EEVblog #1365 - Viewer PCB & Circuit Design Review

[EEVblog]

Dave answers a viewer technical question about why his circuit is acting up, and goes through a design review outlining the possible problems.
Level translation, clocking, schmitt triggers, line driving, loading, buffering, bypassing, and other aspects are discussed.
If you want to see more design review like this please drop a like and comment.

[Rutger]

I like the pcb / schematic reviews, I might have one myself I would like to submit... (an espresso machine controller driven by a rapberry pi and it uses Schmitt trigger for water level controller).

About the current design, is it not possible to use a Schmitt trigger version of the 8-Bit Shift Register like part # SN74HCS594. I am not sure if the footprint is compatible.

[wilfred]

There's another thread "I am done with JLC PCB" that might be suitable for investigation. It is right in your wheelhouse.

I think picking topics off the forum based on likelihood of interesting and instructive analysis is the best approach. Otherwise you could get inundated with repetitive and uninteresting choices.

[Jfcarrer]

In the video you talk about splitting the ground, funnily enough Robert Feranec uploaded a video today talking about how we shouldn't split grounds in most cases. It would be interesting to have your 2 cents on that video.

[Unixon]

There's a problem with these 595 registers that makes it possible to have data-clock race condition between registers separated by a long line. While the first register after MCU could be fed with fresh data on falling clock edges by software, all following shift registers are going to be updated on the rising clock edges due to internal 595 register schematic and this can cause problems on itself with long connections between modules. One of solutions is to use CD4094 shift register which is capable of shifting out data on falling clock edges, but 4094 has no reset line and cannot be powered up cleanly. Another solution is to keep 595 but complement it with a piece of internal schematic of 4094 that does data shifting out of the register on falling clock edges using an additional D-ff and logic gates - this solution allows to use the best of both worlds.

[thm_w]

In the video you talk about splitting the ground, funnily enough Robert Feranec uploaded a video today talking about how we shouldn't split grounds in most cases. It would be interesting to have your 2 cents on that video.

Feranecs video is more about sensitive analog circuitry. I would guess Dave is considering more in terms of very high current designs he's done (say if they had motors, heaters, etc.).
In this case, IMO, 200mA is nothing for a PCB plane like that, splitting won't have significant effect. If you wanted to isolate the Vcc more, you'd throw in a RC filter or possibly diode going to the low current device. But most likely none of those are needed, as noted in the video already.

[Jope]

Texas Instruments has a 74HC595 with Schmitt trigger inputs, the SN74HCS595.