DS1820 to USB connector

[..] [Introduction] [Unraveling DS18X20] [Hardware] [Software] [Downloads] [Acknowledgements] [Copyright/Disclaimers] [Similar projects] [Feedback]

Introduction / What is it

It is a little board with an Atmel AVR (ATtiny45/85) microcontroller that can be used to connect Dallas/Maxim 1wire temperature sensors like the DS18B20 to the USB port of a PC. Some linux software that allows reading the temperature data is included. The whole USB implementation is done in software, using the great AVR-USB library from Objective Development.
In theory, there is no limit on the number of probes that can be connected to the device, but see the software section for some more realism.

Unraveling DS18X20

When you try to get the parts for building this, you may note that Dallas/Maxim has quite a few very very similarily named parts. As I had the same problem, I have put together a little table that explains the differences. Please note that I'm not an expert here, and this table really comes without any warranty. Also, all these devices are 1wire devices, so you should be able to connect them, but to read useful data from them will probably require some minor software changes on the host side.

Part	Explanation/Differences
DS18B20	The one used here: TO-92 case, accuracy ±0.5 degrees (Celsius), user configurable resolution up to 12 bit (=0.0625°C)
DS18B20+	Exactly the same as DS18B20, the + denotes that the product is lead free (this is a requirement in all european countries now, the keyword is "ROHS")
DS18S20	resolution fixed at 9 bit (=0.5°C)
DS1820	resolution fixed at 9 bit, a little faster, but no longer available
DS1822	accuracy ±2 degrees (C)

The Hardware

Schematics and description

The hardware can easily be built on a perfboard / stripboard. A proper board has also been designed and the layout is included in the .tar.gz you find under software.
There aren't really many parts besides the ATtiny and the temperature probes (DS18B20): Only one crystal, two diodes, some resistors and a few capacitors.
The fuses of the ATtiny45/85 should be programmed for external crystal >8 MHz and BOD at 2.7 Volts. Running "make uploadfuses" will give you some hints for programming the fuses. Note that the ATtiny45/85 is run slightly out of spec here: Officially allowed are 10 MHz at 2.7 volts and 20 MHz at 4.5 volts. In this circuit it is run with 15 MHz at around 3.6 volts. AVR-USB can also handle some other clock rates and in the newest version even the internal oscillator, however from the description it seemed to me as if 15 MHz was the most stable option.

The DS18B20 supports a parasite power mode, where it only needs two wires instead of 3. In this mode it will draw the power it needs for operation from the data line. The software supports and detects this mode and will automatically power the probes through the data line when needed. To use this mode, connect the VCC pin of the DS18B20 to GND too. Parasite powered probes will show up with the "P" flag in the hostsoftware status output.

Part list

#	Name	Comments
1	~~ATtiny45~~ ATtiny85	The main microcontroller doing all the work. You should get an ATtiny85 instead of the ATtiny45 I used originally to be on the safe side. See the remarks in the [Software] section. You will also need some programmer to flash the microcontroller with the firmware.
>=1	DS18B20	At least one temperature probe would make sense for a device that is intended to be used for temperature measurement...
2	Diode 1N4001	These are abused to bring the supply voltage down from 5 to 3.6 Volts. Other silicium diodes (e.g. 1N4004, 1N4148) can be used as well, as long as they cause 0.6-0.7 volts of voltage dropout each.
2	Resistor 68Ω
1	Resistor 1MΩ
1	Resistor 1.5kΩ
1	Resistor (<=) 4.7kΩ	This is the pullup for the 1wire bus. You might need to use a resistor with a smaller value for longer cables.
1	Resistor 10kΩ	This one is merely optional, it is a pullup resistor for the ATtinys !reset pin. The ATtiny has an internal pullup, so it's not strictly needed, but supposedly increases stability when the power supply is unstable.
2	Capacitor 100nF	These are used for stabilizing power. They can be larger than 100nF and there can be more than two. But you should put at least one close to the microcontroller and one close to the connection point for the probes.
2	Capacitor 15pF	according to the ATtiny datasheet, everything between 12 and 22 pF should be fine.
1	Crystal 15 MHz	12 MHz can probably be used as well, but that requires a minimal firmware change and hasn't been tested.
1	DIP 8 socket	Recommended, else you would have to directly solder the ATtiny.
1	USB connector	Either use a proper USB jack like in the pictures, or just cut some USB cable open and directly solder the leads.

Pictures

The Software

The Software was written in C. It uses avr-libc and AVR-USB.
Originally, an ATtiny45 was used. When compiled with the standard Ubuntu 8.04 AVR stuff (avr-libc 1.4.7, avr-gcc 4.2.2, and -Os as a compiler switch to optimize for size), the code used around 3900 of the 4096 bytes of flash the attiny45 has. However, other more recent avr-libc and avr-gcc versions generate much larger code, so that the code no longer fits into the ATtiny45. As the price difference is only a few cents, the obvious solution is to use an ATtiny85 instead, which has 8K flash instead of 4. Starting with software version 1.04, this is now the default, and you would have to find a really outdated system to be even able to compile the software for an ATtiny45 anymore.
In theory, you can connect an unlimited amount of temperature probes to the onewire-bus, and read them. However, in reality there are a few limitations:

There is a maximum number of probes supported set at compiletime. You can increase the define in ds1820.c (which defaults to 4), but be aware that each additional probe will require memory (RAM), and you will run out of memory at some point. The actual limit has never been tested because it never was necessary for me, and comes from the ATtiny45 - so you should be able to at least double it with the ATtiny85, and even the ATtiny45 can probably do more than 4.
The length of the bus, and the power used by the probes. If the bus gets too long, signals and power won't be delivered anymore. The same is true for too many probes on the bus.

From some of the success supports I have got, it seems that you need to decrease the pull-up-resistor for the onewire bus in order to support longer cables, e.g. 2 KiloOhm instead of 4.7. This essentially pulls the bus back to its default state faster (sharper signal edges), but also wastes a little more power. There have been reports of a powered probe on 20 meters of cable, and a parasite powered probe on 8 meters - both with different pullup resistors (1kOhm).

The hostsoftware included allows you to query the status of the temperature sensors from the device. The software gets compiled with make hostsoftware - note that you will need libusb(-dev) to compile it. The software has been tested on linux and reported to work on Mac, but should compile/run on any platform where libusb is available, which includes windows.
If you call the hostsoftware with parameter --help it should tell you all you need to know. Here is a short introduction on how the software usually is used:
hostsoftware status - this will list all probes detected at poweron. Output should look something like this:

Device Version is 00.01 Device time is currently timestamp 60742 4 probes supported by device fa serial fl lastseen ticksago lastvalue 28 000001efea9d U 60574 168 22.06 28 000001efe323 U 60635 107 22.12 00 000000000000 0 60742 0.00 00 000000000000 0 60742 0.00

Here, the device has found two temperature sensors. It could handle two more, but only two have been detected.
If you want to make the data from a sensor available via network, you can use the command hostsoftware daemon which will fork a little server process into the background. For example, to serve the data from the probe with serial 000001efea9d on port 7337, you could run:
./hostsoftware daemon '000001efea9d:7337:%T'
The %T specifies the format of the output served on the network port - %T will only print the temperature, while the default is %S %T to print both serial number and temperature.

Downloads

Current version:
Sourcecode v 1.03 (22. December 2011)

Previous versions:
Sourcecode v 1.02 (17. July 2010)
Sourcecode v 1.01 (23. March 2009)

A more recent development version may be available in this gitweb repository.

Acknowledgements

This would not have been possible without AVR-USB (nowadays renamed to V-USB).

Copyright/Disclaimers

All trademarks are property of their respective owners.
Should you try to rebuild anything shown or use anything offered here, you do so at your own risk.
Your mileage may vary.
All plans and the parts of the software that were written by me are released under the GNU GPL Version 2. For details on the licensing of the included AVR-USB library, see the files in directory usbdrv or their homepage.

Similar projects

I know of some projects, where the first one is really a product that is sold, not a project:

Dallas / Maxim sells (proper / professional) 1wire to USB adapters under the name DS9490.
Somebody built a board with an (a lot) bigger AVR and more than one 1wire bus. (page is in german)
A DS1820 attached to a serial port instead of USB (page is in german)
DS18X20 Demo application/library
TempGrid - four 1wire busses with DS18S(!)20 connected to an AVR uC (page is in german)

Feedback?

You can post feedback under the Blog Article referring to this.