Totem Pole Switches and Switch Drivers

Aren't existing switches and switch controls fast enough for most switched power purposes?

Faster is generally better. Faster switching times reduce resistive losses during switching. Faster switching allows tighter control of ON and OFF times. Tighter control reduces spikes and ringing.

How can you eliminate the need for snubbing?

The break-before-make action required of a totem pole switch causes a period when there is no path available for any inductor current. During that period, voltage can rise precipitously. Our fast switches need only a few nanoseconds of break-before-make time. That is not long enough for a troublesome voltage rise.

Doesn't more switching waste energy on gate drive?

Our switch drivers use only marginally more than the minimum theoretic energy needed to drive a given switch capacitance at a given frequency. With efficient switch drivers, higher frequency operation can be more advantagous.

Are you including the inefficiency of the high-side power supply in that calculation?

Our drivers do no need a separate high-side supply.

Doesn't faster operation greatly increase EMI?

Because our switching waveforms are near ideal, the energy goes where it is supposed to, thereby reducing electrical radiation.

How much of a premium must be paid for CogniPower switch drivers?

Likely none. The savings from smaller heat sinks, fewer clamps and snubbers, and from the ability to use lower voltage parts, can exceed any extra driver cost. Add to that the savings of eliminating the high-side power supply, and you should come out well ahead.

Instantaneous AC and DC Current Sensing with Better Common Mode Rejection

Clamp-on probes with Hall Effect sensors are the industry-standard way to measure AC and DC currents. What can you do that they can't?

Those probes need to be derated at higher frequencies. High currents at high frequencies cannot be measured. As the limits are approached, the unit under test is increasingly loaded. Also, Hall Effect sensors aren't particularly accurate, so drift and recalibration can be troublesome.

For good DC accuracy what is wrong with a differential amp and a sense resistor?

At DC, that approach can work quite well, but even the most expensive differential amps lose their CMR rapidly at higher frequencies. In many topologies, there is no way around needing more CMR than conventional differential amplifiers can provide.

How do you get 70 dB CMR at 100 MHZ? That doesn't sound possible.

We use a commutated common mode choke. It is a new, patent-pending technique that achieves most of the common mode rejection in the geometry of the magnetics themselves. Since that geometry is fixed and stable, so is our CMR.

My average current measurement works well enough, why do I need instantaneous current measurement?

SMPCs can be prone to destructive runaway. You need a very fast-acting protection or control mechanism to catch an out-of-limits condition in time to prevent permanent damage. Averaged current measurements are just too slow.

How often do you need to Degauss?

Never.

How often do you need to Rezero?

Never.

New SMPC Control Strategies

Even though PWM is well-understood, PWM controls are tricky. Making small SMPC control changes is perilous. Isn't making major changes highly risky?

Incremental progress has improved SMPCs to the point of diminishing returns. If performance and efficiency are to be taken to new levels, new control approaches are needed.

But, hasn't successful change in SMPC technology historically been incremental?

We have been working on broadening our new technology for years. We haven't skipped the incremental, intervening steps; we just haven't put them on the market.

People talk about “Green” initiatives now, but tomorrow may be different, so why risk unproven technology?

Marketing fads come and go, but the days of unlimited, inexpensive electricity are behind us. Anyway, smoother operation, better transient response, and higher efficiency have value in and of themselves.

Reliability is essential in this business. How do you keep your systems from self-destructing?

Conventional PWM SMPCs are intrinsically unstable. Our controls are intrinsically stable.

What do you mean by "conventional PWM SMPCs are intrinsically unstable”?

The output of any SMPC has a capacitive output filter to smooth the switching effects for achieving a steady voltage. That filter causes lag in the response of the control loop, which causes delayed response to dynamic conditions. The compensation which must be added to get acceptable transient response requires a delicate balancing act.

How do your converters avoid the intrinsic instability of conventional PWM SMPCs?

We can remove the output filter lag through prediction. We don't wait for the output to overshoot or undershoot, and then try to correct. Instead, we can predictively and accurately aim at a moving target.

Your techniques look nice in SPICE simulation, but the real-world is a lot messier than that. Will these devices really work?

The two inventors have 82 years combined experience in electronics, most of that in a pre-SPICE environment. One of the inventors is also a published SPICE expert. We go back and forth between the bench and simulation. We have already built and successfully tested several generations of our new converters and building blocks.

Why haven't these things been done before?

There are two main reasons. First, only recently have switch components improved to point that our approach is practical. Second, SPICE capable enough to simulate SMPCs is recent, and we exercise SPICE beyond its usual limits. Mistakes on the bench yield smoke, which tends to discourage innovation.

New ideas for SMPCs that look good at low voltage and current can come apart for power converters dealing with serious amounts of power. Agreed?

Yes, true. Our second and third generation demonstration systems were 1000 watt systems. That ambitious approach assures us of the practicality of CogniPower designs.

How can you monitor high currents, switched at high speed?

Faster switching of more current pushes conventional high-side current monitoring from troublesome to impossible. Our sixth patent application provides a simple, cost-effective method for monitoring the small voltage developed across a sense resistor in the presence of relatively large and fast common mode voltage swings. Our high-side current monitoring goes all the way to DC and is more than fast enough to keep up with the fastest SMPCs.

Hasn’t stray inductance always stalled efforts to push the limits of SMPCs?

Controlling unwanted inductance is crucial to success. We have proprietary ways to minimize stray inductance.

We've seen the published patents. Aren't your claims far too broad and simple to be seriously considered?

One patent has already issued, with broad claims intact. We have a high degree of confidence that our other broad claims will hold up under scrutiny.