Hello everyone,
I’m working on a project where I need to step down 24V to a stable 5V with a load current of up to 1A. I’ve decided to use a buck converter based on the LM2675 IC.
Here is my schematic made in DipTrace:
Input: 24V (DC)
Output: 5V / 1A
Controller: LM2675
Switching Frequency: 260 kHz
I’m planning to use two capacitors:
Input Capacitor (C1): To smooth the input voltage and suppress noise.
Output Capacitor (C2): To smooth the output voltage and minimize ripple.
I have the following questions and concerns:
Type and value of the input capacitor: I’m considering using a 100 µF electrolytic capacitor with low ESR, but I’ve heard that ceramic or polymer capacitors might be better for high-frequency applications. Which type would be optimal for input noise suppression?
Type and value of the output capacitor: Given the need to minimize output ripple, should I use multiple low ESR ceramic capacitors or a single large electrolytic capacitor? Or perhaps a combination of both?
Impact of switching frequency on capacitor selection: How does the 260 kHz switching frequency affect capacitor choice? Should this be a factor when determining capacitance?
Temperature stability: The system will operate at elevated temperatures (up to 70°C). Which type of capacitor is best suited for these conditions.
I’d appreciate any advice or suggestions! I’m especially interested in hearing from anyone who has experience with similar circuits.
By the way, how do you like DipTrace 5? Has anyone tried working on it yet?
I’m working on a project where I need to step down 24V to a stable 5V with a load current of up to 1A. I’ve decided to use a buck converter based on the LM2675 IC.
Here is my schematic made in DipTrace:
Input: 24V (DC)
Output: 5V / 1A
Controller: LM2675
Switching Frequency: 260 kHz
I’m planning to use two capacitors:
Input Capacitor (C1): To smooth the input voltage and suppress noise.
Output Capacitor (C2): To smooth the output voltage and minimize ripple.
I have the following questions and concerns:
Type and value of the input capacitor: I’m considering using a 100 µF electrolytic capacitor with low ESR, but I’ve heard that ceramic or polymer capacitors might be better for high-frequency applications. Which type would be optimal for input noise suppression?
Type and value of the output capacitor: Given the need to minimize output ripple, should I use multiple low ESR ceramic capacitors or a single large electrolytic capacitor? Or perhaps a combination of both?
Impact of switching frequency on capacitor selection: How does the 260 kHz switching frequency affect capacitor choice? Should this be a factor when determining capacitance?
Temperature stability: The system will operate at elevated temperatures (up to 70°C). Which type of capacitor is best suited for these conditions.
I’d appreciate any advice or suggestions! I’m especially interested in hearing from anyone who has experience with similar circuits.
By the way, how do you like DipTrace 5? Has anyone tried working on it yet?
Read the LM2675 datasheet carefully about choice of inductor, capacitors and layout - don't stray from it if you want reliable operation... Switch mode power conversion is picky about such things.
+1 supplementary ceramic caps, probably X7R or X5R. Some ICs allow you to use exclusively ceramics, but only if the datasheet allows it. Extra resistors for damping may be required.
Polymer electrolytics might an option, but read up thoroughly on their ESR, leakage, and other specs.
Another area to check is the inductor: they often have derating curves showing how the inductance drops off vs current. I usually just pick a nice 'pot' core so it has good shielding.
Polymer electrolytics might an option, but read up thoroughly on their ESR, leakage, and other specs.
Another area to check is the inductor: they often have derating curves showing how the inductance drops off vs current. I usually just pick a nice 'pot' core so it has good shielding.
+2 for adherence to data sheet.
Biggest surprise was awful drop in actual capacitance with rising voltage, especially with ceramics. Huge variation among manufacturers. Test actual C at operating voltage. Data from manufacturers can be difficult to get. If this is a production design, qualify a couple of satisfactory parts and don’t allow substitutes. X7R seemed best and I used an X5R with some reluctance. No experience with polymer.
Biggest surprise was awful drop in actual capacitance with rising voltage, especially with ceramics. Huge variation among manufacturers. Test actual C at operating voltage. Data from manufacturers can be difficult to get. If this is a production design, qualify a couple of satisfactory parts and don’t allow substitutes. X7R seemed best and I used an X5R with some reluctance. No experience with polymer.
You can save yourself a lot of trouble by using one of TI's power modules. For example the LMZ14201. It just requires a few resistors and capacitors to work. You still have to pay attention to the layout, but at least the majority of the work is done for you.
If you have your heart set on the LM2675, I suggest you have a look at WebBench. It's in the Design & Development section of the LM2675 product page. It can be a bit annoying at times, but it is a great tool once you get it running and it's helpful in verifying your design.
Tom
If you have your heart set on the LM2675, I suggest you have a look at WebBench. It's in the Design & Development section of the LM2675 product page. It can be a bit annoying at times, but it is a great tool once you get it running and it's helpful in verifying your design.
Tom
-10 for paralleling small caps with electrolytic capacitors. it is an urban legend (now) born 60 years ago, when E-caps were seriously imperfect.
Nowadays, E-caps are excellent, and those which aren't ultra-low esr add some damping into the circuit, which is often beneficial.
This thread explains most of that:
https://www.diyaudio.com/community/...ps-with-electrolytic-caps.106648/post-2257381
Nowadays, E-caps are excellent, and those which aren't ultra-low esr add some damping into the circuit, which is often beneficial.
This thread explains most of that:
https://www.diyaudio.com/community/...ps-with-electrolytic-caps.106648/post-2257381
A clarification: although a straight paralleling between a ceramic and an E-cap is almost always detrimental, there are cases where "some" paralleling is useful, and even necessary: for example, take a board comprising a number of fast chips, opamps, logic, ADC, DACs, etc.
The board will have a relatively large bypass cap, close to the supply input, but each chip requires a close, effective local bypass to be stable and achieve its stated performance. This means that the global bypass cap (100µF for example) will be supplemented with a number of smaller caps (100nF to 1µF) strategically located.
This is normally not a problem, because the length of tracks is sufficient to damp possible resonances.
If the tracks are too short to ensure an effective damping, a small resistor (0R47, 1 ohm) can be added. A dissipative ferrite bead is also an option, and will contribute to the decoupling
The board will have a relatively large bypass cap, close to the supply input, but each chip requires a close, effective local bypass to be stable and achieve its stated performance. This means that the global bypass cap (100µF for example) will be supplemented with a number of smaller caps (100nF to 1µF) strategically located.
This is normally not a problem, because the length of tracks is sufficient to damp possible resonances.
If the tracks are too short to ensure an effective damping, a small resistor (0R47, 1 ohm) can be added. A dissipative ferrite bead is also an option, and will contribute to the decoupling
Putting 10-100 nF in parallel with the output cap in as SMPS is not a horrible idea to suppress > 1 MHz hash. Going crazy with parallel combinations will hurt more than it helps, though.
Any SMPS that I'm aware of is sensitive to the ESR of the output cap, so you really do want to pay attention there. Don't use a plain electrolytic if an OSCON or ceramic is required by the design.
If the design requires a ceramic capacitor make sure that you take the voltage coefficient into account. High-K dielectrics have a pretty horrid voltage coefficient that often varies from manufacturer to manufacturer and even between component sizes from the same series from the same manufacturer.
One nice thing about WebBench is that it takes all this into account for you. If you wanted to you could just type in your requirements, hit "Create Design", export the BOM, and build it. I usually end up tweaking the designs a bit. I have yet to have a design that WebBench said should work that didn't.
Tom
Any SMPS that I'm aware of is sensitive to the ESR of the output cap, so you really do want to pay attention there. Don't use a plain electrolytic if an OSCON or ceramic is required by the design.
If the design requires a ceramic capacitor make sure that you take the voltage coefficient into account. High-K dielectrics have a pretty horrid voltage coefficient that often varies from manufacturer to manufacturer and even between component sizes from the same series from the same manufacturer.
One nice thing about WebBench is that it takes all this into account for you. If you wanted to you could just type in your requirements, hit "Create Design", export the BOM, and build it. I usually end up tweaking the designs a bit. I have yet to have a design that WebBench said should work that didn't.
Tom
10nF @1MHz is equivalent to 16 ohm, negligible compared to even the worst contemporary E-cap. 100nF is slightly less insignificant, but not much. At higher frequencies, the effect is greater but resonances also come into play, as shown in the measurements.
If you really want to make a difference with a ceramic cap, it needs to be quite large, both to reduce the output impedance and minimize resonance effects. A large initial capacitance will also beat the voltage coefficient and the effect of aging
If you really want to make a difference with a ceramic cap, it needs to be quite large, both to reduce the output impedance and minimize resonance effects. A large initial capacitance will also beat the voltage coefficient and the effect of aging
I tested a 220µ/10V oscon (see the link above), but the series resonance was well below 300kHz, the lower limit of my test at that time. Now, I do not have access to a VNA anymore (I am retired), but I can still test it using traditional methods if it presents an interest
Sometimes it is a mandatory to check the Q-factor of your LC filters.
On the bench we experienced fatal fails of buck converters
when plugging in the cable into the lab supply.
In that case there were several 100nH of lab cable inductance that together
with the ceramic bulk input capacitor on the pcb built a series resonant tank of high Q.
This created lethal voltage overshoots on plugging in the cable
and could be adequately simulated with LTSpice.
Parallelling an electrolytic cap with medium ESR shifted down
the resonant frequency and resonant peak disappeared due to the collapsed Q-factor.
This is an example of beneficial hi-ESR.
On the bench we experienced fatal fails of buck converters
when plugging in the cable into the lab supply.
In that case there were several 100nH of lab cable inductance that together
with the ceramic bulk input capacitor on the pcb built a series resonant tank of high Q.
This created lethal voltage overshoots on plugging in the cable
and could be adequately simulated with LTSpice.
Parallelling an electrolytic cap with medium ESR shifted down
the resonant frequency and resonant peak disappeared due to the collapsed Q-factor.
This is an example of beneficial hi-ESR.