[Diy_efi] Any USB developers out there?

[Torbjörn Forsman]

Texas Instruments has several microcontrollers with integrated USB 
controllers that are popular for hobbyist projects. For example, the 
TUSB3410.
When making high-speed or full-speed USB implementations, the circuit 
layout between the controller's USB pins and the connector is critical. 
It is also critical to follow the controller manufacturer's 
recommendations regarding VCC decoupling, possibly separate ground plane 
for the USB transceiver part of the controller and so on.
Make sure that the clock oscillator runs at the correct frequency with 
maximum ?50 ppm error, use a newly calibrated frequency counter for the 
measurement. It is a good idea to design in a oscillator testpoint with 
a voltage divider, so the clock frequency can be measured without 
disturbing the oscillator. Some USB controllers have a buffered CLKOUT 
pin which can be used for a such measurement.

Best regards

Torbj?rn Forsman

Steve Ravet wrote:
>  
> 
> 
>>-----Original Message-----
>>From: diy_efi-bounces at diy-efi.org 
>>[mailto:diy_efi-bounces at diy-efi.org] On Behalf Of Mike
>>Sent: Wednesday, November 08, 2006 3:48 AM
>>To: diy_efi at diy-efi.org
>>Subject: Re: [Diy_efi] Any USB developers out there?
>>Theres an interesting issue about USB which I've thought 
>>clarifying somewhere without having to wade through numerous 
>>tech docs and that is the low speed backward compatability. 
>>In theory this would allow almost any, even low end single 
>>chip microprocessors to read the serial stream directly in 
>>software at around 19.2 Kbits/sec ( or thereabouts) without 
>>the need for a dedicated hardware USB chip and handle all the 
>>protocols leaving some processing over for the "real stuff" 
>>it is there for in the first place. The consequence clearly 
>>is that any other devices on that USB channel will be also 
>>running low but as most PCs these days have 4 USB ports then 
>>- 'what the heck'...  So if this were the case then a truly 
>>low end opportunity exists for good firmware programmers at 
>>the low end embedded level to interface to engine 
>>electronics. Is 19.2K the lowest with a low granularity or 
>>could 32K, 56K etc be ok if the cpu can handle it ?
> 
> 
> Mike, I think you are talking about bit-banging the USB lines using a
> microcontroller's general purpose I/O lines, right?  I ran across a
> thread about that on the piclist, which I recently joined.  There were
> pointers to a couple software USB implementations but the summary seemed
> to be that adding a software implementation to the firmware already
> required to communicate with the host didn't leave much processing left
> over on the cheap uCs.  A more expensive one might be able to do it, but
> then the incremental cost of interface hardware doesn't matter as much.
> 
> --steve
> 
> _______________________________________________
> Diy_efi mailing list
> Diy_efi at diy-efi.org
> Subscribe: http://lists.diy-efi.org/mailman/listinfo/diy_efi
> Main WWW page:  http://www.diy-efi.org/diy_efi
>