly by V_BATT and not Vcc5 > as noted; ATM_PROGEN will not be tied-high because of an > internal tie-high. > > 3) What will be most helpful is commentary on the selection of > connectors. All that is shown are the pads and at this point, > tin-posts are contemplated (e.g. Molex 53047-- > http://dkc3.digikey.com/PDF/EU042/0037.pdf ) > > 4) I have transceibvers for 2 more lines to signal from pxa > to at128 and > 1 from at128 to pxa255. Preferences? > > 5) Sadly-- no stackability on this go-round. Frankly, I'm > interested in > how the I2C support goes before deciding between the number of choices > for board-addressing amongst daughtercards. > > 6) Choices made: > a) No rs232 level from the atmega to the world, and > b) No uart connection between the atmega and pxa255, and I agree with mgross and think that is not a good choice. > c) The form factor is the same as the Waysmall boards. > I'd suggest to go with a bigger form factor. There Would be more space for additional connectors. I'd suggest further to expose AVRs ISP and JTAG pins. Regards, Andreas |
From: Dave H. <dhy...@gm...> - 2005-02-22 16:21:29
|
> The ATmega128 don't uses default SPI lines for ISP programming. You have to > connect PE0/PE1 to the gumstix if you want to program the AVR that way. > Otherwise you will need a bootloader. PEN is only needed to enter ISP mode > if you cannot be sure that SCK stays low during power-on reset. Usually only > ISP_MOSI (PE0), ISP_MISO (PE1), ISP_SCK (PB1) and RESET are needed. Have a > look at the chapter about 'Serial Downloading' in the datasheet. Good catch! Since PE0/PE1 also correspond to RXD0/TXD0 then I think it makes sense to connect one of the gumstix uarts to these pins. Then, for programming, the uart could be configured as GPIO's and we bit bang the SPI stuff and then switch back to a UART once done. Leaving the SPI connected as originally planned also makes sense, since the SPI bus is much faster (it can go up to fosc/2, and if a 16 MHz crystal is used, then SCK could go as fast as 8 MHz, which would yield 1 megabyte/sec transfer rates (this would requre well written code on the AVR side). -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: David F. <dav...@ya...> - 2005-02-22 13:48:11
|
You might consider the Hirose DF3, DF11,DF13 series since Digikey also sells premade Interconnect wire for those who can't justify tooling. David. --- Gordon Kruberg <go...@gu...> wrote: > 3) What will be most helpful is commentary on > the selection of > connectors. All that is shown are the pads and > at this point, > tin-posts are contemplated (e.g. Molex 53047-- > http://dkc3.digikey.com/PDF/EU042/0037.pdf ) |
From: Holly G. <hg...@ei...> - 2005-02-22 14:34:48
|
I disagree that the waysmall board is too small. To me they are on the big side. For instance, there is tons of real estate on the robostix layout that is either empty or has low density routing only. If the desire was to have more pins from the atmel run out to headers, there would certainly be room to do that within the footprint of the current layout. There are only so many things you are going to be doing concurrently with a little 8MHz auxillary processor, and quite a lot of pins are broken out anyway. Speaking personally, one of the things that attracts me to the gumstix platform is the fact that the boards are not bloated to PC104 or atx size like lots of other embedded boards. Anyways, with the gigantic pitch of the pins on the atmel, you should be able to solder wires onto whichever pins you like since each one is big enough to serve as a decent breakout pad. But as I say it looks like it wouldn't be too annoying to break out more pins to a second row of headers if Gordon thought there was sufficient demand to do so. -Holly >>> c) The form factor is the same as the Waysmall boards. >> >> >> This is too small! If marketing is telling you to keep it that small >> tell >> them the user need access to most all the signals coming from the >> atmega part >> to interface with the robotics gear. Giving up on those just to keep >> it in >> the form factor of the Waysmall board is a crime. > > > Gordon and I will take this to Marketing and beat them up with it -- > thanks > > C > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real users. > Discover which products truly live up to the hype. Start reading now. > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: mgross <mar...@th...> - 2005-02-22 15:34:01
|
On Tuesday 22 February 2005 06:34, Holly Gates wrote: > I disagree that the waysmall board is too small. To me they are on the > big side. For instance, there is tons of real estate on the robostix > layout that is either empty or has low density routing only. If the > desire was to have more pins from the atmel run out to headers, there > would certainly be room to do that within the footprint of the current > layout. There are only so many things you are going to be doing > concurrently with a little 8MHz auxillary processor, and quite a lot of > pins are broken out anyway. We are talking about robot applictions. Moving parts, vibrations, non-optimal execution enviornments, lots of random sensors getting connected and disconnected over time. The hardware will be reused on multiple robot platforms from month to month. The users are big handed hobiest, with questionalble skills with a soldering iron. > > Speaking personally, one of the things that attracts me to the gumstix > platform is the fact that the boards are not bloated to PC104 or atx > size like lots of other embedded boards. Anyways, with the gigantic > pitch of the pins on the atmel, you should be able to solder wires onto > whichever pins you like since each one is big enough to serve as a > decent breakout pad. But as I say it looks like it wouldn't be too > annoying to break out more pins to a second row of headers if Gordon > thought there was sufficient demand to do so. The gumstix size was a negative for me. If I had a choice I would have gotten someting bigger. I realy like the form factor of the OLIMEX LPC-P212X. But it doesn't run linux. The only reason I got my gumstix was the existing software stack, modern Linux kernel and tool chain, working root file system, user community... ***Its all about the software*** The gumstix tool chain / kernel and root filesystem running as a source forge deal + the active mailing list are what sold me. (And the recomendation from Dave Hylands helped too.) > > -Holly > > >>> c) The form factor is the same as the Waysmall boards. > >> > >> This is too small! If marketing is telling you to keep it that small > >> tell > >> them the user need access to most all the signals coming from the > >> atmega part > >> to interface with the robotics gear. Giving up on those just to keep > >> it in > >> the form factor of the Waysmall board is a crime. > > > > Gordon and I will take this to Marketing and beat them up with it -- > > thanks > > > > C > > > > |
From: Dave H. <dhy...@gm...> - 2005-02-22 16:40:34
|
> I disagree that the waysmall board is too small. To me they are on the > big side. For instance, there is tons of real estate on the robostix > layout that is either empty or has low density routing only. If the > desire was to have more pins from the atmel run out to headers, there > would certainly be room to do that within the footprint of the current > layout. There are only so many things you are going to be doing > concurrently with a little 8MHz auxillary processor, and quite a lot of > pins are broken out anyway. It's actually 16MHz :). You'd be amazed at what people are doing with these things (even some image processing - see http://www.jrobot.net/Projects/AVRcam.html) If the pins are brought out in a manner that's convenient, then that means I have to make another board to rearrange everything. If I have to make another board, then there really isn't any advantage to using the RoboStix. For me, if I have to create another board, then I wouldn't bother buying the RoboStix, because creating wiring is really time consuming. I'd just incorporate my AVR into the other board and use a simple interconnect, like the i2c on the back of the waysmall. I guess the Gordon & Craig & Don need to decide which market they want to address with this board. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |
From: Bill B. <wb...@su...> - 2005-02-23 12:47:49
|
It might be an idea to have a robotics board using a dsPIC - 8 channel PWM, 16 channel-12 bit A/D, 30MIPS . . . Bill --------- Bill Bigge wb...@su... +44(0)1273 877621 --------- --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.775 / Virus Database: 522 - Release Date: 08/10/2004 |
From: Holly G. <hg...@ei...> - 2005-02-22 14:40:02
|
.1" headers are like towering skyscrapers! 2mm headers at the largest! Anyone with a halfway decent soldering iron can work with 2mm headers... If you did dual row 2mm right angle surface mount headers on both sides of the board, you could pack an assload of IO onto that one edge, even if one row on each side was all GND. -Holly Nate W wrote: >Some best-of-both-worlds ideas: > >Provide jumpers to tie an Atmel serial port to the gumstix serial port >- with jumpers removed the serial ports are available to use >separately. Are there any readily available connectors with two rows >of pins that can be jumpered when the connector is removed? I.e, to >bind the two serial ports, remove the female connector and jumper the >pins; to use the serial ports separately just attached a female >connector to the male header. I'm pretty sure this would be >straightforward with .1" pin spacing, but it might eat up a lot of >real estate. > >Instead of enlarging the board to expose more I/O, how about just >using readily available connectors with cabling (perhaps ribbon cable) >to allow most of the extra bulk of the robostix-robot interface to be >located away from the robostix itself? > >How about two versions of the board, one waysmall-sized with >higher-density headers and one prototyping-friendly-sized with .1 >headers for everything? Or would that end up costing substantially >more? > >In any case, I think you guys have a great product line on your hands! > > >On Mon, 21 Feb 2005 22:58:24 -0800, Dave Hylands <dhy...@gm...> wrote: > > >>Hi Mark, >> >> >> >>>>Why do you think that the serial port is so important? I'm just trying >>>>to understand the issue. >>>> >>>> >>>> >>>Code reuse. >>> >>>All the software I've used for my robot work has been serial port centric. >>> >>>For instance consider boot loaders. Most existing ones use the serial port. >>>Without serial port communications between the atmel and pxa a new boot >>>loader will need to be written. (This may be a poor example, as a boot >>>loader specific to this system would be needed) >>> >>> >>With the setup Gordon is proposing, you don't need a bootloader, since >>the gumstix will be able to program the AVR without a boot loader, >>using the same technique that the AVRISP. >> >> >> >>>Debug print's from the atmel side will be harder to implement, and will need >>>to use non-standard libraries. On both the PXA side as well as the atmel >>>side. >>> >>>I would like to re-use all the linux host based tools for atmel that are built >>>around the assumption of the serial port being the communication to the >>>atmel. >>> >>> >>So, I agree with you a bit. But I also think it's a waste to tie up >>two serial ports, one on the ATMega and one on the gumstix. >> >>An option that I think would be good would be to take one of the >>serial ports and level shift it to 5v and put it right next to the >>ATMega one so that they could be jumpered together (to do wat you >>want) or be used separately. >> >>It also just occurred to me that with the proposed board, none of the >>serial ports on the gumstix are available, since you can't use the >>Robotstix alongside the Waysmall. >> >>So, I now think that the extra voltage shifters should be used to >>convert the Tx/Rx from one of the gumstix serial ports from 3.3 to 5v. >>Then at least it could be connected to one on the ATMega. >> >>I'd also like to see the other two serial ports brought out to headers >>along with some more GPIO pins from the gumstix. >> >>-- >>Dave Hylands >>Vancouver, BC, Canada >>http://www.DaveHylands.com/ >> >>------------------------------------------------------- >>SF email is sponsored by - The IT Product Guide >>Read honest & candid reviews on hundreds of IT Products from real users. >>Discover which products truly live up to the hype. Start reading now. >>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click >>_______________________________________________ >>gumstix-users mailing list >>gum...@li... >>https://lists.sourceforge.net/lists/listinfo/gumstix-users >> >> >> > > > > |
From: Holly G. <hg...@ei...> - 2005-02-22 14:42:26
|
Also, 2mm dual row headers are compatible with IDC ribbon cable connectors. -Holly > .1" headers are like towering skyscrapers! 2mm headers at the largest! > Anyone with a halfway decent soldering iron can work with 2mm headers... > > If you did dual row 2mm right angle surface mount headers on both > sides of the board, you could pack an assload of IO onto that one > edge, even if one row on each side was all GND. > > -Holly > > Nate W wrote: > >> Some best-of-both-worlds ideas: >> >> Provide jumpers to tie an Atmel serial port to the gumstix serial port >> - with jumpers removed the serial ports are available to use >> separately. Are there any readily available connectors with two rows >> of pins that can be jumpered when the connector is removed? I.e, to >> bind the two serial ports, remove the female connector and jumper the >> pins; to use the serial ports separately just attached a female >> connector to the male header. I'm pretty sure this would be >> straightforward with .1" pin spacing, but it might eat up a lot of >> real estate. >> >> Instead of enlarging the board to expose more I/O, how about just >> using readily available connectors with cabling (perhaps ribbon cable) >> to allow most of the extra bulk of the robostix-robot interface to be >> located away from the robostix itself? >> >> How about two versions of the board, one waysmall-sized with >> higher-density headers and one prototyping-friendly-sized with .1 >> headers for everything? Or would that end up costing substantially >> more? >> >> In any case, I think you guys have a great product line on your hands! >> >> >> On Mon, 21 Feb 2005 22:58:24 -0800, Dave Hylands <dhy...@gm...> >> wrote: >> >> >>> Hi Mark, >>> >>> >>> >>>>> Why do you think that the serial port is so important? I'm just >>>>> trying >>>>> to understand the issue. >>>>> >>>>> >>>> >>>> Code reuse. >>>> >>>> All the software I've used for my robot work has been serial port >>>> centric. >>>> >>>> For instance consider boot loaders. Most existing ones use the >>>> serial port. >>>> Without serial port communications between the atmel and pxa a new >>>> boot >>>> loader will need to be written. (This may be a poor example, as a >>>> boot >>>> loader specific to this system would be needed) >>>> >>> >>> With the setup Gordon is proposing, you don't need a bootloader, since >>> the gumstix will be able to program the AVR without a boot loader, >>> using the same technique that the AVRISP. >>> >>> >>> >>>> Debug print's from the atmel side will be harder to implement, and >>>> will need >>>> to use non-standard libraries. On both the PXA side as well as the >>>> atmel >>>> side. >>>> >>>> I would like to re-use all the linux host based tools for atmel >>>> that are built >>>> around the assumption of the serial port being the communication to >>>> the >>>> atmel. >>>> >>> >>> So, I agree with you a bit. But I also think it's a waste to tie up >>> two serial ports, one on the ATMega and one on the gumstix. >>> >>> An option that I think would be good would be to take one of the >>> serial ports and level shift it to 5v and put it right next to the >>> ATMega one so that they could be jumpered together (to do wat you >>> want) or be used separately. >>> >>> It also just occurred to me that with the proposed board, none of the >>> serial ports on the gumstix are available, since you can't use the >>> Robotstix alongside the Waysmall. >>> >>> So, I now think that the extra voltage shifters should be used to >>> convert the Tx/Rx from one of the gumstix serial ports from 3.3 to 5v. >>> Then at least it could be connected to one on the ATMega. >>> >>> I'd also like to see the other two serial ports brought out to headers >>> along with some more GPIO pins from the gumstix. >>> >>> -- >>> Dave Hylands >>> Vancouver, BC, Canada >>> http://www.DaveHylands.com/ >>> >>> ------------------------------------------------------- >>> SF email is sponsored by - The IT Product Guide >>> Read honest & candid reviews on hundreds of IT Products from real >>> users. >>> Discover which products truly live up to the hype. Start reading now. >>> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click >>> _______________________________________________ >>> gumstix-users mailing list >>> gum...@li... >>> https://lists.sourceforge.net/lists/listinfo/gumstix-users >>> >>> >> >> >> >> >> > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real users. > Discover which products truly live up to the hype. Start reading now. > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click > _______________________________________________ > gumstix-users mailing list > gum...@li... > https://lists.sourceforge.net/lists/listinfo/gumstix-users |
From: Dave H. <dhy...@gm...> - 2005-02-22 16:32:34
|
On Tue, 22 Feb 2005 09:39:24 -0500, Holly Gates <hg...@ei...> wrote: > .1" headers are like towering skyscrapers! 2mm headers at the largest! > Anyone with a halfway decent soldering iron can work with 2mm headers... > > If you did dual row 2mm right angle surface mount headers on both sides > of the board, you could pack an assload of IO onto that one edge, even > if one row on each side was all GND. I'd really like to see 3 rows (power, ground, and signal), and a jumper that allows the power to be connected to an external supply. Even with .1" connectors you can fit 30 pins along one edge. With 2mm you could get 38. Personally, I'd rather go with 0.050" than 2mm. Standard servos come with .1" connectors, so you'd at least want the motor outputs to be .1" Most hobbyists use .1" If you check out all of the stuff from places like: http://www.lynxmotion.com/ http://www.acroname.com/ http://www.pololu.com/ http://www.junun.org/ which are all popular suppliers to the robotics hobbyist, you'll see that they all use .1" headers. -- Dave Hylands Vancouver, BC, Canada http://www.DaveHylands.com/ |