Compact and low cost MySensors + NRF24L01 Board Project

[frenchclem]

Dear MySensors Community,
I'm thinking on lunching a project aiming to making a cheap, compact, and less soldering solution for the Mysensors enthusiastic people. As Mysensors project is aiming at less programming
So the base line is : a $7 all in one Mysensors+NRF24L01
I know the Sensbender board is here, but you must solder a NRF module on top of it, and it's not so cheap.

In more details here is what I have in mind : (nothing new really)

• ATMEGA328P
• NRF24L01+ with PCB Antenna OR Chip Antenna
• ISP programmer Breakout
• Temperature sensor (maybe humidity)
• break out of main GPIO
• power side : Battery supply on 3.3V direct, or 5V external with 3.3 regulator
• SMC component not smaller that 0604
• option on PCB : space for micro usb connector, and SMA connector

I have already quoted a BOM, I'm waiting for PCB and PCB assembly quote from 2 manufacturer. I'm quite confident that this is feasible for 200 or more boards.

the main challenge here, from my point of view is wireless connectivity with chip or PCB Antenna. The rest is open source and all over the web.

Let me know your point of view.

I'll update the schematic I'm working on soon.

[m26872]

Less soldering than Sensbender = completly ready made? Made for less than 50% price? Sounds like a hard challange.

[hek]

As long as you release the design files and the project as open hardware I'm happy with it.

[frenchclem]

@m26872 the most expensive is the Atmega I got a quote at $1.2, but I hope I can get it cheaper. then PCB and NRF chip will cost between $0.6-0.7. The other small SMC don't cost much. Depend now how much I'll get it manufacture for in quite small batch? But let's face it, if we can buy an arduino pro mini and a NRF24L01 module for less than $3-4 in China, then it should be easy to make an all in one board for abour $7-8. Depending on the quotation I'll get, I hope I'll be able to manufacture it in France.

@hek I'm doing this because I just love your project, and I think the cheaper you will give out people ready board, the more the mysensors communitty will grow.
Of course everything will be open! It's already all over the web, the arduino, the NRF, even PCB antenna design. I really don't mean to compete with the Sensbender, I'm really meanning to contribute somehow to the Mysensors Projet.

[tbowmo]

The real price hog on the sensebender is the si7021,prices are around 4-5$ alone. That is what's keeping the prices up

[m26872]

@frenchclem I thought the manufacturing (pcb, assemble, test...) was a high cost, unless you buy in huge quantities.
But great. I really hope you make it!

[frenchclem]

Hello,
Here is draft 1 of the schematics.

I still haven't decided whether the temp sensors will be the famous SI7021 or for cost reason a more basic DS18b20 (about half the price of the SI7021)

For powering the board, I'm planning on having 2 solutions : direct 3.3V (bypass by solder jumper of the regulator) or 5v power supply with a Micro-B

The Antenna I found 2 design :

• http://www.ti.com/lit/an/swru120b/swru120b.pdf : the one used in the schematics because I found a lbr for it)
• http://www.ti.com/lit/an/swra117d/swra117d.pdf : I like this one more, the gain is better, but I've got to create the part in eagle.

I'm still missing the breakout of the main pins.

But I wanted to have a first feedback / point of view : Is EEPROM critical ? Is the temp sensor absolutely necessary ? Same with the Micro USB and Voltage Reg ?

Edit : The Oscillator won't be monted but the solder pad will be here (If I have room for it), I also will remove the capa

[tbowmo]

@frenchclem

Please note that the DS18s20 doesn't work at supply voltages below 3V

also external flash is a really good idea, as are added atsha204a as well..

[frenchclem]

@tbowmo Thanks for shooting out some idea !

Indeed, I read the thread about the birth of the Sensebender which is actually a mine of information and thought about minimal design.

As you mention it while working on the sensebender, the RF will be the hardest part. I can add all the cool feature on the board, if you can't connect it, it's useless.

I'm still waiting for cost estimation from china, and french supplier specialized in IOT, prototyping, and small batch production.
I'm in a design to cost process, so first things first, I want to validate the cost objective from various quotation. Then I will see if I have to make hardware design choice, or if I just can't make it, ...

[Nuubi]

*The *board that should be implemented is just arduino nano connections to NRF module. Very simple, yet, haven't gotten to do it.. anyone?

[frenchclem]

@Nuubi can you explain why you need the :
FTDI USB-to-TTL Serial chip ?
mini B connector is not so used anymore (samsung cellphone use micro USB)

Because beside that, it's what I plan on doing. Oh I might add a temp sensor, but not sure yet.

Then I think it's worth having a thought about it :
EEPROM can allow OTA FW update. which is quite cool, when you what to update your node without taking everything down.
I'm not really convinced with the ATSHA204 tho

[Anticimex]

@frenchclem do you have another security solution planned or do you intend to skip security and completely rely on sw implementations only for security?
I do not recommend that as it makes the board sensitive for memory dumping which would reveal the secret shared keys which in turn would compromise the network. The atsha204 protects from that type of attack.

[korttoma]

What I would like is basically a Sensebender Micro without the Si7021 but with the NRF24L01+ on the same tiny board.

[frenchclem]

@Anticimex not really. My first idea about this board was to say : lot of us by chinese arduino pro, NRF24L01, spend some time soldering everything together, because we are looking for the cheapest solution even if its not the most sexy one, the most secured one, ... So a lot of us, end up designing their own little board, with DIP Atmega, to have something smaller, more sexy, but not secured.

So I guess on my first thought, this board is more for all these people. Now when you start digging, and thinking about it, of course you want to add stuff to it, temp sensors, EEPROM, security.
My ultimate goal is the price, on that I won't deviate.
Then the secondary goal is size, functionality, and evolution : on that I'm open.

• The more I think about it, the more I thinking that the temp sensor is expensive and maybe not mandatory
• EEPROM : I really like the Idea of OTA FW update, plus you can get red of the ISP programming pin. But I'm not sure whether it's working already on mysensors or not (I have not look to be honest)
• atsha204 : As I said, I lot of us don't use HW protection with our DIY solution. But I like the idea of secured communication, so if it's not a cost driver (I have not quoted the chip yet) then it could be part of the board.

By Evolution, I mean that I could throw in some PCB foot print, for SI7021, or other component (on the bottom side for example) so people that want to add more functionality would be able to.

@korttoma I agree, SI7021 might not be absolutely necessary. About the size of the board, there is no way it will be the same size as sensebender ! @tbowmo did an incredible job on that board, plus he used 0402 chip which can be a cost driver for some production. I'll keep it small, but not as small at the sensebender !

[frenchclem]

Small Update : Got the first quote for PCB and PCB assembly in china. That validates that the $7 objective of the board is feasible for 400 boards batch.

I ask for quotation the atsha204, some packaging are obsolete, which one is used on the Sensebender ? Sot23 or SOIC, or TSSOP ?

Also for size reduction, I thing I'll use the ATMEGA QFN32

Now lets do a dummy proto board for antenna benchmarking

[Anticimex]

@frenchclem then why don't just use the existing SenseBender board design files and just order the raw PCB and then solder only the components you want? It's open hardware. You get the footprints to add components later on if you need. To me it sounds like you would like to build a SenseBender but skip a few components to keep BOM cost down.

[frenchclem]

@Anticimex I'm thinking about that option too, as a back up solution if I fail with RF.
Also I think Sensebender is missing voltage regulator and connector so one can choose battery operation or other power supply

[frenchclem]

@Anticimex still thinking about the ATSHA204 and the HW security vs SW security.
If you have a node with the ATSHA204, does the GW needs to have it on board to ?
BTW do you have the link to the security thread so I can update my knowledge on that side ? please

[mfalkvidd]

@frenchclem said:

@Anticimex still thinking about the ATSHA204 and the HW security vs SW security.
If you have a node with the ATSHA204, does the GW needs to have it on board to ?
BTW do you have the link to the security thread so I can update my knowledge on that side ? please

Signing thread http://forum.mysensors.org/topic/1021/security-introducing-signing-support-to-mysensors/1
Software and hardware signing can be mixed, no problem. The biggest difference is that software adds a few kB to the sketch size and that the key can be extracted if someone "borrows" a software-based node.

[Anticimex]

@frenchclem my post is pinned in the development category so you should have no problem finding it. SW and HW signing is compatible so no, you don't have to have hw support in your gw. But sw signing also take more memory. And the quality of the random numbers is much poorer.

[frenchclem]

Thanks I'll read that carefully.

[tbowmo]

@frenchclem

I would suggest to at least add the footprint for ATSHA204A (sot23 housing) on the board. Also with external flash / eeprom, add the footprint..

Another thing, does the price quotation include raw pcb's, all components, and smd assembly? What type of PCB materials have you specified for them to give a quote (RF performance depends on this!). Component quality (specially around the RF parts).

Please remember to give the end users as many options to use the board, as possible.. A feature that you think is not necessary (due to pricing etc.) is the one thing that users want to have on the board. And thus find something else instead.

Also please have a look at mysx connector description https://github.com/mysensors-kicad/documentation/blob/master/TheMYSXconnector.pdf this will be used by more projects in the future..

[ahmedadelhosni]

@frenchclem I have a question please. Why do most people add temp and hum sensors in every node ? I really can't get the benefit of that.
Thanks.

[frenchclem]

@tbowmo I will do my best to please the most ! I'm thinking about the footprint options. I'm also thinking about production option in one batch. If you think about it, in 400 boards, if you use the same Stencil, you can in case 1 populate the full board (EEPROM, temp sensors, ...), in case 2 only temp, and in case 3 the minimum. It's a minor program update to disable one or 2 ref. And then you have 3 types of complete board with three price to make everyone happy. I obviously won't be able to do that with chinese supplier, so I'll talk about that with the French supplier.

I'm actually doing all the quotation with various hypothesis because I don't have yet any design file. I'm working on it. So I have not made a choice on specific ref for the RF part (only the couple of inductance and capacitor for the impedance matching are critical). The PCB will be 1mm thick, and the cooper trace 1oz. There will be no ground plan under the antenna.

Thanks for sharing the MYSX connector, I'll take that into account too. Is there an eagle lib for it ? (I know it's standard dual row pin header, but the pin assignation could be useful to have directly via a lib)

[frenchclem]

@ahmedadelhosni Idk what to tell you, I guess Temp sensor is quite common, and fun thing to have inside and ouside.
Many people like to know what is the weather ouside, and the can also use inside temp sensor to control their heating/cooling system.

[scalz]

@frenchclem: hi.
I thought a little about your project to try to offer availability for people. I think, your project could be interesting, but is the wheel not being reinvented???
Why I say this is because:

• you are not talking about rfm but just in case : atmel+rfm69+eeprom -> moteino
• nrf + atmel : could be fun for those who like nrf. But it is exactly what nrf51822 does, plus BLE...And it is cheap:
  http://www.seeedstudio.com/depot/MDBT40P  nRF51822 based BLE module-p-2503.html?cPath=19_21
• so if you want to derive your board because people want more features, mysxconnector, auth, eeprom.... it's ok for non-battery powered I think...but it won't be optimized.
  If you don't care about this, so it will be sort of sensebender with integrated ant. or if you want to improve it a little, it would be a mix between moteino/sensebender...
• So of course, there is a market for this...when you are noob, you don't know much about sensors, radio, powersavings and not aware about integration, assembling...but when you know this stuff. For proto ok, but with iot you want to integrate then. And you discover that nrf not the best radio, and "sort of arduino mini" is time consuming when assembling node if boards not well thought.
  Then you make like Anticimex, gert, myna, and me, your own board to find a way to optimize time, integration, cost, overall looking, and pleasure to diy. In my case, noob or not, I don't like wiring too much.
• finally, I say this, but my board "MyTinyCamel" (not completely released yet because you will see...) is based on 328p. and I have another 1284p "camel" board because 328p doesn't have lot of mem if you want to play with lots of sensors.. for the moment it's ok, but I think these chip will become features/memory "deprecated" vs others chip. so is 328p a really good choice??? for the moment, maybe, and I stay with this because of availabilty but I have asked atmel for new chips availability...
• did you see mytinycamel? it is exactly what you are trying to make...modularity, footprint, all what you said (but no ant integrated because I wanted to give options rfm or nrf .and bonus it's ultra low power . I have upgraded it with some interesting feature since I have presented it. now I am thinking to ask fabhouse to propagate it for those who can't solder it, maybe in few versions as you thought too...

All what I said, I hope you understand, is positive feedbacks only just my another 2cents thoughts, lol. I don't want to criticize your project as you are right, your idea could help people. it is not because it doesn't fit my needs that it same for others Too bad we didn"t met when I started my projects...

[frenchclem]

@scalz Yes, I know I'm not inventing anything, everything that I propose already exist
the only, and MAJOR difference is the price : I'm following a design to cost process. What does it means : I take available hardware/schematics : NRF, Arduino (Sensebender, and other good board), ... and I try to redesign and make it as cheap as possible.
The nrf51822 is indeed the killer, BUT it's not working yet with MySensors

And I totally agree with you, if you work in HW and if you are not focused on the future (new chip, new protocols, new project, ...) then your project life expectancy is very little

I obviously won't lunch mass production, so I'm running the process with 400 boards batch. So far, only based on hypothesis, meaning its pure fiction, meaning there is a margin of error in my calculation. The more I will progess, the more the margin of error will be important, and the better I will fine tune the final price.
And I'm pretty sure that if we talk about bigger production batch, then the cost of a simple board could be as low as a couple of $

Look at the low cost airline company, they didn't invent anything, they just optimized everything to drive cost down, and now they are very successful company.
Look at the new Raspberry Zero (produced in EU) : given away for free with a magazine, and then sold for $5 !!
I give out those example to illustrate that innovation is not always the key to success and to illustrate that the cheaper you go, the wider spread it will be ! (no need to add : if the quality is here)

And just to be clear, I'm not doing this to earn money, because low cost, and small batch are usually not big money maker.

Now clearly, if you tell me you will manufacture your Mytinycamel, and sell it under $7-10 fully equipped (board + radio) then I'll reconsider the project. and don't worry, if I didn't want to be challenged on the project I won't be here talking about it !
"Too bad we didn"t met when I started my projects..." => It's maybe not too late !? If we are after the same goal, we can maybe add our skills ...

[scalz]

@frenclem: it's great to talk about this
I completely agree with you and like I said we have same goals
in my case, this week I will look more deeply to my bom (already made, and can be cheap if you don't use all features of course. the more you add the more it costs you are right). and ask few fabs to see. I quoted myself 7-10euros without radio of course, and sourcing myself not taking batch in account. As I prefer rfm69, of course it cannot be <10€ but for those reason I added nrf smd (cheap) footprint for those who wants cheap.
oh sorry, I am not saying it is for money. lol, you are right, low cost, good quality, needs enormous batch to make money. but, 400boards batch, wow still lots of money to invest...
yep, nrf51822 is a killer thing, if you like nrf...lol, ok i stop saying this! but I think it will work Mysensors because it is arm0 and tbowmo is working on this chip. can't wait to see his results
rpi zero, impressive the cost they can achieve. for some simple embedded it can be nice. but at the age of iot...no connectivity, too bad! I say this because I have a project yun based. and I like rpi zero footprint.
but to me what could be crazy for battery operated nodes, is same thing nrf51822, but with hoperf.
teamworking...tempting..will think about it.for the moment, Im busy and am already working with someone.

[tbowmo]

For the really low prices (like Raspberry Zero at 5$). It all comes down to quantity! If you produce enough units, an buy components in large stocks, you can save a lot of money..

But in low volume productions, you can't benefit from the same high volume discounts on components, etc.

There is setup fees involved in production.. Think we have a base price of 300$ just for setting up the machine SMD mounting, and start mounting them. Which will be divided out on the boards you make (when making 300 boards, the price is then 1$ per pcba, now if you can make 300.000 boards in one batch, then the setup fee is only a minimal fraction of the unit price)

on the other hand, if you can save 10 cents per board, by using a cheaper capacitor, and produce 1.000.000 boards, then you actually save 100.000$.. while the saving on each pcb is only 10 cents.

[tbowmo]

@frenchclem

Just a thought that popped up in my head.. Have you thought about FCC / CE certifications? If you create a transmitter, that you want to distribute to others, you better have the certificates ready (Even for hobbyists)

Found this thread http://electronics.stackexchange.com/questions/161885/fcc-certification-for-nrf24l01

It might be that nrf modules from China is lacking the FCC certifications, but then "it's not your problem" as a hardware developer, when you let users buy the radio module separately (They can probably buy fcc certified modules from some suppliers..)

According to this, by using ready modules, you can skip FCC certification on transmitters
https://www.lsr.com/white-papers/fcc-guidance-on-transmitter-modules

I'm not saying that this isn't a good idea, it's a great idea! But producing a pcb (with transmitters) isn't just about BOM prices.. There is a lot of other things that you have to take into the equation as well.. And I fear that you haven't investigated all the details fully, and end up facing huge problems later on.

[frenchclem]

@tbowmo
Indeed what you are pointing out is the Achilles heel of the project ! I haven't mention it because I haven't done my research on the subject. Their might be some solutions I will post on that later on.

And you are right once again, I haven't investigated all the details ... **yet **
It's all a project is about, isn't it ? You have an idea, a goal, and then you investigate step by step to see whether you have a good chance of success or not. I use "good chance of success", because in every project you take some hypothesis, you take some risks, because success is never 100% sure.
At the beginning of the project, the chance of failure are very high, but its ok because the investment in very low (just time so far). And the more I'll progress, the lower the risk will be (if I did the investigation work properly) and then it will be ok to invest a little more.

It's also one of the reason why I post in here all of my project, this (all the comments, and feedback) helps me (and the Mysensors community) to evaluate the chance of success Vs the chance of failure. You guys can point out weak point and so on and so forth.

So I'm pretty glad you mentioned that, it shows me my project is of interest.
As mentioned, I will post on that specific crucial part once I'm done evaluating all of my options.

[frenchclem]

Hello there,
So after some reading, phoning, thinking, again and again about FCC/CE Certification here is the plan.
I'm not going to go for the FCC Certification, as it's way too expensive, but I'll run the test myself for CE compliance.
When FCC needs to be done by US agencies, the CE can be self-certified : you run the tests with the appropriate tools, make sure your design respects the requirements, and if so, keep the results in case anyone asks anything.
Now pretty much everywhere, I read that if your product is CE compliant, then your chance of getting the FCC on the first run is very high. Dropping the cost to a "small" $5k or so
So first, CE self certification, and then if needed/wanted/asked/ ... FCC, IT, ... other certifications. Meaning so US selling first just Europe. Are there many US folks using Mysensors ?

To run the test, I'm working with my University which has all the tools, and knowledge to assist me. Could be an interesting project to run with some student ... will see !

So regarding the design, zero risk must be taken regarding the RF part :

• the filtering components will be replaced by a Balun Filter from JOHANSON (http://www.johansontechnology.com/datasheets/baluns/Johanson nRF24L01_nRF24L01+_App_Note.pdf)
• I'll add a shielding on top of the RF part. Still looking for the right standard, low cost, easy mounting shield

I share a great guide from TI about antenna design : http://www.ti.com/lit/an/swra161b/swra161b.pdf
Now I want to try the Inverted F antenna presented in my early post, a Di pole antenna (no more Balun Filter), and Chip antenna and compare them the standard Meandered Inverted-F Antenna (MIFA) we all have on our NRF board !

[scalz]

interesting guide. thx for share

[BigDog]

i am realy exited about this project.
if you need a person to test the stuff and has not much expirients about mysensors and stuff (meby to know if someone with minimal expiriets with mysensors know to make it work ) plz contact me

i have seen manny peoples that want to use mysensors but hasnt got the knowhow to make a good script/programming to make al the conponments work

solding and connection the parts is the simple part to me

greets and Good luck for the project!

Bob

[frenchclem]

Small update here. I've been working on the routing part. it's been years since i played with the nets, and wires on Eagles ... Oh boy, its a mess ! and while placing, moving components, i was thinking , wow, the Balun filter, the NRF24L01 are going to be impossible to hand solder correctly ! and I don't really want to buy a microscope, a $300 soldering station, ...
So I looked around and found something interesting so I post it here, if some makers don't know about it, and are afraid of SMD soldering : OSH also make stencil for you, Kapton stencil for $5 ! https://www.oshstencils.com/index.php

then it's a peace of cake or is it ? https://www.youtube.com/watch?v=1RMtOAHbfvU
Has anyone tried it before ? I'll definitely try it