|
|
| Re: South Ferry Inner Loop Station. (1385573) | |||
 |
|||
| Home > SubChat | |||
|
[ Read Responses | Post a New Response | Return to the Index ] |
|
||
|
Re: South Ferry Inner Loop Station. |
|
|
Posted by Stephen Bauman on Sun Feb 14 21:47:01 2016, in response to Re: South Ferry Inner Loop Station., posted by Jace on Sun Feb 14 18:47:57 2016. wouldn't the additional processors add complexity and with that increase the chances of failure?That depends on how the additional processors are interconnected. The usual way to create reliability (absence of catastrophic failure) is to use redundant elements. See Moore and Shannon's classic paper, "RELIABLE CIRCUITS USING LESS RELIABLE RELAYS" from 1956. my guess is that it [LonWorks] was chosen for simplicity. That's been a losing proposition since the late 1980's. The chip makers have been able to consistently pack more complexity into microprocessors while increasing reliability and decreasing cost. The message protocol is limited by LonWork's simplicity. This means the NYCT will be stuck with LonWorks for the 50+ age of the NTT cars. Finding replacement LonWorks hardware will become increasingly more difficult and expensive. LonWorks will be lucky to be economically viable for 15 years. It's the current nature of the semiconductor business. It used to be worse. New IC's were coming out at a ra[od rate in the late 1960's and early 1970's. It was impossible to design any apparatus that wasn't obsolete by the time the prototype was built. The strategy to overcome hardware obsolescence is to make the message structure hardware neutral. The NYCT message protocol is very much tied to the hardware implementation. And you didn't explain how such a flexible door system would work in practice. I did 20 years ago. Here's an example with regard to the propulsion system. The messages state the throttle position. There may be about 100 different positions (I forget the exact details.) Different model cars have different propulsion controllers that generate different acceleration profiles for the same throttle position. That's why different differet propulsion controllers are not MU'ed. Suppose the message had been accelerate to 20 mph @ 2.5 mph/sec instead. The receiving microprocessors would know how to translate that hardware neutral message into the step sequence on that car's propulsion controller. Different equipment could be MU'ed without bucking. Newer generations (read less expensive, faster, more powerful) of microprocessors could be used on newer cars and still MU. |
|
|
|