switched capacitors for DC to DC

[akis]

Hello

Has anyone built a practical circuit that uses capacitors (not coils) to convert DC to DC? I have experimented a little and have made these topologies so far:

all the below with common ground which is also important

DC -> 555 -> inverter
DC -> 555 -> doubler
DC -> 555 -> +Vs/2 -Vs/2 (ie split a 20V into +/-10V)

I have not taken some of these circuits beyond a few mA on the load.

I am looking for ideas/circuits to step down, eg DC -> 555 -> Vs/2 or Vs/3 as an example, or even variable.

Thanks

[LLLlllou]

Re: switched capacitors for DC to DC

Hello

Has anyone built a practical circuit that uses capacitors (not coils) to convert DC to DC? I have experimented a little and have made these topologies so far:

all the below with common ground which is also important

DC -> 555 -> inverter
DC -> 555 -> doubler
DC -> 555 -> +Vs/2 -Vs/2 (ie split a 20V into +/-10V)

I have not taken some of these circuits beyond a few mA on the load.

I am looking for ideas/circuits to step down, eg DC -> 555 -> Vs/2 or Vs/3 as an example, or even variable.

Thanks

I haven't played with them in about a decade, but Maxim and Linear Technology both make a bunch of "add yer own caps" single chip switched cap. solutions. You might want to check out their data sheets, lots of good info.

[mariushm]

Re: switched capacitors for DC to DC

Wasn't the discussion on EEVBlog good enough for you?

You already got a lot of answers there: http://www.eevblog.com/forum/project...p-down-design/

[akis]

Re: switched capacitors for DC to DC

Hoped I would find something more here. No thanks to you.

[akis]

Re: switched capacitors for DC to DC

I haven't played with them in about a decade, but Maxim and Linear Technology both make a bunch of "add yer own caps" single chip switched cap. solutions. You might want to check out their data sheets, lots of good info.

Thanks. I have found a few not only from MAxim and Linear, but also NS : LM 2751 - but they all seem to be low voltages and currents (eg 5V at 80mA). I have not discovered an equivalent to, say, LM317 but in a switched capacitor topology.

I have built and tested a very simple switched capacitor DC inverter circuit that is good for any voltages (I think it was tested to 15-20V and 500mA), so I was confident that the idea(s) could also be applied/scaled to other DC-DC conversions and currents, especially DC step down.

[mariushm]

Re: switched capacitors for DC to DC

Well, there's no point beating up a dead horse. You got your answers there... this is a forum aimed mainly at repairing hardware, less (hardly) for designing hardware.

You can't do a buck converter with a voltage doubler, or using some switched capacitor scheme. I'm not aware of any circuit capable of doing conversion to lower voltage without using inductors.

[Uniballer]

Re: switched capacitors for DC to DC

You can't do a buck converter with a voltage doubler, or using some switched capacitor scheme. I'm not aware of any circuit capable of doing conversion to lower voltage without using inductors.

It should be possible but may not be practical. For example, a circuit that charged a cap (that was previously discharged) only far enough to get close to the target voltage, then coupled that cap to the output for a while should work, but probably wouldn't give much power and might not be more effective than a linear regulator. A number of such circuits might give more power, but might take too much space compared to a buck converter, etc.

[tom66]

Re: switched capacitors for DC to DC

It should be possible but may not be practical. For example, a circuit that charged a cap (that was previously discharged) only far enough to get close to the target voltage, then coupled that cap to the output for a while should work, but probably wouldn't give much power and might not be more effective than a linear regulator. A number of such circuits might give more power, but might take too much space compared to a buck converter, etc.

It would be exactly as efficient as a linear regulator using ideal switches. If you had realistic nonideal switches then it would waste more power.

You can make a high DC to low DC converter by using pulsed AC into a capacitive divider. Tap off at half way point. But, you must make sure the load is the same on each branch, or the output will drop (or increase) proportionally. No better than a resistive divider, although it can be very low loss.

I did once play around with making a 555 timer into a buck converter, more as a kind of curiosity. I used the on-die transistors to make a synchronous converter, but it turns out that the high current from the inductor and reverse diode current damages them in strange ways. I went through many 555 timers that day.

[akis]

Re: switched capacitors for DC to DC

I also tried simulating the two caps in series and then in parallel, but could not create a suitable circuit (too many switches were needed), was becoming very complex.

I have also many times fallen foul of efficiency - I design something which I think is great only to realise it still spends just as much power as a linear approach would do. Here is a schematic of such an attempt.

Voltage supply is 15V. Output voltage is 4.5V/1A. A 555 generates a fixed frequency of about 7.3KHz. A switching transistor arrangement charges the charge/load capacitor from the power supply. An op-amp controls the duty cycle of the 555 based on a desired voltage and what's actually on the load.

In order to achieve efficiency, the switching transistor should never be in the linear region. Secondly the switching transistor should spend more time being on or off, rather than transiting from one state to the other. If both these conditions are met, then we have good efficiency. Unfortunately as the schematic shows, we are pulsing this transistor so many times, that we spend so much power on it, just as much as if the transistor was driven directly by the op-amp (linear operation).

[budm]

Re: switched capacitors for DC to DC

Did you try using MOSFET with low Rdson? You will need low Vsat if you are going to use transistor to achieve better effi.

[tom66]

Re: switched capacitors for DC to DC

Again, that circuit will have basically the same efficiency as a linear reg, probably less. But it will never be MORE efficient. The reason is simple, if you connect a capacitor to the supply directly it charges up instantly drawing a very high current.mIf you want to limit the output voltage you need an inductive charging source, i.e. a buck converter, otherwise the cap will charge instantly OR the charging will be limited only by the resistance of the power switch which will dissipate power to limit the voltage.

[akis]

Re: switched capacitors for DC to DC

Again, that circuit will have basically the same efficiency as a linear reg, probably less. But it will never be MORE efficient. The reason is simple, if you connect a capacitor to the supply directly it charges up instantly drawing a very high current.mIf you want to limit the output voltage you need an inductive charging source, i.e. a buck converter, otherwise the cap will charge instantly OR the charging will be limited only by the resistance of the power switch which will dissipate power to limit the voltage.

Reading from Maxim's documentation, it should be possible to monitor the charge capacitor's voltage - it should not have to charge to full charge instantly. At least that is how they build their own switched capacitor ICs.

In practice however, as you said, I am discovering that the only way to keep that capacitor from charging to the full voltage, is to interrupt the charge cycle, in other words NOT to allow the pass transistor/MOSFET to open fully. And that right there is the problem, since the transistor is not open fully, you are in the linear region and you waste power.

At least that is what I think is happening.

[tom66]

Re: switched capacitors for DC to DC

Reading from Maxim's documentation, it should be possible to monitor the charge capacitor's voltage - it should not have to charge to full charge instantly. At least that is how they build their own switched capacitor ICs.

That's exactly why every regulated inverter type IC has on resistances of around 20 to 30 ohms. They're designed for low current outputs and also so the output can be regulated by a simple control loop.

If you want a stepdown, you build a buck or flyback.

[budm]

Re: switched capacitors for DC to DC

http://www.edn.com/design/analog/436...-95-efficiency