6 Mar
2012
6 Mar
'12
5:48 a.m.
On 06/03/12 09:57, Mouse wrote:
Indeed. Was it lack of politesse, though, or was it criticism? There
is a very important difference.
It was a question. Professional curiosity.
I asked why they'd used what appeared to be a totem-pole driver IC, when
every floppy drive I'm aware of uses open-collector I/O. I'm not going
to start on the ESD susceptibility of HCTTL, but I've seen HC244s go
into destructive latch-up when subjected to relatively minor ESD spikes.
They need snubber diodes and series resistors if you're going to use
them like this.
I don't like yes-men. I'd rather someone tell me I'd made a mistake if
they spot it... Constructive criticism is fine, and that's how a lot of
potential DiscFerret bugs were detected and fixed in early development.
DiscFerret has had two hardware bugs -- one was a missing resistor
(easily added) on R1 and R2 boards and the other was a missing power
track on R2 boards (caused by the PCB fab ignoring my specifications).
I - and, I suspect, tony - would say that's a good
reason to use
something other than USB as an interface. It's a large part of the
reason I am unhappy depending on USB for anything: it is complex
enough, and ill-documented enough (are the specs even available without
substantial payment and an NDA? I've never looked)
The full specification from PHY level to application layer is open. The
only thing you have to pay for is a Vendor ID allocation. Or if you want
to use the "trident" logo, you have to pay for a Trademark License and
do qualification testing.
In practice? Few pieces of equipment actually use the logo; most either
use the text "USB" (which AIUI isn't trademarked), or just leave the
port unlabelled.
USB has reached a point of ubiquity such that you can point to the
rectangular USB socket on a PC, ask any PC user what it is, and they'll
say... "It's a USB port!"
that, once
now-common chips that implement it are no longer available, USB devices
will become basically useless.
The standard is open. You can implement a USB2.0 Full Speed (not High
Speed 480Mbps -- 12Mbps or 1.1Mbps) host adapter on an FPGA without too
much effort. EHCI, UHCI and OHCI are all documented (EHCI is the I/O
interface for USB2 HAs, UHCI and OHCI cover USB1.1) and standardised.
Sure it's more difficult than RS232 but it can be done.
And if you want RS232 from a Discferret? See that UART port? :)
Swap the communication code for an RS232 implementation. You can
probably keep HandleEvent() as is. The only reason it hasn't been
implemented is because it'd be so slow as to be unusable. If there's
sufficient demand, I'll implement it.
1) Did you use a 74HC244 buffer is the receiver for
the drive cable.
Did you then cause Phil (I think) to be thrown off a forum/facebook
group when he commented on this?
If you'd come to my house and you
continuously make fancy statements
and suggest things (and have done so in other places in the past) it
might happen you get thrown out for not behaving politely. not understand it; my view generally is "costs
more, less flexible,
worse support, what's to like?". Yet people persist in going with the
commercial products. I've never found it explained in a way I can even
understand, much less agree with.
"Nobody ever got fired for buying IBM."
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/