LT3081-based Power Supply Capacitor Question

[Ultron81]

I'm building a 0-20V 0-1A PS based on the LT3081. I have a question about the type of capacitor to use (Film, Ceramic, Electrolytic, etc).

The data sheet recommends a 1uF on the input and a 10uF on the output. From all the examples in the data sheet,  these look to be non-polarized.

There is a note about the dielectric type of Ceramic capacitor in the data sheet also. But, I was wondering if a Film cap would work OK as well? I have some 10uF film capacitors on hand, but would need to pick up some 1uF sizes in either ceramic or film.

Thanks!

[SiliconWizard]

You may like the following reading:
http://www.analog.com/en/analog-dialogue/articles/why-bypass-capacitor-choice-matters.html
https://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf

[Lee Leduc]

Here's a link to the Demo Board for LT3081 complete with parts list.
http://www.analog.com/media/en/dsp-documentation/evaluation-kit-manuals/dc1870af.pdf

[Ultron81]

You may like the following reading:
http://www.analog.com/en/analog-dialogue/articles/why-bypass-capacitor-choice-matters.html
https://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf

Thanks for this!!

Here's a link to the Demo Board for LT3081 complete with parts list.
http://www.analog.com/media/en/dsp-documentation/evaluation-kit-manuals/dc1870af.pdf

Thanks, I was originally going to go this route, but decided to simplify it, leaving out the pre-regulator and just going with a heat sink. OSHPark wanted ~$100 for that board (3 of them), though JLCPCB seemed much cheaper, now that I look. The LT3091, LT8612 and LTC3632 are ~$25 total, too. I decided to go with the LT3081, a 24V SMPS, 2 LM334s (1 for external current source for Voltage set, the other for the 5mA minimum load requirement), and a LMC7660 Charge Pump to return the minimum load current to a negative supply.

[David Hess]

There is a note about the dielectric type of Ceramic capacitor in the data sheet also. But, I was wondering if a Film cap would work OK as well? I have some 10uF film capacitors on hand, but would need to pick up some 1uF sizes in either ceramic or film.

A film capacitor will work fine in place of a ceramic capacitor.

[Bassman59]

I'm building a 0-20V 0-1A PS based on the LT3081. I have a question about the type of capacitor to use (Film, Ceramic, Electrolytic, etc).

Mind the minimum-load requirement.

[Zero999]

I'm building a 0-20V 0-1A PS based on the LT3081. I have a question about the type of capacitor to use (Film, Ceramic, Electrolytic, etc).

Mind the minimum-load requirement.

Which can be catered for using a constant current sink, if it needs to work over a wide voltage range, or a simple resistor if it just needs to work at one voltage.

[coppercone2]

Look at dave-jones recent video with impedance analysis, he tests ceramic and foil capacitors of various kinds to see the curves.

It looks like ceramic and foil are pretty similar resonance points, but you would look for microphonics, dielectric soakage, leakage, voltage coefficent of capacitance, surge current ratings (how does ceramic compare to foil, I am actually asking) and other parameters you might find.

I am kind of thinking actually, it might be better to use a foil one, especially if you are post regulating a switching supply, because that might have some mechanical vibration that can bypass your regulator and enter through your (ceramic) capacitor. The PCB is effectively a mechanical acoustic shunt and the capacitor is an energy converter.


I would like to know how to devise a measurement quantifying the conversion efficiency of capacitors to vibration.

I think you can put mirrors on all three axis to measure the vibration of the capacitor and use its mass to quantify the energy present. I wonder if there is a simple accurate method. And how energy conversion in ceramics acts (does vibration on 1 axis only get converted to energy? Do vibrations on other axis effect it (or the energy absorption of the primary axis if so) and testing the frequency response. 

[SiliconWizard]

Film caps were often used as decoupling caps in the 80s on through-hole boards. The advantage of ceramic caps is the very low ESR and very low cost. And back in the 80s, ceramic caps were still low value. I guess even a 100 nF ceramic cap would have been pretty big back then compared to the multilayer caps we have now.

For a small homebuilt linear PS, it's not likely to make a big difference anyway.

[Ultron81]

Mind the minimum-load requirement.

Thanks, I plan on using a LM334 current source set to 5mA for the minimum load, and putting it across the output and the -5V rail from the 7660 charge pump, to get to 0V output.

After reading about capacitors for LDO regulators, I think I'll stick with ceramic. Should I really be concerned about microphonics from the caps with a voltage regulator?

[coppercone2]

I am starting to think that ceramic performance is a bit over blown with similar values. It has a somewhat sharper resonance but your not really using the resonance in a decoupling cap unless your lucky.

I just mean, don't sweat it too much.

[coppercone2]

https://electronics.stackexchange.com/questions/267590/decoupling-capacitors-low-esr-electrolytic-vs-electrolytic-ceramic-vs-ceramic

[Ultron81]

https://electronics.stackexchange.com/questions/267590/decoupling-capacitors-low-esr-electrolytic-vs-electrolytic-ceramic-vs-ceramic

This graphic is great!

[Zero999]

Mind the minimum-load requirement.

Thanks, I plan on using a LM334 current source set to 5mA for the minimum load, and putting it across the output and the -5V rail from the 7660 charge pump, to get to 0V output.

Unless the negative voltage is used for something else, don't bother with the 7660. Use the circuit I posted previously with two transistors. It's a current mirror, with R1 setting the current. The downside with that is the current will depend on the input voltage, but R1 can be replaced with the LM334 if that's an issue.

After reading about capacitors for LDO regulators, I think I'll stick with ceramic. Should I really be concerned about microphonics from the caps with a voltage regulator?

The microphonics shouldn't be an issue, especially if you add an electrolytic capacitor which will absorb anything generated by the ceramic capacitor.

[Ultron81]

Unless the negative voltage is used for something else, don't bother with the 7660. Use the circuit I posted previously with two transistors. It's a current mirror, with R1 setting the current. The downside with that is the current will depend on the input voltage, but R1 can be replaced with the LM334 if that's an issue.

I was planning on doing a current mirror, but the data sheet specifies to reach 0V output, you need to return the minimum load current to a negative rail (Page 12). I will be using a +5V supply anyway, so I was thinking of using the 7660 for the negative rail.

[David Hess]

C0G/NP0 ceramic capacitors are actually suitable for many precision applications which formerly required film capacitors like sample-and-holds and integrators.