Franz_SchubertSchubert’s Symphony No. 8 in B minor, better known as the “Unfinished Symphony,” was started but never finished even though Schubert lived another six years. The completed portion of the Unfinished Symphony was extraordinarily compelling. Surely intellectual curiosity and passion for one’s own creativity would have beckoned ol’ Schubie to finish, but he didn’t. Why would anyone ever create something so compelling and leave it in an unfinished state? ... “Here are the keys to your new private jet, Mr. Trump. Now would you like engines installed on this aircraft?”

So given all this, I think I know what happened to Schubert. I think he landed a job at a semiconductor company designing high-voltage gate driver ICs. Don’t laugh. The circumstantial evidence is overwhelming, and I’m sure you will agree after I explain the situation. High-voltage gate drivers (HVIC) handle input voltages as high as 1200 VDC…very useful for myriad power-intensive applications. And they have built-in level shifters allowing control by low-voltage ICs, such as microcontrollers (MCUs.) That’s great! And many of them provide high side and low side drivers in a single package. OUTSTANDING! And some of them even have built-in bootstrap circuits. WOOO-HOOO! And…oh yeah, they’re not isolated.


Yeah you heard right. NOT isolated. That means you have to add isolation externally if you want it (and with 1200VDC max input, I think most folks will want it!). Incredible. Such a brilliant beginning, yet such an incomplete ending. How do we explain this? Easy: Schubert’s Symphony No. 8A -- only this time implemented in silicon instead of musical scales. “Schubert, if you’re out there and reading this, ya gotta finish what ya start. I’m startin’ to see a pattern here. “

OK, so what does this all mean for the designer? Well obviously, he or she must figure out a way to add safety isolation to the HVIC driver, which is best added at the driver input.

Here are the options: optocouplers (yuk!), gate drive transformers (urgh!), or CMOS digital isolators from Silicon Labs (I’ll explain why the Silicon Labs aspect is very important later in this blog). Let’s look at the candidates one-by-one. Optocouplers are usually too slow for switch mode power gate drive applications - faster optocouplers are available, but they tend to be cost-prohibitive. And if you read my last blog entry, I’m sure you agree that a root canal job is less painful than designing with optocouplers. Pass. Gate drive transformers eliminate the unfinished HVIC driver symphony, and replace it with relatively bulky, EMI emitting, power-inefficient transformers and associated discrete component reset circuits. Sounds disgusting, but this solution is often more cost effective than the HVIC + optocoupler, and has no problem with speed. On the downside, transformers cannot pass DC or low frequencies, and they usually work best at duty cycles less than 50% unless the designer is up for lots of added reset external components. Pass.

The third choice, adding a Silicon Labs digital isolator to the HVIC gate driver input, is good but there is yet an even better way: ELIMINATE the unfinished HVIC gate driver symphony and replace it with a Silicon Labs ISOdriver, which provides high voltage drive AND isolation in a single package! A complete (read: “finished”) solution with benefits that overwhelm the wimpy aforementioned “competing” solutions. Let’s take a closer look.

Besides being non-isolated, and in spite of my earlier HVIC enthusiasm, there are two other HVIC driver deficiencies worth mentioning: 1) an unfortunate tendency toward parasitic latch-up when exposed to fast output dv/dt transients; and 2) relatively long propagation delay times. The latch-up problem is catastrophic, and is addressed by the manufacturer(s) with a max output dv/dt spec in the datasheet, which is kinda lame…

“Oh and one more thing Mr. Trump, don’t fly your new jet any faster than 60 miles an hour because the wings will latch themselves backwards and you will fall from the sky like a rock. Enjoy your new plane…” Hardly compelling.

The second issue, long propagation time, is not so much an issue in motor control applications because of the low system modulation frequencies used. But if the task at hand involves higher modulation frequencies (which are common in space-constrained systems), you’re better off using carrier pigeons to move electrons from one side of the driver to the other.

In light of this, and as you the reader have most likely already anticipated, Silicon Labs ISOdrivers have NONE of the aforementioned issues because it’s a FINISHED (read: complete) solution…it doesn’t apologize for its lightning-fast 60nS propagation time, its 5kV max isolation rating, it’s single package embodiment, its 25kV per microsecond common mode transient immunity or its remarkably frugal external BOM (as few as 2 VDD bypass capacitors). One package, one complete solution…

“Schubert, are you getting all of this, son?”

And finally, here is the all-import Silicon Labs part as promised earlier: while there are silicon competitors to Silicon Labs’ isolation products, their products have latched outputs, and are highly prone to erroneous operation in noisy power system environments. Silicon is the first and only digital isolator supplier offering continuous wave, on/off keyed isolation products that ensure the device output state unconditionally matches that of the input, as long as both sides of the device are powered. No data errors, no waveform bobbles, no miscues, and no excuses…a finished isolated gate driver masterpiece that will keep your system humming a happy tune for years to come.