Low Power Dissipation Driver Circuit

[JustinH]

Hi all,

I am trying to design a low power dissipation driver circuit. Basically the circuit gets a 1v digital signal which tells it whether to apply a separate DC power source of 100v across a 1nF capacitor. This digital signal is modulated at about 100kHz and will change at about 50kHz. The 100v source is either DC batteries in parallel or DC batteries stepped up with a transformer.

The problem at hand is designing the circuit to use as little power as possible. I've done calculations, and it seems with a simple circuit, 250mW is used on the cap. I've done some research and am interested in energy recovery from the cap when it empties to be reused to fill it when the signal comes to fill it back up. Someone said that perhaps flyback transformers could help, but they weren't too sure. I was thinking perhaps using an LC "tank" to hold the energy while the load cap was discharged, but wasn't sure of the viability. Also, what kind of switching would you guys recommend?

Any ideas would be appreciated, just kind of looking around right now and researching the project.

~Justin

 

[berkeman]

First of all, I think you mean batteries in series to make the 100V. Second, 100V is a dangerous voltage, so you will have to be very careful working with it, and be sure to design your circuit so that no human contact is possible with the high voltage areas. We work with a lot of dangerous voltages in our R&D Lab, and we have fairly strict procedures and rules for how to do things.

At least at first glance, the problem definition is a bit strange. Does it reflect a real-world situation, or is it an artificial lab assignment in school? I can't think of any real-world situations right offhand where you would want to be blasting a 1000pF cap with 100V-to-0V-to-100V squarewaves at around 50kHz. Yikes! I guess with a capacitance that low, it's practical to do this, but I have trouble imagining why.

As to your question, maybe the most efficient way to accomplish the "voltage switching" is to have two caps, one always charged to 100V and the other always discharged. You just switch the appropriate cap into the rest of your circuit based on the digital signal. If the 100V cap's charge isn't bled off much by being connected to the ciruit, then it will take very little power to keep it charged up. Would that work?

 

[JustinH]

Sorry about that, of course I meant series :P. The actual load is a device that I'm approximating as a 1 nF cap. By the nature of this device, I can't switch the load "cap" back and forth, basically I have to charge and discharge the load at about 50k. I'm wondering if there's a way to store the energy discharged from the cap until it needs to be filled back up. I'm kinda just shooting out ideas right now as I'm in the introductory process and just kind of trying to get an idea of where to start. I'm an undergrad doing some research for a professor.

The load cap should not be discharged at all until the digital signal goes low. Basically I'm given the digital signal on one side, the load "cap" on the other side, and I have to design the circuit in the middle. Let me know if you have more ideas, thanks for your help so far.

The main purpose of the design should be to minimize overal power dissipation, as the device will be run on batteries.

~Justin

 

[berkeman]

No I guess it becomes a matter of trying to recover as much energy as possible from the 1000pF cap that is initially charged to 100V. Well, if you could use a superconducting inductor, you could close a switch and let the cap voltage ramp up the inductor current for half of a cycle, and then open the switch to the cap and close a shorting switch on the superconducting inductor. Then when you wanted to charge the cap back up, connect it to the inductor and open the circulating switch on the inductor. The circulating current in the superconducting inductor stores the energy between cycles.

Not practical you say? Well, okay, that's true. Just brainstorming here... You could try to do some variation of that, where you use an inductor or transformer to temporarily store the energy from the capacitor and then transfer it somewhere else. Like if you switch the transfer inductor between the 1000pF cap and some other storage cap... Maybe you could even do a step-up / step-down voltage function with the inductor, so that you don't need to stack up 100/1.5 batteries...

 

[JustinH]

Hi,
It's been a while, but I'm still working on this project. Here's what I'm thinking so far. Basically starting with a low voltage battery, then using a DC-DC converter to get approximately 40V and 80V. Some of the properties have changed. For example, now it's a 500pF capacitor switching at 1MHz and going from 40V(low) to 80V(high).

From what I've read, some kind of adiabatic circuit would seem to work, however most of what I've read of those deal with relatively low voltages. I've been referred to flyback and Cuk converters, but haven't really noticed an application from there. I've heard that something like this has been done with a flyback converter somehow.

Any ideas?

Thanks,
Justin

 

[NoTime]

Odd setup.
What's it supposed to do?

Since you say the device is not a capacitor then you are probably better off doing some system analysis rather than reporting effective input impedance.

 

[JustinH]

The driver circuit is supposed to modulate a laser using a MEMS device that is comparable to a 500 pF cap. The device is essentially a capacitor. Any ideas?

 

[berkeman]

The driver circuit is supposed to modulate a laser using a MEMS device that is comparable to a 500 pF cap. The device is essentially a capacitor. Any ideas?

How precisely do you need to control the MEMS position? If you just bang it with the two voltages 40V/80V, won't you get a lot of ringing at the end of the swings? It seems like you would need some waveshaping on the drive waveform to get critically damped motion of the MEMS element. Is this 1MHz modulation for beam steering, or just OOK data?

 

[JustinH]

Basically when the high voltage is applied across the capacitor, it diffuses the laser light that is reflected back. When less than approximately 55V, the capacitor acts like a mirror. The method of modulation is rather open. Right now I'm working with something that will return to zero after every 1 bit.

The voltage across the MEMS need not be too precise. After about 55V it turns on and starts to diffuse the light, and then as you increase the voltage, there is an exponential rise in diffusion. I'd like to keep my high voltage roughly 80v +/- 10v, and my min voltage maximum of 40v. The light will be modulated going from bright to diffuse.

Something I'm trying to do now is to create a driven resonant circuit that pings voltage between an inductor and a capacitor. I am trying to switch the capacitor in the circuit between the MEMS device, and an actual capacitor of similar capacitance. Basically the sum of the voltages on the two capacitors looks like a sinusoid. The voltage on either of them looks like a series of halfwaves followed by periods of 0 voltage. I'm thinking about using pulse width modulation to keep the peaks around 80v.

Another thing, in the driven resonant circuit that I'm working on, the voltage swings above and below zero, not a problem for the MEMS device as all it recognizes is the square of the voltage, it is independent of positive or negative.

I don't have much experience in switching the capacitors in the resonant circuit. In the simulations I've done in PSPICE, I'm getting about 400mW of loss across my switching MOSFETS. What suggestions could you offer for keeping the switching loss across the MOSFETS low? Should I use something besides MOSFETS to switch?

Thanks,
Justin