0
\$\begingroup\$

In the schematic below there is a sensitive ADC and sensor with 4 MHz SPI lines from the ADC to a 3.3V MCU, along with a 1 MHz boost converter powering an LCD. Could switching noise generated from the boost converter travel backwards through its power supply to affect the analog and digital LDOs? And if so would placement of ferrite beads as shown be a good approach to isolate the LDOs from the noise? And could the analog LDO be affected in the same way from noise in the digital LDO if the ferrites were not in place?

The PCB is 2 layer and has an analog and digital side with a ground plane, I haven't tested it or made it yet, I'm just looking for help on design. Thanks.

ADC LDOs to be used: http://ww1.microchip.com/downloads/en/DeviceDoc/mic5305.pdf

Boost converter: https://www.torexsemi.com/file/xc9141/XC9141-XC9142.pdf

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
  • \$\begingroup\$ The noise is probably not going through the regulators. You most likely have grounding problems with return currents, what does the PCB look like? \$\endgroup\$ – Voltage Spike Aug 19 at 16:26
  • \$\begingroup\$ The PCB is 2 layer and has an analog and digital side with a ground plane, I haven't tested it or made it yet, I'm just looking for help on design. \$\endgroup\$ – wdbwbd1 Aug 19 at 16:30
  • \$\begingroup\$ What are the model numbers of the LDO's you plan on using? \$\endgroup\$ – Voltage Spike Aug 19 at 16:32
  • \$\begingroup\$ The LDOs will be: ww1.microchip.com/downloads/en/DeviceDoc/mic5305.pdf \$\endgroup\$ – wdbwbd1 Aug 19 at 16:35
  • \$\begingroup\$ And boost converter will be torexsemi.com/file/xc9141/XC9141-XC9142.pdf \$\endgroup\$ – wdbwbd1 Aug 19 at 16:36
1
\$\begingroup\$

Could switching noise generated from the boost converter travel backwards through its power supply to affect the analog and digital LDOs?

Yes. However, LDO's are built to filter this out. If you look at the Power Supply Rejection Ratio (PSRR) in the datasheet, it can tell you how much blocking in dB the regulator will provide. If the ripple can be estimated from the switching supply (and on the input of the LDO), one can calculate the amount of ripple that will be on the output of the LDO. (one way to do this is to convert the voltage ripple to dB then subtract the PSRR and convert back to volts)

enter image description here
Source: http://ww1.microchip.com/downloads/en/DeviceDoc/mic5305.pdf

And if so would placement of ferrite beads as shown be a good approach to isolate the LDOs from the noise?

In my designs, I've usually put an LC filter on the switcher to reduce all noise to all LDO's. However, you'll want a ferrite with a lower filter pole, the switcher operates in the 1-4MHz range and the ferrite have 100Ω at 100Mhz It may be better to get ferrites that block at lower frequencies.

And could the analog LDO be affected in the same way from noise in the digital LDO if the ferrites were not in place?

If the digital LDO creates ripple, then yes. LDO's usually don't send noise back to the source, because they have feedback and regulate by burning up the excess voltage as heat, so they are effective at regulation. LDO's don't usually send noise upstream unless there are very large current changes, like switching the LDO's max current on and off.

\$\endgroup\$
  • \$\begingroup\$ Would I have to worry about harmonics as well from the 1 MHz switcher, and how high would they go in frequency? Also, if there is any resonance from the ferrite from any frequencies, would that resonance have any effect on the pure DC that I am trying to deliver to the LDOs? Or will resonance only affect the higher frequencies? I don't have the resonance concept down that well, but I just don't want any effects to the pure flat DC that will exit the bead. Is the LC filter you use placed right before or after the switching regulator to prevent the noise from traveling back through the circuit? \$\endgroup\$ – wdbwbd1 Aug 19 at 17:09
  • \$\begingroup\$ And with the LC filter, did you mean to use an inductor or ferrite with it? \$\endgroup\$ – wdbwbd1 Aug 19 at 18:52
  • \$\begingroup\$ With the LC filter I use regular inductors. A ferrite inductor has more of a complex impedance (due to it's impedance varying over frequency), so if I were to use one I'd probably do a spice analysis on the filter to see what the outcome would be. \$\endgroup\$ – Voltage Spike Aug 19 at 19:05
  • \$\begingroup\$ Low pass filters increase blocking with frequency, so if there are harmonics they will be attenuated more than the switching frequency. There is resonance with LC filters right around the filter pole. The blocking for ferrites can be seen in the frequency graph. \$\endgroup\$ – Voltage Spike Aug 19 at 21:23
2
\$\begingroup\$

Could switching noise generated from the boost converter travel backwards through its power supply to affect the analog and digital LDOs?

The current draw of the boost converter will have a periodic component. If the source impedance of the power supply (including all parasitics) is not zero (it isn't), then this will cause a periodic variation of the supply voltage.

would placement of ferrite beads as shown be a good approach to isolate the LDOs from the noise?

It might be more effective to place a bead at the input to the switching converter. Of course you'd also want to have sufficient decoupling capacitance at the input of the switching converter for it to obtain the current it needs without having to draw it through the ferrite.

Note, however, that many ferrites don't have particularly high impedance at 1 MHz (yours seem to be spec'ed at 100 MHz, for example). You might rather use an ordinary inductor rather than a ferrite bead, to ensure you are blocking the lower harmonics of the switching frequency.

And could the analog LDO be affected in the same way from noise in the digital LDO if the ferrites were not in place?

The digital LDO won't be producing much noise on its own.

It's load, though, will be drawing variable currents as the digital logic switches. This will also produce noise on the overall power supply net, for the same reason the switching converter's periodic current draw does.

Whether this noise is a problem in your system depends how sensitive your analog circuits are, how much decoupling capacitance you provide for the various components, the PSRR of your LDOs, etc.

\$\endgroup\$
  • \$\begingroup\$ Yes the LDOs and boost converter will all be well decoupled, and regarding the suggestion about using an inductor instead of the ferrite, I am just looking for a low pass filter for perfectly flat DC to the LDOs so that seems to make a lot of sense as I was concerned the noise filtering ceiling might be too high with the ferrite. I was thinking that the harmonics would be attenuated but worried the lower frequencies may be skipped. Would just a 1206 chip inductor work for example? And of how many uH do you think approx? \$\endgroup\$ – wdbwbd1 Aug 19 at 16:50
  • \$\begingroup\$ @wdbwbd1, I'd use LTSpice or a similar simulation tool to answer those questions. \$\endgroup\$ – The Photon Aug 19 at 16:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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