I have successfully used a 74LVC1G34DBV to do exactly that, 5V supply, 10u & 
100n on the supply, 499R in series on the output, driven from 3v3 LVTTL.

Run at 5V the WS2812B has Vil/Vih of 0.9 and 3.5 V.  Hence you need more than 
LVTTL or TTL logic levels.  I would be surprised if a pull up was a good 
solution, at 800 kHz you wish transition times on the order of 75 ns : PU 
dissipation ?

The essential spec for a WS2812 buffer / driver device is TTL input levels, 5V 
CMOS output levels : the usu suspects are 74LVC and 74HCT

Martin

-----Original Message-----
From: Adrian Godwin [mailto:[email protected]] 
Sent: 30 July 2025 22:29
To: Martin Bishop <[email protected]>
Cc: General Discussion: On-Topic and Off-Topic Posts <[email protected]>; 
Paul Koning <[email protected]>
Subject: Re: [cctalk] Re: RP2350 5V Input Tollerant Pins

I was thinking specifically of ws2812 addressable leds which have CMOS levels. 
The data stream is fairly slow (~800kHz) and can usually be generated from 3v3 
with the common FET voltage shifter (which also has passive pullup). It's 
sometimes driven with 3v3 but is marginal. I was thinking that it might be 
driven bye a 2350 with a 5V pullup more reliably.



On Wed, Jul 30, 2025 at 11:19 PM Martin Bishop 
<[email protected]> wrote:
>
> Adrian
>
> The summary answer to your questions must be "in general no" : 5V CMOS levels 
> are quite different from TTL levels and open drain /collector signalling is 
> inherently slow.
>
> 5V CMOS, with its very high Vih (typ > 3.5V), is one of those things it is 
> nice not to have to work with.  If you do have to drive it then, in general, 
> logic (level) translation families are necessary.
>
> Contrariwise, 3v3 LVTTL and 5V TTL thresholds are essentially equal Vil at 
> 0v8 and Vih at 2V.  However, 5V pull ups and transients make 5V --> 3v3 
> connections an invitation to smoking : YMMV.
>
> If you have a specific interfacing problem, the essential information to 
> select a way ahead is source chip, sink chip, signal speed, and constraints.
>
> The can being kicked in this thread was the RP2050's 5V tollerant input, 
> which should make it useable for direct connection to TTL, then some apparent 
> "documentation" issues were identified ...
>
> Some "handy primers" are:
> https://www.ti.com/lit/pdf/sdyu001?keyMatch=logic%20families&tisearch=
> universal_search see page 4 for typ logic levels 
> https://www.ti.com/lit/pdf/ssztc76?keyMatch=logic%20families&tisearch=
> universal_search 
> https://www.ti.com/lit/pdf/scyt129?keyMatch=little%20logic&tisearch=un
> iversal_search 
> https://www.ti.com/lit/pdf/scyb018?keyMatch=voltage%20level%20translat
> ion%20guide&tisearch=universal_search
>
> Martin
>
> -----Original Message-----
> From: Adrian Godwin [mailto:[email protected]]
> Sent: 30 July 2025 21:33
> To: General Discussion: On-Topic and Off-Topic Posts 
> <[email protected]>
> Cc: Martin Bishop <[email protected]>; Paul Koning 
> <[email protected]>
> Subject: Re: [cctalk] Re: RP2350 5V Input Tollerant Pins
>
> What about output pins in open-drain mode with a 5V pullup ? Can they provide 
> a 5V CMOS-compatible output ?
>
> On Wed, Jul 30, 2025 at 9:40 PM Paul Koning via cctalk 
> <[email protected]> wrote:
> >
> >
> >
> > > On Jul 30, 2025, at 3:07 PM, Martin Bishop 
> > > <[email protected]> wrote:
> > >
> > > https://www.eenewseurope.com/en/raspberry-pi-spins-its-rp2350-adds-5v-support/
> > >  was my point of departure, doubtless a recycled press release.
> >
> > Yikes.  Assuming that is accurate, Raspberry Pi screwed up very severely by 
> > claiming 5V support in the data sheet without saying it only applies to the 
> > A4 rev.  Any number of people will fry their chips because of this blunder.
> >
> >         paul

Reply via email to