Paralleling capacitors effect on total esr reduction.

[goodguy]

Hi and thanks for taking the time to read this post.

I am trying to decide component values for 2 parallel capacitors, in an effort to reduce esr, which will be for use as power supply smoothing capacitors to be placed after the bridge rectifier.

After examination i noticed that consumer electronic goods generally use 2200uf aluminium electrolytic capacitors bypassed with a one microfarad polypropylene capacitor, the polypropylene operates at a higher frequency at which point the aluminium electrolytic ceases behaving as a true capacitor so cannot continue its filtering role.

But will this reduce the total esr, since both capacitors are operating at different frequencies.

If i took an aluminium electrolytic capacitor with an esr of 2 ohms and bypassed it with a polypropylene with an esr of 0.03 ohms the resultant esr would be 0.06 ohms, or would it?

Many thanks in advance. John.

 

[NeuronsAtWork]

I don't *think* that paralleling caps in the way you are considering will decrease ESR in that way, as I do believe it is frequency dependent as you surmised. Increasing the size of the electrolytic will decrease its ESR, and in my experience there is no hard and fast rule as to size in this application. 2200 uF should be sufficient, but 3300 or 4700 would also work well as long as you size your bridge rectifier to handle the possible inrush current to recharge the cap on each cycle. The second cap is for RFI/EMI suppression, and anything from .1 to 1 uF PP cap should work okay. Make sure the voltage and temp. ratings are adequate and you should be fine. (Note that I'm not an EE but have been an electronic hobbyist for 40+ years and built many many supplies of the type you describe. I've never had an issue using parts in the range discussed, but of course I don't know your application so use best judgement here.)

 

[NascentOxygen]

After examination i noticed that consumer electronic goods generally use 2200uf aluminium electrolytic capacitors bypassed with a one microfarad polypropylene capacitor, the polypropylene operates at a higher frequency at which point the aluminium electrolytic ceases behaving as a true capacitor so cannot continue its filtering role.

But will this reduce the total esr, since both capacitors are operating at different frequencies.

Hi goodguy! http://img96.imageshack.us/img96/5725/red5e5etimes5e5e45e5e25.gif
They are not exactly "operating at different frequencies". They "operate" at the frequency of whatever demands are placed on them by the circuit they are powering. If the powered circuit tries to draw high frequency variations in low-amplitude current, the electrolytic cannot meet this rapidly-changing demand (high ESR) but the plastic dielectric cap can.

If i took an aluminium electrolytic capacitor with an esr of 2 ohms and bypassed it with a polypropylene with an esr of 0.03 ohms the resultant esr would be 0.06 ohms, or would it?

How did you arrive at 0.06? If these figures represent the ESR at the frequency of operation, then 0.03Ω should be about right for the pair. but only for small current levels because the small cap cannot supply large currents. I think you'll find that the big electrolytic has inductance, too, and that acts together with the ESR and C to impose a delayed reaction to rapid changes in current demands, and this is undesirable. https://www.physicsforums.com/images/icons/icon6.gif

 

[sambristol]

The point is that electrolytic capacitors actually act as inductors if the load they are attempting to smooth has any transients on it with high enough frequency (digital circuits are notorious for this). The bigger the capacitance the lower the frequency in ceases to act as a capacitor, in general.

It is always a wise idea to parallel big electrolytics with a capacitor with better performance at high frequency.

A particular example is if you build a regulated power supply with one of the 3 terminal regulators like the 7805. The compensation amplifier in them is good to more than a megahertz (for no good reason). If you use just an electrolytic to bypass a 7805 or similar it can oscillate at rf frequencies and you often end up with nasty rf signals running on top of the dc on your power lines. This can and will cause all manner of unpredictable circuit behaviour as I know from bitter experience.

Regards

Sam

 

[goodguy]

Hi. Many thanks for the replies. It is genuinely appreciated.

@NascentOxygen: I came up with the figure of 0.06 ohms esr after visiting a few websites and briefly noticed the performance specs of electrolytics and polypropylenes; i was looking at prices trying to calculate how much this project would cost me.

I was hoping 2 ohms esr for the electrolytic, in parallel with 0.03 ohms esr for the polypropylene would give a total esr of 0.06 ohms.

On a side issue, i am on a budget and do not want to pay a lot of money for the aluminium electrolytics as they are large capacitors and tend to be more expensive So the cheap ones i am considering have have higher esr and lower performance specs, i was hoping that bypassing with a very low esr polypropylene would bring the total esr down to that of a high performance capacitor and save me some money.

Is this still possible in any way perhaps by changing the value of the polypropylene or the aluminium electrolytic or considering different capacitor types.

What do you think.

Many thanks in advance. John.

 

[NascentOxygen]

You need large electrolytics to absorb and hold the gulps of charge from the rectifiers. To offer a lower impedance to the circuit you are powering, you can parallel those big electrolytics with some smaller good quality ones, e.g., 2200uF, 220uF and 47uF, and one or two plastic dielectric caps. Any circuit you power will have its own power rail capacitors for AF or RF bypassing, so it's probably not worth going to a lot of trouble with that in your power supply.

 

[goodguy]

Hi. Thanks for the replies.

Just one last point:

The value i need is approx 10,000uf at 63volts but a slightly higher voltage or capacitance would be okay.

A google search indicates nichicon pw, kg as being good performers, but i can't find esr values for these two, are there any better. or is it best to get the lowest esr and forget about names.

At the moment i am thinking of two 4700uf in parallel, then 100uf in parallel, then a 1uf polypropylene so that each bandwidth has a capictor dedicated to it.

Or would i do better (lower esr) with say 10x 1000uf in parallel, then a 1uf polypropylene.

Many thanks. John

 

[NascentOxygen]

A google search indicates nichicon pw, kg as being good performers, but i can't find esr values for these two, are there any better. or is it best to get the lowest esr and forget about names.

I know nothing of manufacturer differences. Speaking for myself, I would avoid cheapies made in China, that's all. Electrolytics can fail early, probably the cheaper ones?

At the moment i am thinking of two 4700uf in parallel, then 100uf in parallel, then a 1uf polypropylene so that each bandwidth has a capictor dedicated to it.

Or would i do better (lower esr) with say 10x 1000uf in parallel, then a 1uf polypropylene.

All else being equal, I'd guess the parallel bank would give slightly better performance, but would you notice it? You haven't indicated what your delicate application is where especially low impedance is required.

One thing about parallelling multiple caps--if one dies then your replacement cost is less!

Good luck with your project.

 

[Antiphon]

An EE's turn at it.

You can only successfully lower the ESR by using caps that are similar in value and make. You should really only do this with identical caps.

The reason is that the ESR creates a drop and a heating action in the cap in proportion to the current flowing through it. The current going through the 1uF is nothing compared to the current in the 2200uF so it will not change the ESR of the combination.

Even if you use two caps that are 2200uF, if they are different types with inherently different ESR, they won't divide evenly and you won't get the hoped-for reduction.

 

[goodguy]

Hi. Thanks for the responses.

The capacitors are for use as power supply smoothing capacitors placed after the bridge rectifier, for supplying an audio amplifier i am building.

At the moment i have to finish the power supply, then i am building the chassis and heatsinks to house the project in, finally i will wire it and add input connectors, output speaker connectors and mains supply cable.

From my last project i noticied that the power supply dominated the accuracy of reproduction of the amplifier. I would say that is was responsible for 80% of the sound quality.

The frequeny response needs to be 20hz - 20khz, i need capacitors to supply each bandwidth, with a total capacitance of approximatly 10,000 uf at 63 volts. A 10,000uf capacitor i am considering has a resonant frequency of 5 khz so should be bypassed with a capacitor 10% of its size and each subsequent bypass a further reduction in size of 10% also.

So that component values are 10,000uf bypassed with 1000uf, bypassed with 100uf, bypassed with 10uf, finally bypassed with 1uf. The 10% rule is what i was taught in theory class but as you know real life situations can be different. Advice i have been given from my more experienced peers suggests one 10,000uf capacitor bypassed with a 1uf capacitor may be sufficient if a capacitor with the correct resonant frequency can be found. Aso the question of improved performance by lowering esr also raises its head.

So i am torn between using a series of bypassed capacitors of 10% decreasing size to ensure each bandwidth receives power, or using 10x 1000uf with a 1uf bypass to lower esr.

Many thanks. John.