Review and feedback of proposed Arduino shield (soon please)
andyg (geekscape)
We've got the proposed Arduino shield design on GitHub ... and we
really need to get moving on making some PCB prototypes (for testing)
and then how to get complete boards manufactured. The next 4 months
are going to zip past really quickly ... so, it's vital to get a head-
start now.
http://github.com/lukeweston/Pebble # Check out the .png files
I'm presuming that everyone is comfortable to run with this Arduino
shield for the mini-conf ... if not, please say so now. Don't be
shy :)
If you're happy with the basic design and what it'll enable for the
mini-conf hardware and software tutorials ... it would be good to get
any final feedback on the design now ... so that we can "freeze" the
hardware and move onto software and tutorial presentation preparation
(with everyone involved having a copy of the final board to work
with ... minimize moving targets !).
Some feedback areas ...
- Choice of sensors and actuators
- Electrical design
- Layout
- Anything we've forgotten ?
The overall design criteria are ...
- Very easy for beginners to build themselves (in just one morning of
the miniconf)
- Reasonably inexpensive
- Good base platform / level of functionality for more than just
simplistic projects
- Enables good miniconf software tutorials (during the afternoon)
There is a reasonable amount of prototyping area, so "simple"
individual needs can be catered for by just supplying additional
components. For example, someone might really want to do something
with sound ... that can be dealt with by just adding a speaker and
resistor ... and doesn't require a board re-design (unless everyone
thinks we should have a speaker as part of the standard offering).
If possible, let's see if we can tie down the Arduino shield hardware
design this week ... so we can get cracking with making some prototype
PCBs.
Thanks !
cheers andyg (geekscape)
Trent Lloyd
In my personal experience these kinds of prototyping areas drive me
nuts unless you want to go wire-wrap and I'd prefer a 'bread-board'
style one with two sides. Thoughts?
A piezo could be a useful addition if theres room for it but we don't
need to cover every base.
Regards,
Trent
Trent Lloyd
http://lathiat.net/
Mitch Davis
>
> I'm presuming that everyone is comfortable to run with this Arduino
> shield for the mini-conf ... if not, please say so now. Don't be
> shy :)
I'm not an electronics person, but:
Is there a board overlay to have a look at?
On the overlay, are the signals appropriately marked?
What is the value of VR1?
IC3 and IC6 are relays, right? Can we mark them on the schematic?
Can we include the rating of the relays on the overlay?
Do we need some sort of startup cap and/or pull-up/down on RESET?
Can we swap some of the proto area for some 0.1" pitch dots, and/or
a spot for a standard 16 pin DIP?
For LCA, I would suggest not putting the SMD CPU footprint on the
board: IMO it's something that is too easy to screw up on the day.
Mitch.
Luke Weston
On Aug 25, 3:28 pm, Trent Lloyd <lath...@bur.st> wrote:
> My personal comment is about the prototyping area.
>
> In my personal experience these kinds of prototyping areas drive me
> nuts unless you want to go wire-wrap and I'd prefer a 'bread-board'
> style one with two sides. Thoughts?
>
I think you might be able to - only just - squeeze in one of those
ultra-small breadboard (as sold by Sparkfun for example) into that
prototyping space.
You'd have to keep the LCD mounted above the board, and you'd probably
need to not have any header soldered onto the bottom row of LCD
mounting pads since the breadboard would cover those pads, but it
should barely fit.
This doesn't require any changes to the board layout... just pop the
breadboard on top if you want it.
Cheers,
Luke
Jonathan Oxer
> Is there a board overlay to have a look at?
> On the overlay, are the signals appropriately marked?
> What is the value of VR1?
> IC3 and IC6 are relays, right? Can we mark them on the schematic?
> Can we include the rating of the relays on the overlay?
Those are all good points that relate to making it as easy as possible
to use the board without referring to an external schematic. When Marc
and I were working on v2 or the ProtoShield we took the approach that we
wanted to make everything self-explanatory by looking at the board
alone, so we did things like remove part numbers (who cares if something
is labelled "R3" anyway?) and put values on instead ("1K" is much more
useful information).
On one of my subsequent shields I went so far as to put part of the
actual circuit diagram on the overlay so you could just look at it and
know what is connected to what:
> Do we need some sort of startup cap and/or pull-up/down on RESET?
No, that's taken care of on the Arduino.
> Can we swap some of the proto area for some 0.1" pitch dots, and/or
> a spot for a standard 16 pin DIP?
I don't understand the reasoning behind this. It *is* 0.1" pitch, but as
plated-through-hole (PTH) rather than just dots. If you want to
surface-mount parts onto the prototyping area (something I do all the
time with 0805 parts because they fit perfectly between adjacent pads on
a 0.1" grid) it'll do that nicely, or if you want to put in leaded parts
that'll work fine too. A DIP-16 (or -8, -14, or -18, or whatever) will
go on the prototyping area nicely without requiring a custom area that
prevents it being used for anything else.
> For LCA, I would suggest not putting the SMD CPU footprint on the
> board: IMO it's something that is too easy to screw up on the day.
Maybe I misinterpreted the tone of the original suggestion but I think
it was a bit tongue-in-cheek. I don't think we'll be fitting SMD CPUs on
the day! Showing how to fit simpler SMD parts (0805 resistors, etc)
could well be worthwhile though since they're increasingly common even
in hobbyist projects. I actually prefer hand-assembling SMD now rather
than PTH.
Cheers :-)
--
Jonathan Oxer
Ph +61 4 3851 6600
Geek My Ride! <http://www.geekmyride.org/>
Luke Weston
On Tue, Aug 25, 2009 at 3:20 PM, andyg (geekscape)<an...@geekscape.org> wrote:
>
> I'm presuming that everyone is comfortable to run with this Arduino
> shield for the mini-conf ... if not, please say so now. Don't be
> shy :)
I'm not an electronics person, but:
Is there a board overlay to have a look at?
There should be a .png in the GitHub repository if you don't want to download EAGLE and pull the files down.
On the overlay, are the signals appropriately marked?
Which signals? I don't understand.
What is the value of VR1?
Um, 100k I think. They don't actually matter, you can use whatever you have laying around, as long as it isn't very high or very low resistance. (Low resistance means excessive current, very high resistance means leakage resistance across the PCB introduces error, shifting the effective resistance below the pot's setting.)
Those name/value annotation layers aren't turned on on the .png image I exported from EAGLE. They should be printed on the silkscreen on the finished product, yes. Does that answer your point, and the one below?
IC3 and IC6 are relays, right? Can we mark them on the schematic?
Can we include the rating of the relays on the overlay?
Do we need some sort of startup cap and/or pull-up/down on RESET?
No. (Not on a shield - they're included on the Arduino.)
Can we swap some of the proto area for some 0.1" pitch dots, and/or
a spot for a standard 16 pin DIP?
0.1" pitch dots? Isn't that precisely what the proto area already is?
Can't you just put a DIP chip on the 0.1" pitch proto area pads already? I don't understand what you mean.
Cheers,
Luke
follower
> shield for the mini-conf ... if not, please say so now. Don't be
> shy :)
board (not including optional), as a data point if we assume 30
seconds to solder (place, check, solder, snip, double check) each part
we'd be looking at 15 minutes to assemble in best case scenario. I
think 1 to 2 minutes per part for a beginner would be a lot more
realistic which would indicate an assembly time closer to 30-60
minutes.
In light of this I think it would really pay to ensure the assembly
time is realistic for a bunch of beginners. Does anyone have any
experience on the "average" beginner's time to solder a board together
with more accurate numbers?
One other concern with the "assemble first" then use scenario is it's
a little monolithic and doesn't enable on-the-fly testing which I
think is beneficial in the trouble-shooting stage. Could the assembly
scenario work with a more parallel assemble & use process? e.g. put
the headers on, insert an LED, write the code, run it, then insert a
pushbutton, write the code, run it, etc...
Unless there's a standard way to test the boards at the time they're
assembled what are the odds of fixes/troubleshooting needed in the
software sessions later in the day?
--Phil.
Jonathan Oxer
> In light of this I think it would really pay to ensure the assembly
> time is realistic for a bunch of beginners. Does anyone have any
> experience on the "average" beginner's time to solder a board together
> with more accurate numbers?
That's something we're hoping to discover empirically very soon. The
plan is to have a few beginners come along to a Melbourne Hackerspace
meeting and do a little "Soldering 101" session, then assemble one of
the Pebble shields each. We'll take it casually, see which bits give
them problems, and time how long it all takes.
You make a good point about the number of joints etc to make though.
Perhaps there's a case for having all the boards partially pre-assembled
to some extent just to save time, and let them fit a few of the parts
themselves to get a feel for it and a sense of achievement.
> One other concern with the "assemble first" then use scenario is it's
> a little monolithic and doesn't enable on-the-fly testing which I
> think is beneficial in the trouble-shooting stage. Could the assembly
> scenario work with a more parallel assemble & use process? e.g. put
> the headers on, insert an LED, write the code, run it, then insert a
> pushbutton, write the code, run it, etc...
That's a great idea Phil.
Paul Schulz
>
> On Tue, 2009-08-25 at 18:21 +1200, follower wrote:
>
>> In light of this I think it would really pay to ensure the assembly
>> time is realistic for a bunch of beginners. Does anyone have any
>> experience on the "average" beginner's time to solder a board together
>> with more accurate numbers?
More than happy to help out here.. send me the parts and I'll time
how long it takes me to put it together :-)
> That's something we're hoping to discover empirically very soon. The
> plan is to have a few beginners come along to a Melbourne Hackerspace
> meeting and do a little "Soldering 101" session, then assemble one of
> the Pebble shields each. We'll take it casually, see which bits give
> them problems, and time how long it all takes.
Duh!
Thomas Sprinkmeier
DIO pins on the Arduino have an optional internal pull-up.
If we use DIO pins instead of ADC ones and change S1, S2 and S4 to pull
low instead of high and you can remove resistor R4, R5 and R8.
Another option is to keep the resistors and put a cap in parallel as a
simple de-bouncing circuit (saves doing it in software).
Can the SPST reed relays be changed for SPDT ones like
http://parts.digikey.com/1/parts/756899-relay-reed-spdt-no-nc-5vdc-dip-d1c05c.html
(though 25mA for the coil is a bit hungry, not sure how much the SPST
draws)
The second flyback diode seems a bit far from the coil. Will it fit
between the relay and the voltage reg (mounted vertically perhaps?)
Can we add an LED to one of the DIO lines for some instant gratification
blinkenlights?
The power LED is a bit close to the LDR.
I've never used ZigBee, pardon me if these questions are a bit dumb(er
than usual)....
Does it need an antenna?
Is there a Linux-friendly ZigBee widget?
Thomas
--
Trying to child-proof the world makes us neglect the
more important task of world-proofing the child.
Vik Olliver
> I'm presuming that everyone is comfortable to run with this Arduino
> shield for the mini-conf ... if not, please say so now. Don't be
> shy :)
I was actually serious about putting an SMD CPU on the flip side.
Beginners can ignore it - or populate it at home.
Vik :v)
Mitch Davis
>>
>> I'm presuming that everyone is comfortable to run with this Arduino
>> shield for the mini-conf ... if not, please say so now. Don't be
>> shy :)
>
> a spot for a standard 16 pin DIP?
I take back my comment: I was thrown by the different pad sizes.
(Goes back to being shy)
Mitch.
Mitch Davis
>
> I actually prefer hand-assembling SMD now rather than PTH.
Jon, you'll do anything to use your super dandy
was-a-garden-fork-in-a-previous-life SMD point weight, won't you! :-)
Jonathan Oxer
> (Goes back to being shy)
Please don't do that! Anyone on this list should feel comfortable
speaking up if they have a different opinion or spot a potential
problem.
David Z
On Aug 25, 5:20 pm, "andyg (geekscape)" <an...@geekscape.org> wrote:
> http://github.com/lukeweston/Pebble # Check out the .png files
between 10uF and 100uF tantalum, not non-polarised 100nF caps. Also,
do you actually mean the LD1117AV33 instead?
If they're actually supposed to be 10-100uF polarised, then they might
be a bit tall to sit under the ZigBee module.
I think! :D
Luke Weston
You're right... LD1117 is what I meant, with three 1's. LD1117V33C to be specific.
Well... we're a bit tight for physical space here, the 3.3V rail current consumption isn't really high, the only 3.3V load (the Xbee) is very close to the voltage regulator, and in many cases we won't even be using a XBee at all, so I think we can get away with using something small.
Maybe I'll swap around the 0.1 uF on the input with the one on the output, so the capacitor on the output is sticking out from under the Xbee, and change it to a 10 uF 10V or so through-hole tantalum, which will have the same 0.2" lead spacing and won't be much larger. In reality I reckon the 100 nF would be fine, but in the interests of professionalism, I'll follow the datasheet.
Luke Weston
Hi All,
DIO pins on the Arduino have an optional internal pull-up.
If we use DIO pins instead of ADC ones and change S1, S2 and S4 to pull
low instead of high and you can remove resistor R4, R5 and R8.
Another option is to keep the resistors and put a cap in parallel as a
simple de-bouncing circuit (saves doing it in software).
That's a good comment :-)
Moving the buttons to use software-pull-up DIO pins would require moving, say, the LCD driving pins to use analog pins, which would require re-routing those traces across the board. OK, fair enough, you could do it, but then we would break the (admittedly not widely used at present) defacto standard for the pin assignments which has been followed consistently on the couple of preliminary prototypes that exist at present. Well, that's not a huge deal, but then again a couple of resistors aren't a huge deal either personally.
Let's remember the assumption that this project is supposed to be accessible to users who don't have much experience with soldering and assembling electronics.
It's also fair to assume that most if not all LCA attendees have at least some programming experience, even if they have no soldering experience.
Which would you say would be the easier option for users in this context... soldering a few extra capacitors for hardware-implemented debounce, or writing a simple few lines extra
to implement software-based debounce? I would say most would find software debounce easier.
Can the SPST reed relays be changed for SPDT ones like
http://parts.digikey.com/1/parts/756899-relay-reed-spdt-no-nc-5vdc-dip-d1c05c.html
(though 25mA for the coil is a bit hungry, not sure how much the SPST
draws)
Does anyone remember what the AVR's spec is for the maximum IO pin current draw? 25 mA might be pushing it, but I don't have a datasheet in front of me and I don't remember.
The 10-15 mA for a SPST reed relay is fine.
The second flyback diode seems a bit far from the coil. Will it fit
between the relay and the voltage reg (mounted vertically perhaps?)
Does it really matter that it's not immediately adjacent to the relay?
Can we add an LED to one of the DIO lines for some instant gratification
blinkenlights?
Hmm... on A5, which we're currently not using, and put on the board next to the XBee status LED? That sounds pretty easy. Will do :)
The power LED is a bit close to the LDR.
It's actually the Xbee activity indicator LED... but in any case, that's a valid point. But there's not too much flexibility to move it, since you don't want it hidden under the LCD.
It can be moved to the right a little bit though, which I'll do. If in doubt, mount a little opaque shield around the LDR to shield it.
Remember that (a) LEDs emit most of their light in a forward direction through the lens, and minimal light out the side, and (b) the LDR isn't providing really quantitatively precise data anyway.
I've never used ZigBee, pardon me if these questions are a bit dumb(er
than usual)....
Does it need an antenna?
Is there a Linux-friendly ZigBee widget?
(b) You're just talking to a UART on the device from the Arduino (or your PC via the FTDI chip if you remove the AVR), so I see no reason why it's OS-dependant.
Admittedly I've never played with Xbees, but someone else (like, say, Andy G.) will know more.
Cheers,
Luke
Jonathan Oxer
> Does anyone remember what the AVR's spec is for the maximum IO pin current
> draw? 25 mA might be pushing it, but I don't have a datasheet in front of me
> and I don't remember.
> The 10-15 mA for a SPST reed relay is fine.
Cheers :-)
Mitch Davis
>
>> If we use DIO pins instead of ADC ones and change S1, S2 and S4 to pull
> Moving the buttons to use software-pull-up DIO pins would require moving,
> say, the LCD driving pins to use analog pins, which would require re-routing
> those traces across the board.
It may not be feasible or desirable, but one option might be to have
some cut-and-bridge points so we can choose whether to use the digital
or analog inputs for the pushbuttons.
Another option is to just use one analog input, and wire each switch
in series with a carefully chosen resistor. That way, we can
multiplex multiple buttons onto one input, and decode which button was
pressed (or even simultaneous multiple button presses) in software.
Mitch.
Thomas Sprinkmeier
>
> On Tue, Aug 25, 2009 at 11:04 PM, Thomas Sprinkmeier
> <thomas.sp...@gmail.com> wrote:
>
> Hi All,
>> (using DIO for switches instead of ADC)
>
> Well, that's not a huge deal, but then again a couple of resistors
> aren't a huge deal either personally.
I just hate 'wasting' perfectly good ADC's on simple digital input.
Also, some of the DIO's can drive interrupts
( http://www.arduino.cc/en/Reference/AttachInterrupt ), might be fun
playing with.
Having said that, I really like Mitch's idea of a 'resistor ladder' to
multiplex the switches onto a single ADC. Not only saving pins, but a
neat hack!
http://en.wikipedia.org/wiki/Resistor_ladder
> Can the SPST reed relays be changed for SPDT ones like
> http://parts.digikey.com/1/parts/756899-relay-reed-spdt-no-nc-5vdc-dip-d1c05c.html
> (though 25mA for the coil is a bit hungry, not sure how much
> the SPST
> draws)
>
> Does anyone remember what the AVR's spec is for the maximum IO pin
> current draw? 25 mA might be pushing it, but I don't have a datasheet
> in front of me and I don't remember.
> The 10-15 mA for a SPST reed relay is fine.
The nominated SPST relay is 10mA
http://www.jaycar.com.au/productView.asp?ID=SY4030
The current limit per pin is 40mA, total not to exceed 200mA
http://arduino.cc/en/Main/ArduinoBoardDuemilanove
So, 25mA is doable I guess, but depending on what the other pins are
doing might cause problems.
I just like the idea of having an 'on-by-default' option on the relays.
> The second flyback diode seems a bit far from the coil. Will
> it fit
> between the relay and the voltage reg (mounted vertically
> perhaps?)
>
> Does it really matter that it's not immediately adjacent to the
> relay?
Potentially, if a digital input is left in 'high impedance' mode there
might just be enough EMI to trigger it. If that input is used to trigger
the relay then you can set up a nice oscillator :-)
But yeah, a 10mA reed relay shouldn't cause any problems, just being
paranoid...
> Can we add an LED to one of the DIO lines for some instant
> gratification
> blinkenlights?
>
> Hmm... on A5, which we're currently not using, and put on the board
> next to the XBee status LED? That sounds pretty easy. Will do :)
>
> The power LED is a bit close to the LDR.
>
> It's actually the Xbee activity indicator LED... but in any case,
> that's a valid point. But there's not too much flexibility to move it,
> since you don't want it hidden under the LCD.
> It can be moved to the right a little bit though, which I'll do. If in
> doubt, mount a little opaque shield around the LDR to shield it.
>
> Remember that (a) LEDs emit most of their light in a forward direction
> through the lens, and minimal light out the side, and (b) the LDR
> isn't providing really quantitatively precise data anyway.
Good point.
Can the LDR be swapped with the 1820?
> I've never used ZigBee, pardon me if these questions are a bit
> dumb(er
> than usual)....
> Does it need an antenna?
> Is there a Linux-friendly ZigBee widget?
>
> (a) No, but you can get variants with an external antenna connector if
> you want to for extra performance. All variants are pinout-compatible
> and interchangable.
> (b) You're just talking to a UART on the device from the Arduino (or
> your PC via the FTDI chip if you remove the AVR), so I see no reason
> why it's OS-dependant.
> Admittedly I've never played with Xbees, but someone else (like, say,
> Andy G.) will know more.
I was wondering if there was a USB-Xbee dongle (sort of like USB-RS232,
or USB-IRDA).
Thomas
David Z
On Aug 26, 9:05 pm, Luke Weston <reindeerfloti...@gmail.com> wrote:
> On Wed, Aug 26, 2009 at 3:06 PM, David Z <tgk...@gmail.com> wrote:
> > The caps around the LD117AV33 look wrong. I'd have expected something
> > between 10uF and 100uF tantalum, not non-polarised 100nF caps.
>
> consumption isn't really high, the only 3.3V load (the Xbee) is very close
> to the voltage regulator, and in many cases we won't even be using a XBee at
> all, so I think we can get away with using something small.
of the things it would be good to get across to people is datasheets
have numbers, and maybe those numbers are useful to start from. And
that people will crib from this for maybe their own ideas. :)
The traces are short enough I think an output cap only will be fine.
The datasheet (going from memory) marked the input side as optional
anyway, and it'll be less of a problem given we're feeding it
regulated 5V from the main board.
IMO :) Electronics n00b. Be wared! :)
andyg (geekscape)
So that the Arduino will boot and run properly with or without a
serial connection ...
Put a 100Kohm pull-down resistor on Arduino digital pin 0 (Rx).
This is so that noise on a floating digital pin 0 isn't interpreted by
the boot-loader as new firmware to download, thus preventing the
Arduino from simply running the existing firmware (as it should).
cheers andyg (geekscape)
Trent Lloyd
Have you ever actually had problems with this?
I never have... don't forget it's connected to the FTDI chip which
should hold it at an appropriate level presumably.
Trent
Jonathan Oxer
> On 28/08/2009, at 9:33 AM, andyg (geekscape) wrote:
> > Put a 100Kohm pull-down resistor on Arduino digital pin 0 (Rx).
> Have you ever actually had problems with this?
I have. My web-services light switch worked fine on the bench connected
to USB, then when I stuck it in the wall and ran it off POE it wouldn't
boot. Took it back out, worked fine. Put it back in, failed. Plugged in
the laptop to debug it in-situ, worked fine. Drove me insane until I
figured out it was noise on RX whenever the USB wasn't plugged in, and a
resistor solved the problem immediately.
http://jon.oxer.com.au/blog/id/307
Cheers :-)
--
Jonathan Oxer
Ph +61 4 3851 6600
andyg (geekscape)
On Aug 28, 11:44 am, Trent Lloyd <lath...@bur.st> wrote:
> Have you ever actually had problems with this?
original) Arduino boards.
Based on various forums, I've seen that other people have this problem
as well.
Given that it is a single resistor, I don't think that it is much of a
problem or cost to add.
Anyone who doesn't need it (or doesn't like it) ... can simply leave
the resistor out ... no harm, no foul.
I presume that you've just checking ... and don't have a serious
problem with this suggestion ?
cheers andyg (geekscape)
Luke Weston
Pebble feature creep status 28/8/09
------------------------------------------------------
Add Sparkfun mini breadboard footprint, check it will fit (DONE)
Swap position of LDR with DS18B20 (DONE)
Add der spitzensparken blinkenlichten LED on Analog 5 (DONE)
Move diode closer to relay. (DONE)
Change 3.3V regulator output cap to 10 uF Ta (DONE)
(I know, I know, revised board not pushed to GitHub yet)
Add 100k pull-down on Digital 0. (to be done soon)
Pushbutton multiplexing (three-bit R-2R DAC) on a single analog input
(to be done soon)
Consider SPDT relays? (Need to see if suitable relays exist)
Improve silkscreen component labels (to be done)
Thomas Sprinkmeier
>
> Pebble feature creep status 28/8/09
> ------------------------------------------------------
>
> Add Sparkfun mini breadboard footprint, check it will fit (DONE)
>
> Swap position of LDR with DS18B20 (DONE)
>
> Add der spitzensparken blinkenlichten LED on Analog 5 (DONE)
One of the PWM pins? nice touch :-)
> Move diode closer to relay. (DONE)
>
> Change 3.3V regulator output cap to 10 uF Ta (DONE)
>
> (I know, I know, revised board not pushed to GitHub yet)
>
> Add 100k pull-down on Digital 0. (to be done soon)
>
> Pushbutton multiplexing (three-bit R-2R DAC) on a single analog input
> (to be done soon)
I had a brain-fade on that. The R-2R ladder requires SPDT switch to pull
the resistor to ground or the rail. The current pushbuttons (effectively
SPST) won't do.
Something like
http://www.altronics.com.au/index.asp?area=item&id=S1493
only low-profile (can't seem to find any)
The current SPDT ones won't give us a nicely spaced voltages, but
something like the attached circuit should let us tell which buttons
have been pressed (gimme a while to work on the resistor values...)
> Consider SPDT relays? (Need to see if suitable relays exist)
The header above the relays is 6 long, GND, +V and the two N/O contacts.
To use SPDT relays this header would either have to drop the ground and
voltage rails (undesirable), grow by 2 (enough space?) or maybe use a
double-row header, otherwise the N/C contact won't be easily available.
Starting to sound like too much trouble....
Thomas
--
Perl is gaffa tape of IT. It is like The Force: it has a
light side and a dark side, and it holds the universe together
Thomas Sprinkmeier
> Pushbutton multiplexing (three-bit R-2R DAC) on a single analog input
> (to be done soon)
SPST rather than SPDT shouldn't be a problem, see attached spreadsheet
and previous circuit.
Even with the first R values that came to mind the minimum step will be
39, a good 5 bits to spare on the ADC :-)
DISCLAIMER: It's late, I'm tired, YMMV....
Thomas
Luke Weston
The files in the GitHub repository have now been updated reflecting
most of the things we've discussed so far.
I tidied up the schematic a little too.
Please have a look and tell me what you think.
Again, there are a pair of *.png files there for convenience if you
don't have EAGLE installed.
You'll have to excuse my dodgy use of a transistor to represent the
DS18B20, I don't have a library part for it, and I thought it would be
important to add the "D" shaped TO-92 footprint on the PCB silkscreen,
so people don't solder the device on the board in the wrong
orientation and let the smoke out.
>
> To use SPDT relays this header would either have to drop the ground and
> voltage rails (undesirable), grow by 2 (enough space?) or maybe use a
> double-row header, otherwise the N/C contact won't be easily available.
>
> Starting to sound like too much trouble....
>
package.
Right now, I will leave the SPST relays as is.
I have just put down 390 k, 220 k and 100 k as the resistor values for
the pushbuttons for the moment... spaced apart by factors of two. I
hadn't read your above post on the subject when drawing that up... and
I don't have OpenOffice on this machine, anyway, so I can't read it. :-
(
As long as the resistors are changing in resistance significantly from
one step to the next in a non-linear way, so that the different bits
have different significance than others, the exact values aren't too
crucial, it's just a matter of a lookup table with only eight
precalculated values to map a particular value on the ADC to one of
the eight possible combinations of button states.
Cheers,
Luke
Luke Weston
Also, perhaps it would be good to clean up the spam on this group and
lock it down a little to keep the spam away.
We used to have exactly the same issue with the CCHS group.
Cheers.
Thomas Sprinkmeier
>
> The files in the GitHub repository have now been updated reflecting
> most of the things we've discussed so far.
> I tidied up the schematic a little too.
>
> Please have a look and tell me what you think.
Nice! I like the sectioning in the schematic, makes it much easier to
follow.
IIRC A5 is an ADC port (i.e. analogue in).
The LED should be running off D9, D10 or D11 (the PWM ports).
> I have just put down 390 k, 220 k and 100 k as the resistor values for
> the pushbuttons for the moment... spaced apart by factors of two. I
> hadn't read your above post on the subject when drawing that up... and
> I don't have OpenOffice on this machine, anyway, so I can't read it. :-
exported to PDF.
The resistors values I had (which I picked out of the air, not according
to any resistor series) gives slightly more distinguishable levels, but
either should work.
Thomas
Trent Lloyd
On Sat, 2009-08-29 at 06:26 -0700, Luke Weston wrote:The files in the GitHub repository have now been updated reflectingmost of the things we've discussed so far.I tidied up the schematic a little too.Please have a look and tell me what you think.
Nice! I like the sectioning in the schematic, makes it much easier to
follow.
IIRC A5 is an ADC port (i.e. analogue in).
The LED should be running off D9, D10 or D11 (the PWM ports).
On newer Arduino boards (including the Mini and BT) with the ATmega168 chip, this function works on pins 3, 5, 6, 9, 10, and 11. Older USB and serial Arduino boards with an ATmega8 only support analogWrite() on pins 9, 10, and 11.
Luke Weston
used with no risk of other devices on the pins it uses interfering
with its operation.
But I guess adding a LED in parallel on pin 9 should be OK in that
context, right?
Thomas Sprinkmeier
I don't know anything about the Ethernet shields, but I've got a mad
scheme to uses the unused Q1 on the 4094:
Connect the LED to Q1.
Connect Qs to Q via a resistor.
With D3 in high-impedance mode the 4094 becomes a ring-buffer.
To change the LED (and no other outputs) set D3 to input (i.e.
high-impedance), shift 7 times, then set D3 to the desired state of the
LED and shift again.
Note that we can now read the 4094 (potentially an extra 8 bits of
RAM!!)
Should work unless someone has a fancy ISR to updated the LCD.
The LCD driver will probably nuke Q1 (unless it's patched to shift it
back in) so the LED will be in an undetermined state whenever the LCD is
updated.
Thomas
Jonathan Oxer
>
> Also, perhaps it would be good to clean up the spam on this group and
> lock it down a little to keep the spam away.
I did that yesterday at Andy's suggestion. I'm now manually moderating
all group signups plus the first messages sent by new users.
Cheers :-)
Jon
Vik Olliver
> Note that we can now read the 4094 (potentially an extra 8 bits of
> RAM!!)
If you want to do that kind of thing, just add a place for an I2C memory
chip. SMT if you like - it's only 8 pins!
Vik :v)
Luke Weston
>
> Connect the LED to Q1.
> Connect Qs to Q via a resistor.
>
there's no 'Q' pin.
Also, what sort of resistance did you have in mind? What's the
function of the resistor?
After this bit is implemented, I think we're pretty close to the point
where we can consider the initial design finished.
Cheers,
Luke
Thomas Sprinkmeier
>
>
>>
>> Connect the LED to Q1.
>> Connect Qs to Q via a resistor.
>>
>
> Which pin on the 4094 did you mean by 'Q'? As far as I can tell,
> there's no 'Q' pin.
I don't have the spec in front of me (on holiday!), but IIRC the 4094
has an output used to daisy-chain further 4094's.
> Also, what sort of resistance did you have in mind? What's the
> function of the resistor?
>
> After this bit is implemented, I think we're pretty close to the point
> where we cn consider the initial design finished.
100k or so should do.
The idea is that when the arduino output is in high-impedance (i.e
input) mode, the daisy-chain output is looped back into the data
input, turning the 4096 into a barrel shifter. We can clock it 7
times, effectively shifting the output left by one.
When the arduino is in high or low mode it overrides the 100k resistor
and allows us to put whatever data we want into the 4096.
So.... arduino to high-impedance, clock 7 times (i.e. shift backwards
one), set the desired state of the LED and clock again. Then enabe the
4096 output, the LCD pins (Q2 to Q8) should be unchanged while Q1 is
as desired.
Totally impractical, of course, but it appeals to my warped sense of humour :-)
Thomas
> Cheers,
> Luke
Luke Weston
Cheers.
Thomas Sprinkmeier
> Do you mean the serial output pin, Qs ? The serial outputs Qs and Qs' are
> the only other outputs the device has got.
yes.
I grabbed a copy of the spec sheet. from the truth table it looks like
we'd want Q`s rather than Qs.
on the falling clock edge it'll load the state of Q7, and on the next
rising edge it will pass that on to Q1 (assuming it's not overridden
by the arduino)
The line "Connect Qs to Q via a resistor." in my previous email should
have been "Connect Q`s to DataIn via a resistor.", sorry about any
confusion.
Thomas
--
Quis custodiet ipsos custodes?
Luke Weston
D is connected to the Arduino, and it's also connected to a 100k
resistor. The other end of that resistor is connected
to Qs'. Q1 is connected via a 100R resistor to the LED's anode, and
the LEDs cathode is connected to ground.
Should be right. I have implemented this on the schematic and routed
it on the PCB.
Can somebody double check for me please? Or just wait until I've
pushed the files to GitHub so you can see it.
Actually, it would not hurt to have someone else run eyes over the
whole lot as a quick check.
Unfortunately I can't push to GitHub over the uni firewall... it's
essentially impossible to use anything that isn't http.
It will have to wait until I'm home.
Anyway, we should pretty much be all done now. :)
Sorry about the delay in finishing this and pushing it out... so much
hacking and so little time.
Cheers,
Luke
Thomas Sprinkmeier
> OK.
>
> D is connected to the Arduino, and it's also connected to a 100k
> resistor. The other end of that resistor is connected
> to Qs'. Q1 is connected via a 100R resistor to the LED's anode, and
> the LEDs cathode is connected to ground.
>
> Should be right. I have implemented this on the schematic and routed
> it on the PCB.
>
> Can somebody double check for me please?
glad to.
Thomas
Thomas Sprinkmeier
OK, sorry. I kinda spaced, I though you migth send some eagle files or
something.
The description above sounds right.
That should allow us to light the LEDat will, while at the same time
allowing us to preserve the state of those control lines for the LCD.
Any chance of getting a prototybe board to play with? There was
mention of needing noobs to assemble them to time how long it takes, I
voulenteer :-)
Thomas
Thomas
Luke Weston
When we make a couple of prototype PCBs I think we can send you one :)
Luke Weston
Board layout and schematic files now updated in GitHub repo.
Mitch Davis
Thomas Sprinkmeier
> You live in Adelaide, right?
yes.
I work at Mile End, conveniently close to town.
> When we make a couple of prototype PCBs I think we can send you one :)
That would be great, thanks!
Are you buying components as well?
I'd be more than happy to buy a set off you, it would save me chasing
around getting my own.
I've got tha latest git files, not time to look at them now though
(school run :-)
Thomas
Thomas Sprinkmeier
> On Wed, 2009-10-21 at 05:07 -0700, Luke Weston wrote:
>> You live in Adelaide, right?
>
> yes.
> I work at Mile End, conveniently close to town.
>
>> When we make a couple of prototype PCBs I think we can send you one :)
>
> That would be great, thanks!
> Are you buying components as well?
> I'd be more than happy to buy a set off you, it would save me chasing
> around getting my own.
On second thoughts... Do you want me to get bits?
I've got the BOM from git, geting a bunch shouldn't be much harder
than getting one set
Thomas
Luke Weston
correct... doing that now :)
Luke Weston
github repo.
Cheers,
Luke
Luke Weston
So, can anyone identify anything else that stands between us and now
declaring the board design finished for now?
I guess now we may contemplate making a couple of PCBs, and also
buying some components.
The readme file on gitHub has now been updated with the full and up to
date BOM.
These are the components required to build one, however, I'm guessing
we initially want to build two or three at this stage for thorough
testing and stuff?
I guess whoever is willing to put up the capital for the parts gets to
keep the prototypes once they're built and tested?
Here's the BOM, anyway...
The following components are the more "special" components, which we
will probably have to order from an offshore supplier such as Sparkfun
or Digi-Key:
1 x LCD module (Cheap 20 x 4 ones from Hong Kong eBay seller are
recommended)
1 x 40-pin 0.1" female header socket strip (To mount LCD on board)
1 x LD1117V33C 3.3V LDO voltage regulator, TO-220 package
1 x Mini breadboard (Sparkfun PRT-07916) (optional)
2 x 10-pin 2.0 mm female header sockets to suit XBee modules
1 x DS18B20 temperature sensor
1 x Arduino, of course.
1 x XBee module
Combining with someone's existing order from one of those suppliers,
if possible, means less shipping costs for everyone.
All the rest of the components can just be had from Jaycar.
7 x 100 k (all resistors standard through-hole)
1 x 100 R
1 x 390 k
1 x 220 k
1 x 4.7 k
1 x 69 R
1 x 4094 shift register IC
1 x LDR
2 x 3 mm LEDs
2 x 1N4004s
2 x 5V reed relays (DIP form factor)
1 x 100 nF monolithic through-hole capacitor
1 x 10 uF 16 V Tantalum through-hole capacitor
1 x 100k linear pot, eg. Jaycar RP8518
1 x 100k linear trimpot, eg. Jaycar RT4366
4 x momentary pushbutton switches
1 x 16 pin IC socket
2 x 14 pin IC socket
6-pin 0.1" Molex KK type locking PCB header
2 x 40-pin 0.1" break-away male pin header strip
Cheers,
Luke
Mitch Davis
>
> I guess whoever is willing to put up the capital for the parts gets to
> keep the prototypes once they're built and tested?
Yeah :-)
> 1 x LCD module (Cheap 20 x 4 ones from Hong Kong eBay seller are
> recommended)
I am happy to purchase these, if it would help. How many do we get?
(I'm in for one)
Mitch.
Thomas Sprinkmeier
> These are the components required to build one, however, I'm guessing
> we initially want to build two or three at this stage for thorough
> testing and stuff?
> I guess whoever is willing to put up the capital for the parts gets to
> keep the prototypes once they're built and tested?
At least now I know what to say when people ask me what I want for
Christmas :-)
Put me down for one.
Thomas
Trent Lloyd
I volunteer to make one as well.. I'll supply all my own parts but I
will need the PCB if you are going to manufacture some?
Also, at this stage looks like I'm probably not going to make it to
LCA2010 at all. Then again I said that last year until 3 days before
it started...
Thanks,
Trent
Mitch Davis
On Thu, Oct 22, 2009 at 9:19 PM, Luke Weston <reindeer...@gmail.com> wrote:
>
>
> So, can anyone identify anything else that stands between us and now
> declaring the board design finished for now?
>
> I guess now we may contemplate making a couple of PCBs, and also
> buying some components.
As discussed last night, Thomas and I are planning to run a Pebble
Parts Program. I have spent the day scouring the net to find the best
prices, and we will source all the parts locally, except for the LCD.
The price of a "package" (being all the components needed to build
Pebble, minus the Arduino, the Pebble PCB, the mini-breadboard, and
the XBee and the XBee sockets) comes to $30.56 minus shipping. Once
we know quantities, we'll be able to work out the extra amount for
having the parts shipped to us, and optionally, the amount to ship
them to the person who has signed up for our program.
(Our spreadsheet is currently at
http://spreadsheets.google.com/ccc?key=0AsTwCG-wTDDBdDAtLTFHcWFFZGk5bTF0d082S2pTUFE&hl=en
. I'll make it editable by all to allow sign-ups in the next day or
so)
I have some questions about the parts list:
The 10uF tantalum cap: Does it have to be tantalum, as $1.20 (Jaycar)
is a bit steep. Would a regular electrolytic ($0.26) do?
JP1 and JP2 are the 2 x 16 pin female LCD sockets. Do the LCDs we got
last time from "square-man" fit the board?
Are there labels for the 2 x 16 pin male pin headers to connect to the
shield? Are there labels for the XBee sockets? (Do we need them?)
Which LED? http://au.farnell.com/jsp/search/productdetail.jsp?sku=1581118
vs http://au.farnell.com/jsp/search/productdetail.jsp?sku=1581127 .
Only difference is brightness vs viewing angle. I kind of think go
for the latter - it's dimmer, but has a wider viewing angle.
You've specified a 69R resistor. 69 is not a preferred EIA value; 68
is. Freudian slip?
What colour switches should we get? All the same? Different colours?
If you have time, we'd appreciate your feedback.
Mitch and Thomas.
Mitch Davis
>
> I have some questions about the parts list:
And some more questions:
Can we change the designators for the relays on the schematic from
IC4/IC6 to RY1 and RY2?
As far as I'm aware, on the schematic, VR1 (analog input) uses a
trimpot symbol, but is a conventional pot, whereas VR2 uses a
conventional pot symbol, but is really a trimpot. (Going out on a
limb here, maybe I'm wrong)
Might the XBee LED be better off near the XBee?
Finally, an appeal to viewers: Are you able to do better on price for
any of the components on our spreadsheet? In particular, the pot and
the relays...
Mitch.
Luke Weston
> As discussed last night, Thomas and I are planning to run a Pebble
> Parts Program.
> The price of a "package" (being all the components needed to build
> Pebble, minus the Arduino, the Pebble PCB, the mini-breadboard, and
> the XBee and the XBee sockets) comes to $30.56 minus shipping.
they're not very expensive, and have them soldered in even if you're
not using an XBee.
That way, you might decide to get an XBee later on, or you might
borrow one from a friend to temporarily try out an XBee application.
If you've already got the sockets soldered onto your board, you just
pop it into the socket and off you go.
>
> (Our spreadsheet is currently athttp://spreadsheets.google.com/ccc?key=0AsTwCG-wTDDBdDAtLTFHcWFFZGk5b...
> The 10uF tantalum cap: Does it have to be tantalum, as $1.20 (Jaycar)
> is a bit steep. Would a regular electrolytic ($0.26) do?
diameter of the capacitor to make sure they're acceptable, but they
should be.
Is it possible to source tantalum caps from a different supplier at a
better price? To be honest, for this kind of thing if you really want
the best possible price when you're buying a significant number of
little components, Jaycar can be a bit overpriced.
> JP1 and JP2 are the 2 x 16 pin female LCD sockets. Do the LCDs we got
> last time from "square-man" fit the board?
ones we get have two rows of interface pads, i.e. the same as that
batch of them which we bought before.
> Are there labels for the 2 x 16 pin male pin headers to connect to the
> shield? Are there labels for the XBee sockets? (Do we need them?)
(b) No, because in terms of the schematic the sockets don't exist,
only the XBee module does.
Some interpretation is required when parsing the Eagle-generated BOM
list to convert it into a meaningful "shopping list".
> Which LED? http://au.farnell.com/jsp/search/productdetail.jsp?sku=1581118
> vshttp://au.farnell.com/jsp/search/productdetail.jsp?sku=1581127.
> for the latter - it's dimmer, but has a wider viewing angle.
I really don't know. Pick whatever you want.
If the individual builder would like to install a 500-Candela uber
ninja light-up-the-room-at-night blue LED, that's up to you.
> You've specified a 69R resistor. 69 is not a preferred EIA value; 68
> is.
>
> What colour switches should we get? All the same? Different colours?
>
> If you have time, we'd appreciate your feedback.
A concise list of everything I should edit or fix in the Eagle files
would be nice. (I note the email following this one. Is there anything
else? Change the 69R to 68R, noted.)
Luke Weston
pushbutton switches instead.
Have you considered getting all the components that you're getting
from Jaycar from Farnell instead, instead of a little from A and a
little from B?
This would be easier to order and likely cheaper.
Some components you might like to order in a large pack to take
advantage of economy of scale, even if they aren't all required for
Pebble units at the moment.
Eg. there is an economy of scale if you buy a pack of 20 DS18B20s from
Clint's Components, you're getting them for $2 each.
If you bought that, myself and/or Andy and/or other HackerSpace
members would surely "buy out" the rest of the pack of 20, once you've
taken off the number needed for Pebble boards.
In fact, I would propose buying a 20 pack of these, having all regular
HackerSpace attendees chip in a share of money, and keeping them for
communal use in the group, since they're such a nice device, and
commonly used in Arduino-type projects. Everybody building a Pebble
needs one, everyone building an Aiko gateway needs one, everyone just
getting started with Arduino probably wants one... you get the idea.
Clint's Components has 0.1 uF monolithic capacitors in packs of 50 and
500 with good economies of scale. Jaycar sells a 100 pack also, but I
don't recall how much it is.
Similarly, since these are very, very common in other digital
electronics projects, I think it is well worth considering buying a
big pack and having everyone in the hackerspace chip in a little money
as well as the price-per-capacitor corresponding to the ones you need
for orders for Pebble kits.
Buying them one at a time from Jaycar is a ripoff.
If everyone who's a Hackerspace regular chips in say $2, times say 10
people, that's $20, so you get a pack of 500. That's 50 capacitors per
person, which is a large supply for general hacking and
experimentation at the Hackerspace. The price per capacitor is 4
cents, so that's the cost which is cost of the capacitors for each
Pebble kit - a significant cost saving. That 4 cents from every Pebble
kit order can probably just be put back into the HackerSpace petty
cash fund for communal drinks/food/components, if we choose to follow
this kind of model for purchasing common components.
PS: That Clints' Components guy sells lots of really awesome stuff, at
good prices. Very cool. Everyone go and look them up if you haven't
already done so.
Mitch Davis
On Wed, Oct 28, 2009 at 6:31 PM, Luke Weston <reindeer...@gmail.com> wrote:
>
>> As discussed last night, Thomas and I are planning to run a Pebble
>> Parts Program.
>
> Oh, the alliteration.
Thought you'd like it.
> Regular electrolytic is OK. Check both the pitch of the leads and the
> diameter of the capacitor to make sure they're acceptable, but they
> should be.
Noted, and updated. 5mm pitch leads are hard to get. I've put in a
5mm body with 2mm leads, but I think it's not too hard to spread 2mm
legs to 5mm.
> Is it possible to source tantalum caps from a different supplier at a
> better price? To be honest, for this kind of thing if you really want
> the best possible price
That was the cheapest tantalum I could find. And re wanting best
possible price, hold that thought.
> I recommend including the XBee sockets in the "standard kit" since
> That way, you might decide to get an XBee later on, or you might
> borrow one from a friend to temporarily try out an XBee application.
I agree in principle, good idea. (Me, I know I never want to do it)
> they're not very expensive, and have them soldered in even if you're
> not using an XBee.
They add $2 (7%) to the board, as I wasn't able to find many suppliers
of it. Me, I'm cost conscious, and I'd rather spend the $2 on
something else. The good thing about the spreadsheet is that it lets
the cost conscious do their thing, while others can make an all
singing, all dancing one.
> If the individual builder would like to install a 500-Candela uber
> ninja light-up-the-room-at-night blue LED, that's up to you.
If they want to install such a LED, they better not select the tinsy
one in the spreadsheet, as they'll waste 8c! :-)
>> What colour switches should we get? All the same? Different colours?
>
> I have no idea.
Ideas anyone? (Oops, bike-shed question)
> Don't use Jaycar SP0720 switches, they're crap. Maybe use tactile
> pushbutton switches instead.
I tried the ones on Andy's prototype. Yep, they're crap. Can anyone
suggest an alternative?
> Have you considered getting all the components that you're getting
> from Jaycar from Farnell instead, instead of a little from A and a
> little from B?
Yes.
> This would be easier to order and likely cheaper.
No. The prices that are there represent the best I could find from
several sources, roughly factoring in postage. I think the board
would be 40% more expensive if sourced at one supplier, regardless of
supplier, as each supplier has lower prices in some areas but not
others. For example, Farnell's lowest price on basic pin headers is
about $5, whereas Rockby is about 60c. The opposite situation occurs
often. There's only four or five suppliers, and Thomas and I are
happy to do the ordering (and shouting when it's inevitably wrong).
PLEASE, if you think you can find suitable components cheaper, find
them and tell us! (Especially the pot and the relays)
> Some components you might like to order in a large pack to take
> advantage of economy of scale, even if they aren't all required for
> Pebble units at the moment.
At the moment, the aim of this exercise is to get a handful of boards
built as quickly as possible, and as cheaply as possible. The prices
I got today are based around building that handful. We can always
rejig it for version 2.
You had some good things to say about building up an in-house
inventory for the use of Hackerspace members (and as a reminder, not
all readers of this list are CCHS members). I agree with what you're
saying there: It would be cool to seed people with the DS18B20s, and
by buying in bulk, we can pass the savings onto members. But that's
out of scope of the present exercise that Thomas and I are doing (see
previous paragraph). But it's a very good point and we can take that
up IRL or online at CCHS.
> PS: That Clints' Components guy sells lots of really awesome stuff, at
> good prices. Very cool. Everyone go and look them up if you haven't
> already done so.
Indeed. I got an email reply from him tonight: He's started up with
the express focus of helping hobbyists.
Luke, please, if you have time, can you run down the components I've
chosen and see if they're suitable for our use?
Thanks, see you soon,
Mitch.
andyg (geekscape)
On Wed, Oct 28, 2009 at 6:31 PM, Luke Weston
<reindeerfloti...@gmail.com> wrote:
> I recommend including the XBee sockets in the "standard kit" since
> That way, you might decide to get an XBee later on, or you might
> borrow one from a friend to temporarily try out an XBee application.
experiment with Zigbee. Mesh networking will become more prevalent
(especially for device sensor networks) ... and this is an excellent
opportunity to give people (a little nudge) to experience something
that they might not try by themselves. I believe this is the whole
essence of paying a significant amount of money and making the effort
to go to LCA ... which is to get these sort of learning experiences.
I'd really like the default to be that the Zigbee headers are included
as a standard part of the kit, because (with 2 mm pitch) they aren't
an easy item to acquire (compared to ordinary headers).
Okay, good ... we have agreement !
The prototypes must test what is going to LCA ... and not having the
initial alpha-testers testing *all* aspects of the board would be
incorrect process at this point.
And, that's fine ... change the default just for your order. If you
don't want to try / test the Zigbee functionality, that's cool. But,
please don't cause the whole group of alpha-testers to not test all of
the board. We'll also need to know the time it takes to build the
*whole* board, for the purposes of scheduling.
No disrespect intended ... but, I hope that you are the exception and
that everyone else will welcome the opportunity to play with mesh
networking.
On Wed, Oct 28, 2009 at 6:31 PM, Luke Weston
<reindeerfloti...@gmail.com> wrote:
> they're not very expensive, and have them soldered in even if you're not using an XBee.
accounting for all the hours that has gone into this).
It's probably 0.2 % of the cost of going to LCA (if you are from NZ)
and probably < 0.1% of the cost of going to LCA, if you are from
Australia.
On Oct 29, 1:26 am, Mitch Davis <m...@afork.com> wrote:
> as I wasn't able to find many suppliers of it.
others the effort of sourcing this part.
On Oct 29, 1:26 am, Mitch Davis <m...@afork.com> wrote:
> Me, I'm cost conscious, and I'd rather spend the $2 on something else.
I'm working on the basis that people are attending the LCA mini-conf
to get a high quality, value for money learning experience. Note: I'm
not interested in just wasting people's time and money either.
I'll be taking along additional Zigbee modules, so that people can
experiment with mesh-networking on the day ... and then choose whether
to acquire Zigbee modules or not. I believe that experience will be
worth $2 to many people.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
-
I really appreciate your cost-sensitive approach and the significant
effort that you put into sourcing the cheapest components. We've all
benefited from that. And, I don't mean to trivialize that important
engineering aim.
However, in this specific case, I'd like to balance the cost criteria
against the broader experience that we are trying to provide at the
LCA mini-conf and the need to test every aspect that we include in the
prototype.
cheers andyg (@geekscape)
Thomas Sprinkmeier
If you really don't want to pay for an XBee I'll throw it in for free
provided you solder it in and tell us how long it took to assemble the
thing (after all, getting these times is part of the point of the
exercise).
I found alternatives for the caps, diodes and relays at Farnell. They
had trimpots as well but I couldn't figure out which ones to get.
When it comes time to order I'll coordinate with Mitch and might go
for larger/cheaper. I'd hate to see the effort Mitch put into this
going to waste 'cos we end up arguing over a few $$$'s here and there
and end up running out of time.
I think everyone should have access to the spreadhseet now, have at it!
Thomas
andyg (geekscape)
On Oct 29, 10:07 am, Thomas Sprinkmeier <thomas.sprinkme...@gmail.com>
wrote:
> When it comes time to order I'll coordinate with Mitch and might go
> for larger/cheaper. I'd hate to see the effort Mitch put into this
> going to waste 'cos we end up arguing over a few $$$'s here and there
> and end up running out of time.
> I think everyone should have access to the spreadhseet now, have at it!
thanks.
How are you determining that "everyone" has access to the
spreadsheet ?
Perhaps emailing the "everyone" list to this group would be a good way
for us to check whether anyone has been missed.
cheers andyg (@geekscape)
andyg (geekscape)
On Oct 29, 10:26 am, "andyg (geekscape)" <an...@geekscape.org> wrote:
> Perhaps emailing the "everyone" list to this group would be a good way
> for us to check whether anyone has been missed.
list. Sorry, I didn't mean to imply people's email addresses.
Perhaps just use first names and last name initial.
cheers andyg (@geekscape)
Thomas Sprinkmeier
>> for larger/cheaper. I'd hate to see the effort Mitch put into this
>> going to waste 'cos we end up arguing over a few $$$'s here and there
>> and end up running out of time.
>
> The effort that *everyone* has put into this.
My bad!
I meant specifically the effort Mitch put into getting quotes etc for
all the gear.
I realise a whole bunch of people have put a whole lot of effort into
getting this far, thanks to you all amd apologies for any offence.
Thomas
Thomas Sprinkmeier
> thanks.
done.
> How are you determining that "everyone" has access to the
> spreadsheet ?
mostly wishful thinking as it turns out...
I'm new to google docs, I though the invitation sent to the lsit the
otherday was enough.
live an learn.. beats just getting older :-)
> Perhaps emailing the "everyone" list to this group would be a good way
> for us to check whether anyone has been missed.
Failing that, email me (off-list if you prefer) and I'll add you to the list.
Thomas
andyg (geekscape)
On Oct 29, 10:40 am, Thomas Sprinkmeier <thomas.sprinkme...@gmail.com>
wrote:
> done.
Cool, thanks.
> > How are you determining that "everyone" has access to the spreadsheet ?
HughB, ScottP ... (anyone who is likely to attend/assist the mini-conf
or has provided significant feedback and may want to build one).
Even if someone doesn't want to build this particular prototype ...
they may want to get more seriously involved in one of the later
stages (PCB build or complete assembly purchase / test).
> Failing that, email me (off-list if you prefer) and I'll add you to the list.
1) Create a folder called "LCA2010 Pebble" (or similar)
2) Move the "PebbleBits" spreadsheet into that folder
3) Share the folder with those that want access
This means that as additional documents get created and put into the
"LCA2010 Pebble" folder you win two ways ...
a) You don't have to keep adding people to every single new document
(a real pain)
b) When someone new joins, you only have to share the folder with them
in a single action, not each document
cheers andyg (@geekscape)
Luke Weston
There's no harm in ordering components though.
Also, Thomas and Mitch have been doing a great job, much appreciated :)
Mitch Davis
I think Andy's right: I've been viewing this too much as an exercise
to get me a toy, and I've taken my eye off the main goal, which is to
proof what we'll be taking to NZ. Count me in for the other things
too.
Thomas, thanks for looking after the Google Docs side of things and
for finding better parts. (Have you plugged them in?)
We need some kind of signoff on the parts so we can go ahead and order
them. Luke, are they ok?
Assuming the parts are ok, when do we close the order window? I was
thinking Saturday night. That way, it should be enough time for the
LCDs to arrive for each of us to assemble on the weekend of Nov 7/8.
Mitch.
Thomas Sprinkmeier
> for finding better parts. (Have you plugged them in?)
I added them to a new column ("Alternate Source")
> We need some kind of signoff on the parts so we can go ahead and order
> them. Luke, are they ok?
>
> Assuming the parts are ok, when do we close the order window? I was
> thinking Saturday night. That way, it should be enough time for the
> LCDs to arrive for each of us to assemble on the weekend of Nov 7/8.
I figure not much would happen on sunday, why not get people an extra
day to decide and then order first thing monday?
Thomas
Mitch Davis
<thomas.sp...@gmail.com> wrote:
>
>> Thomas, thanks for looking after the Google Docs side of things and
>> for finding better parts. (Have you plugged them in?)
>
> I added them to a new column ("Alternate Source")
Good. After feedback here: Should be buy some of both and try them
for our prototypes? (Esp important for fit issues) Or should we
decide on one and go with it? (I'm trying to keep ordering simple)
> I figure not much would happen on sunday, why not get people an extra
> day to decide and then order first thing monday?
Ok. May mean the LCDs don't come in time for the weekend, but since
they'll just clip on, no drama.
Mitch.
Luke Weston
of exceptions.
You want the Jaycar SY4030 relays or similar. You need the ones that
look like an IC DIP package.
Jaycar has some different green round ones which are slightly cheaper.
These are *not* suitable.
I know they're ~$4 each, but maybe you can find DIP style 5V reed
relays somewhere else cheaper?
The Jaycar round colorful buttons are crap, don't use them.
Jaycar SP0603 buttons or similar would be a better choice.
For the 6-pin relay connector, what I originally had in mind was a
0.1" (2.54 mm) pitch Phoenix-type screw down terminal block. However,
if this is not available or deemed too expensive, I guess a Molex
style locking pin header is OK.
Jaycar has the HM3416 for $0.60 cheaper than the $1.10 of the
currently listed component.
I note you listed some strips of 16-pin male pin headers from Rockby
for the LCD pins.
Might it be cheaper to buy standard large strips of 40 break away pin
headers and break them up yourself?
Remember that you will also need 0.1" header pins, a further 30 or so,
to put on the shield so you can plug in the Arduino.
Other than that everything looks good. :)
Thomas Sprinkmeier
>
> I have looked at the spreadsheet. Most parts look fine, with a couple
> of exceptions.
>
> You want the Jaycar SY4030 relays or similar. You need the ones that
> look like an IC DIP package.
> Jaycar has some different green round ones which are slightly cheaper.
> These are *not* suitable.
> I know they're ~$4 each, but maybe you can find DIP style 5V reed
> relays somewhere else cheaper?
I added a column for alternate parts (you have to scroll right quite a
bit to get there)
are these suitable:
$3.32 http://au.farnell.com/coto-technology/8l01-05-001/relay-dip-reed-spst-5v/dp/1081659
> The Jaycar round colorful buttons are crap, don't use them.
I looked at Farnell but couldn't figure out which ones were suitable.
> Jaycar SP0603 buttons or similar would be a better choice.
>
> For the 6-pin relay connector, what I originally had in mind was a
> 0.1" (2.54 mm) pitch Phoenix-type screw down terminal block. However,
> if this is not available or deemed too expensive, I guess a Molex
> style locking pin header is OK.
> Jaycar has the HM3416 for $0.60 cheaper than the $1.10 of the
> currently listed component.
I'll add it as an 'alternate'
> I note you listed some strips of 16-pin male pin headers from Rockby
> for the LCD pins.
> Might it be cheaper to buy standard large strips of 40 break away pin
> headers and break them up yourself?
>
> Remember that you will also need 0.1" header pins, a further 30 or so,
> to put on the shield so you can plug in the Arduino.
good idea, I'll get a few strips
> Other than that everything looks good. :)
Thomas
Luke Weston
Mitch Davis
>
Luke thanks for all the good oil on the parts.
> You want the Jaycar SY4030 relays or similar. You need the ones that
> look like an IC DIP package.
Noted as an alternative (see below).
> Jaycar has some different green round ones which are slightly cheaper.
> These are *not* suitable.
I just chose those ones because I saw them on the prototype :-)
> Jaycar SP0603 buttons or similar would be a better choice.
Done.
> For the 6-pin relay connector, what I originally had in mind was a
> 0.1" (2.54 mm) pitch Phoenix-type screw down terminal block.
Done. Cheaper too.
> I note you listed some strips of 16-pin male pin headers from Rockby
> for the LCD pins. Might it be cheaper to buy standard large strips of 40 break away pin
> headers and break them up yourself?
Sure. It's cheaper to buy them as 16-way from Rockby, than DIY from
40-way from Jaycar. But if you can find cheaper, please do.
> Remember that you will also need 0.1" header pins, a further 30 or so,
> to put on the shield so you can plug in the Arduino.
I listed them on row 11 as "Shield headers", and it works out cheaper
to do it that way than buy 40-way. But once again, if you can do
better, please go for it. (Curiously enough these are insanely
expensive at Farnell)
On Thu, Oct 29, 2009 at 4:50 PM, Thomas Sprinkmeier
<thomas.sp...@gmail.com> wrote:
>
> are these suitable:
> $3.32 http://au.farnell.com/coto-technology/8l01-05-001/relay-dip-reed-spst-5v/dp/1081659
Done.
> I looked at Farnell but couldn't figure out which ones were suitable.
Hee, welcome to my world :-)
BTW I just noticed I only had 1 14-pin socket, not 2 (fixed now). Can
someone double check the quantities to make sure I haven't made a
similar mistake elsewhere?
Mitch.
Luke Weston
The ratio of those two columns should be the number of component x that are required to build one board, right? Or am I not interpreting it correctly?
Thomas Sprinkmeier
Correct.
Enter the number of sets you want components for on the second sheet and
the third sheet should tally the number of components.
Thomas
Mitch Davis
Well, it should, but (previously) it didn't. All fixed now. Nice pickup!
Mitch.
Mitch Davis
>
> So, can anyone identify anything else that stands between us and now
> declaring the board design finished for now?
I am concerned that the between-IC-pin tracks will present a soldering
problem for those with less experience, especially around the 4094.
The track between pins 2 and 3 of the relay can be avoided by bringing
the track from pin 6 of the relay out to the left of the relay, rather
than up the middle.
The track between pins 9 and 10 of the 4094 can be avoided by routing
the track to the left of pin 9.
Is it possible to reduce the between-pin tracks around the 4094 by
using more vias? (Newbies will be getting PCBs from pcbkart, so the
vias will already be plated)
Mitch.
Luke Weston
>> I am concerned that the between-IC-pin tracks will present a soldering problem for those with less experience, especially around the 4094.
>> Is it possible to reduce the between-pin tracks around the 4094 by
using more vias?
Most of those can't practically be changed, due to the pinout of the LCD and the 4094.
I don't think successfully soldering a 0.1" through-hole DIP chip with a single 16 thou track going between two pads, on a solder-masked board, is an unacceptably hard soldering exercise.
If someone absolutely can not do it, someone at LCA will be able to help them.
>> The track between pins 2 and 3 of the relay can be avoided by bringing...
Done. (I know, done does not mean exported to new images and pushed to repo yet, but I will in a minute.)
>> The track between pins 9 and 10 of the 4094 can be avoided by routing...
Relay designators: done.
Trimpot symbols: can not be changed without significant difficulty because they're set in the libraries corresponding to those parts.
XBee LED: must be towards the bottom of the LED so that it's not obscured by the LCD module in place.
Luke Weston
Luke Weston
Mitch Davis
> I would like to declare the hardware design frozen until
> we've actually made a board and built one.
Sounds great. You've put a lot of effort into it!
Mitch.
Mitch Davis
An update on the Pebble Parts Purchasing Program. If you signed up
for it, please read this carefully.
We now have all of the parts, except for the LCDs. I have done all
the component sorting and the accounting.
For CCHS members, my plan is to distribute parts at Tuesday night's
CCHS meeting. If you're not a member (Thomas and Trent), or you are
but won't be there on Tuesday night, I'll snail mail them to you. (If
I don't already have your address, and you'd like it mailed, please
reply to me with it. I'll tell you later how much postage is). Or
you can come get them from me any time (0411 532 516)
There have been a few minor complications with parts. First, the 100k
LDR companion resistor has been replaced with a 10k resistor (but you
still get the old 100k resistor). Second, the sockets we proposed for
the LCD weren't suitable. What we're now doing for the LCD connection
is to split a 16-way IC socket into two and solder these in-line to
the Pebble PCB, then solder pin headers into the LCD PCB, then push
these LCD pins into the split IC sockets. It's not a great solution
mechanically, but it feels ok and it's all I could do with parts from
Jaycar and Dick Smith. You get to keep the original socket, and we
can work out a better mechanical combination next time.
Regarding accounting, the original formula which computed your total
had a bug: It only considered one of each part, not the cost of a set
(for example, for one Pebble board, it added up 1 14-pin socket, not 2
as it should have). This has increased the price around %10 on the
original estimate.
Second, now that we have the parts, I have been able to calculate the
true cost of each component, taking into consideration overbuy to get
a better rate, GST, and courier costs. The new formula for
calculating your total is based on the true cost of the components.
When I add up what Thomas and I outlaid, and compare it to the sum of
what people ordered based on these true prices, the difference is
about 10c (and no, there's no fudge factor :-)). So I have confidence
in the calculations. What with the formula error, the overbuy, GST
and postage, as we said would happen, the true price is around 25%
higher than we initially calculated.
Luke Weston <reindeer...@gmail.com> wrote:
>
> Eg. there is an economy of scale if you buy a pack of 20 DS18B20s from
> Clint's Components, you're getting them for $2 each.
> If you bought that, myself and/or Andy and/or other HackerSpace
> members would surely "buy out" the rest of the pack of 20, once you've
> taken off the number needed for Pebble boards.
now have the following spare components, valued at $17.60:
9 x 10uF, 14 x 100n, 3 x 1N4004, 3 x 14-pin sockets, 4 x 16-pin
sockets, 3 reed relays, 1 x 24 pins' worth of PCB header.
See column M in the spreadsheet, and column E for prices. Mail me if
you want spares, and I'll update the spreadsheet and put the spares in
your parts bag. Thomas will buy out any spares not spoken for.
So, regarding payment, on the first sheet you can see how much you're
up for, and on the second sheet (row 40), why. I think the easiest
and cheapest way is to pay me via bank transfer. Please email me for
my BSB and account name/number. If you want to pay me cash at CCHS,
please make it exact.
Once you've paid us, Thomas and/or I will update the money columns to
show that you've paid. (I'll sort out Thomas' share of the money
separately)
Thank you for taking part! We had had a lot of fun doing it, and we
hope you are happy with the results.
Mitch and Thomas.
Luke Weston
minor improvements to the layout, and a change to the Xbee pads which
should make it much nicer to drill and solder.
The revisions are really very minor compared to the existing board, so
there's nothing really wrong with the existing boards and artwork, but
still, I recommend the new board artwork be used for new board
fabrication.
Cheers,
Luke
Luke Weston
Some points regarding assembly and manufacturing and components:
1) The blue 16-pin headers aren't much good. Use standard black female
header strips, from Altronics or Sparkfun for example.
2) Parts kits should include 1 resistor for the resistor associated
with the LDR - not both one 10k and one 100k. Either that, or the
presence of both resistors should be documented so you don't end up
with an Ikea style left over resistor which is potentially confusing.
3) Soldering components on the top layer, getting underneath the ICs,
etc, is hard. However, this is not relevant on most cases, only on the
first couple of hand made PCBs without through-hole plating.
4) Some of the early hand made boards may have some holes which aren't
large enough, for example for the voltage regulator. Some holes may
have to be re-drilled out to 1 mm.
5) It would be better to obtain, in future, 0.1 uF capacitors and 10
uF capacitors that have a 5.08 mm pitch, not a 2.54 mm pitch.
6) Certain components, particularly the IC sockets, must be soldered
first, so you can get under them to solder on the top layer. With
other components surrounding them, it's extremely hard. Again, this is
only relevant if you don't have a professionally fabricated board. You
must use machined pin IC sockets, not standard cheaper IC sockets, so
you can solder the exposed parts of the pins to the top copper layer.
Standard IC sockets are fine in the kits with proper PCBs, though.
7) The relays are quite tall and almost hit the top of the LCD when
the LCD is plugged in, It might be better to not use IC sockets for
the relays.
8) You must use a right-angle type for the 6-pin connector for the
relays - otherwise the LCD will block it when the LCD is installed.
8) The 10 uF electrolytic capacitor may be too tall and may hit the
bottom of the LCD. One can either leave a little bit of exposed length
in the capacitor's leads and fold it down on the board - or use a
tantalum capacitor which isn't as tall.
9) The XBee module should be a chip-antenna version, not a wire
antenna version. (Not sure about the external antenna version.) Again,
this is an issue with clearance below the LCD.
Luke Weston
work... you have an assembly bug :)