[NTLK] MP2x00 Internal Interconnect Breakout

Jake Bordens jake at allaboutjake.com
Tue Jan 13 18:21:11 EST 2015


>I will not use one of the dumb USB-to-RS232 converters, but instead use 
>an LPC1347 CPU, which incidentally costs the same, but comes with an ARM 
>Cortex MCU core running at 72 MHz. It has more than enough pins to 
>connect to every pin on the connector, and more (much more) than enough 
>power to handle USB and Ser1.

Cool idea.  Bonus points if you can get the external clocking to work as 
an option for really fast (2 megabit transfers).  I would love to see 
something better than 119200bps.

>It has *one* feature that made it my favorite: it can be programmed 
>through the USB port alone without any special software.

I haven't used this specific LPC, but I have used the mbed, and it is very 
nice.  I think the 1347 is supported by the mbed toolchain as well.

>- fixing the timing bug that requires slowdown.exe on MSWindows

Interested to know more about this, since you've mentioned it before.  I 
guess I'm a Mac user and don't experience this?   I thought that slowdown 
helps in poorly coded desktop apps that use no-op loops to delay.   The 
delays are too quick on modern computers.  But the USB-to-serial thing 
should be communicating at a fixed baud rate, so I don't see why it would 
be an issue for modern computers.  Matched baud clocks even on faster 
processors should communicate at the same speed.  I guess I just don't 
understand the issue, and I want to make sure I'm not missing something 
that I should be concerned with.

>And finally for the crazy part. We have access to the otherwise unused 
>serial port 3, so why not use it and add some functionality?

One crazy idea I had was to control a Si5351 clock generator and make an 
overclocking "implant" that could be switched by serial command.  In 
theory, the Si5351 could be made to switch frequencies cleanly, without 
glitching and causing a crash.  With your custom timing, serial comms 
could even be made to work at the odd baud rates you caused by 
overclocking.

I'm still partial to the WiReach module. 
(http://www.connectone.com/?page_id=3527) --especially if we could do 
external clocking and ~2mbit speeds. An all-in-one dream of an internal 
wifi card with USB would be damn sweet.  The chip would be able to select 
for USB or WiFi-PPP.  I think the big constraint in this case is physical 
space inside the case.



More information about the NewtonTalk mailing list