Does This Circuit Make Sense for Variable Voltage to Constant Voltage?

The answer is in here someplace...

http://en.wikipedia.org/wiki/Rectifier

I think this is called a synchronous rectifier. And they are used to get rid of the large PN junction voltage drop, and power burner, in modern switch mode power supplies.

At least I think that's what your circuit is.

Is the above circuit practical? What would be a good voltage sensing switching mechanism, rather than a time based switching mechanism?

"Is the above circuit practical? "

In a word no...

In many words

you have all the gates tied together but each fet has a different Source Voltage so your FETS will have a varying VGS related to the charge of the capacitor attached to it.

If you over come this by using apropriate High side Fet drivers than your Caps still wont Charge as they have no "return" path the negative side of each cap is isolated by the FET inbetween each capacitor.

It apears that you are trying to charge a bank of caps in parallel than discharging them in series. which would give you a voltage multiplier effect.

"It apears that you are trying to charge a bank of caps in parallel than discharging them in series."

Yea thats what. If the capacitor has positive plate connected to voltage, and the negative floating would it charge?

I don't understand what you are trying to achieve.

I think you need to consider where the power rails are connected to the various elements of the circuit.

What is a bridge rectifier with only two terminals?

Try using a free schematic capture package like the one included with LTspice:-

http://www.linear.com/designtools/software/

to make your circuits a little more readable.

The two terminals are the point where we get the pulsating DC.

Let me get my welding goggles and I can watch you turn it on.

Is this a practical circuit? Well the problem is that you do not provide us anywhere near enough information to answer this question. You do not tell us what the circuit is supposed to do. You do imply that this is to make a power supply but you do not tell us if this will step up/down from the diode bridge voltage. You also do not tell us what frequency the pulse generator produces and how if at all it relates to the frequency your bridge rectifier will be producing. Remember a full bridge will double the fundamental frequency along with producing a DC component. The filtering capability of a capacitor stores the energy between cycles. Your bridge output will not be just a DC value.

Another significant oversight in your information, you do not indicate at all what the threshold voltage is for these FETS. I realize that many casually consider a FET as a switch but adding the critical information of VGS_th will quickly reveal a problem with this circuit.

VGS=VG-VS

VG is the voltage referenced to ground out of your function generator. VS is the voltage at the source of each FET referenced to ground. Only when VG>VGSth does an N channel FET conduct.

"Does This Circuit Make Sense for Variable Voltage to Constant Voltage?" -

Your original question above sounds like converting AC to DC, which can be done quite successfully with a rectifier-bridge and filtering, or switching power-supplies that are more efficient and easier to filter.

As already noted by others, your circuit has plenty of practical design problems but let's assume that those can be fixed;

So what are you really trying to achieve? Your circuit looks like an attempt to re-invent a switching power-supply of the Capacitor Charge Pump type, that would multiply the input voltage x4.

Unfortunately, without any output regulation, the output will track the input variations! Switching power-supplies usually have complicated, pulse-by-pulse regulation to achieve a stable output. Also, charge-pump schemes can deliver only limited amounts of power compared to switching power-supplies that employ inductors. Keep on studying...

@All

The idea was to attempt to convert any random waveform to a DC. Could someone go through the synthesis of such a circuit?

In my mind, the basic functionality would be as follows:

1. Convert waveform to pulsating DC using rectifier

2. Allow capacitors in a branch to charge to certain voltage value, discharge when they reach that value.

3. When value of capacitors in discharging branch reduce below specified voltage value, set them back to charge, while discharge from another branch of capacitors which were set to charge earlier.

Your bridge rectifier will convert a random waveform to the absolute magnitude of the waveform, minus diode drop. There is one caveat, the random waveform generator must be capable of driving the load on the other side of the diode bridge.

As I said earlier, your circuit has far to many things unspecified to properly analyze in detail. The most critical missing information is the capacitance values, the threshold voltage of the FET and the frequency of the pulse generator.

However, I do see a clear problem. In your capacitor charge cycle, (top red trace and bottom blue trace) this is not a closed loop. The capacitors will not charge! If you complete your circuit diagrams by adding your voltage sources and output load impedance, this will become obvious.

Thanks Redfred,

Forget the circuit posted above. I understand its wrong. Take my post before this and imagine that I am trying to build such a circuit as mentioned.

If I was to change the waveform to its magnitude form my the bridge rectifier (consider diodes ideal, so the 0.7 V drop doesn't exist), and then make a switching scheme to charge a branch of capacitors while another discharges so as to give constant DC output, how would one make such a circuit?

The switching scheme should be such that its senses the voltage values on the capacitors and switches, to discharge them in series while, the other branch charges at that point. When the capacitors of the first branch reduce by a particular value, the other branch should be set to discharge in series, while the first one goes back into charging.

Hey Jay,

what your describing sounds to me like a Power factor correction circuit.

To generate a DC voltage from any AC wavefor you only need to rectify and filter with a cap. (or inductor)

If you rectifiy any waveform and apply that to a capacitor the cap with only a small load will charge to the peak voltage and current will only be drawn when the rectified waveform voltage exceeds the voltage accross the cap. the rest of the time no current will be drawn from the circuit generating the waveform.

To draw a constitant current from the source rquires something along the lines rectifying it then chopping the resultan waveform up and feedingthat into an inductor (which should step the voltage up) rectify the output and feed that into a cap.

The trcky bit comes in with the control of the duty cycle of the chopper circuit which will be a function of the output current and the input voltage waveform.

Have a look at the opperational discription of Power factor cobtrol circuits and see if that describes the outcome you are trying to acheive.

What you continue to describe implies to me a voltage multiplier circuit. (TI has a nice theoretical paper you can study.) The elegance of this design is that only one time source signal is required. The analysis of how this circuit should function is very straight forward. Your proposal has two time based signals. So without any information of how your function generator is intended to perform in reference to the other wave form (synchronously, asynchronously, which one is faster) there is little one can say about the function.

Another circuit you might be trying to analyze and fabricate is a switched capacitor design to make a resistor. This hybridization of analog and digital theory is a complicated circuit that requires a good understanding of circuit theory to understand. You will notice that class notes only refer to switches opening and closing. A FET can be easily made to act like a switch under the correct conditions. One must now know what those FET limitations are and if the anticipated voltages and currents will fall in that region for the chosen FET.

OK, for anyone still following this thread I have a question for you to ponder.

A resistor is a very simple two lead component with no reference to ground. The simple version of a switched capacitor equivalent to a resistor must be grounded and is much more complicated than a simple resistor. Non-grounded switched capacitor circuits are even more complicated than the grounded form but they do exist. So why would anyone bother to make a complicated device to replace a simple device?

There is more than one correct reason to answer my question. I know a few of them. I will share my answers later if somebody makes an attempt to answer.

Integrated Circuits make use of many complicated and non-intuitive circuits, because the implementation of these "imitation" components and functions takes less real-estate on the IC than the real thing. Also, since many IC's are discrete-time/switched/digital anyway - these solutions are a natural choice.

Another example of non-intuitive circuits are the gyrator/impedance-inverters that would as an example, imitate a large inductor by inverting the impedance of a relatively small capacitor, using an analog OpAmp circuit.

That's one of the answers I'm aware of. I'm going to hold off for now before I divulge the other answer I know. I'm hoping that somebody might provide an answer I didn't think of.

OK, you asked for it...

These implementations can be produced and reproduced with much higher precision than a normal resistor, and have a bonus feature of being switching-frequency dependent, so slight adjustments to the frequency (multiply/divide) will tweak the "resistance" value. Sounds crazy enough to go to that length but obviously valuable for some products, filters for example.

"Integrated Circuits make use of many complicated and non-intuitive circuits, because the implementation of these "imitation" components and functions takes less real-estate on the IC than the real thing"

Is that like MOSFETs being used as capacitors, are you referring to the same thing?

Actually the use of the gate to drain capacitance of a MOSFET is an actual capacitor not the "imitation" components miklosh is talking about. The silicon oxide insulation between gate and drain acts as the dielectric and the reasonably well controlled overlap areas of these two form a real capacitor. The "imitation" components are the use of operational amplifier circuit topologies like a gyrator or general impedance converter with a MOSFET gate capacitance as the capacitor and switched gate capacitors for the resistors to fabricate any impedance desired.