Sometime years ago, I had heard (IIRC on this forum) that 1-wire technology is patented, such that I can't implement my own slaves. I've not come up with the patent yet, and Maxim hasn't responded to me yet. Anyone know if this is true, if the patent is still valid, and what the general details are?
I want to bi-directional communication to a PIC application, and have only 1 wire available. I also have no room for additional (1-wire slave) chips, but that would be pointless anyway as that slave chip would need more than one wire for communicating with the PIC.
If this is not doable, I may end up rolling my own protocol that can work with a single wire, but I'd need to know what specifically is patented with respect to 1-Wire technology.
Here's a 1998 discussion on Dallas 1 wire appolication. Refence 1
cites a 1997 paper. That's 13 years - about at the upper limit of any
patent that may have applied originally.
Half duplex on a single pair was used since time immemorial and hald
duplex with ground return was used by Noah in the ark ((had problems
due to poor ground).
> Sometime years ago, I had heard (IIRC on this forum) that 1-wire
> technology is patented, such that I can't implement my own slaves.
> I've not come up with the patent yet, and Maxim hasn't responded to me
> yet. Anyone know if this is true, if the patent is still valid, and
> what the general details are?
>
> I want to bi-directional communication to a PIC application, and have
> only 1 wire available. I also have no room for additional (1-wire
> slave) chips, but that would be pointless anyway as that slave chip
> would need more than one wire for communicating with the PIC.
>
> If this is not doable, I may end up rolling my own protocol that can
> work with a single wire, but I'd need to know what specifically is
> patented with respect to 1-Wire technology.
>
> Cheers,
> -Neil.
>
>
On 23/09/10 14:57, PICdude wrote:
> Sometime years ago, I had heard (IIRC on this forum) that 1-wire
> technology is patented, such that I can't implement my own slaves.
> I've not come up with the patent yet, and Maxim hasn't responded to me
> yet. Anyone know if this is true, if the patent is still valid, and
> what the general details are?
You probably couldn't find it because it wasn't applied for by Maxim. It was applied for by Dallas Semiconductor:
A system architecture which provides efficient data communication, over a one-wire bus, with a portable data module which does not necessarily include any accurate time base whatsoever. The time base in the module can be extremely crude (e.g. more than 4:1 uncertainty). An open-collector architecture is used, with electrical relations defined to absolutely minimize the drain on the portable module's battery. The protocol has been specified so that the module never sources current to the data line, but only sinks current. The protocol includes signals for read; write-zero; write-one; and reset. Each one-bit transaction is initiated by a falling edge from the host. The time base in the module defines a delay, after which (in write mode) the module tests the data state of the data line. In read mode, after a falling edge the module does or does not turn on its pull-down transistor, depending on the data value. Thus, the host system, after the falling edge, attempts to pull the data line high again, and then tests the potential of the data line to ascertain the data value.
..... Bit of a mouthful, but there you go. Won't expire for another ten years, near as I can tell.
> If this is not doable, I may end up rolling my own protocol that can
> work with a single wire, but I'd need to know what specifically is
> patented with respect to 1-Wire technology.
In a word: "Everything". From the iButton MicroCan and ESD protection circuitry to the 1Wire control protocol. All in that patent I linked above.
I seem to recall someone from Dallas stating that they don't enforce the patent with respect to 1-Wire Master devices (for obvious reasons) but they will take action against anyone making slave devices that are claimed to be 1-Wire compliant -- each 1-W device has a 64-bit unique serial number, and Dallas didn't want S/N collisions (because this is the one way to REALLY screw things up in a 1-Wire bus).
IMHO there is a fair bit of prior art. If you took the signal from MSF (the Rugby / Anthorn time signal), converted to baseband and sped it up, you'd end up with something similar to 1-Wire. It's basically pulse-width modulation. If all you've got is one wire, PWM is pretty much the only game in town.
If you don't claim it to be 1-Wire (call it PWM-Bus or something) and don't implement some of the more esoteric bits of the protocol (i.e. use it as a physical-layer with your own link-layer protocol) you might be able to get away with it in the same way lots of companies got away with making I2C EEPROMs... call them something else. 2-wire, MicroWire, "two wire addressable serial bus"...
Ob disclaimer: this is an Engineer's Opinion. Seek the advice of a lawyer specialising in intellectual property law before you go designing anything...
> Of PICdude
> Sent: 23 September 2010 14:58
> To: EraseMEpiclistspam_OUTTakeThisOuTmit.edu
> Subject: [EE] Implementing 1-wire slaves
>
> Sometime years ago, I had heard (IIRC on this forum) that 1-wire
> technology is patented, such that I can't implement my own slaves.
> I've not come up with the patent yet, and Maxim hasn't responded to me
> yet. Anyone know if this is true, if the patent is still valid, and
> what the general details are?
>
> I want to bi-directional communication to a PIC application, and have
> only 1 wire available. I also have no room for additional (1-wire
> slave) chips, but that would be pointless anyway as that slave chip
> would need more than one wire for communicating with the PIC.
>
> If this is not doable, I may end up rolling my own protocol that can
> work with a single wire, but I'd need to know what specifically is
> patented with respect to 1-Wire technology.
You can do simplex UART comms over a single wire if you drive the wire
with an open collector type interface. Obviously the USART which is
transmitting on the bus also receives it's own data, so it's relatively
easy to detect collisions on a byte level (or bit level if you bit bash
the UART).
Regards
Mike
=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================
It never ceases to amaze me how quickly you research these things... I need to learn your google skills. I'd go look for a tutorial, but that really is my problem there, isn't it :)
FWIW, IIRC a patent was valid for 17 years, and recently has been changed to 20 years. I need to verify that, but if so, I should be good. I'll report back after some reading.
> 1993 patent appears to be 5210846
>
>
> Here's a 1998 discussion on Dallas 1 wire appolication. Refence 1
> cites a 1997 paper. That's 13 years - about at the upper limit of any
> patent that may have applied originally.
>
> http://archives.sensorsmag.com/articles/0698/wir0698/index.htm
>
> Half duplex on a single pair was used since time immemorial and hald
> duplex with ground return was used by Noah in the ark ((had problems
> due to poor ground).
>
> Atmel think they/you can do it
>
> http://www.atmel.com/dyn/resources/prod_documents/AVR274.pdf
>
> 2000
>
>
> http://www.sensorsmag.com/sensors/humidity-moisture/a-1-wire-humidity-sensor-1080
>
> 1997
>
> Feb. 1997. "Transmitting Data and Power over a One-Wire Bus,"
> Sensors, Vol. 14, No. 2:48-51.
>
> Text from 1997 paper here
>
>
> http://www.datasheetarchive.com/datasheet-pdf/03/DSA0036838.html
>
> Related
>
>
> http://blog.onlinecomponents.com/post/buddy-can-you-spare-an-i-o-pin-try-a-1-wire-interface
>
> Patent litigation 1998, re 1993 patent
>
> http://www.uspto.gov/web/offices/com/sol/og/1998/week30/patrequ.htm
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> On 24 September 2010 01:57, PICdude <@spam@picdude3KILLspamnarwani.org> wrote:
>> Sometime years ago, I had heard (IIRC on this forum) that 1-wire
>> technology is patented, such that I can't implement my own slaves.
>> I've not come up with the patent yet, and Maxim hasn't responded to me
>> yet. Anyone know if this is true, if the patent is still valid, and
>> what the general details are?
>>
>> I want to bi-directional communication to a PIC application, and have
>> only 1 wire available. I also have no room for additional (1-wire
>> slave) chips, but that would be pointless anyway as that slave chip
>> would need more than one wire for communicating with the PIC.
>>
>> If this is not doable, I may end up rolling my own protocol that can
>> work with a single wire, but I'd need to know what specifically is
>> patented with respect to 1-Wire technology.
>>
>> Cheers,
>> -Neil.
>>
>>
> ...
> US Patent number 6,108,751, granted August 22, 2000.
> PDF here: http://www.freepatentsonline.com/6108751.pdf
> ...
> .... Bit of a mouthful, but there you go. Won't expire for another ten
> years, near as I can tell.
Wow, that's odd -- a different patent #, and a few years later than the one Russell found.
So does this mean that there was a correction (possibly because of the Lexmark lawsuit) and a new patent number? Guess I'll find some answers when I read these.
> ...
> IMHO there is a fair bit of prior art. If you took the signal from MSF
> (the Rugby / Anthorn time signal), converted to baseband and sped it up,
> you'd end up with something similar to 1-Wire. It's basically
> pulse-width modulation. If all you've got is one wire, PWM is pretty
> much the only game in town.
> ...
When I think of PWM, I think of unidirectional signals with many different duty-cycles, such as RC servo control signals. From what I remember of 1-wire from a few years ago, there was a time slot for requesting and reading back whether a bit is '0', then another slot for requesting and reading back if that same bit is a '1', so it did two checks for each bit.
> If you don't claim it to be 1-Wire (call it PWM-Bus or something) and
> don't implement some of the more esoteric bits of the protocol (i.e. use
> it as a physical-layer with your own link-layer protocol) you might be
> able to get away with it in the same way lots of companies got away with
> making I2C EEPROMs... call them something else. 2-wire, MicroWire, "two
> wire addressable serial bus"...
Or I can just create my own protocol. What I'm doing is not rocket science, and fairly low speed, so I can do some sort of serial with the master telling each slave when to send over it's data. I don't plan to implement unique serial numbers, because the only thing I didn't care for with 1-Wire is that the order of the slaves (such as multiple temp sensors) was unknown until after they were connected. Yes, it can be corrected in software (with the user indicating which one is sensor 1, sensor 2, etc) or by physically swapping sensors after that. And I don't need it to be 1-wire compatible.
On Thu, 23 Sep 2010 16:13:23 -0700, "PICdude" said:
> Or I can just create my own protocol. What I'm doing is not rocket
> science, and fairly low speed, so I can do some sort of serial with
> the master telling each slave when to send over it's data. I don't
> plan to implement unique serial numbers, because the only thing I
> didn't care for with 1-Wire is that the order of the slaves (such as
> multiple temp sensors) was unknown until after they were connected.
> Yes, it can be corrected in software (with the user indicating which
> one is sensor 1, sensor 2, etc) or by physically swapping sensors
> after that. And I don't need it to be 1-wire compatible.
Hi Neil,
I did a thing with one of our products that has a wired remote using one
wire for bidirectional data. In normal operation the data rate was very
low, like 150 bps. But it could run at high speed so during manufacture
or repair it gets hooked to a computer serial port so it can get
calibrated. On the computer side I just made an adapter with one diode
in it, tying RXD and TXD together. The device was always a slave so
either the remote or the computer was master.
You could do something like that. As long as you always have only one
master that keeps it from getting complicated.
> Or I can just create my own protocol. What I'm doing is not rocket
> science, and fairly low speed, so I can do some sort of serial with
> the master telling each slave when to send over it's data. I don't
> plan to implement unique serial numbers, because the only thing I
> didn't care for with 1-Wire is that the order of the slaves (such as
> multiple temp sensors) was unknown until after they were connected.
> Yes, it can be corrected in software (with the user indicating which
> one is sensor 1, sensor 2, etc) or by physically swapping sensors
> after that. And I don't need it to be 1-wire compatible.
>
> Thanks,
> -Neil.
>
The serial number is also used to let a master find out which devices are present on the bus using the Rom Search algorithm. So even if the master does not have prior knowledge about the the serial numbers of the devices that are connected to it's 1-wire bus, using the Rom Search algorithm and the unique serial numbers it can find this information.
There are even some new temperature sensors that uses an extra wire which can be used to find out in which physical order the sensors are located on the bus.
/Ruben
==============================
Ruben Jönsson
AB Liros Electronic
Box 9124, 200 39 Malmö, Sweden
TEL INT +46 40142078
FAX INT +46 40947388 RemoveMErubenTakeThisOuTpp.sbbs.se
==============================
> There are even some new temperature sensors that uses an extra wire which can
> be used to find out in which physical order the sensors are located on the bus.
Real men [tm] use TDR :-).
I once did a job where somebody wanted me to use TDR to do just this
as part of a commercial many many station multi drop system
(bidirectional multistation signalling all over an inductive powering
loop). I looked at the TDR timings and then assured them that there
were far easier ways to achieve the desired result. (This lead to my
only US patent - since expired. Not at all marvellous as patents go -
but fun to have had and good precedence for power and signalling
transfer over inductive loop systems. I must dig it out and look at
the date - maybe I predate Maxim :-) ).
There are historically numerous systems which derive slaves' addresses
by polling and random duration backoff on collision algorithms, or
which even assign addresses to unruly addressless rabble at start up
using similar means.
I suspect, that Maxim's concern is systems which talk to their
products or which are talked to by Maxim master systems, and that they
have neither interest or claim in anything which just uses 2 wires
(signal plus return to convey bidirectional signalling and power - or
just signalling, but is incompatible with their systems. But, I may be
wrong.
On 24/09/2010 08:06, RussellMc wrote:
>
> There are historically numerous systems which derive slaves' addresses
> by polling and random duration backoff on collision algorithms, or
> which even assign addresses to unruly addressless rabble at start up
> using similar means.
Ethernet on a coax comes to mind surely as a famous "1 wire" multi-party half duplex bus with all unique but unknown addresses (MAC) and 10Mbps.
With a well known
random duration backoff on collision algorithms
If you did AX.25 on a wire, rather than over the air it would also achieve the same. There are AX.25 implementations (some may even be open source on a PIC that use external modem, hence the UART could be interfaced via open collector to a multi-party one wire bus.
> > There are historically numerous systems which derive slaves'
addresses
> > by polling and random duration backoff on collision algorithms, or
> > which even assign addresses to unruly addressless rabble at start up
> > using similar means.
>
> Ethernet on a coax comes to mind surely as a famous "1 wire"
multi-party
> half duplex bus with all unique but unknown addresses (MAC) and
10Mbps.
But this doesn't power the remote devices over the same wire, which I
suspect is a feature of the 1-wire system that they want to protect (I
haven't checked the patent to see if this is part of the description).
-- Scanned by iCritical.
On 24/09/2010 09:42, spamBeGonealan.b.pearcespamBeGonestfc.ac.uk wrote:
>>> There are historically numerous systems which derive slaves'
> addresses
>>> by polling and random duration backoff on collision algorithms, or
>>> which even assign addresses to unruly addressless rabble at start up
>>> using similar means.
>> Ethernet on a coax comes to mind surely as a famous "1 wire"
> multi-party
>> half duplex bus with all unique but unknown addresses (MAC) and
> 10Mbps.
>
> But this doesn't power the remote devices over the same wire, which I
> suspect is a feature of the 1-wire system that they want to protect (I
> haven't checked the patent to see if this is part of the description).
Pretty simple to power over the same wire if you use Manchester encoding.
You can then either use a RF filter and carrier for data over arbitrary desired DC, or just use the average voltage of Manchester encoding and make peak = Vdd x2 and have LC filter...
narwani.org> wrote:
