Just a little something I cooked up ;) no pictures yet, but here's a wall of text about it...
I've got hand drawn schematics made up for an EOMA-68... er... I'm calling it a Carrier Board for now (the PCB that the CPU card goes into). This one uses only through-hole components, and all but two (the PCMCIA slot and, oddly enough, the Ethernet jack) are very cheap. The idea is that someone like me who is rather a bit of a dunce with the soldering iron can still put it together in a dedicated weekend, if so inclined. *That is, a person with fairly beginner-level hobby skills can buy a fistful of parts and a CPU Card, etch a PCB (or get one from somewhere) and after a few hours of lead fume inhalation, has a complete computer in their hands.* That's a huge gift, I think, to the Maker community, not to mention the technically-inclined poor folk out there (I *know* I'm not the only one!)... seriously, it sounds like good stuff to me.
My two rules for designing were (1) no surface mount anything at all period end-of-story, and (2) use as many very standard parts as possible. Every component can be had at Mouser Electronics in single unit quantities.
The Ethernet jack has the magnetics built in, so it's (unfortunately) the most expensive part on the board -- but I couldn't find a through-hole Ethernet transformer... I've probably also omitted some necessary things out of simple ignorance (I have a hunch that there's more to the USB connection than four wires, a power supply and data feed, and the connector itself, for instance). I'm more budding hobbyist with this stuff than anything else -- but hey, you gotta start somewhere, right? ;)
The only thing I don't like is that it still requires a custom PCB unless one wants to do some very creative dongle-making... probably doable but it'll be very ugly in a number of ways. That said, I'll be quite surprised if this design cannot get away with using a single-sided PCB -- meaning any shmuck who can get to eBay can order the supplies to make the board at home if they want to. (Sounds a little like me!)
If anyone wants to try reading my horrible chicken scratch I'll send out a link to a scanned copy, otherwise I'll try and work up the motivation to move it into my graphics software (CorelDRAW X3), since, although I have a copy of Kicad, I never really bothered to learn how to use it properly...
Any interest at all?
On Tue, Oct 15, 2013 at 8:38 PM, Christopher Havel laserhawk64@gmail.comwrote:
Just a little something I cooked up ;) no pictures yet, but here's a wall of text about it...
I've got hand drawn schematics made up for an EOMA-68... er... I'm calling it a Carrier Board for now (the PCB that the CPU card goes into). This one uses only through-hole components, and all but two (the PCMCIA slot and, oddly enough, the Ethernet jack) are very cheap. The idea is that someone like me who is rather a bit of a dunce with the soldering iron can still put it together in a dedicated weekend, if so inclined. *That is, a person with fairly beginner-level hobby skills can buy a fistful of parts and a CPU Card, etch a PCB (or get one from somewhere) and after a few hours of lead fume inhalation, has a complete computer in their hands.*That's a huge gift, I think, to the Maker community, not to mention the technically-inclined poor folk out there (I *know* I'm not the only one!)... seriously, it sounds like good stuff to me.
With the exception of the Through Hole components, we're right there with you... here was my idea I am currently in the process of building right now.
http://rhombus-tech.net/community_ideas/carrier_board/
Regarding the Ethernet component, I have a Gigabit Ethernet jack that has magnetics and is through hole that is only $3. Also, I have a Hirose PCMCIA connector that is ALSO very cheap ($3.69), and is through hole as well. ALSO, it can be had for free from "NeedASample.com". The board I mentioned above will be listed as opensource and free to download and modify.
My two rules for designing were (1) no surface mount anything at all period end-of-story, and (2) use as many very standard parts as possible. Every component can be had at Mouser Electronics in single unit quantities.
My consideration to this would be, while Through Hole is definitely easy, SMD is ALSO easy (within reason), it is only perceived as complicated due to the size differences. SMD components aren't necessarily more fragile, and they are certainly manageable for the common person. I have taught several high school classes and freshman college courses on SMD soldering by hand with commonly available tools. My most common recommendations, 1) Clean TIP, 2) Flux, 3) Steady but abundant heat availability. Also, if soldering SMD, for passive components, it's easier to stay within the 0805-1206 range, and TQFP/SOICS, QFN and BGA ARE doable by moderate-average skilled people, but just require more patience.
The Ethernet jack has the magnetics built in, so it's (unfortunately) the most expensive part on the board -- but I couldn't find a through-hole Ethernet transformer... I've probably also omitted some necessary things out of simple ignorance (I have a hunch that there's more to the USB connection than four wires, a power supply and data feed, and the connector itself, for instance). I'm more budding hobbyist with this stuff than anything else -- but hey, you gotta start somewhere, right? ;)
The only thing I don't like is that it still requires a custom PCB unless one wants to do some very creative dongle-making... probably doable but it'll be very ugly in a number of ways. That said, I'll be quite surprised if this design cannot get away with using a single-sided PCB -- meaning any shmuck who can get to eBay can order the supplies to make the board at home if they want to. (Sounds a little like me!)
Most boards that require high current, or impedance matching will need to
be 2 layer at a minimum, and 4 layer is recommended for the larger component counts. A simple carrier board like you are describing has already been made
http://rhombus-tech.net/community_ideas/micro_engineering_board/ , and it consist entirely of through hole components.
The Second Revision to this board will be going out for prototyping this week. Pictures and Details to follow this weekend.
If anyone wants to try reading my horrible chicken scratch I'll send out a link to a scanned copy, otherwise I'll try and work up the motivation to move it into my graphics software (CorelDRAW X3), since, although I have a copy of Kicad, I never really bothered to learn how to use it properly...
Any interest at all?
There is TONS of interest, and the market is there. My company, and others, will be releasing several iterations very soon, all open source, and easy to purchase, build, or assemble. They will be easy to send to places like BatchPCB, SeeedStudio, or even your local hackerspace (for those that can do 2 layer etching).
I have given 3 Presentations in the last 2 weeks, they have all been received with very warm and interested queries, with things like "Do you know what I could use this for?! When can I buy it? Can I have this sample?! How many can you sell me?"
Your ideas are very much welcomed, and if you like, I, or any number of people on here, would be more than happy to look at what you've got. Or you can post images to imgur and include the links via a new topic thread to let everyone see.
And if you have any questions, please feel free to call or email me any time, or post a question to the list, and someone should get back to you shortly.
And welcome to the list/world of EOMA and embedded ARM.
Christopher Thomas Firemoth Industries, LLC - Owner christopher@firemothindustries.com cell: 214-458-5990
Here's the image --> http://i.imgur.com/EeIByb5.jpg
That's a scan of Ye Olde-Fashioned Hand Drawn Electronic Schematics. It's a pretty minimal setup, TBH, one USB port for a powered (or not) hub, a simple (if a little clumsy) Mims-inspired momentary-switch power on-off setup, and R-2R ladder DACs (and some transistor based level-shifter buffers) to a VGA port. Ethernet port is a bit weird -- it's really a Gigabit+POE connector where I've tied the POE bits to ground and used only half the transformer. (The GPIO ports to Ethernet LEDs are strictly optional...) I'd LOVE the part numbers for your cheaper parts -- the PCMCIA card slot I found was over $6 and the Ethernet jack was IIRC like $14.50 (geez louise!!).
You could actually build this on perfboard (like the crap at Radio Shack). You'd need to have custom dongles for the card slot and SATA port, and the USB port would be on a header -- but that's not hard. Use a pair 80wire IDE cables from a (slightly) older computer, cut them up to make a pair of 68wire cables, tie every other wire to ground on both sides... use the leftovers for the SATA connector in the same way. It would probably work OK-ish as long as the video resolution didn't go too high...
On 10/15/2013 10:58 PM, Christopher Thomas wrote:
On Tue, Oct 15, 2013 at 8:38 PM, Christopher Havel <laserhawk64@gmail.com mailto:laserhawk64@gmail.com> wrote:
Just a little something I cooked up ;) no pictures yet, but here's a wall of text about it... I've got hand drawn schematics made up for an EOMA-68... er... I'm calling it a Carrier Board for now (the PCB that the CPU card goes into). This one uses only through-hole components, and all but two (the PCMCIA slot and, oddly enough, the Ethernet jack) are very cheap. The idea is that someone like me who is rather a bit of a dunce with the soldering iron can still put it together in a dedicated weekend, if so inclined. *That is, a person with fairly beginner-level hobby skills can buy a fistful of parts and a CPU Card, etch a PCB (or get one from somewhere) and after a few hours of lead fume inhalation, has a complete computer in their hands.* That's a huge gift, I think, to the Maker community, not to mention the technically-inclined poor folk out there (I *know* I'm not the only one!)... seriously, it sounds like good stuff to me.With the exception of the Through Hole components, we're right there with you... here was my idea I am currently in the process of building right now.
http://rhombus-tech.net/community_ideas/carrier_board/
Regarding the Ethernet component, I have a Gigabit Ethernet jack that has magnetics and is through hole that is only $3. Also, I have a Hirose PCMCIA connector that is ALSO very cheap ($3.69), and is through hole as well. ALSO, it can be had for free from "NeedASample.com". The board I mentioned above will be listed as opensource and free to download and modify.
My two rules for designing were (1) no surface mount anything at all period end-of-story, and (2) use as many very standard parts as possible. Every component can be had at Mouser Electronics in single unit quantities.My consideration to this would be, while Through Hole is definitely easy, SMD is ALSO easy (within reason), it is only perceived as complicated due to the size differences. SMD components aren't necessarily more fragile, and they are certainly manageable for the common person. I have taught several high school classes and freshman college courses on SMD soldering by hand with commonly available tools. My most common recommendations, 1) Clean TIP, 2) Flux, 3) Steady but abundant heat availability. Also, if soldering SMD, for passive components, it's easier to stay within the 0805-1206 range, and TQFP/SOICS, QFN and BGA ARE doable by moderate-average skilled people, but just require more patience.
The Ethernet jack has the magnetics built in, so it's (unfortunately) the most expensive part on the board -- but I couldn't find a through-hole Ethernet transformer... I've probably also omitted some necessary things out of simple ignorance (I have a hunch that there's more to the USB connection than four wires, a power supply and data feed, and the connector itself, for instance). I'm more budding hobbyist with this stuff than anything else -- but hey, you gotta start somewhere, right? ;) The only thing I don't like is that it still requires a custom PCB unless one wants to do some very creative dongle-making... probably doable but it'll be very ugly in a number of ways. That said, I'll be quite surprised if this design cannot get away with using a single-sided PCB -- meaning any shmuck who can get to eBay can order the supplies to make the board at home if they want to. (Sounds a little like me!)Most boards that require high current, or impedance matching will need to be 2 layer at a minimum, and 4 layer is recommended for the larger component counts. A simple carrier board like you are describing has already been made
http://rhombus-tech.net/community_ideas/micro_engineering_board/ , and it consist entirely of through hole components.
The Second Revision to this board will be going out for prototyping this week. Pictures and Details to follow this weekend.
If anyone wants to try reading my horrible chicken scratch I'll send out a link to a scanned copy, otherwise I'll try and work up the motivation to move it into my graphics software (CorelDRAW X3), since, although I have a copy of Kicad, I never really bothered to learn how to use it properly... Any interest at all?There is TONS of interest, and the market is there. My company, and others, will be releasing several iterations very soon, all open source, and easy to purchase, build, or assemble. They will be easy to send to places like BatchPCB, SeeedStudio, or even your local hackerspace (for those that can do 2 layer etching).
I have given 3 Presentations in the last 2 weeks, they have all been received with very warm and interested queries, with things like "Do you know what I could use this for?! When can I buy it? Can I have this sample?! How many can you sell me?"
Your ideas are very much welcomed, and if you like, I, or any number of people on here, would be more than happy to look at what you've got. Or you can post images to imgur and include the links via a new topic thread to let everyone see.
And if you have any questions, please feel free to call or email me any time, or post a question to the list, and someone should get back to you shortly.
And welcome to the list/world of EOMA and embedded ARM.
Christopher Thomas Firemoth Industries, LLC - Owner christopher@firemothindustries.com mailto:christopher@firemothindustries.com cell: 214-458-5990
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
chris this is awesome. suggestion: if you have the experience and knowledge could you look at selecting and designing in suitable buffers on the RGB/TTL lines? there's a schematic from olimex which does this (doesn't use R-2R uses 1x 2x 4x 8x etc. resistors) which i'd like to use but it's an "attribution" license which is incompatible with the GPL.
l.
I'm... I'm blown away. YOU guys think MY stuff is awesome...? Holy /crap/. I'm the smallest fry in the bin.
I've got a lifetime total of maybe an hour of futzing with microcontrollers -- and I'm being generous there! 90% of what I know how to make is discrete logic or transistors-only or relay crap. Probably my most complicated circuit ever is a relay adder-subtracter that has a rotary phone dial as input. (Never have been able to build it -- I'm actually fairly poor so being able to build anything is an event of near-legendary proportions...)
Seriously. I'm the sort of electronics hobbyist that has to look up circuits on the 'Net and then tweak them or combine them to get what I want. I'm not sure I'm even qualifiable as "little league" here -- I'm the hardware equivalent of a script kiddie, maybe. I think I might actually be a level or two below that...
The schematics that you say are "awesome" are a combination of six parts Wikipedia, five parts general Internet (with tweaking), one part pinouts[dot]ru, one part from a Forrest Mims book, three parts sheer unadulterated guesswork, and two parts straight off the EOMA-68 wiki page. I can't say myself if it's awesome or not, but to me it looks like a total kludge :P
Oh yeah -- one idea I had to make assembly even easier (definitely need a PCB for this, hand wiring it would be a nightmare) -- socket the R-2R ladders -- so one DIP-16 socket for the input 2Rs and another for the column of Rs and the 2R at the end. Makes things more compact, and it's super easy to replace resistors if they're faulty someday.
If you can link me to the Olimex schematic, I can try to understand it and adapt it, but I'm not promising anything. I'm a very visual person so words describing it will probably not work anywhere near as well -- I'll probably just be extremely confused. I can look up pinouts and read datasheets and put it together like a jigsaw puzzle in my head, and that's the best way for me 'cause that's what I'm used to ;)
On 10/16/2013 5:02 AM, luke.leighton wrote:
chris this is awesome. suggestion: if you have the experience and knowledge could you look at selecting and designing in suitable buffers on the RGB/TTL lines? there's a schematic from olimex which does this (doesn't use R-2R uses 1x 2x 4x 8x etc. resistors) which i'd like to use but it's an "attribution" license which is incompatible with the GPL.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Wed, Oct 16, 2013 at 4:59 PM, Christopher Havel laserhawk64@gmail.com wrote:
I'm... I'm blown away. YOU guys think MY stuff is awesome...? Holy crap. I'm the smallest fry in the bin.
no you're not! the whole purpose of the exercise is to bring people like you a huge "component" around which boards exactly like the one you designed can be made up.
I've got a lifetime total of maybe an hour of futzing with microcontrollers -- and I'm being generous there! 90% of what I know how to make is discrete logic or transistors-only or relay crap.
ah. what are you like at op-amps? do you know anything about the LM13700 (absolutely fascinating chip with an amazing history) i need to use it as a microphone amplifier with voltage-controlled volume.
Seriously. I'm the sort of electronics hobbyist that has to look up circuits on the 'Net and then tweak them or combine them to get what I want.
uhhh... chris? i'm not sure i should tell you this but that's exactly how i put the schematics together for the flying squirrel :)
The schematics that you say are "awesome" are a combination of six parts Wikipedia, five parts general Internet (with tweaking), one part pinouts[dot]ru, one part from a Forrest Mims book, three parts sheer unadulterated guesswork, and two parts straight off the EOMA-68 wiki page. I can't say myself if it's awesome or not, but to me it looks like a total kludge :P
*cackle*.
If you can link me to the Olimex schematic,
https://www.google.co.uk/search?num=40&q=a10s+olimex+schematics+pdf
try that - you want the "micro"
l.
Nevermind, I found the schematic on their wordpress blog. I think I've got them beat -- how about hardware-level switching between 18b and 24b color, using a single GPIO pin? <g>
Don't worry about Olimex, my idea uses different buffer chips, in a different way, and the buffer scheme is really the only part of the design that they "own" (if at all -- there's no license info or anything on the blog post...) so I think we should be in the clear there.
I'll have schematics up in a bit...
On 10/16/2013 5:02 AM, luke.leighton wrote:
chris this is awesome. suggestion: if you have the experience and knowledge could you look at selecting and designing in suitable buffers on the RGB/TTL lines? there's a schematic from olimex which does this (doesn't use R-2R uses 1x 2x 4x 8x etc. resistors) which i'd like to use but it's an "attribution" license which is incompatible with the GPL.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Wed, Oct 16, 2013 at 5:50 PM, Christopher Havel laserhawk64@gmail.com wrote:
Nevermind, I found the schematic on their wordpress blog. I think I've got them beat -- how about hardware-level switching between 18b and 24b color, using a single GPIO pin? <g>
Don't worry about Olimex, my idea uses different buffer chips, in a different way, and the buffer scheme is really the only part of the design that they "own" (if at all -- there's no license info or anything on the blog post...) so I think we should be in the clear there.
I'll have schematics up in a bit...
star.
what i'll then do if that's ok with you is put the bits of it that are different from the MEB (because, effectively, the design you've come up with is near-identical to the MEB except the addition of VGA) into KiCAD. would a GPLv3+ license be ok with you?
l.
Never dealt with licensing, so whatever you think is best works for me. I'm pawing through that PDF you pointed me to, just to be safe -- if it's any help the revised VGA schematic I'm drawing up now uses 74HC125's for the buffer-drivers. Gimme another half hour or so and I'll have you another scan <g>
One other thing I thought of -- would it be permissible to tie a GPIO in with my power-on logic setup (in between the Mims-debounced pushbutton and the improvised T flipflop) so that the system could at least turn itself off...? Right now you'd need a Win95 style "it is now safe to turn off your computer" screen to tell the user to hit the pushbutton. Ack. (Also, if there's a good way to do that circuit with an IRF510 instead of an IRF9510 I'd love to know -- IRF510s can be had at Radio Shack, but the IRF9510 is almost hen's-teeth even at Mouser...)
I'm afraid I'm useless with opamps. I've used a 741 once and it didn't work >.< of course when I stuffed a 386 in there, that DID work -- was a super duper simple amplified speaker in a reuseable coffee cup for a relative who never used it because by the time it was done she had a different car with an aux-in jack so she didn't need it.
Oh, yeah... "flying squirrel"...? What the heck...?
On 10/16/2013 1:37 PM, luke.leighton wrote:
On Wed, Oct 16, 2013 at 5:50 PM, Christopher Havel laserhawk64@gmail.com wrote:
Nevermind, I found the schematic on their wordpress blog. I think I've got them beat -- how about hardware-level switching between 18b and 24b color, using a single GPIO pin? <g>
Don't worry about Olimex, my idea uses different buffer chips, in a different way, and the buffer scheme is really the only part of the design that they "own" (if at all -- there's no license info or anything on the blog post...) so I think we should be in the clear there.
I'll have schematics up in a bit...
star.
what i'll then do if that's ok with you is put the bits of it that are different from the MEB (because, effectively, the design you've come up with is near-identical to the MEB except the addition of VGA) into KiCAD. would a GPLv3+ license be ok with you?
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 10/16/2013 01:45 PM, Christopher Havel wrote:
Oh, yeah... "flying squirrel"...? What the heck...?
The Code Name for the KDE based tablet. This will be the one of the first feature products to use the EOMA-68 Cards.
http://rhombus-tech.net/community_ideas/kde_tablet/news/
LOVE the name :D that's great... KDE as in K Desktop Environment, or is there something else out there using the same acronym...? I'm more partial to GNOME2 myself :P
On 10/16/2013 1:54 PM, Scott Sullivan wrote:
On 10/16/2013 01:45 PM, Christopher Havel wrote:
Oh, yeah... "flying squirrel"...? What the heck...?
The Code Name for the KDE based tablet. This will be the one of the first feature products to use the EOMA-68 Cards.
On 10/16/2013 01:58 PM, Christopher Havel wrote:
LOVE the name :D that's great... KDE as in K Desktop Environment, or is there something else out there using the same acronym...? I'm more partial to GNOME2 myself :P
Same KDE, they need a device on which to show case their Plasma Active Tablet environment.
Nice thing is, with an GPL compliant card, you'll be able to port, or get port any distribution / desktop you wish. It's all down to dollars, time and/or effort.
On 10/16/2013 1:54 PM, Scott Sullivan wrote:
On 10/16/2013 01:45 PM, Christopher Havel wrote:
Oh, yeah... "flying squirrel"...? What the heck...?
The Code Name for the KDE based tablet. This will be the one of the first feature products to use the EOMA-68 Cards.
The initial attempt at a device named 'Vivaldi' ran into complications. I don't know any of the details really, but I believe it had to due with GPL violation on the part of the SoC vendor. Someone more knowledgeable can fill that in, or you can search the list archive for previous answers.
Here, have a scan, guys! --> http://i.imgur.com/f2lbm3R.jpg
There's the VGA interface. Any tweaks you want to add, add.
By the way, I saw on the rhombus-tech site that there'll be possibly an x86 CPU Card -- I'm looking forward to that.
I'm a devout Puppy Linux fanatic and there isn't much in ARM territory with that distro. There's a version for the RasPi that works if your software skills are good and you have strong enough convictions, and someone's devving a version for A10 based stuff but it's in Alpha... my software skills suck (and that's on a good day!).
"make" doesn't build applications for me, it builds error messages and noisy fireworks that are showy but unproductive. Sometimes it'll actually give me a broken application. I consider myself pretty lucky at that point. Once in a while tho it goes off the deep end and tries to put together a magic chocolate factory or some crap like that... it usually doesn't get very far... usually I get that only when I have to use both "make" and "configure" or something like that.
I'm happiest with the x86 stuff -- watt-hog that it is, relative to ARM, it "just works", you know? I can just tack on LibreOffice or Wine or Chrome or whatever and it works as long as I'm not trying to exceed the capabilities of the distro or something equally stupid.
On 10/16/2013 2:07 PM, Scott Sullivan wrote:
On 10/16/2013 01:58 PM, Christopher Havel wrote:
LOVE the name :D that's great... KDE as in K Desktop Environment, or is there something else out there using the same acronym...? I'm more partial to GNOME2 myself :P
Same KDE, they need a device on which to show case their Plasma Active Tablet environment.
Nice thing is, with an GPL compliant card, you'll be able to port, or get port any distribution / desktop you wish. It's all down to dollars, time and/or effort.
On 10/16/2013 1:54 PM, Scott Sullivan wrote:
On 10/16/2013 01:45 PM, Christopher Havel wrote:
Oh, yeah... "flying squirrel"...? What the heck...?
The Code Name for the KDE based tablet. This will be the one of the first feature products to use the EOMA-68 Cards.
The initial attempt at a device named 'Vivaldi' ran into complications. I don't know any of the details really, but I believe it had to due with GPL violation on the part of the SoC vendor. Someone more knowledgeable can fill that in, or you can search the list archive for previous answers.
On Wed, Oct 16, 2013 at 7:47 PM, Christopher Havel laserhawk64@gmail.com wrote:
Here, have a scan, guys! --> http://i.imgur.com/f2lbm3R.jpg
just spotted something... on the other image... what was it.... oh yes! ethernet. you have to connect up all 8 wires - it's part of the spec that CPU cards will be up to gigabit and up to USB3.
There's the VGA interface. Any tweaks you want to add, add.
awesome.
By the way, I saw on the rhombus-tech site that there'll be possibly an x86 CPU Card -- I'm looking forward to that.
yeahh, me too. but... someone has to come out with a low-power x86 processor first!
l.
My understanding, which was based on the Wiki, was that gigabit and USB3 would be implemented later. Guess the Wiki's a little out of date. (No big.)
Just a thought on the x86 bit... I actually have a PicoITX board (super doober cheap eBay find!) that uses the Atom Z530 + NM10 chipset. It's Poulsbo PITA graphics but some Linuces do work, including a couple versions of Puppy. Oh right -- the PSU is a 5v 4a model...
Just a few thoughts on that sort of config...
Atom Z530 = 2.0w TDP NM10 PCH = 2.1w TDP
...so a Z530 + NM10 = 4.1w TDP.
According to CPU-World, the max power dissipation ever for a Z530 would be 4.64w. According to the NM10 datasheet (DC Characteristics on p225) and a bit of math, the maximum power dissipation in the S0 state would be 2.794w. That's a total of 7.434w, leaving a ton of room (electrically speaking) for a fan.
NM10 datasheet here --> http://www.intel.com/Assets/PDF/datasheet/322896.pdf
Z530 "Recommended Customer Price" is $65.00. For the NM10 it's $20.00. (USD of course.) That said, there's a seller on eBay with the NM10 at $9 with free shipping -- although I can't vouch for their reputability! Worth noting also is that the Z530 CPU is not to be found by itself on eBay either. The eBay listing for the NM10 is here --> http://www.ebay.com/itm/400565337435
So maybe look at that config...? or is Intel incompatible with you guys for licensing reasons or something?
On 10/16/2013 3:53 PM, luke.leighton wrote:
On Wed, Oct 16, 2013 at 7:47 PM, Christopher Havel laserhawk64@gmail.com wrote:
Here, have a scan, guys! --> http://i.imgur.com/f2lbm3R.jpg
just spotted something... on the other image... what was it.... oh yes! ethernet. you have to connect up all 8 wires - it's part of the spec that CPU cards will be up to gigabit and up to USB3.
There's the VGA interface. Any tweaks you want to add, add.
awesome.
By the way, I saw on the rhombus-tech site that there'll be possibly an x86 CPU Card -- I'm looking forward to that.
yeahh, me too. but... someone has to come out with a low-power x86 processor first!
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 10/16/2013 05:16 PM, Christopher Havel wrote:
My understanding, which was based on the Wiki, was that gigabit and USB3 would be implemented later. Guess the Wiki's a little out of date. (No big.)
The first card will not implement it, true. But later cards may, so you should design your board to support USB3 and GigE so as be forward compatible.
...that does make sense...
On 10/16/2013 6:55 PM, Scott Sullivan wrote:
On 10/16/2013 05:16 PM, Christopher Havel wrote:
My understanding, which was based on the Wiki, was that gigabit and USB3 would be implemented later. Guess the Wiki's a little out of date. (No big.)
The first card will not implement it, true. But later cards may, so you should design your board to support USB3 and GigE so as be forward compatible.
On Wed, Oct 16, 2013 at 10:16 PM, Christopher Havel laserhawk64@gmail.com wrote:
My understanding, which was based on the Wiki, was that gigabit and USB3 would be implemented later.
not quite correct. *in the future* there will be other CPU Cards with those features. the spec *allows* for those. so if you design an I/O board which only has 10/100, it's not going to be desirable or useful with a CPU Card that has gigabit capabilities, is it?
and the one thing that EOMA68 is designed for is with "long term investment" in mind. you buy one or more chassis' and you keep them in service until they literally fall apart of old age. .. and you upgrade the CPU Cards to make sure they stay useful and current.
... so if you design a chassis with only USB2 or 10/100 ethernet, how is it going to be useful to anyone in 2-3 years time?
Just a thought on the x86 bit... I actually have a PicoITX board (super doober cheap eBay find!) that uses the Atom Z530 + NM10 chipset. It's Poulsbo PITA graphics but some Linuces do work, including a couple versions of Puppy. Oh right -- the PSU is a 5v 4a model...
yeah, exactly. the closest that intel gets to "low power" is an ULV 600mhz processor that uses 2.5 watts. that's excluding the southbridge chip!
Just a few thoughts on that sort of config...
Atom Z530 = 2.0w TDP NM10 PCH = 2.1w TDP
...so a Z530 + NM10 = 4.1w TDP.
yeah exactly. now try putting that on a 43 x 75mm board in an enclosed space - don't forget that DDR3 RAM will add another 0.75 watts (assume 2x 32-bit @ 800mhz). and then there's the inefficiency of the PMIC. now you're well over 7 watts.
According to CPU-World, the max power dissipation ever for a Z530 would be 4.64w. According to the NM10 datasheet (DC Characteristics on p225) and a bit of math, the maximum power dissipation in the S0 state would be 2.794w. That's a total of 7.434w, leaving a ton of room (electrically speaking) for a fan.
you can't get a fan in a 5mm space. and it has to go on the CPU Card (the fan) because otherwise it's necessary to add a fan into the *other* side of the spec... and that's a cost that's completely unnecessary *and* means that when you get into mass-volume you just stuffed a) the profit margins b) the reliability.
So maybe look at that config...? or is Intel incompatible with you guys for licensing reasons or something?
not at all. intel just simply don't have a product that works. they've been told. i've tried reaching out to AMD: it's like climbing a glass wall. i even contacted RDC and they honestly don't believe in their own product because nobody's trying to buy it. and the RDC IAD100 series also needs a companion IC and a 3D GPU to make it comparable.
l.
They make 5mm thick 40mm fans. I have a couple. Not hard to adapt that to a blower -- the jerry-rig way is to cover what'd normally be the exhaust with a bit of thin plastic and then clip off one of the four sides of the fan housing, presto, you got one blower that works fine, altho you'll also have a 40mm square area on the PCB that you can't use on one side -- or use one of these and have a proper hole in the PCB --> http://www.ebay.com/itm/321090918766
Either way, as long as you've got good venting it'll work.
Also... if you're talking about the Celeron M ULV 600 I have a board with one of those and it's useless. I can't get even a fairly light version of Puppy to work on it without really rattling my patience fast... DSL might fare slightly better but I can't stand the look. Talk about "Made For Windows 98"... >.< ugh.
PicoITX is a little larger than what you're working with -- 72*100mm -- but it still works and it works well. I think it's doable with a nice copper heatsink and a blower that doesn't ever cut off... although you've got a point re: the VRMs... they like to heat up.
What if you used the PCMCIA shell as the VRM heatsink...? Make it aluminum and require some vent holes near it on the carrier board housing... not as cheap or strong as steel but it'll probably do, especially if the VRM is efficient and switching and not nasty old linear crap like a 78xx is...
On 10/16/2013 7:50 PM, luke.leighton wrote:
On Wed, Oct 16, 2013 at 10:16 PM, Christopher Havel laserhawk64@gmail.com wrote:
My understanding, which was based on the Wiki, was that gigabit and USB3 would be implemented later.
not quite correct. *in the future* there will be other CPU Cards with those features. the spec *allows* for those. so if you design an I/O board which only has 10/100, it's not going to be desirable or useful with a CPU Card that has gigabit capabilities, is it?
and the one thing that EOMA68 is designed for is with "long term investment" in mind. you buy one or more chassis' and you keep them in service until they literally fall apart of old age. .. and you upgrade the CPU Cards to make sure they stay useful and current.
... so if you design a chassis with only USB2 or 10/100 ethernet, how is it going to be useful to anyone in 2-3 years time?
Just a thought on the x86 bit... I actually have a PicoITX board (super doober cheap eBay find!) that uses the Atom Z530 + NM10 chipset. It's Poulsbo PITA graphics but some Linuces do work, including a couple versions of Puppy. Oh right -- the PSU is a 5v 4a model...
yeah, exactly. the closest that intel gets to "low power" is an ULV 600mhz processor that uses 2.5 watts. that's excluding the southbridge chip!
Just a few thoughts on that sort of config...
Atom Z530 = 2.0w TDP NM10 PCH = 2.1w TDP
...so a Z530 + NM10 = 4.1w TDP.
yeah exactly. now try putting that on a 43 x 75mm board in an enclosed space - don't forget that DDR3 RAM will add another 0.75 watts (assume 2x 32-bit @ 800mhz). and then there's the inefficiency of the PMIC. now you're well over 7 watts.
According to CPU-World, the max power dissipation ever for a Z530 would be 4.64w. According to the NM10 datasheet (DC Characteristics on p225) and a bit of math, the maximum power dissipation in the S0 state would be 2.794w. That's a total of 7.434w, leaving a ton of room (electrically speaking) for a fan.
you can't get a fan in a 5mm space. and it has to go on the CPU Card (the fan) because otherwise it's necessary to add a fan into the *other* side of the spec... and that's a cost that's completely unnecessary *and* means that when you get into mass-volume you just stuffed a) the profit margins b) the reliability.
So maybe look at that config...? or is Intel incompatible with you guys for licensing reasons or something?
not at all. intel just simply don't have a product that works. they've been told. i've tried reaching out to AMD: it's like climbing a glass wall. i even contacted RDC and they honestly don't believe in their own product because nobody's trying to buy it. and the RDC IAD100 series also needs a companion IC and a 3D GPU to make it comparable.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Thu, Oct 17, 2013 at 1:46 AM, Christopher Havel laserhawk64@gmail.com wrote:
They make 5mm thick 40mm fans.
that would be fine for the type III PCMCIA cards which we're *considering* for future products. however as they'd be 10mm thick there's no way they'd go into any of the existing 5mm products (but the other way round would be ok).
so we're not ruling it out, but it's going to need to be client-driven. (for example up-front paid for, in cash, full development NREs and a cash order for the actual products or for that client to accept the trade-off of product NREs and fees as well).
;/
Schematic for the improved VGA circuit is DONE. I'll go scan it right now (after I fix the one bug I just realized I made).
This is the Olimex design I found --> http://olimex.wordpress.com/2012/06/12/low-cost-lcd-to-vga-adapter/
Again, mine's completely different. To enable 18b VGA you need one GPIO. Adding 24b capability requires a second GPIO. The circuit as drawn uses GPIO3 and GPIO4 to not conflict with the earlier schematics which use GPIOs 0,1,2 to drive the Ethernet port LEDs. Not that which GPIOs are used matters. Oh right -- the enable signals are active low, because electronics are almost always like that :P
Just a thought -- a few extra transistors and resistors here and you could have it work with 8/12/18/24b color as well -- but you'd need four GPIOs (base would be 8b, with 12/18/24 selectable with up to 3 more GPIOs).
On 10/16/2013 1:54 PM, Scott Sullivan wrote:
On 10/16/2013 01:45 PM, Christopher Havel wrote:
Oh, yeah... "flying squirrel"...? What the heck...?
The Code Name for the KDE based tablet. This will be the one of the first feature products to use the EOMA-68 Cards.
On Wed, Oct 16, 2013 at 6:45 PM, Christopher Havel laserhawk64@gmail.com wrote:
Never dealt with licensing, so whatever you think is best works for me. I'm pawing through that PDF you pointed me to, just to be safe -- if it's any help the revised VGA schematic I'm drawing up now uses 74HC125's for the buffer-drivers. Gimme another half hour or so and I'll have you another scan
<g>
:)
One other thing I thought of -- would it be permissible to tie a GPIO in with my power-on logic setup (in between the Mims-debounced pushbutton and the improvised T flipflop) so that the system could at least turn itself off...?
yeah i don't see why not *shrugs*. although... to be honest there's a self-power-down method in software anyway: you just ask the AXP209 to go "bye bye" and it does :)
Right now you'd need a Win95 style "it is now safe to turn off your computer" screen to tell the user to hit the pushbutton. Ack. (Also, if there's a good way to do that circuit with an IRF510 instead of an IRF9510 I'd love to know -- IRF510s can be had at Radio Shack, but the IRF9510 is almost hen's-teeth even at Mouser...)
I'm afraid I'm useless with opamps.
well... at least i'm having fun with them. currently quite a lot of "bzzzzt" i have to get rid of off the mic - i'm looking instead at tiny bzzt at the mic a *HUGE* bzzzzt post-amplified :)
I've used a 741 once and it didn't work
.< of course when I stuffed a 386 in there, that DID work -- was a super
duper simple amplified speaker in a reuseable coffee cup for a relative who never used it because by the time it was done she had a different car with an aux-in jack so she didn't need it.
doh!
Oh, yeah... "flying squirrel"...? What the heck...?
first name someone came up with
l.
ons 2013-10-16 klockan 10:02 +0100 skrev luke.leighton:
chris this is awesome. suggestion: if you have the experience and knowledge could you look at selecting and designing in suitable buffers on the RGB/TTL lines? there's a schematic from olimex which does this (doesn't use R-2R uses 1x 2x 4x 8x etc. resistors) which i'd like to use but it's an "attribution" license which is incompatible with the GPL.
Have you tried asking about reusing that part under a GPL license?
Regards Henrik
arm-netbook@lists.phcomp.co.uk
-
Christopher Havel -
Christopher Thomas -
Henrik Nordström -
luke.leighton -
Scott Sullivan