I think I remember that around 90 cm to 1 m is suspected to be the maximum pratical length, due to the way I2C signalling works. But it would be interesting to use this method to test this...
Steve Barker from HiTechnic (http://www.hitechnic.com/) told me that they ended up with 90cm as the longest cable in their 'Extended Connector Cable Set' because of problems with I2C beyond 1 metre.
Now, I'm not an electronics boffin, but I think he said that cables longer than 1 metre had problems with induction, and bringing it back to 90cm gave them a 10% (or so) safety margin.
This is also why there will probably never be a female-female cable joiner... )-:
Well, maybe Karl will read these comments and add one of his own to let us know his results. I know that even cat-5 cables has limits for its own lengths, so it doesn't surprise me that this type of cable will have its limits.
Anonymous said…
I'm going to try different lengths, I've just thought up an easy way to do this. More on that later.
Meanwhile. This page says that the main limit is the capacitance of the cable of 400pF: http://www.i2cchip.com/i2c_connector.html
Not being an electronics type I'm not sure how that relates to the cat5 cable, I found a cat5e specification of 5.6nF/100m (nominal). Whatever that means.
Writing as an, ahem, 'electronics boffin', or at least a guy whose degree was 50% electronics:
It is a long time time since I worked my way through the I^2C (I squared C) spec, but 90 cm sounds like a good place to stop.
The touch sensors do not use the I^2C bus, so they can probably stretch further. The motors do not use the I^2C bus, but the tachometer inputs are similar.
Philips created the I^2C interface as a cheap, efficient way to connect chips and sub-assemblies, (e.g. in TV set). It was not designed to run over long distances.
In mechanical terms I^2C is like a short lever with only a small movement. The further away you are from the lever the harder it is to tell whether your piece of string is being tugged upwards or downwards.
Other signaling schemes such as Ethernet over CAT cable and the RS485 signals on Port 4 of the NXT brick are better for long distances. In mechanical terms they push in one place whilst pulling in a corresponding second place.
Karl - your success with longer cables may depend on: (a) cutting out nearly all of the cheap phone cable, and (b)which wires are connected to each twist pairs in the CAT-5 cable. Pins 1 & 2 should be on the same pair. As should 3 & 6, and 4 & 5.
5.6nF/100m means a worst case of 5600pF capacitance over 100 metres of cable. So if 400pF defines the cut off point you may be able to achieve about 7m.
I am skeptical that this will be reliable. Especially as the cables are likely to run near electrically noisy Lego motors.
Comments
--
Brian Davis
Now, I'm not an electronics boffin, but I think he said that cables longer than 1 metre had problems with induction, and bringing it back to 90cm gave them a 10% (or so) safety margin.
This is also why there will probably never be a female-female cable joiner... )-:
Rob
Meanwhile. This page says that the main limit is the capacitance of the cable of 400pF: http://www.i2cchip.com/i2c_connector.html
Not being an electronics type I'm not sure how that relates to the cat5 cable, I found a cat5e specification of 5.6nF/100m (nominal). Whatever that means.
It is a long time time since I worked my way through the I^2C (I squared C) spec, but 90 cm sounds like a good place to stop.
The touch sensors do not use the I^2C bus, so they can probably stretch further. The motors do not use the I^2C bus, but the tachometer inputs are similar.
Philips created the I^2C interface as a cheap, efficient way to connect chips and sub-assemblies, (e.g. in TV set). It was not designed to run over long distances.
In mechanical terms I^2C is like a short lever with only a small movement. The further away you are from the lever the harder it is to tell whether your piece of string is being tugged upwards or downwards.
Other signaling schemes such as Ethernet over CAT cable and the RS485 signals on Port 4 of the NXT brick are better for long distances. In mechanical terms they push in one place whilst pulling in a corresponding second place.
Karl - your success with longer cables may depend on:
(a) cutting out nearly all of the cheap phone cable, and
(b)which wires are connected to each twist pairs in the CAT-5 cable. Pins 1 & 2 should be on the same pair. As should 3 & 6, and 4 & 5.
5.6nF/100m means a worst case of 5600pF capacitance over 100 metres of cable. So if 400pF defines the cut off point you may be able to achieve about 7m.
I am skeptical that this will be reliable. Especially as the cables are likely to run near electrically noisy Lego motors.
Apologies for the long essay.
ttfn,
Tony N