1. Home
  2. Electronics Design
  3. Diode bypass cap

Diode bypass cap

We generally add bypass caps to temporarily supply power to an IC for various reasons. The only problem with this is that the bypass caps will also supply power to any other ic's and components around them.

By adding a diode in series with the supply voltage in front of the caps when can force the power to be supplied only to the IC. This has the effect of still providing power to the IC and preventing reverse polarity from damaging the IC(theoretically at least).

But conversely they also do not supply power in a broader region which might be beneficial. The diodes also cost more and waste power.

I have a board consisting of many LED's and several constant current LED drivers(with a few other things that are not important for this discussion). I'm trying to determine if I should add the diodes or not. I am bypassing the drivers with several caps from 0.1uF to 1uF(since I need packages of 402 because of placement issues). What I'm thinking is that the LED supply is not as important as the driver supply since if any type of very quick power dips occur the drivers would be potentially reset while the LED's would just get dimmer.

I'm not sure if it all really justifies all the diodes though as it might simply be a waste? The main thing I like about them is that it allows the bypass caps to only bypass the IC's they are designated for and not end up supplying LED's which is not important(or as critical).

But of course it could be wasting the caps since in some circumstances maybe the LED's need more "bypassing" than the IC's yet the diodes prevent this. I will sprinkle some larger caps across the boards for flavor.

Reply to
Jon Slaughter
Loading thread data ...

And, since their forward voltage is going to vary (at least somewhat) as a function of the current flowing through them, they will tend to reduce the quality of the voltage regulation for the IC.

Well... I wouldn't solve the problem in the way you're thinking, for a number of reasons.

For one thing, your approach tries to buffer the Vcc to the ICs against the voltage dragdown from the LEDs, but it does nothing to protect the other half of the equation: ground. Even with the diodes in place, the current flowing out of the LED cathodes can produce a significant amount of "ground bounce" - the ground voltage reference near them will jump upwards by many millivolts.

This can cause all sorts of signalling problems for any ICs which share that particular ground region.

Rather than use diodes, what I would suggest is this:

- Have two completely separate voltage-regulation domains (i.e. two regulators, fed from the same filtered-DC raw supply). Run one Vcc to the ICs, and another to the LEDs. Depending on the LED drivers you use, you might even be able to run the LEDs from a much more loosely-regulated supply, or even directly from the raw filtered DC with no active regulation.

- Run separate LED-current returns (PC-board traces or wires) from the LED cathodes back to the system ground point. Don't let the LED return current flow through PC board traces which are also carrying the ground current for the low-leve logic ICs. A guy I knew some years ago spec'ed out an LED display board for a local tech museum, and specified separate ground returns. The fellow he contracted the construction out to, ignored his instructions, and tied all the ground wiring into a single bus. The board didn't work reliably - if one tried to turn on too many LEDs simultaneously, the sudden change in current would ground-bounce the control ICs and they'd glitch or reset.

--
Dave Platt                                    AE6EO
Friends of Jade Warrior home page:  http://www.radagast.org/jade-warrior
  I do _not_ wish to receive unsolicited commercial email, and I will
     boycott any company which has the gall to send me such ads!
Reply to
Dave Platt

ll

ct

ght

n).

402

is

er

ust

the

p

ybe

.

It would help to know what you call a driver chip. Is the driver chip the current source feeding the LED, does it gate a current source, or is it the current sink for a matrixing scheme.

"Driver" chips, i.e. silicon that will be sinking or sourcing current, will have hysteresis on the logic inputs. The chip designer knows the ground and supply will be wiggling, so hysteresis is a given. I rather not mess with multipe supplies or this diode scheme unless you can convince yourself you will never force the drivers into latch-up, i.e. forward bias the protection diodes on inputs or the parasitic diodes of the fets on the output.

I think I've mentioned my pet peeve about board designers using pfets to gate chip power when you provide them with a damn shutdown pin. This series diode or dual supply is in the same league. An attempt to solve a non-problem that creates a potential latch-up situation.

Reply to
miso

..want to save some space? Then put the SMT Rs and Cs on their side...

Reply to
Robert Baer

ll

ct

ght

n).

402

is

er

ust

the

p

ybe

.

If you want some isolation you can add a resistor rather than a diode.

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.