I would like to suggest that a set of header holes be on the pcb before the gold fingers as follows...
2x32 0.7mm holes spaced 1.0x1.0mm to take a 2x32 1x1mm header. While these pin headers are not cheap, at least that gives us the ability to use either the 1x1mm headers and solder direct to a daughter pcb, or to a female header (I only found smt mount females with a quick look) on the users pcb.This would only lengthen the pcb by ~3mm.
Hi
My internet is too slow for the video for some reason - are there any links to stills of the board you are all talking about - I'm pulling my hair out (whats left) trying to guess.
I will make some "from scratch" boards, but probably only for boards that diverge from the modules capabilities (ie need more I/O's, don't need SDRAM), or when a customer orders a large quantity of boards.
Of course this is subject to change, and depends on how well the modules will work on the designs I have in mind for them, and also on being able to rely on its connector staying a Parallax supported standard for years (same for the modules).
Having a nice small module take care of the most challenging surface mounting is very attractive for trying multiple designs, and should be very attractive for many of Parallax's customers.
Does this mean no one else will be making boards of their own? I would think that by only Parallax only selling one or two different boards that it would open the market for third party boards. I expect that Parallax will be marketing a BOE type of board with a breadboard, etc in short order.
Hi
My internet is too slow for the video for some reason - are there any links to stills of the board you are all talking about - I'm pulling my hair out (whats left) trying to guess.
I would bet that Parallax will produce something akin to the BOE for the P2 as the motherboard, then let 3rd party developers come up with all sorts of designs.
I'm personally not very interested in the carrier board (remember the old BASIC Stamp carrier boards?). I don't want to be married to the SDRAM chip and the I/O it consumes.
My initial idea is for a board that does realtime telepresence with the Occulus Rift HMD. I've been looking for parts to fill the need and think I might have most of that covered.
The board would have 2 VGA cameras for stereo capture at 30FPS, higher FPS at lower resolutions. It would then transmit the image data in realtime to a PC via Ethernet.
I'm seriously thinking this would be a good application for PoE too, eschewing external power connections.
Oh, and the board would be equipped with servo drivers to do pan and tilt for an immersive experience.
Why do you think Parallax will only produce a board akin to the BOE? In the second post on this tread Chip stated that Parallax will.. make a little module with the Prop2, Flash, crystal, 1.8V regulator, and SDRAM.
I'm talking about a motherboard, the backplane that the aforementioned module plugs into. Obviously they don't want to leave it entirely up to the end user to build an eval board, but they don't want to be stuck making 5 different boards either. They will produce a carrier board and probably 1 motherboard.
PCIE connectors are easy to find in a variety of options, but many are expensive at small volumes. As an example, Samtec makes PCIE in vertical, right angle and edge connector formats. The edge connector format is the easiest way to keep your overall PCB height at a minimum if you choose to use this Parallax core module concept. The edge connector PCIE-064-02-F-D-EMS2 runs $3.22.
Personally, my opinion on the "P2 core board" is mixed. I like the idea of the circuit arrangement (heck, I suggested nearly the same thing in the P2 marketing thread), however, I am not in agreement with the interconnect choice. There are so many interconnect choices out there designed for high speed IO that it would make sense to review many others instead of PCIE. Going with goldfingers on it does reduce the cost of the PCB, but at an expense of added connector cost for the main PCB in a format that is not always considered user friendly. Because of this, I can see a multitude of breakout boards for the module popping up once the "core PCB" design is locked in and released. One breakout board design I can envision is a simple one: A PCIE connector broken out to standard 0.1" pitch headers.
I would prefer a castellated module with 0.1" spacing on 3 or 4 sides.
From my perspective, Chip's whole drive towards a PCI-E 4x connector was the 27 cent cost in 1k quantities.
I actually argued they design the board with all the I/O brought to a PCI-E 8x connector (if married to such) and offer the board in SDRAM and san-SDRAM variants.
I agree that this form factor may be difficult for some, but it's hard to argue with the 27 cent connector cost. The edge connector *is* the bees knees for compact though.
Chip is also keen on keeping the module very small in footprint, castellated pins would require a much larger footprint.
All of that said, populating the P2 onto a different board shouldn't be that difficult if you accept the requirement to use a reflow oven. I took the plunge and bought a desktop unit for a < $300 and now I'm unafraid.
Chip is also keen on keeping the module very small in footprint, castellated pins would require a much larger footprint
As a 0.1" spaced design like I mentioned, yes it would be larger, but if they are truly concerned about size, it will depend on the method chosen. At work we populate a module with 42 castellated pins that is half the size of this design. It's using pins spaced at 50 mils and can be easily soldered by hand.
All of that said, populating the P2 onto a different board shouldn't be that difficult if you accept the requirement to use a reflow oven. I took the plunge and bought a desktop unit for a < $300 and now I'm unafraid.
Completely agree! At some point there were several negative comments on the package chosen due to lead spacing, but with the right equipment, prototyping is not a challenge. Many people will be able to hand solder the P2
I still think that someone will develop a baseboard with added area/features that will become the P2 standard for development. I think there will be products that are a basic design that will use this module to save their overall development costs and complexity, but in the long run, I don't see it being heavily used. It's just too easy to make a better solution for a particular need. The P2 package and required circuitry are not difficult to design in from scratch.
From my perspective, Chip's whole drive towards a PCI-E 4x connector was the 27 cent cost in 1k quantities.
Quite a few others also follow this pathway - it is a good way to get a lot of IO in a small space, at very low cost.
Often they also give some 0.1" headers, and so that covers more users.
For low IO counts, 0.1" is ok, but for more access, the edge connector is used.
Completely agree! At some point there were several negative comments on the package chosen due to lead spacing, but with the right equipment, prototyping is not a challenge. Many people will be able to hand solder the P2
Maybe Parallax could make a "Learn to solder the Propeller2" kit. The kit would have a number of (non-functional) boards with the P2 footprint on it, and a bunch of dead/empty P2 style packages that could be used for practice. I don't think I would want to practice on a $10-$15+ chip.
This has a leg in both camps, Raspberry Pi & Arduino
It looks to have level shifters on the Arduino side, which avoids the big issue of 3V IO on that footprint.
Backstory: Today, I read a political article online that I thought made some good points. Then I read some of the comments with opposing views that also made some good points. That reminded me that one never has thought something over well until one has tried to consider or examine existing arguments on the other side. I think that this is applicable to the connection choice(s) for the Prop module.
But other than the price of the connector being considered, I don't think we've produced a plain listing of other advantages of using a connector in the first place. I guess one rather obvious advantage would be that a module could be easily replaced (if need be) or moved to another system, as it wouldn't be soldered in place. I'm guessing that there are other advantages.
Now switching to the other side if I may, I realize that Parallax has indicated that they don't want to produce a lot of board types, but I don't see the harm in producing a version of the module that could be mounted flat, even if it were somewhat bigger. While adding to Parallax's inventory burden and so on, it might serve the community well and drives some sales, even if the "big dogs" will produce their own boards and/or modules. Of course, I'm glad to hear that we can all learn and equip ourselves to reflow solder, but it's nice to have options (and I don't relish becoming a human pick-and-place machine for passive components for anything more than prototypes).
Over on iFixit, I was looking at the photos of a disassembled OUYA console and controller. For the controller, there's one of those small Bluetooth modules with castellated connectors on three sides (approaching 50 "pins" in total) directly soldered to a "user" PCB. It seems to be a pretty good solution for that usage scenario (particularly where folks don't want to deal with the radio stuff). Anyway, perhaps some people think such connections are unbecoming appearance-wise, but they do appear to work. No, they wouldn't allow for module removal (not easily, anyway). Well, there are probably other solutions available that would allow for flat mounting.
At this point, I'm too ignorant to know what I want (but at least I know that I don't fully know). Maybe I need an Apple-type company to tell me what I want. But I can't help thinking that a module that could be mounted flush-and-flat would be desirable (in low-profile or space-constrained designs).
But a socket would have some advantages, too. But here's the thing: if the pin pitch is tight, then it increases layout and fabrication difficulty if working on a prototype (with the board not sourced from a PCB house). Anyway, I can see a right-angle socket being desirable. Yeah, there'd likely be another circuit board under it, which might seem wasteful, but it could have some circuitry, etc. I don't think most designs will require a vertical-oriented multi-socket bus like an old Apple II or similar (though some might).
I'm just trying to be another voice (or "bark box") for the little dogs. Obviously, word-of-mouth experience from us small folks will occasionally rise up to the big dogs, and maybe even some small folks will have projects that go on to sell chips in quantity. It's great to encourage Americans (and other countries' citizens) to be inventive, but if there's not a relatively smooth path that lets them test market a design by producing a small batch without a sizable investment, then the lesson being taught might end up being counterproductive.
Having said all that, I'm thrilled that Parallax will be making a module, of whatever sort. And I know many members here will fill in any voids. But having a standard module or two directly available from Parallax has some advantages in terms of long-term supply stability. Besides, it could be argued that the member community could better spend its time focusing on base boards, not modules. But whatever happens, it looks like there's good times ahead.
Here's a couiple of photos of the right angle Samtec PCIE-064-02-F-D that turned up this morning.
The height of this connector is 9.4mm / 0.370" above the backplane circuit board. I'm not sure whether the P2 would end up facing up or down.
The pin spacing is very easy for hand soldering - 4 rows in a 2.5 x 2.0mm grid
Here's a couiple of photos of the right angle Samtec PCIE-064-02-F-D that turned up this morning.
The height of this connector is 9.4mm / 0.370" above the backplane circuit board. I'm not sure whether the P2 would end up facing up or down.
The pin spacing is very easy for hand soldering - 4 rows in a 2.5 x 2.0mm grid
The pin spacing is 2.0/2.5/2.0=6.5 x 2.0mm and appears the same as the vertical socket with the mounting feet in a different position.
Still seems to be the cheapest connectors around.
If mounting holes are added to the pcb, a 1.0x1.0mm grid for 1mm pin headers (2x32) gives a minimum pcb requirement of ~0.15".
Using 2mm pin spacing requires 4 rows of 2mm pin headers and gives a minimum pcb requirement of ~0.3". ANd this will be significant in the whole size.
I don't like either, but I still think a grid is necessary - so perhaps the 1x1mm grid is the best???
I think that connector is a good choice in many ways. Perhaps it is just wishful thinking on my part but it seems that SMT packages are moving to metric pin/pad/ball spacing. From a board layout perspective it would really be nice if all the other parts (or at least as many as possible) were metric pin spacing as well. Not having to switch between metric and inch grids while laying out a board would make it easier.
I was looking at DigiKey. They're probably available much cheaper elsewhere. I work for Digi, and many SOMs use that type of connector. It includes latches to hold down the back side of the board, and since the board levers into place, the pins make great contact and just won't loosen.
Comments
2x32 0.7mm holes spaced 1.0x1.0mm to take a 2x32 1x1mm header. While these pin headers are not cheap, at least that gives us the ability to use either the 1x1mm headers and solder direct to a daughter pcb, or to a female header (I only found smt mount females with a quick look) on the users pcb.This would only lengthen the pcb by ~3mm.
http://nw.foxconn.com/search/Product_Details_Report.asp?P_PN=2EG63213-D2DL-4F&P_type=Device-to-PCB connector&P_Family=Edgecard Connector&P_Series=PCI-Express Connector&searchTypeID=4
here's a photo of its 52 pin brother
tubular: Pricing will be interesting.
I am convinced this will be a nice connection. A little backplane is in order
My internet is too slow for the video for some reason - are there any links to stills of the board you are all talking about - I'm pulling my hair out (whats left) trying to guess.
Dave
I will make some "from scratch" boards, but probably only for boards that diverge from the modules capabilities (ie need more I/O's, don't need SDRAM), or when a customer orders a large quantity of boards.
Of course this is subject to change, and depends on how well the modules will work on the designs I have in mind for them, and also on being able to rely on its connector staying a Parallax supported standard for years (same for the modules).
Having a nice small module take care of the most challenging surface mounting is very attractive for trying multiple designs, and should be very attractive for many of Parallax's customers.
and more infos in
this thread.
Andy
Thanks
Dave
I'm hoping that the carrier can be credit card size max or below, and 0.10 standard spacings, even double or triple rows.
At these densities, there is no reason for specific connectors, just headers for breakout on all sides.
Height doesn't matter too much (here), you can always flip it downward if the XY spacings stay ON GRID.
I hope that the layout guys remember to stay on grid this time, no 0.050 offsets like plagued me with the new(er) servo boards.
These are supposed to be EASY for people to work with.
Bill Henning, you did say "Hard Metric" a few posts back, shame on you.
Let's encourage Parallax to stay away from another Babel this time.
jack
I'm personally not very interested in the carrier board (remember the old BASIC Stamp carrier boards?). I don't want to be married to the SDRAM chip and the I/O it consumes.
My initial idea is for a board that does realtime telepresence with the Occulus Rift HMD. I've been looking for parts to fill the need and think I might have most of that covered.
The board would have 2 VGA cameras for stereo capture at 30FPS, higher FPS at lower resolutions. It would then transmit the image data in realtime to a PC via Ethernet.
I'm seriously thinking this would be a good application for PoE too, eschewing external power connections.
Oh, and the board would be equipped with servo drivers to do pan and tilt for an immersive experience.
Why do you think Parallax will only produce a board akin to the BOE? In the second post on this tread Chip stated that Parallax will.. make a little module with the Prop2, Flash, crystal, 1.8V regulator, and SDRAM.
I have no idea what you are talking about.
Personally, my opinion on the "P2 core board" is mixed. I like the idea of the circuit arrangement (heck, I suggested nearly the same thing in the P2 marketing thread), however, I am not in agreement with the interconnect choice. There are so many interconnect choices out there designed for high speed IO that it would make sense to review many others instead of PCIE. Going with goldfingers on it does reduce the cost of the PCB, but at an expense of added connector cost for the main PCB in a format that is not always considered user friendly. Because of this, I can see a multitude of breakout boards for the module popping up once the "core PCB" design is locked in and released. One breakout board design I can envision is a simple one: A PCIE connector broken out to standard 0.1" pitch headers.
I would prefer a castellated module with 0.1" spacing on 3 or 4 sides.
I actually argued they design the board with all the I/O brought to a PCI-E 8x connector (if married to such) and offer the board in SDRAM and san-SDRAM variants.
I agree that this form factor may be difficult for some, but it's hard to argue with the 27 cent connector cost. The edge connector *is* the bees knees for compact though.
Chip is also keen on keeping the module very small in footprint, castellated pins would require a much larger footprint.
All of that said, populating the P2 onto a different board shouldn't be that difficult if you accept the requirement to use a reflow oven. I took the plunge and bought a desktop unit for a < $300 and now I'm unafraid.
"The 2.00mm dual row ones would be easy to solder!"
You said ''metric''.
I don't even mind that really, I was mostly concerned that they stay
on grid (XY) this time so you can rotate and invert the header.
If it's 0.10 spacings on grid, then it's doubleplusgood.
jack
Bill was talking about the PCI headers footprint which is 2mm spacing:
not sure, what you will rotate and invert here...
Andy
A 2.0mm header (and a few mounting holes) would be better than none at all for stacking.
A 0.10 inch (2.54mm) header wouldn't route nicely would it?
As a 0.1" spaced design like I mentioned, yes it would be larger, but if they are truly concerned about size, it will depend on the method chosen. At work we populate a module with 42 castellated pins that is half the size of this design. It's using pins spaced at 50 mils and can be easily soldered by hand.
Completely agree! At some point there were several negative comments on the package chosen due to lead spacing, but with the right equipment, prototyping is not a challenge. Many people will be able to hand solder the P2
I still think that someone will develop a baseboard with added area/features that will become the P2 standard for development. I think there will be products that are a basic design that will use this module to save their overall development costs and complexity, but in the long run, I don't see it being heavily used. It's just too easy to make a better solution for a particular need. The P2 package and required circuitry are not difficult to design in from scratch.
Imagine you need some very expensive hardware to make it easy...
Quite a few others also follow this pathway - it is a good way to get a lot of IO in a small space, at very low cost.
Often they also give some 0.1" headers, and so that covers more users.
For low IO counts, 0.1" is ok, but for more access, the edge connector is used.
The Infineon XMC eval boards are examples of this
http://www.infineon.com/cms/en/product/microcontrollers/development-tools,-software-and-kits/xmc-development-tools,-software-and-kits/channel.html?channel=db3a304335b504400135be64c7da68b0
Maybe Parallax could make a "Learn to solder the Propeller2" kit. The kit would have a number of (non-functional) boards with the P2 footprint on it, and a bunch of dead/empty P2 style packages that could be used for practice. I don't think I would want to practice on a $10-$15+ chip.
http://www.coocox.org/epi.html
This has a leg in both camps, Raspberry Pi & Arduino
It looks to have level shifters on the Arduino side, which avoids the big issue of 3V IO on that footprint.
But other than the price of the connector being considered, I don't think we've produced a plain listing of other advantages of using a connector in the first place. I guess one rather obvious advantage would be that a module could be easily replaced (if need be) or moved to another system, as it wouldn't be soldered in place. I'm guessing that there are other advantages.
Now switching to the other side if I may, I realize that Parallax has indicated that they don't want to produce a lot of board types, but I don't see the harm in producing a version of the module that could be mounted flat, even if it were somewhat bigger. While adding to Parallax's inventory burden and so on, it might serve the community well and drives some sales, even if the "big dogs" will produce their own boards and/or modules. Of course, I'm glad to hear that we can all learn and equip ourselves to reflow solder, but it's nice to have options (and I don't relish becoming a human pick-and-place machine for passive components for anything more than prototypes).
Over on iFixit, I was looking at the photos of a disassembled OUYA console and controller. For the controller, there's one of those small Bluetooth modules with castellated connectors on three sides (approaching 50 "pins" in total) directly soldered to a "user" PCB. It seems to be a pretty good solution for that usage scenario (particularly where folks don't want to deal with the radio stuff). Anyway, perhaps some people think such connections are unbecoming appearance-wise, but they do appear to work. No, they wouldn't allow for module removal (not easily, anyway). Well, there are probably other solutions available that would allow for flat mounting.
The iFixit Article: http://www.ifixit.com/Teardown/Ouya+Teardown/14224/2
Bluetooth board: http://d3nevzfk7ii3be.cloudfront.net/igi/fYsbUSAKb5h66Ual.huge (possibly click to enlarge)
At this point, I'm too ignorant to know what I want (but at least I know that I don't fully know). Maybe I need an Apple-type company to tell me what I want. But I can't help thinking that a module that could be mounted flush-and-flat would be desirable (in low-profile or space-constrained designs).
But a socket would have some advantages, too. But here's the thing: if the pin pitch is tight, then it increases layout and fabrication difficulty if working on a prototype (with the board not sourced from a PCB house). Anyway, I can see a right-angle socket being desirable. Yeah, there'd likely be another circuit board under it, which might seem wasteful, but it could have some circuitry, etc. I don't think most designs will require a vertical-oriented multi-socket bus like an old Apple II or similar (though some might).
I'm just trying to be another voice (or "bark box") for the little dogs. Obviously, word-of-mouth experience from us small folks will occasionally rise up to the big dogs, and maybe even some small folks will have projects that go on to sell chips in quantity. It's great to encourage Americans (and other countries' citizens) to be inventive, but if there's not a relatively smooth path that lets them test market a design by producing a small batch without a sizable investment, then the lesson being taught might end up being counterproductive.
Having said all that, I'm thrilled that Parallax will be making a module, of whatever sort. And I know many members here will fill in any voids. But having a standard module or two directly available from Parallax has some advantages in terms of long-term supply stability. Besides, it could be argued that the member community could better spend its time focusing on base boards, not modules. But whatever happens, it looks like there's good times ahead.
The height of this connector is 9.4mm / 0.370" above the backplane circuit board. I'm not sure whether the P2 would end up facing up or down.
The pin spacing is very easy for hand soldering - 4 rows in a 2.5 x 2.0mm grid
That looks pretty good. Almost $2.00 USD for a connector (Newark) is pretty rough.
An unstuffed dual-row header in Chip's board is still a good idea - a mounting hole is necessary either way.
The pin spacing is 2.0/2.5/2.0=6.5 x 2.0mm and appears the same as the vertical socket with the mounting feet in a different position.
Still seems to be the cheapest connectors around.
If mounting holes are added to the pcb, a 1.0x1.0mm grid for 1mm pin headers (2x32) gives a minimum pcb requirement of ~0.15".
Using 2mm pin spacing requires 4 rows of 2mm pin headers and gives a minimum pcb requirement of ~0.3". ANd this will be significant in the whole size.
I don't like either, but I still think a grid is necessary - so perhaps the 1x1mm grid is the best???
Has anyone thought of using MiniPCI instead of PCI? What about SODIMM? Both are small card-edge connectors that are easily available in right-angle.
AMP Connector that is 144 pins, right-angle, and $6.35 for 1, $4.23/ea in 500 qty.