Subject:
|
IR Performance -- 'standard' vs 'fast' (2400 vs 4800) and RIS 1.0 and 2.0
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 25 Nov 2005 06:29:42 GMT
|
Viewed:
|
1577 times
|
| |
|
|
Several recent posts about infrared signaling performance
with and without fast download (2400 vs 4800 baud, with and
without message 'complement' bytes) on 1.0 and 2.0 bricks.
Here's some results of testing five COM Towers and five USB
towers with many different RIS 1.0 bricks and one RIS 2.0
brick.
There are electronic differences between the bricks and
towers in their infrared (IR) receiver circuitry.
* Not all bricks and towers have the same performance.
* RIS 1.0 and 2.0 have different electronics for infrared
support. 1.0 uses an IR receiver optimized for a 38 KHz
infrared carrier. 2.0 uses an IR receiver optimized for a 76
KHz carrier. "Optimized" means they have a bandpass filter
at that frequency.
* Serial towers use IR receivers optimized for 38 KHz. USB
towers use 76 KHz IR receiver.
* When transmitting on a 38 KHz carrier, this usually works
on a 76 KHz IR receiver. I guess the filter also lets 38 KHz
leak in.
* Another quirk in the IR receiver chips is that they need
at least six complete cycles of carrier to recognize a
"one". If you do the math, you'll find that there is between
seven and eight carrier cycles in one bit time at 4800 baud
transmission; so there's not a real big margin for error.
* There also appears to be differences between the two tower
types in their IR transmitters. The serial towers put out a
better signal that generate fewer brick reception errors
than the USB towers.
I've tested ten different RIS 1.0 bricks at quad speed
performance using multiple serial towers and USB towers. The
results are:
* Serial towers give near perfect performance at 4800 baud
(and 38 KHz carrier). Only 3 of 10 bricks worked with USB
towers.
* I then analyzed where the errors occurred. Almost all of
the errors were with the brick failing to recognize the
message.
* They predominant error condition was failure to recognize
the first byte of an message - typically a framing error or
parity error occurred.
* 1.0 bricks do not work when using a 76 KHz carrier
frequency.
I next looked at ways to improve the messaging performance
using the available configuration options:
* USB tower has options to allow you to set the transmit
power. I couldn't find a power level setting that improved
the performance.
* USB tower has options to specify a custom IR carrier
frequency. I tried using custom frequencies between 38 and
76 KHz to see if performance would improve hoping that this
might increase the "tolerance" on need for six complete
carrier cycles to recognize a "one". I combined this with
increasing the transmit power to compensate for the
'bandpass filter". Again, I couldn't find a setting that
improved performance.
* Putting the brick and tower in a cardboard box doesn't
work to improve quad speed messaging performance.
So, bottom line is:
* Serial towers appear to give near perfect performance on
both 1.0 and 2.0 bricks using 38 KHz carrier.
* USB towers and 1.0 bricks are often incompatible at 4800
baud transmission. If you're not lucky enough to have a 1.0
brick that works at 4800 baud, then you'll simply have to
give up and use 2400 baud.
* If you're using a 2.0 brick and a USB tower then you want
your PC software to use the 76 KHz carrier frequency instead
of 38 KHz. I don't know if the various IDEs allow this as an
option. This works perfectly on my system but I only have
one 2.0 brick to test on so I can't generalize this.
* 1.0 bricks don't work with 76 KHz carrier. So if you can
configure the carrier setting, make sure it's set
appropriately for the type of brick you own.
* What I've done is given up on reliable quad speed
downloading and fallen back on a dual speed solution. My
downloading uses 2400 baud but without the complement bytes
to get a dual speed performance. This works very robustly
with no noticeable errors.
|
|
Message has 1 Reply:
Message is in Reply To:
 | | Re: RCX Question
|
| (...) I've never come across this phenomenon - mine have all worked first time. All of the RCX versions are the same except for the external power socket on the 1.0's. In terms of software, firmware, etc they are all utterly identical. (19 years ago, 24-Nov-05, to lugnet.robotics)
|
6 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
Active threads in Robotics
|
|
|
|
|
|
