1
\$\begingroup\$

I'm not sure if this is a stupid question, but I'd still prefer to ask than regret it later. I'm designing a PCB with the CJ7805 which, like the LM7805, can supply a steady 5 volts. In this case, it's to power a Teensy 3.6. There's a recommendation to place a 100nF capacitor between the Vin pin of the Teensy and ground. I already have capacitors (100nF and 330nF) situated around the CJ7805 as specified by the datasheet. I was thinking of placing a capacitor for the Teensy, but I was wondering if that would affect the overall performance of the CJ7805 since it would change the net capacitance? Or maybe capacitors work differently when placed far apart on a PCB? Would I even need to place additional capacitors on the PCB when it's already being monitored by one capacitor? I've seen boards use soo many capacitors before, and I'm getting a little confused. I'm new to PCB design so I'm still trying to find and understand the proper ways of keeping noise to a minimum. Any help would be appreciated. Thank you!

Are these capacitors okay to keep?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ The teensy's microcontroller never sees these 5V – they're just used to feed a linear voltage controller on board (the LP38691). Hence, unless you need 5V to power a USB device attached to the teensy, that voltage really doesn't have to be overly exact at 5V – my guess is that 3.8 V to 10 V will probably work fine. \$\endgroup\$ Commented Sep 28, 2019 at 9:51

4 Answers 4

1
\$\begingroup\$

The regulator needs a bypass capacitor to be stable. The Teensy needs one too. The farther a capacitor is, the less effective it is. If there is 1cm between regulator and Teensy, you may only need one. If they are 1 meter apart, you definitely need two. Since they are on same PCB just draw both and you can choose later to leave one out.

\$\endgroup\$
3
  • \$\begingroup\$ I see! Thanks for the advice! I'm sorry if I sound picky, but I'm wondering if there's a particular formula that can tell when to use two or more? Or is this a skill you just develop with time, kinda like an engineer's 6th sense? Thanks again for the tip;) \$\endgroup\$
    – Espresso
    Commented Sep 28, 2019 at 9:53
  • \$\begingroup\$ there's no particular formula, since you decided to use a LM7805 knockoff that comes with no proper datasheet that would give one. However, as mentioned in my comment under your question, it's likely you don't even need the 7805 at all – the teensy has all the regulation you need. \$\endgroup\$ Commented Sep 28, 2019 at 10:04
  • \$\begingroup\$ With capacitors, typically more is more. And by doing (and reading the datasheets) you learn what's enough. \$\endgroup\$
    – TemeV
    Commented Sep 29, 2019 at 20:25
0
\$\begingroup\$

The Datasheet is not comprehensive, but it's safe to guess the chip is a 7805 - alike linear regulator. The 330nF and 100nF (non-electrolytic) capacitors are mandatory to guarantee that the regulator is stable. They should be Low ESR and must be placed as close to the regulator as possible. But at the same time, there is no harm in adding extra capacitors. Adding an electrolytic (10 uF would be fine) capacitor as a bulk capacitor could help to decrease ripple on the input and output.

\$\endgroup\$
0
\$\begingroup\$

Teensy schematics show it already has a capacitor on the 5V input, so you don't need to add one on your board on the Teensy pins.

You should simply use the caps specified by your voltage regulator datasheet.

If your board has more 5V chips, they may also require their own decoupling.

\$\endgroup\$
0
\$\begingroup\$

When thousands of CMOS switches simultaneously switch x pF of charge between those switches and either Vdd or Vss inside any CMOS chip there is a ns current spike on the rails. If that spike is supplied from a remote regulator, then using 0.5nH/mm for each track, you must consider the consequences of emissions and induced voltage spike from this inductance V=LdI/dt.

Thus if you know dI/dt on supply IC noise ( which you can do with a 50mV shunt R) and you have computed the supply trace or plane inductance, then you can correlate the improvement if any by putting a low ESR ceramic cap across the IC device such that dV/dt =Ic/C +Ic*ESR/dt(?) Now you must estimate the ESR of all those 25~50 Ohm switches inside in parallel that are switching synchronously or use your lab result dI from a 50mV current Rsense to see if the new RC low pass filter will improve supply ripple and compare with Voltage source ESR of both caps.

I realize this is complex but not impossible. Let's say it's for insurance and won't hurt or take the time to measure it someday for a simple IC , complex IC on a PCB , and make your own Rule of thumb with good 200MHz BW measurements.

Conclusion

Always satisfy both the load and source capacitance criteria for optimal EMI emissions and susceptibility for conducted and radiated noise.

  • Use both caps unless the gap between CMOS and regulator > the size of the IC away.
  • BUT if your IC is actually a PCB (Teensy) with caps already , then this is redundant.

It may work without both, but logic works great until it fails ;) -- and this is a SNR supply fix to a probability sensitive result when it comes to noise induced spikes exceeding spec tolerances ...

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.