20 Aug
2012
20 Aug
'12
11:33 a.m.
On Aug 19, 2012, at 2:04 AM, Mouse wrote:
>> In fact, for what it was designed to do, it works amazingly well.
>> It's not even "new" anymore, so we can't use that excuse.
>
> The biggest problem I have with it is that it requires a full-fledged
> CPU or a comparable amount of custom silicon to speak at all - it is
> not amenable to "throw together a breadboard lashup from discretes"
> implementations. Of course, this is a purely personal point of view;
> there's only one other person here I'm reasonably sure will agree with
> it, that being tony.
That is cetainly one of my main objections to USB in a lot of cases --
it's over-complex for the job in hand.
Well, not the only one. That's my primary issue with USB, as it is
with PCI. But I do recognize that once speeds get into the tens
of MHz, things rapidly fall out of the "hobbyist" range, so much
Sure _if you need that sort of data throughput_. I didn't complain about
the USB interface on the Diskferret becuase the amount of data to
transfer was large, and USB was about as simple/complex as any other
interface which could transfoer that data i na reasonable time.
But when you have soemthing that doesn't send large amounts of data, then
USB _is_ over complex. There are simpler interfaces that will do the job
as well.
I do find the idea of building-in a USB to async serial IC to be mildly
unpleasant. My point being that the async interface can clearly handle
the data throughput _and it is likely to be able to be used with many
other types of computer. It's easy to convert a USB host to connect to an
Async peripehral (the FTDI IC being high up the list...) it's much, much,
harder to do the reverse. I hate designs that make things deliberately
difficult.
-tony