gnupic: DIY USB programmer ?


Previous by date: 8 Jan 2005 20:21:06 +0000 Re: Problems compiling gpsim, Diego Belotti
Next by date: 8 Jan 2005 20:21:06 +0000 Re: Problems compiling gpsim, Peter L. Peres
Previous in thread: 8 Jan 2005 20:21:06 +0000 Re: DIY USB programmer ?, Byron A Jeff
Next in thread: 8 Jan 2005 20:21:06 +0000 Re: DIY USB programmer ?, Manuel Bessler

Subject: Re: DIY USB programmer ?
From: ####@####.####
Date: 8 Jan 2005 20:21:06 +0000
Message-Id: <26537.1105215637@www18.gmx.net>

> On Sat, Jan 08, 2005 at 08:59:42AM +0100, ####@####.#### wrote:
> > Hi:
> 
> Hi.
> 
> > tree points to clearify:
> > USB FTDI chip:
> 
> > This chip is full programmable and it's possible to get full control.
> >Timing tests has shown, that non streamed bitbang mode give a resolution
> of
> >2ms and streaming bitbang mode give a resolution of 1ms.  To archive
> this,
> >the D2XX dll must be installed and used. This dll is available for
> windows
> >and linux. Take a look at the D2xx programmer manual on the ftdi website.
> 
> But the question is are the requests guaranteed to be serialized? If I
> send
> a turn on bit 1, turn off bit 2, turn off bit 2, turn off bit 1, within a
> 1ms window will you get all 4 transistions.

The transaction is byte based. The byte is translated to the 8 input/output
pins
according to the direction setup, like any ordinary 8 bit MCU port.
If you write a byte and read a byte, the byte is sampled at the same
intervall,
every 1ms or any other timebase you have specified. The BM chip don't
support values
faster as 1ms. The C chip has a serial engine, that support faster times.
So, if you turn on bit 1, turn off bit 2, turn off bit 2, turn off bit 1,
bit 1 is turned on after 2ms regeadless of using the internal fifo and
streaming or not.
If streaming is not used, the timing is 2ms instead of 1ms using the 1ms
intervall time on setting up the communication.
one ms later bit 2 is turned off.
one ms later bit 2 is turned off
one ms later bit 1 is turned off.

The important point is, that the vdi driver must be deinstalled and de d2xx
driver
must be installed to perform this under windows. Under linux this is
similar.

> 
> Again I don't think that the FTDI chips are the solution for the casual
> user. Too slow from your report here, too difficult to manage in the
> raw surface mount package, and too expensive when bundled into a module.
> 

Yes, you'r are in right, but for a project, that already have planned to use
FTDI chip for USB usage, what is better as using this chip to perform the
initial (bootloader) programming. 

> > 
> > USB to serial converter:
> 
> > The high price converter ~40$ have 5 or 8 drivers as opposed to max232
> that
> > has 4.  The lower priced only four line drivers.  Neverless thees works
> > great, if a appropiate schematic is choosen, but you must know, if the
> > converter is dce or dte. With a voltmeter it's possible to check this. 
> The
> > windows function call must be used to drive the pins. The direct
> register
> > access will not work.  Here are some example of low cost, or low line
> count
> > programming interfaces, with or without level converters. Mostly, the
> > transistors and r/c circuits can be exchanged with a max232.
> > http://www.yusoft.by.ru/pic/images.htm
> 
> Same question for the USB to serial converter as for the FTDI chips: How
> fast (or can you) wiggle the modem control lines? What are the timing 
> relationship of the modem control lines to each other and to TX? Can you 
> guarantee these relationship among all of the common USB to serial
> converter cables?
Using the windows api, i can manipulate control signals and this are
guaranted to
action between any tx frame.
So i can setup dtr as data, and trigger the the data with the TX line. The
only
rule that i must observe, is that i generate only one square wave with the
tx line and that the data pin is sampled, if the tx goes from high to low,
so i need a 
line inverter. Just as a case, the max232 perform this automatically.
I can even use some of the jdm programmers when exchanging tx from making a
reset to
the clock line and using a sw that observes the above rule.
Under unix, i know how to make this with a ftdi232xx chip, i don't know if
the same
ioctl's works on any or the most other usb to rs232 converters.

> 
> It would require some extensive testing to answer these questions.
> 
>  
> > If you know some ready available low cost usb2rs232 converter that uses
> the
> >ftdi  chip (10-15$) let me know.
> 
> No clue. Also I have no interest in limiting a circuit to a particular
> cable.
This was for my personal interrest. You have claimed, that the cheap usb to
rs232
converters use this chips. I have only seen converter in the 35$ range using
this chip. 


> 
> 
> >  For a pic programmer using only the rs232 tx line, this is the complete
> >circuit:
> >
>
http://web.gramlich.net/projects/extensible_programmer/emdp2/bootstrap/rev_a/index.html
> 
> 
> Now that's interesting. But just the discussions we've been having I think
> that it's way too compilcated for the task it's designed to do. 
You have proposed this. Ready available parts, ... .
> bootstrap
> programmer is one and done. Build it quick, use it simply, verify outside
> of the programmer. Wayne has the same basic idea, dumb bootstrap loader to
> put a bootloader into the part. But once the bootloader is in the part,
> there
> is not need for the programmer anymore. 
Yes, Wayne surely have esagerated to design a pbc for it. On the other side,
the
circuit is simple, the parts are not exotic. The sw Wayne has written for it
works
and is tested. 

I personally prefer to use one control line and the tx line to program the
pic device, but hey, if you want  to use tx only for some reason, do it.
Wayne has designed a on shoot multivibrator that works. Use it for your
design if you prefer. If you design is compatible with his, you can use his
SW too.
Naturally give hime some Credit.
It was a hint to you for you'r intent to make a simple programmer using only
one
line.
However it's clear, that with such a programmer, reading back the data is
too
difficult and it's only for setting up a bootloader.

> 
> So keep it simple. Design a programmer that works with any USB to serial
> cable, doesn't depend on any internal chipset or modem control signals.
> 
> I'm going to post this to the list to keep the discussion going.

> 
> BAJ
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####
> 

-- 
+++ Sparen Sie mit GMX DSL +++ http://www.gmx.net/de/go/dsl
AKTION für Wechsler: DSL-Tarife ab 3,99 EUR/Monat + Startguthaben

Previous by date: 8 Jan 2005 20:21:06 +0000 Re: Problems compiling gpsim, Diego Belotti
Next by date: 8 Jan 2005 20:21:06 +0000 Re: Problems compiling gpsim, Peter L. Peres
Previous in thread: 8 Jan 2005 20:21:06 +0000 Re: DIY USB programmer ?, Byron A Jeff
Next in thread: 8 Jan 2005 20:21:06 +0000 Re: DIY USB programmer ?, Manuel Bessler


Powered by ezmlm-browse 0.20.