Minimal design thoughts
-
I'm thinking about spinning up the good old eagle cad program again, and start creating sensor boards (Like so many others here :))
Only thing is, that unlike (almost) everyone else, that are using normal arduino micro boards, I'm thinking about skipping that part, and create my "own" arduino board, thinking about the following:
- use atmega 328 tqfp (That's doable for home soldering :))
- pads for a 32Khz crystal, for timing purposes
- Standard atmel programming header.
- ftdi header like on arduino mini, so could use arduino bootloader (could be some details about the baudrate though, as it's running off internal RC oscillator)
- Using Si7021 integrated humidity / temperature sensor, footprint for bmp180 barometric sensor
- skip power converters, run everything off a battery directly (disable brownout detection in atmel)
- SPI flash / e2prom onboard. Using same connections as moteino from lowpowerlabs, so should be able to use same bootloader for updates to firmware
- pinheader for connecting radio module
- Pinheader for D3-D7 on one side, and SDA/SCL + A0/A1 and power on the other side
- Formfactor is 30x17.5mm (almost same size as the NRF24 module).
- ATSHA204A sot23 footprint on board (Connected to A3). Can be used for SHA256 key exchange and verification
- status LED (on A2)
Any thoughts about this? Or should I wait for the all in one mysensors battery sensor board? (When will that be available?)
/ Thomas
(Spec updated January 28th, 2015)
-
I'm thinking about spinning up the good old eagle cad program again, and start creating sensor boards (Like so many others here :))
Only thing is, that unlike (almost) everyone else, that are using normal arduino micro boards, I'm thinking about skipping that part, and create my "own" arduino board, thinking about the following:
- use atmega 328 tqfp (That's doable for home soldering :))
- pads for a 32Khz crystal, for timing purposes
- Standard atmel programming header.
- ftdi header like on arduino mini, so could use arduino bootloader (could be some details about the baudrate though, as it's running off internal RC oscillator)
- Using Si7021 integrated humidity / temperature sensor, footprint for bmp180 barometric sensor
- skip power converters, run everything off a battery directly (disable brownout detection in atmel)
- SPI flash / e2prom onboard. Using same connections as moteino from lowpowerlabs, so should be able to use same bootloader for updates to firmware
- pinheader for connecting radio module
- Pinheader for D3-D7 on one side, and SDA/SCL + A0/A1 and power on the other side
- Formfactor is 30x17.5mm (almost same size as the NRF24 module).
- ATSHA204A sot23 footprint on board (Connected to A3). Can be used for SHA256 key exchange and verification
- status LED (on A2)
Any thoughts about this? Or should I wait for the all in one mysensors battery sensor board? (When will that be available?)
/ Thomas
(Spec updated January 28th, 2015)
@Thomas-Bowman-Mørch Please open source the design if you do! Would be interesting to see how small you can make it.
I'd still like to see space for a 3.3V regulator of some sort though. Even if it means a few more components, at the very least 2x capacitors and 1x inductor. And the nrf24 probably shouldn't hang right over the inductor, even if it is shielded, seeing as there have been so many iffy issue with that chip. Even if the atmel can run at really low voltages, many sensors, like the DS18B20 don't seem to be able to. With a regulator, we'd have much more freedom to choose from 1xAA/AAA or 2xAA/AAA etc.
-
With a regulator you have indeed more freedom but it is less minimal and .. there are already other boards doing it that way.
My question would be how exactly you like to power the board.
I see the nRF24L01 can run from 1.9 Volts and the ATmega328 is said to work down to 1.8V but what about the sensors you are planning to use?
-
I haven't been looking too deeply into the different sensor datasheets yet. But yes, it seems like the Dallas chip doesn't like voltages below 3V. What a bummer..
Main purpose was actually to make things as small as possible, but still be able to hand solder things.
/ Thomas
-
Have been thinking about this in the past and what could be done is (apart from searching another sensor) to only generate the 3V when the sensor is about to be read and switch it off afterward.
Not really making things "as small as possible" but as making things as less power hungry as possible.
-
I'm thinking about spinning up the good old eagle cad program again, and start creating sensor boards (Like so many others here :))
Only thing is, that unlike (almost) everyone else, that are using normal arduino micro boards, I'm thinking about skipping that part, and create my "own" arduino board, thinking about the following:
- use atmega 328 tqfp (That's doable for home soldering :))
- pads for a 32Khz crystal, for timing purposes
- Standard atmel programming header.
- ftdi header like on arduino mini, so could use arduino bootloader (could be some details about the baudrate though, as it's running off internal RC oscillator)
- Using Si7021 integrated humidity / temperature sensor, footprint for bmp180 barometric sensor
- skip power converters, run everything off a battery directly (disable brownout detection in atmel)
- SPI flash / e2prom onboard. Using same connections as moteino from lowpowerlabs, so should be able to use same bootloader for updates to firmware
- pinheader for connecting radio module
- Pinheader for D3-D7 on one side, and SDA/SCL + A0/A1 and power on the other side
- Formfactor is 30x17.5mm (almost same size as the NRF24 module).
- ATSHA204A sot23 footprint on board (Connected to A3). Can be used for SHA256 key exchange and verification
- status LED (on A2)
Any thoughts about this? Or should I wait for the all in one mysensors battery sensor board? (When will that be available?)
/ Thomas
(Spec updated January 28th, 2015)
@Thomas-Bowman-Mørch I am thinking about the same thing but with power converter.
I have a board with a Mini Pro and flash memory chip right now but want to get rid of the Arduino (not the chrystal though based on input from several sources).
I would even want to make the NRF24 on board instead of using a module.
The same for the converter.Getting rid of the Arduino is a no-brainer imho but the on-board NRF24 and converter are less easy to do...
-
I would even want to make the NRF24 on board instead of using a module.
The same for the converter.That sounds good and will likely come with better reliability.
The board layout for the NRF24 should not be too hard, I think you can get the design for the Nordim Semicondutor site. More problematic can be the soldering and tuning the RF, although I am not convinced the standard boards has good tuning done.
What makes the convertor less easy you think?
-
@marceltrapman Originally I was thinking about making everything on a single board (also with the nrf24 on board). But the modules are dirt cheap, and I don't like to hand solder the NRF chip (too small footprint).
So that's the reason to just plug it in as a module.
@daulagari Had been thinking about the exact same thing, after discovering the problem with DS18x20 and supply voltage. I stumbled uppon MCP1640 from microchip, which can go into pass-through when disabled. And then probably a TPS27081A from TI to disable power to the sensors, when in sleepmode.
The thought was that when it powers up sensors for measuring, it also ramps up the supply voltage to the atmel. (Actually there is a application note describing this somewhere on microchips site)
But then we are back to putting on inductors etc (and finding reliable sources for them). Damn, I sometimes miss my old job at a hardware design house, where we had a fairly large stock of 'generic' components in the house for experiments.
/ Thomas
-
I would even want to make the NRF24 on board instead of using a module.
The same for the converter.That sounds good and will likely come with better reliability.
The board layout for the NRF24 should not be too hard, I think you can get the design for the Nordim Semicondutor site. More problematic can be the soldering and tuning the RF, although I am not convinced the standard boards has good tuning done.
What makes the convertor less easy you think?
@daulagari said:
What makes the convertor less easy you think?
Thinking out loud (and for myself) I lack the skills to decide what components to use.
@Thomas-Bowman-Mørch
But the modules are dirt cheap
I agree but there are pro's as well and if price is the only con I would go for it :)
and I don't like to hand solder the NRF chip
Use a rework blower. Works well...
-
I see "multiple" cons with embedding the nrf chip:
- Components too small to handsolder (yes you could use rework blower, but then you must buy one in order to populate the boards yourself).
- RF design is not for kids :) antenna design, impedance matching etc. (There must be a reason to why we considered the RF guys as gurus, when I worked at the design house)
- They are still cheap :)
the only pro to embedding it, is actually to make things relatively smaller, than with two modules packed together.
And taking RF design into consideration, then I don't think it is possible to make things that much smaller, because the RF is as it should be, can't be smaller.. (other than maybe go with 4 layer boards, in order to get solid gnd plane) Placing components on both sides of the PCB. And then the price for the PCB goes up, comparing it to 2 layer pcb's
/ Thomas
-
@Thomas-Bowman-Mørch
it seems like the Dallas chip doesn't like voltages below 3V. What a bummer..
Are you married with the Dallas chip? ;-)
Had a look at the Internet and there are at least low-voltage temperature sensors, see for example the DS75LV.
The thought was that when it powers up sensors for measuring, it also ramps up the supply voltage to the atmel. (Actually there is a application note describing this somewhere on microchips site)
Why also "power up" the Atmel?
-
@Thomas-Bowman-Mørch
it seems like the Dallas chip doesn't like voltages below 3V. What a bummer..
Are you married with the Dallas chip? ;-)
Had a look at the Internet and there are at least low-voltage temperature sensors, see for example the DS75LV.
The thought was that when it powers up sensors for measuring, it also ramps up the supply voltage to the atmel. (Actually there is a application note describing this somewhere on microchips site)
Why also "power up" the Atmel?
@daulagari said:
Are you married with the Dallas chip? ;-)
Had a look at the Internet and there are at least low-voltage temperature sensors, see for example the DS75LV.
DS75LV has an accuracy of +/- 2 degrees celcius, while DS18s20 have +/- 0.5 degrees. Also, I really like the ability to have as many sensors hanging on the same single wire, as you desire.
I also had a look at the DHT22 (I know, there probably are other chips out there for doing combined humidity and temperature measurements) but that is also spec'ed to minimum 3.3V operating voltage.
Why also "power up" the Atmel?
To keep the logic levels the same for sensors and the atmel, And for switching to a higher working frequency when it wakes up for doing measurements (if necessary).
-
DS75LV has an accuracy of +/- 2 degrees celcius, while DS18s20 have +/- 0.5 degrees.
Yes also saw that and meant another IC, the DS620.
Also, I really like the ability to have as many sensors hanging on the same single wire, as you desire.
"For distributed-sensing applications, it is multidroppable with three address pins that allow up to eight DS620s to operate on a single bus."
Why also "power up" the Atmel?
To keep the logic levels the same for sensors and the atmel, And for switching to a higher working frequency when it wakes up for doing measurements (if necessary).
That does not sound like a "minimal design" ;-)
-
DS75LV has an accuracy of +/- 2 degrees celcius, while DS18s20 have +/- 0.5 degrees.
Yes also saw that and meant another IC, the DS620.
Also, I really like the ability to have as many sensors hanging on the same single wire, as you desire.
"For distributed-sensing applications, it is multidroppable with three address pins that allow up to eight DS620s to operate on a single bus."
Why also "power up" the Atmel?
To keep the logic levels the same for sensors and the atmel, And for switching to a higher working frequency when it wakes up for doing measurements (if necessary).
That does not sound like a "minimal design" ;-)
@daulagari said:
DS75LV has an accuracy of +/- 2 degrees celcius, while DS18s20 have +/- 0.5 degrees.
Yes also saw that and meant another IC, the DS620.
Also, I really like the ability to have as many sensors hanging on the same single wire, as you desire.
"For distributed-sensing applications, it is multidroppable with three address pins that allow up to eight DS620s to operate on a single bus."
Still it's i2c bus, which means 4 wires (GND, VCC, SDA, SCL), while DS18x20 is 2 wires (GND and Data/VCC)
Why also "power up" the Atmel?
To keep the logic levels the same for sensors and the atmel, And for switching to a higher working frequency when it wakes up for doing measurements (if necessary).
That does not sound like a "minimal design" ;-)
I know, that kinda blew the 'minimal design' away (a little), because I discovered that the sensors that I had in mind need at least 3V (some even 3.3V). If the sensors need 3.3V, and I need a powerconverter onboard, then I might as well make it possible to ramp up the voltage to the atmel.
Anyways, main design details is still in my head. Need to get it on paper soon..
/ Thomas
-
This NRF24 module and even cheaper ones (also mini 10 pcs 6 USD) could be an addition/alternative for minimal radio design.
-
Why also "power up" the Atmel?
To keep the logic levels the same for sensors and the atmel, And for switching to a higher working frequency when it wakes up for doing measurements (if necessary).
That does not sound like a "minimal design" ;)
I have studied the datasheets for a couple of sensors, and the atmega now, and come to the conclusion that if one needs to use 3.3V sensors, one need to boost the supply voltage to the atmega as well.
Absolute maximum ratings on any input pin for the atmega is VCC + 0.5V. This means, that if bat voltage is 1.8V, maximum high voltage is 2.3V on the input pin of the atmega.
Also, the minimum high voltage of ds18b20 (I know, I keep comming back to this one :)) is 3V (or 2.2V if running with local power). Which means that you need to have the atmel chip running on the same voltage as the sensors.
So it won't work if the atmel is powered directly from batt, and sensors are running of a converter.
Of course I could find alternative sensors, but I really like the ds18x20 as you only need 2 wires to the chip, if you need "remote" sensing (I have one place where I want to measure 5-6 temperatures on the central heating distribution that eases up things, that I only need 2 wires to each temperature sensor)
But let's see, I'm still searching for different sensors.
/ Thomas
-
Think that I have prety much settled for a Si7021 as temperature / humidity sensor, instead of DS18b20
Ok, it's a tiny chip (3x3mm DFN6 package) so it might be a beast to solder. But it goes down to 1.8V on VCC, so well suited for battery operation, together with the atmel and radio.
I have been arround a couple of iterations with DS18b20 and step-up, because I need one node to be able to measure multiple temperatures. But this is 1 node, whereas the rest only needs temperature / humidity, which can be covered with Si7021. I have also thought about throwing in the opportunity to use a PIR sensor on the nodes as well, just for the fun of it. But then I need the step-up, as the PIR sensor needs 3V minimum.
Right now, board dimensions with Si7021, 328p, spi flash, no stepup, is down to 15x21mm (smaller than the radio module). Might throw in a footprint for bmp180 as well, so I could equip one of the sensors with barometric sensor.. The complete sensor could fit on the end of a 2xAA battery holder (they are roughly 15x32x60mm).
-
Right now, board dimensions with Si7021, 328p, spi flash, no stepup, is down to 15x21mm (smaller than the radio module).
Nice, that sounds really like a minimal module!
-
Sounds very interesting! Though personally I'd rather have a good regulator / boost (even if it is in a separate module) than having to source particular components that will run at 1.8V.
How do you plan to integrate the nrf24? The "standard" or the "mini" version?
I would love it if the MySensors web site could have a department of open source pcb schematics just like it has models for 3D printing.
Soldering sounds tricky, you'd either have to reflow or heat the underside of the DFN6 on the pcb. With extra flux and paste.
