0
\$\begingroup\$

I have a simple question, and I want to ensure that nothing unexpected happens. I'm currently working on a Raspberry Pi-based robot project with a Pi camera, and I plan to power it with a battery, even if the battery only lasts for around 0.5 hours. For this, I intend to use two 18650 batteries connected in series along with a buck converter (LM2596) to step down the voltage to 5V.

While conducting my research, I came across two potential issues related to the buck converter:

  • Output Ripple: I initially considered using a 10uF capacitor to address this concern. However, I discovered that a 220uF capacitor is already included in the module, as shown in the picture, so I won't be adding any additional capacitor.

  • Electromagnetic Interference (EMI): To address this issue, I plan to enclose the entire buck converter in a box, as depicted in the image. This step is particularly important because the buck converter will be situated very close to the Raspberry Pi, and since I'll be using WiFi for communication, I want to avoid any potential interference with the communication signals.

I would like to confirm if everything seems in order, or if I may have missed something or if there's a better way to handle these concerns. Any insights or recommendations would be greatly appreciated.

Buck converter with a box

\$\endgroup\$
12
  • 1
    \$\begingroup\$ Is your box made out of plastic or metal? Even if metal, it will only have some effect on the radiated emissions from the module it self, it won't do anything to conducted EMI. \$\endgroup\$
    – winny
    Jul 25, 2023 at 8:10
  • \$\begingroup\$ I apologize, but I'm not quite clear about the term 'conducted EMI.' I assume you are referring to the EMI generated by the wires connected to the regulator and going to the Raspberry Pi, in addition to the radiated emissions from the regulator's inductor. Regarding radiated emissions, a metal shield or metal box will indeed have a substantial impact. Even a thin metal sheet can efficiently attenuate the EMI generated by the inductor, significantly reducing its effects. However, dealing with conducted EMI appears to be more challenging, especially when it comes to using shielded wires. \$\endgroup\$
    – M-125
    Jul 25, 2023 at 12:57
  • \$\begingroup\$ Unless you have high-mu metal, you’re only shielding the E-fields. The H-fields will pass right though it. Shielding the cables won’t stop CM and DM noise from propagating and poison what’s connected to them, just stops them from radiating. You want to solve all EMI issues at the source. \$\endgroup\$
    – winny
    Jul 25, 2023 at 13:04
  • \$\begingroup\$ I will use a metallic box made of galvanized steel with a zinc coating, and I plan to incorporate some ventilation holes to facilitate heat exchange. Additionally, I will utilize shielded wire to power up the Raspberry Pi from the regulator. Are these measures sufficient, or do I need to take any other precautions? \$\endgroup\$
    – M-125
    Jul 25, 2023 at 15:52
  • \$\begingroup\$ Depends on what EMI standard you need to comply to and how the rest of the circuit looks. Galvanized steel is low-mu so won’t do anything to shield the leaking magnetic field from your unshielded inductor. The question is if you need it in the first place. I don’t want to sound condescending, but it’s a beginners trap to think about shielding first. Yes, they do have a place and use, but only after you have covered the fundamentals like layout, decoupling and filters. \$\endgroup\$
    – winny
    Jul 25, 2023 at 16:03

5 Answers 5

4
\$\begingroup\$

Your biggest problem is that all these cheap "LM2596" modules use counterfeit chips. To put it bluntly, they don't work. They use the cheapest high-ESR capacitors available, so voltage ripple is extremely high. Usually the inductor saturates because the fake chip switching frequency is lower than a real LM2596.

In addition even the real LM2596 would be a bad fit because it has quite high input-output dropout voltage: with 5V at the output you'd need at least 7-7.5V at the input (including inductor losses) so you won't be able to use the whole capacity of your two series batteries. Also it has high idle power, and low efficiency, so it is not ideal for battery use.

Here's a much better choice of buck converter from a good manufacturer.

It will output 5V with only 5.5V at the input, has high efficiency, low idle power, shielded package... You need to mount it on a PCB with a solid ground plane, and if you worry about EMI, the datasheet provides a schematic for an input filter on page 6. The package is a bit annoying but still hand solderable with a hacked footprint.

For an easier to use package you can check this one, also good efficiency, but startup voltage is a bit higher at 6.5V.

\$\endgroup\$
3
\$\begingroup\$

The regulator's inductor can cause some EMI issues if it is an unshielded type. This EMI generally isn't up the region that would affect 2.4GHz band. Of greater concern is coupling into the camera module, causing noise in the sensor.

Check with the vendor to see if the inductor is shielded. If not, you could possibly replace it with one that is. Also, you can face the inductor-side (top) of the board away from the sensor (The PSU board hopefully has a ground plane that can function as a shield.) I'd be wary of putting it in a box unless you deal with its heat.

To check the effectiveness of your EMI strategy, you can make a poor-man's near-field EMI sniffer by connecting your scope ground lead to a probe tip to make a small loop antenna. This will show stray inductive fields.

The second issue is common-mode noise from the PSU coupling into the camera. You can suppress most of this by using a ferrite on the power supply output.

\$\endgroup\$
7
  • \$\begingroup\$ Considering enclosing the converter in a box, as mentioned in a previous comment, my plan is to cover its internal surface with aluminum foil to create a shielding effect for EMI. Additionally, I will create some holes in the box to facilitate heat dissipation. I believe the temperature inside the box will not become excessively high, based on a comparison video I watched earlier. The buck converter's temperature was only about 10 degrees Celsius higher than the ambient temperature, which seems manageable. \$\endgroup\$
    – M-125
    Jul 25, 2023 at 0:43
  • \$\begingroup\$ Will using a ferrite on the power supply output still be necessary even after implementing the shielding to ensure that the wires themselves do not impact the camera sensors? If so, do you recommend placing the ferrites very close to the regulator module, potentially even inside the box, or do you suggest placing them at the end of the wire connected to the supply of the Raspberry Pi? \$\endgroup\$
    – M-125
    Jul 25, 2023 at 0:45
  • \$\begingroup\$ Rather than use a shield box, you can also interpose a metal plate between the PSU and the rest of the circuit. This will catch any stray fields. Do you need the ferrite? I don't know, but it is a good way to suppress any common-mode noise coming from the supply. It should be placed close to the PSU for best effect. \$\endgroup\$ Jul 25, 2023 at 0:51
  • \$\begingroup\$ If placing the ferrite close to the PSU provides the best filtering effect, why is it commonly placed at the end of the wires of electronic devices? Why is the ferrite bead, for example, placed inside the mobile charger box? \$\endgroup\$
    – M-125
    Jul 25, 2023 at 1:03
  • \$\begingroup\$ I'm not sure what you mean by 'mobile charger box.' At any rate, the ferrite placement It depends on the problem the ferrite is intended to solve. If the PSU is the problem, it's placed closer to it on the cable; if it's the powered device, it's placed at the device end. In your case your concern is PSU ground noise making its way into the r-Pi and to the camera, so it's best to suppress it as close to the PSU as possible. \$\endgroup\$ Jul 25, 2023 at 1:19
2
\$\begingroup\$

You need a few things. You need a BMS or undervoltage circuit so the regulator shuts off before the batteries reach an undervoltage condition which can kill the batteries, typically this occurs at 3V on most 18650's.

There is EMI, the regulator will put out RF at the switching frequency and harmonics. A shield should block most of it. Usually the LM2596 operates at 150kHz so its unlikely to interfere with devices that operate at higher frequencies

\$\endgroup\$
1
  • \$\begingroup\$ Considering the first case of using an undervoltage circuit, thank you for bringing it to my attention; I completely forgot about it. Here is a simple circuit that I simulated. This circuit automatically powers off the source when its voltage drops below 6V. Since 6V corresponds to approximately 3.0V for each battery, it serves as the cutoff voltage for the 18650 batteries. \$\endgroup\$
    – M-125
    Jul 25, 2023 at 0:03
2
\$\begingroup\$

First of all, all answer I give are on an as-is basis.

Output ripple, SMPS have them, you can't do much about that. Adding capacitance can be just as much of an issue. Ripple is normal and you need to live with it. Otherwise, add an LDO to it's output.

EMI: Plastic box don't do anything to EMI. Shielding is a last resources since it is expensive and complex. Your SMPS, by it's nature, is noisy. The closer the specs of your SMPS to your design, the less noisy it will be, but it will always emit. My question would be first, did you see any interference between both? If not, don't over think it!

\$\endgroup\$
5
  • \$\begingroup\$ Regarding the output ripple, I believe it's best not to increase the output capacitance to avoid potential stability issues with the regulator. As for considering an LDO (Low Dropout Regulator) as an alternative, I wish I could go that route, but currently, I am unable to find an LDO with a dropout voltage less than 1.1V in the market. Regarding the plastic box, my plan is to cover its internal surface with aluminum foil. This will act as a shielding measure to significantly reduce any potential Electromagnetic Interference (EMI). \$\endgroup\$
    – M-125
    Jul 25, 2023 at 0:16
  • 1
    \$\begingroup\$ A 220u capacitor will have a relatively high ESR (effective series resistance) and so won’t have much effect on smoothing high frequency conducted emissions. It would be wise to add one or more smaller capacitors. I’d go for 100nF, 10nF and 1nF to provide attenuation over a wide frequency range. \$\endgroup\$
    – Frog
    Jul 25, 2023 at 9:06
  • \$\begingroup\$ The 220uF capacitor is already present by default in the module. So, are you suggesting that it's preferable to desolder the existing capacitor and replace it with a smaller one? \$\endgroup\$
    – M-125
    Jul 26, 2023 at 2:15
  • \$\begingroup\$ I already have a 100uF 10V Tantalum capacitor of type B. Is this suitable? \$\endgroup\$
    – M-125
    Jul 26, 2023 at 2:24
  • \$\begingroup\$ Honestly, I'll stick to my point, did you saw any noise issue? If yes, add an LDO, if not leave it as is. \$\endgroup\$
    – Julien
    Jul 26, 2023 at 13:45
0
\$\begingroup\$

Another potential source of disturbance is input current distortion caused by the converter.

The current waveform can cause radiation in the converter power cable.

Make sure your converter has an effective filter that smoothes the input current, or observe the current to be sure.

\$\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.