Review request from a beginner with a simple design for autistic kids

[JensAndree]

If you are reading this - thank you! (if I by posting this break any rules in this forum then I'm sorry in advance. I read the FAQ and tried to comply as much as I could)

Below is a long text describing what I intend to do and in order to get the best help I've tried to describe the project as much in detail as I can. I think I know my way around Eagle now on a basic level but I'm still crawling with regards to my skills in electronics design, but I try to learn.

I've posted a link to the board and schematic files made in Eagle 6.4 in the end of this post.

Timstock Light - What is it and the background:







This is a device for autistic kids in order to visualize time connected to tasks in real life.
The buttons are from left 5, 10, 15 and 20 minutes and they each start a countdown which is visible by leds turning off from right to left.
When a countdown has started it is only possible to cancel it by pressing button 2+4 for 5 seconds but this is not demonstrated to the users. (autistic people can cheat too )
The reason for keeping it as slim as possible is to be able to stick it inside a folder next to a list of activities that needs to be done. Each activity has a value (time+colour) that tells the user how long time he/she should spend on the activity. (this methodology is already used globally) Some users have it on their desk, in the pocket, inside the desk and so on...

It is technically not more complicated than your average arduino blinky button starter kit, but for some it will be very important and should never fail.

It should also be as simple as possible with regards to components and their costs. The commercially available tools like this today are really bulky, heavy and VERY expensive. This is a low-cost alternative that I will release as open source as soon as I have a verified prototype.
I have also thought about robustness and I don't want components to be easily ripped off... (hence why the buttons are thru-hole in my design since I guess they are more sturdy.)
If it's not too expensive I might get a plastic film printed that'll protect the PCB and components on the front but the goal is to keep costs at reasonable level without going crazy. I can't estimate if 100 of these will be built or 100'000 so the design is for fewer and to be hand assembled for now. (older autistic kids can even do it as a kit!)

I am doing this project since many kids and schools simply can't afford to give these tools to kids who need them and the goal is to make it so cheap they can just hand them out to whoever need one. Parents should be able to buy them for their autistic kids without getting ruined and those who are able to build his/her own hardware (or variant) should be able to do so - as long as they don't sell it for a profit! (I will sell it for whatever they cost me to make and I will not take any profit from it. If someone wants to donate for future work then that's ok.)

I've contacted the original designer and I'm allowed to do this and I did a software only (Android) version more than two years ago which recycles old phones into tools instead - available for free. This is very popular but phones are bulky and this is where this project started.

Usage:
Device waits for input from either of the four buttons and starts a countdown when a button is pressed.
The countdown is visualized with the leds and will counted down to 0. (all leds are lit when started and one by one is turned off to represent accurate time that passes)
The countdown will be possible to reset by pressing two buttons (2 and 4) for 5 seconds.
The board is powered by 3v battery on the back.
When not in use it will run in power consumption mode. (a power switch will reset the counter hence why it's not in the design)

Design:
The reason for using ws2812b is both less components and customization possibilities. (+future features) According to spec 3v should be enough to drive the leds and less bright would be better but this I have to test before I know if it's the right way to go. I have used this led before and it's so simple to use - just wire up as many as you want and have power for!
ISP header mounted on the back for both programming and debug features during development. Will not populate on finished product but want the ability for now.
Will ISP header work with current pin-usage on the attiny85?
Is it necessary with one 15pf cap per led or can this be done with one cap with different/same value on a common vcc "rail" to the leds? (took the layout from ws2812b documentation) It is just a cap between gnd and vcc where led led is powered and although schematic seems logical, reality is really just a bunch of caps between gnd and vcc towards the leds.

I have no formal training in electronics design but I'm a tinkerer at home and try to learn new things in life. I have made a few pcb's in the past and I've made the same mistakes as anyone else but since this project is purely financed out of my pocket I hope someone could be kind enough to review my design in order to get something early that actually work as intended when the components are soldered in place. I can't go through too many iterations before I have something I can give the kids.

I have not done anything with the silk screen stuff yet so text etc are in wrong places.
Everything apart from leds and the four buttons can be moved around but I prefer to have the battery centered on the backside and slightly towards the top in case someone want to put it on a desk or table.
I prefer to stick with the attiny85 since I'm rather used to writing code for those microcontrollers and I've never worked with PIC, but if you think this solution is foobar then change I will.

I'd love to receive feedback for the design and suggestions for improvements. I know my skill is on beginner-tinkerer level, hence why I ask for help.
Perhaps my design needs drastic changes, then please feel free to implement them. I'll include the name(s) on the board of people who aided in the design!
Making something quickly at home on protoboard is one thing, making something for the public is way different and here I simply lack the skill.

Routing is something I still struggle with and watching Dave do it makes me wanna forget everything I know and stick to writing code - but I try as best I can. Sadly have none to talk to locally since I work only with software and they don't really talk hardware at all

Why am I doing this? I have two autistic boys and every little piece of electronic gadget that can help them costs stupid money! (one small whyteboard with leds and a digital clock I got the a few years ago - $1200!!!)
I've written tons of software to help them throughout the years and if I do something that I think is useful for others I put in the hours and release them for free. I've done a few hardware projects on protoboard but this needs to be for real and I don't want to give up due to my incompetence and too much time spent doing the wrong things...

Here is the zipfile containing board and schematic.
http://arrendegarden.se/bilder/TimstockSlim.zip

TL;DR
I need help with review (or redesign) of a board I've designed for autistic kids. It's a visual aid for showing time counting down from 5, 10, 15 or 20 minutes to 0.
It will be released as an Open Source project and sold as non-profit in order to make it available to those who might not afford what's available on the market today.
I'm a PCB and electronics noob but I want to become better.

Many thanks in advance!
I hope my lengthy description didn't ruin my chances but this is an important tool for thousands of kids and adults today and I want more to be able to get it!

Jens, bearded Swede

[kizzap]

A couple of things:

First: don't use 6mil traces if you don't need to. Manufacturers will charge you more for it.

Second: Utilize a ground pour (polygon at desired outline, rename to GND). It makes routing a little easier.

Third: Move the battery holder from bottom to top, It means the board won't rock. Which is always good. It will also roughly half the height of the board. Which is also good.

Fourth: if the board is being routed, and the board is not going to be encased, round the edges. I know for a fact that the PCB material from OSH park is quite sharp on the edges.

Fifth: Might be a minor issue, but there is no decoupling cap on the board for the IC. You have one for every LED though.

Sixth: Eagle has this little nasty habit of using a proportional font when it should default to vector, considering that is what it outputs.

Seventh: DONT USE AUTOROUTER   

As it stands, I grabbed the board files you had there, and ran them through Eagle and got how I would lay out the boards. I kept your basic placement, but changed a couple of things. one you might need to pay attention to is the ISP header. Firstly, I changed the pads to octagons, as I needed space between those pads to run traces. I also flipped the component back to the top side, as you shouldnt be soldering a header into there, so it makes it easier to program, as you aren't required to hold the board at a weird angle. I also added a decoupling cap for the uController.

You can find the board files in the attached zip

-kizzap

[JensAndree]

WOW! Thank you kizzap!

The battery will probably(?) have to go back to the bottom side since these kids are infamous for "fiddling" with anything that sticks out and a battery like that will probably disappear in 2 minutes, but the rest of the layout is really nicely done!
If the solution can be as cheap as I hope it to be then we can stick them inside folders etc and be available for everyone. If a more robust version is needed then I could possible put a shell around it but for low-cost I wanted to test a bare PCB and see if it could work?

I have loads to learn yet but by comparing my initial "layout" with your professional one - I have to say that I probably learnt more by comparing the two than I've done so far!

The hw design is not based on an already existing solution, it's based on an existing methodology which I've been allowed to copy. (there is a lot of work behind the layout and function)

I will examine the files in more detail now but wanted to express my big thanks for the help!!!

[amwales]

Can you add some mounting holes.
You may want to think about putting this in an enclosure.
A small simple 3D printed case, maybe with a simple clip mechanism on the back would be very cheap to produce in small quantities. Larger quantities and you want to look at some form of injection molding I guess.

[KJDS]

I would find a standard injection moulded abs plastic enclosure. it will make it look so much better and increase the robustness significantly.

[JensAndree]

I really hope not to use a case since the thickness is critical but this will be shown during usability testing... (one requirement is that it should be able to glue/tape it to a page in a folder in order to use it with an activity schedule on the opposite page, hence why it should be as thin as possible.)
I have thought about printing a "sticker" with a window for the LEDS as protection as well but currently I need to sign of on the layout and usability.

There can always be an option to have it in an enclosure but that's something I have to look at after the form-factor and usability has been "passed"!

I realized that the 3.3v from the battery wasn't enough for the data pin for the LEDS so I have now put a 3.3-5v boost converter on the backside, behind the battery.
I made a few more minor changes as well and these boards are currently on its way from OSH Park.

Thank you for the tip about mounting holes! They can be useful for more than screwing it in place inside an enclosure so that's something I shall make room for.
I just need to remember to isolate the mounting holes from the ground plane.

I have documented the whole project including design, code and philosophy, sadly only in Swedish but I will translate it ASAP for the international readers.

Once again, thanks for all the help! When I have something running I'll update this thread!

[kizzap]

A thought on the enclosure: perhaps you could get an enclosure that clips over the board to cover the majority of the components. Makes things a little neater, especially if you can keep the battery on the top side.

I am assuming you are using a switching boost converter? can I suggest uploading an image of the board layout for that so we can check it for you? SMPS can be a pain to get right.

-kizzap

[JensAndree]

After I put in the boost converter I read that they are rather "picky" regarding the design so I guess I'll have to see if my implementation works or is flonked...
It is a big step from doing things on protoboards and designing low-power smd stuff, but I've learned so much already and kizzaps routing help taught me loads and I have re-routed a few other of my projects since with much better layout and routing - including proper ground-planes - thank you yet again for this!

Here is my layout of the boost converter and it is perhaps not exactly according to spec and I guess I will find out this coming week when the first boards arrive.







I made the layout on the backside since the front is rather busy, and perhaps this was a bad choice if the switching boost converter is sensitive, but if so then I've learned something new and will make the necessary changed for next revision.

I do appreciate all the help and feedback since I'm learning as I take each step

[JensAndree]

wow, i agree, alot of these "aids" equipment, they cost stupid money really. i have been through a slightly different version of what you are going through, and i guess it is an on going fight to find a better balance in having some quality of life (or better version of it) for everyone in that particular circle. in some countries, it will really cost you an arm and leg, even to seek official medical support even in gov medical agencies ...

i have some random thoughts, 

1)would it be better if they see actual numbers in counting down? or there are numbers stickered to each LED indicator? ie --> does it make sense to put in extra subcircuit for a 7 segment drive?

2)would it make sense to spend say $5 more on components and make it operate from 1x 1.5v cell? long run savings vs just using a single cell convenience?

not sure if those are worth thinking about (or maybe i over looked something), but i was just trying to visualize it virtually if i am trying to use it somehow.

It is common to want to resort to numbers or other forms of identifying the "counting" but one has to keep in mind that many of the users of this tool - which has existed in a -very- expensive form for many years - don't see things in numbers or letters without losing focus and it is much cleaner, and simpler, by just looking at the "length" of the light-bar and see it getting smaller until it is gone - hence the time is up.
We try to use these tools on very young kids who are autistic or have ADHD and are unable to grasp the concept of time and by doing this as simple as possible they quickly learn how to do something for 5, 10, 15 or 20 minutes depending on which button you pressed to start it.
The buttons also have different colours since many of the kids can't relate to a number so a colour is instead connected to a task or an exercise. (blue, green, yellow and red)

An autistic person, especially a child, has a very different way of thinking and although I today understand many of the things my two sons think - I still have no actual clue how they think, only why. One thing is for sure - they react to so many more details in their environment as a "normal" do and we have learned that "less is more" for these kids since it gives them fewer choices of how to interpret the information.

I can probably write several books on the subject but will not deviate from the task, at least not in this forum

The reason for the current battery-setup is purely based on what is available in most stores here compared to what's required to run the electronics. (although I might use a 20 mm battery in the next version?)
The WS2812B are there simply because I had some at home and they allow for "any" colour to be used, not just red, or whatever colour you pick - plus you don't need shift registers etc and they are very simple to use! (they are also relatively cheap to buy.)
I have made another version in Eagle that uses an low-voltage attiny and low-powered leds that would require less powerful batteries but it quickly becomes more complicated with regards of controlling the leds with shift-registers etc...

If this design proves unreliable from a power consumption point of view I will resort to one of the other solutions I have but I hope I can get it to work since it has the fewest components - and allows for individual adaption.

If I can get this version to work then it will be so much cheaper for the end-users and everyone will be able to afford atleast one! Schools can give them to whoever needs one without having to fight for extra budget etc... I have got a few "official" tools for my kids and the money they costed was both stupid and disgusting... I don't mind if people earn a living on what they make - after all we can't do everything for free, but why suddenly an electronic device cost 100 times more just because it is for someone with a handicap is for me NOT ok and this is why I have decided to make this available to the public as a non-profit Open Source project. I have made quite a lot of tools for my kids and I try to make them available for free as and when I have the time and whilst all of them so far has been pure software project this is the first that is also hardware.

I do this so others won't have to pay $160 or more when they can pay $15 for the same - or better. (or build it themselves )

Thank you for you comments

[kizzap]

After I put in the boost converter I read that they are rather "picky" regarding the design so I guess I'll have to see if my implementation works or is flonked...
It is a big step from doing things on protoboards and designing low-power smd stuff, but I've learned so much already and kizzaps routing help taught me loads and I have re-routed a few other of my projects since with much better layout and routing - including proper ground-planes - thank you yet again for this!

Here is my layout of the boost converter and it is perhaps not exactly according to spec and I guess I will find out this coming week when the first boards arrive.

http://www.arrendegarden.se/bilder/boost-3.jpg



http://www.arrendegarden.se/bilder/boost-2.jpg

I made the layout on the backside since the front is rather busy, and perhaps this was a bad choice if the switching boost converter is sensitive, but if so then I've learned something new and will make the necessary changed for next revision.

I do appreciate all the help and feedback since I'm learning as I take each step

Do you want the good or the bad news first?

I'll assume good news.

The layout of the controller isnt terrible. It isnt great, but at least there is a resemblance of star grounding for the device. added to that, the main ground return path hits the capacitors first.

The bad news? I have no idea where you plan to get a 68uF capacitor that can fit a 0805 sized pad. even 10uF is going to be pushing it. It is possible to get a 68uF tant that fits in a 0805 pad, however it is only rated to 3V, which imo is too tight a margin ( I would be suggesting to have the voltage rating at least 6.3V.) I haven't looked into the sizing of the inductor, but I have a feeling that it might also be too small.

-kizzap

[JensAndree]

Do you want the good or the bad news first?

I'll assume good news.

The layout of the controller isnt terrible. It isnt great, but at least there is a resemblance of star grounding for the device. added to that, the main ground return path hits the capacitors first.

The bad news? I have no idea where you plan to get a 68uF capacitor that can fit a 0805 sized pad. even 10uF is going to be pushing it. It is possible to get a 68uF tant that fits in a 0805 pad, however it is only rated to 3V, which imo is too tight a margin ( I would be suggesting to have the voltage rating at least 6.3V.) I haven't looked into the sizing of the inductor, but I have a feeling that it might also be too small.

-kizzap

Yes, I did wonder about the caps since they seemed very small, and I know where it went wrong as well... :p

For some reason I must've entered 68pf into DigiKey search instead of 68uf... doh! (or picked the wrong value in the drop-down list... epic fail regardless!)
The good news is that I will never do that mistake ever again and "learning by doing" is by far the best method I've come across the last 40 years... hehe.

SMD is totally new to me and I fell into the trap big-time, but this will have to be addressed in the next version of the board.
Normally I'd prototype something on breadboard, then protoboard and finally perhaps a real pcb. With smd it relies on skill and I now know more than I knew a couple of weeks ago!

The inductors I've ordered also seems very small and inductors are yet to be included in my skill-set so this was merely picked by chance...

Code: [Select]

Inductance Tolerance Current Rating Current - Saturation DC Resistance (DCR)
47µH ±20% 300mA 320mA 1.608 Ohm Max

I assume this inductor is more in line with what I should use instead? If it is then it makes sense.
http://www.digikey.se/product-detail/en/CR54NP-470LC/308-2212-1-ND/3947435

I have now looked at the schematics, and more importantly, the board layout of SparkFuns NCP1402 5v boost converter and I see much beefier components.
Note to self: check for existing solutions after own attempt in order to avoid stupid mistakes in the future!
I will use a similar 5v boost converter on a breakout board to test my first prototype boards so it's not a total failure, but I do have more work to be done for sure!

Thank you for noticing this - perhaps blatantly obvious - beginners mistake and pointing it out! It would've taken me some headscratching with a multimeter before finding out my mistake for sure... My implementation of the boost converter is way out of range!

I grew up long before the internet was what it is today (although I was an active user in ISCA from 1989!!!) and I learned 6502 and 68k assembler by reading books and trying until I got it to work. Today I seldom google for the solution since I don't believe I can learn as much from that compared to finding the solution myself. This does require me making some stupid mistakes along the line but I learn from the mistakes and today I have learned an important lesson! (actually more than one lesson, but one is rather "bigger" than the others )

I hope you don't mind me trying to learn by not asking for details regarding every step of the process? I do however appreciate being corrected and please don't refrain from pointing out my mistakes since I do want to progress as well. I also understand that what I'm trying to do is above my skill-level, but that's something I have to overcome - and with your help so far I have learned more and I'm grateful for the time you spend on helping me!

Next version I'll post for review before ordering in order not to make another stupid mistake since the turnaround-time for the pcb's are rather long... And I accept that I can't iterate through every noob mistake since then this will take forever to complete.

Perhaps I should take a minute to explain why I have chosen the design and the components.

The WS2812B leds can blink fast enough for POV (persistance of vision), and they require very little components to run and minimal pins on a microcontroller. I have made quite a few projects with WS2812A and WS2812B for interactive "paintings" (I have them framed behind a canvas and a small camera detects the environment and change depending on colour, brightness and movement) and they have become my favourite RGB leds. I've written a library that allows me to control a lot of them individually in a very fast manner. They have a ~18mA constant current draw (50mA at full brightness)
In a perfect world I'd like to have 20 leds on this board but I could only fit 10 and by using different colours I have created a different methodology of displaying decreasing time that will be easy for someone with a cognitive handicap. (I am first with this idea and is something I've kept a secret until now since I'd hate if someone copied it before it's released in order to "copyright" it to stop me. There are "upset" people right now since I do this as a non-profit project...)

The data-line on the WS2812 needs +5v according to spec but I have run them as low as +3.3v. Using 3v batteries will not be enough so that's why i added the boost converter.
First I planned to only boost the data line but it would be much simpler to boost VCC to +5v and I decided to test and see what kind of performance I would get with regards to power consumption. Perhaps there are more suitable level-shifters for this task but I have very limited experience in this area so please let me know if there are better solutions to the problem?

By making the tool small enough, older autistic people - who might want to hide the fact that they need a tool like this - can have it hidden, or actually get away with it since it will look like an electronic toy instead of a "handicap-aid". It can also be hung around the neck in a string so the younger user can have it with them whatever they do. It must also be able to mount inside a folder next to the daily schedule, a task list or a step-by-step list with different times specified for each step. The usage is wide and often individually adapted but the available ones today are both really big and they draw unwanted attention.

I will also add a piezo membrane for audible feedback. (I hoped to do without this initially but my latest survey from the people who use these wanted this feature as well.)

A minimalistic plastic case is something I will look into since I more and more realise that I need to protect the board from the environment.
The clip-on suggestion is really good and could perhaps be made without costing too much? It can be made as an optional extra for those who need one "on the go".

I hope it makes some sort of sense?

Many thanks kizzap, and you others, for taking your time to assist this beginner in the world of electronics!

[kizzap]

Yes, I did wonder about the caps since they seemed very small, and I know where it went wrong as well... :p

For some reason I must've entered 68pf into DigiKey search instead of 68uf... doh! (or picked the wrong value in the drop-down list... epic fail regardless!)
The good news is that I will never do that mistake ever again and "learning by doing" is by far the best method I've come across the last 40 years... hehe.

SMD is totally new to me and I fell into the trap big-time, but this will have to be addressed in the next version of the board.
Normally I'd prototype something on breadboard, then protoboard and finally perhaps a real pcb. With smd it relies on skill and I now know more than I knew a couple of weeks ago!

The inductors I've ordered also seems very small and inductors are yet to be included in my skill-set so this was merely picked by chance...

Code: [Select]

Inductance Tolerance Current Rating Current - Saturation DC Resistance (DCR)
47µH ±20% 300mA 320mA 1.608 Ohm Max

I assume this inductor is more in line with what I should use instead? If it is then it makes sense.
http://www.digikey.se/product-detail/en/CR54NP-470LC/308-2212-1-ND/3947435

The best way to pick if the components are suitable, is to go through the Maths! If you look through the datasheet for the converter you chose, it gives you a table of formulae. Go top to bottom through them, and it will give you the components you need. Find the appropriate components (digikey or where-ever) and then you can start worrying about the PCB layout

have a watch of this video:

I have now looked at the schematics, and more importantly, the board layout of SparkFuns NCP1402 5v boost converter and I see much beefier components.
Note to self: check for existing solutions after own attempt in order to avoid stupid mistakes in the future!
I will use a similar 5v boost converter on a breakout board to test my first prototype boards so it's not a total failure, but I do have more work to be done for sure!

Thank you for noticing this - perhaps blatantly obvious - beginners mistake and pointing it out! It would've taken me some headscratching with a multimeter before finding out my mistake for sure... My implementation of the boost converter is way out of range!

I hope you don't mind me trying to learn by not asking for details regarding every step of the process? I do however appreciate being corrected and please don't refrain from pointing out my mistakes since I do want to progress as well. I also understand that what I'm trying to do is above my skill-level, but that's something I have to overcome - and with your help so far I have learned more and I'm grateful for the time you spend on helping me!

Next version I'll post for review before ordering in order not to make another stupid mistake since the turnaround-time for the pcb's are rather long... And I accept that I can't iterate through every noob mistake since then this will take forever to complete.

We all make mistakes on our PCBs, hell I only just recently caught an error on a PCB I designed, two weeks after sending the design out to be made. Thankfully I have already figured out the error, and have a planned work around in place

A good place to look for the best way to lay out components that you are not sure of is the datasheet. Many smart people spend weeks on getting that design perfect, so why not copy it? It's not like the company who designed the part is going to complain, I mean hell, they are getting sales.

Perhaps I should take a minute to explain why I have chosen the design and the components.

The WS2812B leds can blink fast enough for POV (persistance of vision), and they require very little components to run and minimal pins on a microcontroller. I have made quite a few projects with WS2812A and WS2812B for interactive "paintings" (I have them framed behind a canvas and a small camera detects the environment and change depending on colour, brightness and movement) and they have become my favourite RGB leds. I've written a library that allows me to control a lot of them individually in a very fast manner. They have a ~18mA constant current draw (50mA at full brightness)
In a perfect world I'd like to have 20 leds on this board but I could only fit 10 and by using different colours I have created a different methodology of displaying decreasing time that will be easy for someone with a cognitive handicap. (I am first with this idea and is something I've kept a secret until now since I'd hate if someone copied it before it's released in order to "copyright" it to stop me. There are "upset" people right now since I do this as a non-profit project...)

The data-line on the WS2812 needs +5v according to spec but I have run them as low as +3.3v. Using 3v batteries will not be enough so that's why i added the boost converter.
First I planned to only boost the data line but it would be much simpler to boost VCC to +5v and I decided to test and see what kind of performance I would get with regards to power consumption. Perhaps there are more suitable level-shifters for this task but I have very limited experience in this area so please let me know if there are better solutions to the problem?

By making the tool small enough, older autistic people - who might want to hide the fact that they need a tool like this - can have it hidden, or actually get away with it since it will look like an electronic toy instead of a "handicap-aid". It can also be hung around the neck in a string so the younger user can have it with them whatever they do. It must also be able to mount inside a folder next to the daily schedule, a task list or a step-by-step list with different times specified for each step. The usage is wide and often individually adapted but the available ones today are both really big and they draw unwanted attention.

I will also add a piezo membrane for audible feedback. (I hoped to do without this initially but my latest survey from the people who use these wanted this feature as well.)

A minimalistic plastic case is something I will look into since I more and more realise that I need to protect the board from the environment.
The clip-on suggestion is really good and could perhaps be made without costing too much? It can be made as an optional extra for those who need one "on the go".

I hope it makes some sort of sense?

All I can find about those LEDs with the on-board driver is, truth be told, a basically useless datasheet.

As a general rule, your HIGH and LOW signals will operate as a percentage of your Voltage supply. In fact, some chips will actually start to complain (releasing magic smoke in some cases!) if your input voltage on an IO pin is greater then VCC. I dare say, if they worked at 3.3V, don't change it.

Many thanks kizzap, and you others, for taking your time to assist this beginner in the world of electronics!

No problem, always enjoyable trying to solve a problem

[JensAndree]

Thank you kizzap!!!
If, and when, this project is finished I will send you a Timstock Slim (as it's called) as a memory of this perhaps not so usual project, but who will help many people out there.

I have now studied and read more on DC-DC converters and I think I understand why they work as they do, but I have a long way to go before I can fully understand the details and know enough to design a solution like the SparkFun NCP1402 breakout board with good performance... Luckily I find this both challenging and interesting so it's a really great hobby I've stumbled upon!

Are there any special things to bare in mind when designing with boost converter with regards to CE certification and similar? I assume I have to get it certified if I want to be able to sell it legally?
I know this will be expensive but I think I can apply for a grant that would cover these costs... (unless there is some other way to legally sell "homemade" electronics without certification?)

I often think digital electronics (microcontrollers, integrated circuits, logic gates, transistors etc) is so much simpler than analogue electronics (audio filters like RIAA and "old" things) but that's without adequate experience and I often lack generic basic skills that would be beneficial at this point in time.
I did learn ohms law and similar things in school but I only learned the difference between a npn and a pnp -transistor last year and I often build simple circuits on a breadboard and using my DSO to probe and see what actually happens, and why?
I'm going to start to read entry-level books in electronics design to better understand the building-blocks of the modern electronic components we use today since I now know that it is not enough to understand how to wire up and program an atmega microcontroller and what value resistors I need to use for a blinketyblink project, I need to know the whole chain and better understand how to do this without the microcontroller!

Last year I made a "robot" with the kids and I didn't have any h-bridges at home to control the motors with so I made my own with some protoboard and some transistors, resistors and diodes. If I hadn't done this I would never have learned how they worked and this was very helpful.
I have also built an 555 on a wooden board with nails and airwires to both understand it and it also looks great! It only blinks a few leds but I could probe it in all places and that's how I started to understand transistors the first time. This I recommend to all beginners out there - it's a fun and simple build and it's something you can be proud of later
I think I used 20 transistors in all (11 npn and 9 pnp) and a handful of resistors. I admit I didn't understand how it would work whilst building it and I did have a problem with the threshold comparator since it got way too hot initially but after some tweaking it worked. Sadly none of my friends understood why I did it, what it was and why I used all those components just to blink a single led, but I was proud of what I made and the kids thought it looked cool.

I have now changed the schematics and redone the layout with the correct passives and I will post it for review/giggles here later but first I shall assemble a board as soon as they arrive in mail and see if my design actually works, and if there are any major alterations that needs to be done?
I will both try a SparkFun NCP1402 as power source with the chosen battery and directly with a bench supply so I can see how low voltage I can drop to and still make it work according to plan.
Next batch of boards will correct the noob mistakes I made the previous round but I am happy I did these mistakes, otherwise I wouldn't have learned anything - and this is important since there are many many more things I'd like to build in the future! (I have always dreamed about making my own Amiga on my own pcb! I have the schematics but it would be a massive undertaking and just sourcing some of the components would be cumbersome... In a few year perhaps I could do something like that but I first have to master a 2-layered pcb before even attempting a 4-layered )

One final question:
Today I use all the pins on the attiny85 and I need to add a piezo "speaker" in order to tell the user with a sound as well as blinking the leds.
This is simple to do codewise (change a pin to output instead of input and then back again after playing the sound) but since the available pwm pins are connected to the tactile buttons as well I worry that the piezo membrane will accidentally trigger a "button press" if the user drops the board or in other way cause any form of pressure on the mounted piezo.
I know there are piezo membranes that comes in a plastic housing which would make this less of an issue but since I want it flat that's out of question at this point in time.

Is it possible to limit the "output" of the piezo but still being able to make it sound? I know I can test this but I thought I check what you think about this before the weekend when I hopefully should be able to make the first prototype...
I would prefer not to swap the attiny85 to a attiny84 since the board is already rather full on the top side at the moment but will do that change if needed.

Anyhow, yet again thanks for all the help and support!!!

Now I shall try to not think about the white stuff that fell from the sky this evening... Spring finally arrived a couple of weeks ago and snow is something I really did not want to see again i many months!!! I wish I could buy a house i Australia where I could live between October and April but that's sadly not feasible...

[kizzap]

Wall o' text!

No need to send one my way

As far as CE certification goes, I have no idea. At a guess they would be more worried about small parts then any electronics circuitry. There might be a provision for radiated noise, but I think that is covered under other standards. But as I said, I have absolutely zero idea about that. I am certain that there are projects on tindie that haven't passed any CE certifications.

If you want to save a couple of buttons, I can see a way to connect 4 buttons to 3 pins if that helps? It does involve the use of a couple of diode though...



Basically what you need to do is put interrupt routines on whatever pins you connect PIN1 or PIN3 to, which then read PIN2, and chose what to do depending on the state of PIN2, with it being HIGH or LOW, thus telling you what button was pressed. Frees up a pin for you As a side note, I would try use a small signal diode here, one with the smallest forward voltage drop you can get. I believe 0.3V is possible. IIRC you had a couple of hidden menus, so this might screw up that, as you would only be able to effectively use one button at a time.

As for lowering the volume of a speaker, again, I have no idea, but I suspect it might be as simple as lowering the input voltage going across the speaker.

Hope this helps a bit

-kizzap

[amwales]

If you need to reduce the number of IO pins you are using, this is a great document
http://ww1.microchip.com/downloads/en/DeviceDoc/01146B.pdf

1. buttons - it should be possible to scan many buttons with a single ADC pin.
TIP #5 Scanning Many Keys With One Input

2. LEDs - You should be able to address LEDs using Charlie Plexing
TIP #2 Input/Output Multiplexing

[JensAndree]

Thanks for the suggestions!

The example kizzap shows is totally new to me and really cunning! Very clever and this I shall remember for future layouts when IO is sparse...

There are "hidden" button combinations for cancelling the timer and setup so bitbanging will not work either, but I will read the pdf amwales linked and see what goodies lies behind that.
One possible way out of the problem is to pick a pwn pin that is connected to a button, but a button which is not part of the combinations (I can make it that way) and then maybe I can get away with it..?

[rant]
The pcb's was still not in the mail today so I suspect the Swedish customs are holding it for the traditional two weeks before they open it just to discover that it's not tax ovation or similar.
Even if they clear your shipment you still have to pay an admin fee since they opened it! This is really stupid since they shouldn't have checked it in the first place and now I have to pay $18 just to have it delayed another two weeks...
[/rant]

The leds have a serial interface so they only require one pin regardless of how many leds you daisy-chain. This is one of the reasons why I almost always use WS2812B and in the past LPD8806 since they are so simple to use. vcc + gnd and one data line which is chained from data out to the next leds data in etc etc, that's it.
Unless you have hundreds of leds one arduino can easily drive the whole thing without external power supply.
They are bought straight from factory in China and although some people have had less than perfect yield sometimes, I have never had a bad led so far and the "paintings" I have at home which I made with WS2812A haven't been turned off for over a year and they are equally bright today.
They used to be quite expensive but the cost for the WS2811 (led driver chip that is built into every led) has really come down the last 6 months and if you buy bulk the price is very reasonable!

I wish my pcb's would be here now and although I love the quality and ease of use of OSH Park I need to find another fab-house in Europe somewhere, just to have faster turn-around times.
Alternatively I could try to etch my own boards? I have a laser scanner and a laminate machine at home so I guess I can easily transfer the ink to bare copper boards.
I have to check what etching acids are available here though since we no longer can buy virtually any acid due to "the war on terror".

What are homemade pcb's like to reflow with hot-air stations and solderpaste? Do you have to pre-tin the pads or will it work with only flux?
This is an area where I have no experience in yet apart from a few things I did really long time ago but that was made with UV lamps and photo etching.

The time in the northern hemisphere tells me to go to bed now since it's still in the middle of the week but I will first have a quick look at the pdf document

Once again, thanks for all the help!

[JensAndree]

A quick update.

The mailman has delivered the boards and soldering has commenced. The first board was populated without an attiny85 since they can be a pig to reflash so I wired up a breakout board that connects to an Arduino Leonardo for simpler software handling.

Everything works fine and I've bypassed the boost converter, and guess what... It runs just fine on 3.3v so there's a slim chance it runs on 3v as well, but I will test this later tonight with a proper bench power supply.
I'm running the LEDs at 10% capacity in order not to blind the kids and the amperage seems very reasonable unless I turn all colours on max, then it draws appx 230 mAh, but that's never going to happen.

Last time I did smd reflow with hot air was back in 1988-90 I think so the first caps were removed and after some cleanup I tried again - this time with much less solder.
The board looks fine and was easy to assemble but I will have to get someone to assemble them if the orders starts to pile in, unless I make a solder stencil and get my speed up?

If I can do without the boost converter I only have components on the top side and that will make assembling much easier.

Next step is to get test code on an attiny85 which runs a fixed routine so I can get performance data out of it. At the moment I have no clue how long it will be able to run on a battery, actually I don't know if it'll run at all on a battery at 3v and 35-40mAh capacity and in hindsight I should've gone for a 20mm coincell battery from the start since most of them have 225mAh capacity...
Learn by doing

I will try to run some more tests tonight after the kids have gone to bed but the day has been successful and promising

[kizzap]

1. buttons - it should be possible to scan many buttons with a single ADC pin.
TIP #5 Scanning Many Keys With One Input

Just quietly. That is brilliant. IMO much better then the solution I came up with. Still gives an issue when you need to use more button then one though. Jens could possibly get around the hidden feature list by splitting the buttons in half.

A quick update.

The mailman has delivered the boards and soldering has commenced. The first board was populated without an attiny85 since they can be a pig to reflash so I wired up a breakout board that connects to an Arduino Leonardo for simpler software handling.

Everything works fine and I've bypassed the boost converter, and guess what... It runs just fine on 3.3v so there's a slim chance it runs on 3v as well, but I will test this later tonight with a proper bench power supply.
I'm running the LEDs at 10% capacity in order not to blind the kids and the amperage seems very reasonable unless I turn all colours on max, then it draws appx 230 mAh, but that's never going to happen.

Last time I did smd reflow with hot air was back in 1988-90 I think so the first caps were removed and after some cleanup I tried again - this time with much less solder.
The board looks fine and was easy to assemble but I will have to get someone to assemble them if the orders starts to pile in, unless I make a solder stencil and get my speed up?

If I can do without the boost converter I only have components on the top side and that will make assembling much easier.

Next step is to get test code on an attiny85 which runs a fixed routine so I can get performance data out of it. At the moment I have no clue how long it will be able to run on a battery, actually I don't know if it'll run at all on a battery at 3v and 35-40mAh capacity and in hindsight I should've gone for a 20mm coincell battery from the start since most of them have 225mAh capacity...
Learn by doing

I will try to run some more tests tonight after the kids have gone to bed but the day has been successful and promising

You have an ISP header on the board. It is literally all you need to program the chip!

Not surprised about the thing working fine at 3.3V

Solder stencils will speed up your assembly time, however I wouldn't organise getting one till you finalise your final board, unless you wanted to try OSHstensils (I have no idea what they are like btw)

[JensAndree]

Hi all!
Sorry for going AWOL for a while but my I've been to the hospital in order to finally get the go-ahead for the surgery I've needed for the last 4 years... They have finally decided that I do need surgery now and I should get a date shortly!
I fell down some stairs in Feb 2010 and damaged three disks and I have since tried to get surgery to fixate the disks and free the trapped nerves and after years of rehab they finally agreed to cut me open! It will probably happen later in May and this I very much look forward to since I've been unable to walk for the last 2.5 years and not work at all for the last 6 months...
We always try to pretend that Sweden has the best healthcare system in the world, but reality says the opposite.
The last couple of weeks have been particularly bad and I'm stuck in bed for 22+ hours every day and that's not human... 

Enough about that and back to the world of electronics!

I have built two units and done a lot of testing in order to ascertain the power consumption and battery life and it's been less than great...
The LEDS are also so bright that I have to run them at 5% effect and I think this is perhaps not the best solution after all... They draw quite a lot, more than I can afford I think. (I also have a heating issue on C1, the first capacitor that connects to VCC in the first LED and this I have not yet figured out, but I have not checked it with a multimeter yet to see if too many volts, or amps, is running through since I only discovered it the same day as I had to go to the hospital and I've been stuck in bed ever since.)
Newer batches are also reported to require +5v VCC and the difference between the voltage of VCC and the data lines are required, hence why I would have to stick to the boost converter anyway... (unless I redesign the whole thing with smaller single-colour LEDs or something down that line)

Before I make a decision to work around my problems or bin it I have started to look at a different design and I wonder if you have any suggestions on good led drivers?
The simplest way would be to use a couple of 74HC595 shift registers I guess but perhaps there are better components out there? I have used the WS2011 drivers but VDD min @ 4.5v makes them useless in this project.

The goal is to make it as good as possible with regards to power consumption but not without making the tool too expensive...
It should run @+3v or less and with a single 20 mm button-cell battery. (not the 12.5 mm I initially went for)
I can solder 0805 without any problem but many of the LEDs with low current consumption are way smaller and probably too hard for me to hand solder, and this would force me to have them assembled in a factory, which I can't fund for the first batches...
Do any of you know any hand-solder-able (hot air or iron) smd LED's that are about 5mA in current-test? 0805 would be great but I can't find anything at DigiKey.

I should be able to make a new PCB, order and receive it just before it's time to go under the knife so any input is more than welcome!

I have learned A LOT by building this first prototype and with the knowledge I have now a different approach would be better, and now understanding the real power requirements compared to what I "though" would work.

Sorry for yet again pestering you but I would really love to see this project through because there are many people who really would benefit from this tool.
I'm leaning towards a change towards "stupid" LEDs and shift registers but in the end of the day I don't have enough experience to make all the right decisions and I hope you can help me to make the best choice for the next step?

Many thanks in advance

p.s the original inventor to the Timstock (early 80's I think) contacted me a few weeks ago and he loved what I'm doing and if there ever was a legal obstacle down the road it is now gone!

TL;DR
Ditch the current solution and go with smaller leds and shift registers? Which leds? Other solutions?

[SL4P]

I love your purpose and interest in developing this.
Congratulations.

Without reading all the posts, I did see that you are trying to achieve minimum thickness, as well as removing the fiddle factor for users.

How about flipping the board over - putting all the components on the rear, and only having the buttons and LEDS peeking forward up through routed holes in the PCB.

That way the front is completely clear, no protrusions at all, except the buttons (and your label).
It would be nice to think about a different power source - perhaps a rechargeable cell, and a USB (or other) plug - which would allow you to epoxy dip the whole board and 'everything' - with just the LEDs and button tops peering up through the board. (indeed- You could use membrane switches on the top surface for even less complexity)

Cheers & good luck - you should put this on kickstarter
- and deserve to make a bundle that you can reinvest in further disability aids.

[theatrus]

You can get Li coin rechargeable cells which are series limited, but their capacities are really low. Would need to do the calculations on it to see if the lifetime is ok (as LEDs are power hungry compared to everything else).

As for continuing this work, even though I have no need for this product, I would help back any fundraising effort. I'd suggest using the funding to get a nice injection molded case made.

[JensAndree]

Had to take a longer hiatus than expected since my back problems turned worse than ever before, and it has recently been discovered that I, apart from three dodgy disks, also have a fractured pelvic bone - which I've had for more than four years! Surgery is now possible and on the way and with a bit of hope I should be able to recover somewhat within the next few years! 

Enough about that. I've been doing tests with the WS2812B and a 20 mm coin-cell battery and I just couldn't get any sort of performance out of it so I have instead gone for 1206 smd leds, and doubled the number of leds to 20 in order to increase the perceived feedback from time counting down. These leds draw a lot less current so battery will last a lot longer - and this was a critical requirement. I have added three 74HC595 shift registers to control the leds.

Sadly it won't be able to shift colours but that's not a critical function and it always was just for fun.

I have also increased the size of the board in order to cater for the bigger footprint of a 20 mm battery holder instead of the badly chosen 12 mm one I had in the first design, and the boost converter has been removed.

Since I made a lot of changes and aquired more footprint I replaced the microcontroller with a attiny84 and now have plenty of IO for the buttons. No impact on cost, and the new leds made the BOM cheaper!


Illustration of the current prototype.

I also decided to learn how to etch my own PCB's so I can speed up prototyping, and my first attempt with a laser printer and a clothes iron turned out really well! Even the really fine traces I did to test what could be made turned out great. Making vias on home-made boards is time consuming but I make somewhat larger vias than normal (40 in Eagle) and then I drill a tiny hole which I put a small wire through, leave a tiny excess on both ends and fold + solder to the traces on both ends seems to work just fine.
Making my own boards have been very rewarding since I really feel like I'm developing something and not just drawing something in Eagle

Sadly I haven't taken any pictures at the moment but I will update with pictures as soon as possible.

My Eagle skills have also improved a lot by going through the work and pointers kizzap gave me and this I'm forever thankful for, but there is loads to learn still but making everything yourselves teaches you what works and what doesn't.

One funny side-story though...
My spinal injury means that I spend most of the days in bed in order to have enough strength to pick up the kids after school and spend some quality time with then before they go to bed.
This means that I have a lot of stuff on trays next to the bed, and this includes soldering stations and oscilloscope/DSO etc etc... (and yes, I'm single )

When I'd made my first pcb with the new 20 mm battery holder and the new leds I wrote the basic code that would drive the whole solution and implemented all the rules etc.
I try to pwm the leds in order to save battery and I've never tried this with shift registers, nor know even if it works, but unless you try you don't know.

Anyhow, the code didn't seem to work so I sat there, in bed, with my oscilloscope trying to look at what was coming out of the shift registers and nothing made sense... Leds would go on when they shouldn't and all sorts of weird things happened every now and then... I spent days trying to figure out what was happening and suddenly I realised that I was measuring one led - which was blinking when it shouldn't - and the damn thing wasn't on? I mean no battery and no USB cable - and yet one of the leds lit up every now and then?

WTF? Poltegeists? 

I put everything away and decided that I probably had taken too much morphine and that it was a dream, but it wasn't.

Since I can't bend down due to my injury I don't wear socks any more. When sitting in bed I sometime put one of my feet on the floor to stretch it out. Wooden unpainted floors.
I live in an old farmhouse (1880-ish) with non-earthed mains @ 230 v, 50Hz.

Somewhere in the house someone before me must have swapped a zero and phase, or a mouse have been poking around where they shouldn't have, and since the house is a tad damp the floors actually conduct one of the phases.
One foot on the floor, me holding the pcb, oscilloscope grounding the led = leds turning on without power!

First I laughed really hard since it was such a stupid fault I'd discovered but then I got concerned since I need to find the electrical fault, and anyone who's been poking around an old house with old installations know that anything could've happened the last 70 years. The 50 Hz noise that I had seen so many times in my oscilloscope was also explained at the same time.

As an added bonus I can now test my leds with just bare feet and an oscilloscope! It is just about perfect to light them up 

TL;DR
Been away being ill a long time. Made major changes to design and made own pcb's. Discovered that I have dodgy electrics in the house and bare feet gave bizarre effects!

[Zeta]

the new pcb looks great. although I would place all parts and battery in the back and maybe use capacitive sensing touch buttons in the front so there is no component exposed in the front. you should think about somehow protecting the battery, locking it in a case or something. Coin cell batteries should never be accessible to small kids, they are letal. just imagine one of the kids taking the battery off the device and eating it. very dangerous.

[JensAndree]

the new pcb looks great. although I would place all parts and battery in the back and maybe use capacitive sensing touch buttons in the front so there is no component exposed in the front. you should think about somehow protecting the battery, locking it in a case or something. Coin cell batteries should never be accessible to small kids, they are letal. just imagine one of the kids taking the battery off the device and eating it. very dangerous.

I did try capacitive buttons way back when I started tinkering with this but a side-effect of autism, especially with young children, is that they often have poor fine motor skills and require rather large buttons in order to be able to press them with accuracy. Older kids learn this but in order to cater for as many as possible I have to use tactile switches, and I have a few different ones at home that fits the footprint on the pcb. The buttons are also colour coordinated for those who need it and they are 5-blue, 10-green, 15-yellow and 20-red. (international standard)

The current layout is for proof of concept and will not be final. I have made so many changes that I now need to make a few boards with the new components and run tests on them.
The final layout will have the leds towards the center of the board and I will shift most components to the back again I think, but it depends whether I will make a case or not. The ultimate goal is to make it as cheap as possible in order for everyone to be able to afford one and the one that is available on the market costs appx $250 at the moment - and this is all profit in someones pocket... 

My goal has always been to keep it under 100 SEK, which is around $15, and at the moment I can... I think...?
Regardless it'll be a lot cheaper - and Open Source! (as soon as I have everything ready for release)

The Battery is the only part that would call for an enclosure and I suspect having an injection moulded custom case will be rather expensive, so it will always be optional.
The current battery holder I use require (almost) brute force to remove the battery, but if someone can do it - kids can

I have kept the back clean since it must be easily attached to the inside of a folder etc and stay there, but the IC's and the battery holder are rather slim so it should work just fine and it won't be too bulky.

This device is used not only as a simple timer, it is used heavily throughout the day, and all activities are connected to a colour - which dictates how many minutes they should spend on an activity, and sadly the concept of time is something autistic people have to learn mechanically, but they never really learn to "feel" time nor the relation of time, and this is where this device is so useful.

An added bonus of using a bare-pcb gadget is that it doesn't look like it's an aid for someone with a handicap - and this is important to the kids as they are getting a bit older.

Many years ago I made an Android version (free of course!) so they could hide it in public, and I'm just about to release the same functionality for Sony Smartwatch2.
The Android version is made to work on Cupcake and onwards in order to recycle old phones into useful tools, for free - and that's where this all started.

Now I need to be satisfied with the chosen components, current consumption and coverage of all requirements and use-cases. When that is done I can start doing a final layout.
I will at some stage ascertain the costs of a custom enclosure, and I have searched for something existing but not even come close, so no luck with that.

I have learned loads by doing this project and I have received so much help and constructive critique from this forum that I will continue to update this thread until it's done!

I will not take a profit from this, rather the opposite, since I think it's fundamentally wrong to produce tools for handicapped people - especially kids - with grotesque profit margins.
I think it is better to sell cheaply to many instead of expensive to few, and if I can show that it is possible to do this then perhaps others can follow lead?
By having it Open Source opens up for custom variants for those with specific needs as long as they know someone who can modify the design. An adapted computer keyboard e.g. for someone who can only type with the aid of a stick in their mouth can easily cost $1000+ but only cost a fraction of that to manufacture... If I can poke a hole somewhere in that market then I will be pleased!

Thank you for your comments!