Re: An interesting chip for USB applications

From: Wolfgang Moser (
Date: 2006-10-07 23:38:44

Hello Marko,

Marko Mäkelä schrieb:
> A friend of mine recently told me about the MAX3421E from Maxim
> <>.
> He ordered some samples, but the estimated delivery time is in March 2007.
> Also, there are virtually no programming examples yet.
> The chip implements a USB interface, and it is controlled via SPI bus.
> It seems to support enough primitives to implement any USB protocol.

to me this chip looks like just another USB to any other
general availability serial bus. Beside RS232 (ft232bm or
pl2302) or 8-bit parallel fifo (ft245) a SPI this time.

> With some restrictions, it can even play a USB master.  Possible
> applications of that could be connecting a USB mouse, keyboard,
> memory stick, or a USB-equipped PC to a microcontroller, which would
> in turn emulate some Commodore peripheral.

Yeah, but the soft-USB cores are a high competitor in
the field of low-bandwidth USB peripherals, don't you

> by virtually all operating systems that support USB.  Another nice
> thing about this chip is that it adds USB support to any microcontroller
> that talks SPI.

Whenever it comes to USB for µCs I prefer a single chip
solution. Actually I don't prefer currently, since I did
not finish any USB project of my own beside some very
basic 8 LEDs soft-PWM driven "scroller"; I _would_ prefer
a single chip solution.

At first there's the Cypress EZ-USB series with the FX
and FX2 µCs. These are backed by an 8051 µC core with some
_very_nice_ extensions that no other 8051 derivative does
own (I really love the both autoincrementing pointers
with some undocumented features to find in outdated docs,
explaining that SFRs [some sort of "zeropage" for 8051]
can be used when internal RAM is accessed via the APTRs).
Please let me tell you that I really fell in love into
Cypress' 8051 implementation beside the USB feature; that
much that I consider to use these controllers for non-USB
projects! And yes, true, 8051 really is a very twisted
architecture, causing heavy brain damages, much too long
development times and more side effects professional
programmers always try to circument.

Secondly there are Microchips PICs with builtin USB cores

I heard about some Cons regarding the USB implementation,
but that may be bad rumors only from people not having
fully understood the specs or the concepts.

> For simple projects, USB-to-RS-232 converters could be a better choice.
> However, I have had very bad experience from the drivers for FTDI chips,
> and I do not think that the PL-2303 (which has excellent driver support
> on Linux and Windows) can be ordered in small quantities.

For the sort of very simple protocol converters like
USB to IEC or such I could think of an USB chip that
incorporates a programmable state machine with
additional "facilities" (RAM in/out FIFO, programmable
input/output/output driver).

Well, Cypress EZ-USB FX series already does contain
such a programmable state machine with up to 5 states.
It is needed to drive external circuits like memory
busses with USB-2.0 High Speed of 40MB/s.
I wouldn't get too overoptimistic because programming
of that state machine is _very_ complex, when you have
to obey all the nuts and bolts for a reliable transfer
engine. On all I have read about it last year it would
be worth some effort in learning how to do it, if I
think on the applications that may become possible
then (pumping bytes into a C64 via DMA).

The dual chip solutions always have one clear
advantage over the integrated USB chips: You don't
need any USB programming knowledge, when the chip
manufacturer delivers proper USV drivers as well. So
you can concentrate on you preferred µC architecture.


       Message was sent through the cbm-hackers mailing list

Archive generated by hypermail pre-2.1.8.