From: Andreas Zwirtes <zwirtes_at_gmx.de>
Date: Mon, 14 Oct 2002 19:30:51 +0200

Oh, don't be sorry to answer, this is a discussion. There's nothing
really implemented by now.
If it sometimes sounds like I'm totally fixed on one standpoint, that
might be a language issue. I'm not very good at social talk in english,
since I mostly read and write technical english. Just beleive me, I'm
not fixed on one standpoint. I just have a little screw loose, that
keeps me arguing untill I'm convinced. If this is going anyone on his
nerves, just tell me. I do NOT want to discredit anybody's ideas, indeed
I'm very happy that so many people almost have very good ideas. If I
embarassed somebody, please tell when when and how, and I'll avoid that
in the future. This thing here is very much fun for me and it should be
for everone else. Also maybe I didn't get the message, did I?

Ok, now to the technical issues:

Here are some new infos, that calpefrosch and me figured out.
It seems that some recorders have a 1kOhm resistor pushing to Vcc of the
processor (which is 3.3V). I don't have that resistor. So we have two
different types of interfaces to deal with. Also i don't know, if mine
isn't defective. I tested communication and it works, so i think this
"failure" is intended. We now have ports that are floating and ports
that are strongly pushed to Vcc. Our hardware should support both.

It would be very helpfull, if everybody who can do this measures the
voltage on the remote pin without anything connected (no remote or
headphone, since they draw current or short circuit). If you don't know
how, please don't do it, because you might have to disassemble your
player/recorder. Everybody who know how and has some time to do it,
please do and mail me (or the list) the details of the device and the
result in volt (measured to ground). That would help very much, thanks!


Stuart Tedford wrote:

>Sorry, I have to answer this...
>
>
>Don't you mean "we _should_ limit the interface", surely having the
>resistors in the remote is limiting the remotes interface - rather than
>having a standard UART interface?
>
Now I understand. You want to connect the remote to other devices,
right? Ok, didn't get that. We can do both. The resistors on the PCB can
be a setup option, so you are free to choose. Either you put the
resistors in the cable and connect three or four pins (depending on the
design), or you solder then to the PCB and use two pins. My view has
been from the Archos side and my subject was to use the cable for as
many different circuits as possible. And to make it as easy as possible,
since I think PCBs are easier to populate than soldering resistors to
cables. At least there is a SMD-processor, that needs to be soldered.
That's much more complicated. Anyway, if someone wan't his SMD-parts
soldered, just ask me. I can do that for you.

>I agree that short circuit protection would be great, but I don't see how it
>needs to have the resistors on the remote rather than the cable - but I'm
>not a hardware guy.
>
The charger circuit and the short circuit protection are now somewhat
obsolete. Physically they work, but there is a short "short circuit" to
ground every time you plug or unplug the remote. Also, if you connect a
standard headphone, the communication pin is always shortened to ground,
as long as the headphone is plugged in. Therefore it's too dangerous to
set the processor pin to high. If someone unplugs his remote in that
moment, the processor I/O-port will very likely blow up. Maybe there are
some people out there, who brave the danger and nevertheless want to do
it. So we should keep the possibility of energy saving in mind, but
don't enable it by default.

>I guess for something this small and limited, there is no point in providing
>a standard UART interface becasue it probably wouldn't be very useful for
>controlling a PC or something, so yeah, I agree with you here - size it the
>most important thing in this case.
>
But where do you want connect the two-resistor interface? Even a
PC-interface uses three transistors for bidirectional mode (maybe it can
be done with two) and lots of other parts. I may be wrong, but every
interface I can think of uses different circuits and therefore an
individual interface. One thing that is possible, would be a universal
interface with different options (as said before) where anybody can
choose from by populating the PCB with different parts or sawing areas off.

>RJ-11 (telephone plug) is quite small - and certainly small enough for a
>remote sporting an LCD display.
>
Using RJ11 is a yery good idea. I searched for connectors in some
catalogues. The only alternative that I found are mini-DIN plugs (like
SVHS), but they are almost bigger that RJ11. Are there RJ11 plugs with
housing? I also would like to have a miniature remote without LCD, but I
think I'll solder the cables directly to it. All other types should use
plugs, I suppose.

>I also agree that the remote must initiate/maintain the whole protocol. It
>is too dangerous to have the Archos toggleing the port pin if it isn't
>guaranteed what is connected at the other end - It is more comms overhead,
>but I'd rather not blow up my Archos.
>
>I guess my main point is, why bother designing something for just one fixed
>purpose, when it would be relatively easy to allow it to be used for others?
>The adaptor cable was just one idea - if there is a better way to make the
>remote more flexible for non-hardware people then I'm all ears.
>
I hope we are on the way to the same track :-)
What do you think about the above?

>Oh, and what I meant by a new protocol being trivial... I should have said
>_relatively_ trivial. Becasue it would only be a software change in the
>remote - rather than a hardware and software change if you wanted to connect
>to a new device. And depending on the device - the required protocol might
>be quite simple.
>
Yes, the protocol of the original remote for example is really simple.
Just wondering, why they needed so many parts. I'm experimenting with a
bidirectional PC-interface at the moment. If anybody is interested in
details (hardware & software configuration), feel free to ask. Well, no,
I don't have transmission routines running on the archos yet. Any help
would be appreciated! The main issue is to toggle the port 9600 times a
second between "push to ground" and "open / floating" in an RS232
manner. I don't know, if there's an obsolete timer with a high prority
interrupt. That would be the easiest way to do it.


Andreas
Received on 2002-10-14