EEVblog Electronics Community Forum

General => General Chat => Topic started by: aqarwaen on February 26, 2020, 11:52:45 pm

Title: would it be possible design any psu without any electrolytic capacitors?
Post by: aqarwaen on February 26, 2020, 11:52:45 pm
I just wonder how hard and how much it would cost design any psu without any electrolytic  capacitors?lets say psu is any 750w and don't use any electrolytic fluid capacitors.would it possible to design any psu what used and capacitors instead electrolytic fluid capacitors?
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: angrybird on February 27, 2020, 12:02:15 am
This is a wonderful topic to get some seasoned power supply guys arguing over  :-DD

What power supply architecture are you considering?  What is your input/output voltage?  Is it fixed output voltage?  Do you have specific transient response or fault tolerance requirements?  Any objections to a high frequency design?  Depending on architecture, are you capable of winding your own magnetics?
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Circlotron on February 27, 2020, 12:34:46 am
For an AC mains fed PSU, if you are going to use capacitors of relatively small value you are not going to be able to keep the voltage from falling to near zero when the AC mains crosses zero. Unless of course you can tolerate a battleship size iron core inductor in there somewhere.

One solution might be to use a 3-phase full wave rectified source because the voltage never falls to zero.

(https://www.electronics-tutorials.ws/wp-content/uploads/2018/09/power98.gif)
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: dmills on February 27, 2020, 12:57:47 am
Three phase input makes it easy to drop the bus cap as long as power factor is not too much of an issue.

If you really want smooth DC on the bus, go with the old 12 pulse rectifier technique, needs a star-delta transformer of rather weird ratio, but does work.

The low voltage output caps are possibly harder, so some kind of highly polyphase design might be indicated to get the required total value down far enough. Could be that voltage mode (spit!) would actually be easier here for all that it makes basically EVERYTHING else worse.

Got to admit, I don't really get the hate on for elcos, a good quality part derated suitably for temperature and ripple will go a good 20 years or so if you derate with that intention in mind, just got to do the sums and them make sure purchasing don't screw you over.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: twospoons on February 27, 2020, 01:32:45 am
Well, the simple answer is "yes", with the caveat that it would be more expensive and physically larger than a PSU built using electrolytics.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: AG6QR on February 27, 2020, 04:27:10 am
If the supply is supposed to deliver relatively constant current, filtering can be done using lots of inductance, combined with a bit of resistance. 

Another possibility is to use a very (impractically) large number of small-value non-electrolytic capacitors in parallel, in place of what would ordinarily be an electrolytic capacitor.

It's possible to use an op-amp with a capacitor to produce a "virtual inductor". It should be equally possible to use an op-amp with an inductor to produce a "virtual capacitor".  I've never heard of anyone actually doing this, because typically, capacitors are cheaper and smaller than inductors.

I wouldn't think any of these strategies would be particularly practical; there are good reasons why power supplies have electrolytic capacitors.  Electrolytic capacitors have advantages in areas of expense, size, and weight.  But as a thought experiment, or an "I want to do this to prove it's possible" project, it should be possible.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Ben321 on February 27, 2020, 04:44:01 am
It could theoretically be possible. If the frequency of the switching supply were high enough (switching transistor driven by a 1MHz or higher frequency oscillator) then you could easily get away with using a 1uF filter capacitor, which is available as a non-electrolytic type, instead of something like a 470uF capacitor, which is only available as an electrolytic type. The problem would be RF interference in the AM broadcast band or other RF bands governed by the FCC. The amount of filtering done to keep the RF signal from escaping from the power supply through either the DC side or AC side wires, would mean that additional components would be necessary, like RF chokes on the AC input wires and DC output wires.

With a normal switching supply, running at about 50kHz, that doesn't correspond to an RF band that's actually used for any communications systems, so no additional filters are needed, and the wires (both on the AC input and DC output sides of the power supply) are allowed to radiate RF in these low frequency bands.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: twospoons on February 27, 2020, 04:57:03 am
Doing some back of the envelope calcs ...

first stage would be a boost PFC to 400V bus. That lets you pull current over most of a mains cycle.
I'm going to say 50Hz system, so the bus cap needs to hold enough energy to last for 0.01 seconds.  Its actually even shorter than that thanks to the PFC, but for arguments sake I'll use 0.01s
For 750W output, thats 7.5J in 0.01s.
So lets store 10J for some margin. 10J @400V is 125uF.   Thats doable => 5 x ~25uF 500V film caps. At a cost of roughly 5 x the cost of electros
Then buck down at  high frequency to reduce output capacitor size. This is assuming you want something like 12V, much lower than the bus voltage.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: TerraHertz on February 27, 2020, 06:09:33 am
Absolutely! Three phase input rectifier, high frequency PFC switching circuit to HV bus with only small HV polyester + ceramic caps, standard HF switchmode running from that bus.

With only single phase mains, not so much. Unless you went mechanical, like a rotary converter or something else providing energy storage via kinetic energy.  Or batteries.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Berni on February 27, 2020, 07:20:00 am
That is a very loose question.

Technically a AC motor driving a brushed DC generator fits as an answer. That not only does not contain electrolytic capacitors but doesn't contain ANY capacitors at all (And also contains no resistors or semicodnuctors)

But yeah there is no reason you couldn't take any switching PSU, rip all the electrolytics out and replace them with film or ceramic caps of the same value. The resulting PSU will still have similar performance but be way bigger, heavier and significantly more expensive.

There is a reason electrolytic capacitors are used, they are a very good way of providing large amounts of capacitance. Its not planned obsolescence to make your PSU die after 2 years once its out of warranty. I have equipment from the 80s that still runs fine on its original caps. Its just a matter of using properly rated and quality made capacitors if you want your PSU to have a long life.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: shakalnokturn on February 27, 2020, 08:03:51 am
Switching frequencies have gone higher since the 80's, while SMPS can be made reliable no recent consumer designs even go to the trouble of adding the few extra components to take the high dI/dt off the electrolytics that would allow even the Yaego and worse to last over 2 years.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: brabus on February 27, 2020, 08:19:21 am
(…) I have equipment from the 80s that still runs fine on its original caps. Its just a matter of using properly rated and quality made capacitors if you want your PSU to have a long life.

^This.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: exe on February 27, 2020, 10:54:17 am
A non-electrolytic cap and a capacitance multiplier? Though, considering the load, it should be a beefy device. I can't even guesstimate heat dissipation without a simulator.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Psi on February 27, 2020, 11:06:54 am
I just wonder how hard and how much it would cost design any psu without any electrolytic  capacitors?lets say psu is any 750w and don't use any electrolytic fluid capacitors.would it possible to design any psu what used and capacitors instead electrolytic fluid capacitors?

Depends,  are these ideas breaking your rules?

- Using solid polymer electrolytic caps (you said no electrolytic 'fluid' capacitors)
- Using a combination of ceramic caps and super capacitors to replace the bulk storage of the electrolytics
- Cheap china tantalums :)   (might be a bit explodey but it works, for a little while anyway...)
- China has started using rechargeable lipo cells in disposable products because they are cheaper than primary cells sometimes.
So I wonder if you could use lipos to replace the bulk storage of the electrolytic caps. Operate the cells at 3.8V continuously for long cell life.

hm.. If PCB space was not limited, i wonder where the sweet spot would be for paralleling hundreds of small ceramic caps

Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Kjelt on February 27, 2020, 12:02:38 pm
I just wonder how hard and how much it would cost design any psu without any electrolytic  capacitors?lets say psu is any 750w and don't use any electrolytic fluid capacitors.would it possible to design any psu what used and capacitors instead electrolytic fluid capacitors?
Since your requirements don't demand that it is a DC psu, and you say "any psu", any (non-regenerated) AC psu will fullfill your requirements  :)
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: richard.cs on February 27, 2020, 12:29:23 pm
Got to admit, I don't really get the hate on for elcos, a good quality part derated suitably for temperature and ripple will go a good 20 years or so if you derate with that intention in mind, just got to do the sums and them make sure purchasing don't screw you over.

Sometimes 20 years isn't enough. Actually after quite a bit of research we failed to find any electrolytic manufacturers that would guarantee them to work beyond 15 years at any temperature - it was the longest they were confident the seals would last for. There are some ludicrously expensive electrolytics with glass to metal seals but they're not affordable for most applications (I've never actually seen one but they're in the digikey catalogue).

As a result we've done a number of completely electrolytic-free designs for a particular customer, with an expected minimum service life of 35 years. These are mostly in the 150 W region and tend to use quite a lot of expensive ceramics, high operating frequencies and oversized magnetics compared to common designs.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: T3sl4co1l on February 27, 2020, 12:46:54 pm
Sure.  What's your size and cost budget?

When you see the alternatives, you may rethink your perspective on electrolytics. :)

Tim
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: NiHaoMike on February 27, 2020, 01:13:12 pm
- China has started using rechargeable lipo cells in disposable products because they are cheaper than primary cells sometimes.
More like they found something better to do with the rejects rather than recycling them or using them to scam unaware buyers.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: tom66 on February 27, 2020, 01:54:00 pm
Absolutely.  But the 50Hz AC is a problem.

It depends on what you are powering.  Many cheap LED lamps don't use a capacitor on the 400V side, because the 100Hz lamp flicker is acceptable (if potentially annoying). 

Many electric vehicle chargers now do not provide a constant current output to the vehicle battery. It is modulated with the AC mains, which reduces the power factor circuit requirements as well.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: jesuscf on March 02, 2020, 04:55:46 am
Once I saw a 7500VDC @ 20kW power supply built with a three phase rectifier (made of many smallish diodes connected in series and parallel).  It didn't have any capacitors.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: Berni on March 02, 2020, 06:18:00 am
Once I saw a 7500VDC @ 20kW power supply built with a three phase rectifier (made of many smallish diodes connected in series and parallel).  It didn't have any capacitors.

Yep uses the trick described by Circlotron up there earlier https://www.eevblog.com/forum/chat/would-it-be-possible-design-any-psu-without-any-electrolytic-capacitors/msg2938666/#msg2938666 (https://www.eevblog.com/forum/chat/would-it-be-possible-design-any-psu-without-any-electrolytic-capacitors/msg2938666/#msg2938666)

The catch is that you don't always have 3 phase power available to you. Some houses don't even have the street itself wired for 3 phase power.

Some trains also use DC power. Back before semiconductors the massive amounts of DC current needed ware delivered by mercury arc rectifier tubes in sizes larger than a refrigerator. Its basically a giant glorified neon bulb where the combination of mercury and carbon electrodes makes electron emission in one direction happen much easier, creating a diode. To make it multiphase simply multiple carbon electrodes are inserted making these things very common with 3 or 6 anodes (6 is for full wave rectification)
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: dietert1 on March 02, 2020, 04:38:33 pm
Usually the larger electrolytic capacitors last longer. If you replace the small ones up to 100 uF by ceramic multilayer caps or film caps, the device will probably last some decades. Recently i recapped a Keithley 148 nanovoltmeter made in 1965, so that's 55 years now. It had many 100 uF electrolytics (Sprague) that were all in good condition. One 22 uF cap was dry with an ESR of 68 Ohms. This is a low power instrument running cool inside, at least most of it.

The main enemy of electrolytic capacitors are close-by power resistors running at several hundred °C. I mean the durability of a power supply is a matter of thermal design. This is well known to those sad engineers who need to implement premature failure for commercial reasons.

Regards, Dieter
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: blueskull on March 02, 2020, 04:43:06 pm
With single phase, no, unless you want to use stupidly large film or ceramic caps.

However, it is possible do build a single phase PSU without wet caps.

You can put the reservoir cap on the secondary side, which usually is at lower voltage, within the range of polymer caps.

Something like a single stage isolated PFC+ripple cleaning buck or a fixed ratio isolated converter+secondary PFC+ripple cleaning buck.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: David Hess on March 02, 2020, 04:55:09 pm
It is completely feasible to design power supplies and regulators which require no electrolytic capacitors except in the case of a single phase AC input where the minimum capacitance is determined by the required holdup time which results in large values of capacitance.  And even there, it can be done but with a lower level of practicality; for instance, an inductive input filter could be used instead.

Military, aerospace, and high reliability (long life) power supplies and regulators are sometimes designed to use only ceramic, film, and tantalum capacitors that lack inherent wear-out mechanisms.

first stage would be a boost PFC to 400V bus. That lets you pull current over most of a mains cycle.
I'm going to say 50Hz system, so the bus cap needs to hold enough energy to last for 0.01 seconds.  Its actually even shorter than that thanks to the PFC, but for arguments sake I'll use 0.01s
For 750W output, thats 7.5J in 0.01s.
So lets store 10J for some margin. 10J @400V is 125uF.   Thats doable => 5 x ~25uF 500V film caps. At a cost of roughly 5 x the cost of electros

An active PFC stage is a great way to lower the required input bulk capacitance by increasing the conduction angle however using it this way compromises the power factor.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: SiliconWizard on March 02, 2020, 05:01:34 pm
The main enemy of electrolytic capacitors are close-by power resistors running at several hundred °C. I mean the durability of a power supply is a matter of thermal design. This is well known to those sad engineers who need to implement premature failure for commercial reasons.

Yup. Electrolytic caps are very sensitive to temperature, which is a particular problem in power supplies.

And yes, sometimes this is used on purpose. My Samsung LCD TV (well, the power supply board) was exactly designed like this. There were a handful of electrolytics that were mounted right against heatsinks. A couple of them went dry within a couple years. The fix was just a few bucks for new caps (which I tried to mount a little further away from the heatsinks), but this is irritating.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: T3sl4co1l on March 02, 2020, 05:16:30 pm
An active PFC stage is a great way to lower the required input bulk capacitance by increasing the conduction angle however using it this way compromises the power factor.

You can get the best of both worlds using an active capacitor multiplier -- essentially, use a smaller capacitor at a higher ripple fraction, by application of a synchronous buck converter (controlled to act as, I think, a step-up transformer relative to a virtual midpoint).

In combination with poled ceramics (e.g. Ceralink), this enables record low volumes (e.g. Google Little Box inverter from some years ago).

Tim
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: chris_leyson on March 02, 2020, 05:28:32 pm
I once designed a 6W 48V to 5V converter with nothing but X7R ceramic caps on the primary and secondary side. It was designed to run in an 85C+ environment 24 hours a day and I didn't want to use electrolyics. Doable but expensive.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: blueskull on March 02, 2020, 05:29:45 pm
I once designed a 6W 48V to 5V converter with nothing but X7R ceramic caps on the primary and secondary side. It was designed to run in an 85C+ environment 24 hours a day and I didn't want to use electrolyics. Doable but expensive.

48V AC or DC?
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: chris_leyson on March 02, 2020, 05:33:16 pm
It was DC so I guess it doesn't count  :(
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: SiliconWizard on March 02, 2020, 05:34:03 pm
Whereas it's certainly relatively easy (if a bit expensive) for a DC/DC converter, as was mentioned earlier, PS from single-phase mains for anything but very low power are much trickier, and unless you're ready to design something ridiculously expensive and bulky, it's often a compromise with the power factor (which in turn may make it not compliant with current regulations.)
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: David Hess on March 02, 2020, 06:41:30 pm
An active PFC stage is a great way to lower the required input bulk capacitance by increasing the conduction angle however using it this way compromises the power factor.

You can get the best of both worlds using an active capacitor multiplier -- essentially, use a smaller capacitor at a higher ripple fraction, by application of a synchronous buck converter (controlled to act as, I think, a step-up transformer relative to a virtual midpoint).

In combination with poled ceramics (e.g. Ceralink), this enables record low volumes (e.g. Google Little Box inverter from some years ago).

I have never had to take advantage of it but for a given size of capacitor, doubling the working voltage while halving the capacitance doubles the energy storage (0.5CV^2).  So stepping the AC input voltage up further allows for a physically smaller input capacitor but also requires handling a higher ripple voltage.

A couple years ago I looked into this for SSD power loss protection and while it would work, it is not necessary because the energy density of existing non-super capacitors is sufficient for SSDs and in some you can see a bank of solid tantalum capacitors just for this purpose.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: schmitt trigger on March 02, 2020, 06:56:30 pm
A rotary converter?
https://www.nycsubway.org/wiki/Rotary_Converter_Power_Technology (https://www.nycsubway.org/wiki/Rotary_Converter_Power_Technology)

Now seriously, like others have mentioned, it is feasible but impractical because of size and cost constraints.
Title: Re: would it be possible design any psu without any electrolytic capacitors?
Post by: larrybl on March 02, 2020, 07:34:41 pm
I have a 28 VAC 6A Center taped Xformer with a bridge rectifier, works great for testing 12 VDC things like 12V Solenoid's, 12V electric clutch, 12V incandescent lamps, No caps involved.

The normal charging system on a Garden Tractor consists of a stator and either a regulator or single diode to provide power to drive electric clutch and charge the battery.