several commercial i2c sensors to connect at 1 NXT port?

[HaWe]

yes, I know, at least gyro and accelerometer would be a good choice.
but 3 sensors is still almost nothing compared to possibly up to 127 devices at 1 port acc to i2c standard.

It's really poor that we cant have this with devices made for NXT usage.

[mightor]

All robotics platforms come with their strengths and weaknesses. You can either accept them and make the most out of what you have or pick another platform. There is no lack of alternatives that might do a better job at the whole I2C thing. However, those have their weaknesses, too.

I will send a mail to Jørgen Vig Knudstorp and let him know you are not impressed with the I2C functionality in the NXT.

- Xander

[HaWe]

I doubt that he cares ;)

but actually it's not the NXT i2c functionality - it's simply the lack of not having free adjustable sensors for disabling pullups and configuring different bus addresses -
that's a deficiency of the sensors' abilities, not of the NXT brick.

But I'm looking forward to a fruitful professional dialogue! :D

[mightor]

You must keep in mind that extra functionality in a sensor means extra development time, which equals additional development costs. The number of people who want to have more than 4 sensors on their NXT is probably quite small, compared to the number of people who are happy with the 4 they can use. If you take the number of people who want more than 4 and you subtract the number whose problem is not solved with the HiTechnic Sensor MUX, then you are left with a really, really small group, which, unfortunately for you, includes you.
I'd love to have a sensor port that can do SPI as well or a proper hardware accelerated I2C implementation, rather than this bit-banged stuff that crawls along, when compared to the 100/400KHz clock speeds most I2C buses go at.
The VEX Cortex has a lot more sensor and motor ports, including one dedicated I2C port. The I2C port is apparently fast enough to be used for encoders but I don't have all the technical details just yet. I'm due to receive some of these new encoders and I'll let you know how it goes.

- Xander

[HaWe]

if it had been planned correctly according to the standards right from the start this wouldn't be an issue and we wouldn't have to discuss now.

I honestly doubt that only few people would like to have more than 4 sensors, or there wouldn't be a demand for sensor (or motor) multiplexers or use of a 2nd or 3rd brick by BT or RS485 to multiply the number of I/O's. Maybe your guess is true just for freshmen having a new basic kit.

I also doubt that disabling pullups e.g., by a microswitch (just short-circuiting the 2 resistors) would cost noticably much money, maybe just 10ct for hardware and 10ct for thinking about it. On the other hand, 40-50$ for a sensor should include a way for adjusting the bus addresses by software.

But I'm really excited to start an exchange of views with Jørgen about that.

[mightor]

Don't forget that the target market for the NXT, as defined by LEGO's marketing department, is a 13 year old boy. I think it is safe to say that you fall slightly outside that category

The NXT's design is rife with compromises, that's part of making a commercial design As for adding DIP switches or something similar to a design, it really adds to the overall manufacturing costs. I think you underestimate the amount of effort that goes into making a sensor and overestimate the amount of profit these companies make.

Oh, Jørgen V got back to me and said "Oh no, not Doc H again! Tell him to buy an Arduino and learn to solder!". So there you have it. You should check out this place: http://www.watterott.com/, it's a good German online store. I buy there all the time.

- Xander

[HaWe]

I doubt that he had ever heard of me, it's a nice story though. But what MS can do, can Lego do as well.
And no, no soldering ever, and no Arduino at all as long as something like Java will be within 2 miles around it.
But I will spend Jorgen 20cts (10 for thinking, and 10 for building) for each produced lego sensor with a dip switch if I'll get 10 percent from the profit of every additionally sold sensor.

[mightor]

I've heard stories that would make your jaws drop that involve cost cutting measures You would be amazed at some of the things they do to save 10 cents

Oh and Jørgen says you should really stop following him around to ask him about I2C, he's about this -> <- close to getting a restraining order on you. It's not funny anymore. You're scaring his kids.

- Xander

[aswin0]

Doc,

You're trolling again. Why?
1 you do not come up with a real problem
2 you can use a MS sensor together with another sensor on the same port. This increases the possibilities
3 you can remove the pull ups from some sensors if you really need to
4 you probably can use two sensors both having 82k pull ups on a single port as this still give a pull up value of 41k
5 you started bashing java before anyone else mentioned it

Btw. The number 127 refers to the number of different addresses, not to the number of sensors that can be attached to one port. Every system will hit a limit of some sort long before 127 sensors are attached. On the NXT the limit might be reached a bit sooner. Who cares.

[HaWe]

you are not serious accusing me of trolling, are you?
I was just replying to Xanders posts - nothing else, just the same as I'm now replying to your post.

It was Xanders OT idea bringing Jorgen personally (I never asked him about anything! ) and some fictive company price policies into play, not my.
Also the OT suggestion of soldering using a Java based Android - again OT.
The same issue was about your OT post about the dexter sensor which was not in any relationsship to my question,
and you will not pretend to judge if I have a real problem with the number of sensors I actually need to use.

My question has been answered.

If I am approached directly, I will surely have the right to respond, or not?