[Noisebridge-discuss] [dorkbotsf-blabber] 3.3v serial-parallel shift register with open-collector output?

Anders Nelson anders.k.nelson at gmail.com
Wed Jun 20 20:50:23 UTC 2012


Hi Tymm,

Thanks so much for the detailed input!

Yeah it seems like the 74AHC595 from NXP responds to a VinH of 2.1v at 3v
supply voltage for a ratio of 0.7. If I apply that ratio to 4.2v I get a
VinH of 2.94v which ought to work when I throw 3.3v at it. Provided of
course my supply is stable enough... More caps!

=]
--
Anders Nelson

+1 (517) 775-6129

www.erogear.com


On Wed, Jun 20, 2012 at 1:24 PM, Tymm Twillman <tymmothy at gmail.com> wrote:

> In similar situations I've generally thrown a cheap npn transistor on
> the uC outputs & switched that with the GPIO.  Tristating should be
> relatively ok, but the npn transistor will generally give you more
> current -> faster transitions (dumping the charge on the fet's gate
> more quickly).  Hmm; at 250ns... if things have to actually switch
> that quickly you may have issues getting enough charge through the
> pull-up resistor on the fet's gate to turn it off in time.  if it's
> really important to meet timing, you'll want to calculate the
> worst-case timing based on the capacitance of the fet's gate, the gate
> threshold voltage and the resistance to the high rail... capacitance
> is much higher generally with larger fets
>
> Tristating, depending on the microcontroller, should be plenty fast on
> the control side of things if you are able to set all the pins on the
> port at once and can pre-compute the new values... though there's
> still potentially the above fet switching issue... also with almost
> any uC you'll be limited to somewhere 40ma or less, which may also
> extend the fet switching time.
>
> There are assorted serial interfaced driver chips that one can use,
> using fets or bipolar outputs... tpic6b595 and others... though
> they'll generally be quite a bit more expensive than driving via gpio.
>
> other approach would be to run e.g. a non-open-collector 74hc or
> similar shift register at 4.2v -- worth checking data sheets but in
> most cases your 3.3v signaling should work fine as an input, and the
> full swing on the output can help switch the fets more quickly.
> 74ac's will also give more drive current, which can also help speed
> switching times.
>
> On Wed, Jun 20, 2012 at 12:57 PM, Anders Nelson
> <anders.k.nelson at gmail.com> wrote:
> > Heyoo,
> >
> > I've become rather exasperated in my search for an 8-bit serial-parallel
> > shift register with open collector outputs. See, I'm trying to switch the
> > gates of some P-FETs that act as high-side current switches for the rows
> in
> > a scanned LED matrix. The LED bus voltage is 4.2v and all the ICs run on
> > 3.3v.
> >
> > I found the SN74LS596, but it's an ancient part that is not only enormous
> > but runs at 4.5v-5.5v which won't work in my system.
> >
> > I could use a microcontroller and just toggle the tri-state status of the
> > pins but I worry about how fast the device will respond - It will have to
> > blast out the I/O change within a blanking cycle lasting 250ns. That
> > requires between 4 and 8 MIPS which is not too powerful but this is a
> > low-cost product. =/
> >
> > Any suggestions are most appreciated!
> > --
> > Anders Nelson
> >
> > +1 (517) 775-6129
> >
> > www.erogear.com
> >
> >
> > ----------
> >
> > dorkbot: people doing strange things with electricity
> > http://dorkbot.org/dorkbotsf/
> > SUBSCRIPTION AND UNSUBSCRIBE OPTIONS HERE:
> > http://music.columbia.edu/mailman/listinfo/dorkbotsf-blabber
> >
> > ----------
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.noisebridge.net/pipermail/noisebridge-discuss/attachments/20120620/f091ea04/attachment-0003.html>


More information about the Noisebridge-discuss mailing list