[AHRMS] hardware progress (was: introduction)

[Robert F. Scheer]

On Thu, 2008-03-13 at 00:11 -0700, Zac Wheeler wrote:
> Robert F. Scheer wrote:
> > The only reason to use Eagle is it is free and has a large user-base.
> > 
> > You mentioned an open-source layout tool.  If you'd rather use that, I
> > certainly don't object, especially if it cuts down your time invested.
> > 
> > If it's a commercial tool you prefer, what is it?
> 
> PADS + Orcad would blow the budget. I'll keep rolling with Eagle. It is
> popular, in the same way and for much the same reasons as x86. Just
> venting a little :)
> 
> My main gripe with Eagle is that it is more labor intensive than it
> needs to be to create parts, especially if they are only for schematic
> use (and maybe in the end not used for anything). There are numerous
> other smaller nags but I'll get over it.

Been there.  Hate that.  Have you seen all of the Sparkfun Eagle files
btw?  There's a ton and more.

> 
> Just wanted to post that I am still cranking on it although I missed the
> deadline I set for myself. I was reading the USB spec and noticed that
> we are only guaranteed +/-5% on the 5V supply. And that's before the
> cable. Ended up adding an inductorless (very little EMI to bother our
> sensors) DC/DC boost + linear regulator to get us a decent +5V from the
> USB 5V which we need for the Z gyro and ADC.
> 

Inductorless must mean capacitor pumping.  Interesting solution.

I have to dump some on USB here myself.  I'm at the last stage of
integrating Tarzan on the bench (ok for the 3rd big time) and am using
USB to serial converters for all the i/o.  I have these cute little ftdi
units you can get at Sparkfun (sometimes) and they are remarkably good
fits into the usb port. (No - I don't own any part of Sparkfun!)

http://www.sparkfun.com/commerce/product_info.php?products_id=8551

Anyway, USB might be good for the desktop and it may be a good
electrical and logical interface but it s**ks as a mechanical interface
on a robot.  The cabling is too bulky and too heavy.  The standard plugs
are too big and heavy.  The mini and micro plugs are also too big and
heavy.  But the serious problem is the contact reliability.  It's not
good enough imo for a bouncing robot, especially with the weight of the
connectors and cables tugging on the connectors.  

And now we see USB isn't good enough as a power source for the AHRMS.

I also found that it wasn't good enough to power my Maxbotix sonar array
(the momentary short from the 1800uF cap on the sonar power rails would
reset the serial port).  Sure there are ways around that but it was
easier just to slap in a Dimension Engineering 5V switcher.  

It's important not to throw out the only good solution but I'm wondering
whether any of the serial interfaces might be better to standardize on
for robots in general.  Are you all familiar with the Bioloid interface?
I think it is based on 12V and supports 1Mbps and many party-line units
sharing the line.  I'll look into it more soon.

> The downsides are more power used and ripple issues to address. Upside
> is it provides a guaranteed 5V instead of hoping for 4.8V (this is
> basically a requirement) and allows for an easy dual-power input
> arrangement. We can either power direct off USB or take an input voltage
> up to 37V (only theoretically; things will get hot and shut down or
> cause thermocouples and thermal noise in reality). We should be able to
> do a standard +12V without an issue though.
> 
> Mark, have you figured out how to request 500mA from the USB host yet?
> (Mostly) Kidding ;)

The ftdi modules pass 500mA I think.  Didn't test it carefully.

- Robert

> 
> Zac
>