Getting Hysterical over the Hysteretic Buck Controller

National and TI both offer hysteretic control ICs, which have the very rapid response times required for modern CPU loads and have thus had a revival in popularity. However, the non-constant switching frequency and sensitivity to output capacitor characteristics are aggravating, and National seems to be pushing what they're calling 'emulated current mode' more now. For one example, see LM3475, LM3489, LM3485 for examples of hysteretic controllers. Paul Mathews

article:

hysteretic

on:

National and TI both offer hysteretic control ICs, which have the very rapid response times required for modern CPU loads and have thus had a revival in popularity. However, the non-constant switching frequency and sensitivity to output capacitor characteristics are aggravating, and National seems to be pushing what they're calling 'emulated current mode' more now. For one example, see LM3475, LM3489, LM3485 for examples of hysteretic controllers. Paul Mathews

------------- I came across the NatSemi numbers above.. Those IC's drive P FETS..Prefer N FET drive.

Checked out TI's hysteretic controllers and found the TI "monster" 28 pin TPS5210 hysteretic controller for powering microprocessors. Wayy to many goodies for my app.

I'll keep hammering on the TI and NatSemi sites.

Thanks..

interest in Hysteretic controllers..

hysteretic control ICs, which have

Of course, power supply controllers intended for CPUs are complex these days, so have high pin-count. The low-end types from National target low-cost and small size, so they drive P-FETs. There's usually no good reason to avoid P-FETs for low voltage applications these days. Tiny and inexpensive PFETs are avaiable from IR, Fairchild, Infineon, etc. The NFET alternative, for buck controllers, usually requires a charge pump to derive gate drive bias, which has its own costs. Again, if you do go the hysteretic route, take care to observe cautions about output cap characteristics. This may include your bypass caps, unless you isolate the load with an additional inductor. Paul Mathews

Of course, power supply controllers intended for CPUs are complex these days, so have high pin-count. The low-end types from National target low-cost and small size, so they drive P-FETs. There's usually no good reason to avoid P-FETs for low voltage applications these days. Tiny and inexpensive PFETs are avaiable from IR, Fairchild, Infineon, etc. The NFET alternative, for buck controllers, usually requires a charge pump to derive gate drive bias, which has its own costs. Again, if you do go the hysteretic route, take care to observe cautions about output cap characteristics. This may include your bypass caps, unless you isolate the load with an additional inductor. Paul Mathews

-------------- "Hide and show" text??? Ooops...don't know if I'm breaking some netiquette..

The N fet drive is preferred because the buck hysteretic controller is going to be forced into use as a Cuk controller.. (It's like looking for a Cuk in a hay stack of Bucks. ) The Vd in my app is about 300Vpeak. Id is about 2A. I'm no pro so...I do pad that and use an overated 500V Nmos just to be sure. If I recall P MOSFET gets pricy in this performance area.

Unfortunately, I'm starting to lean toward glueing a bunch a chips together to get the functions I need. Might be better than searching and forcing a chip into something it's not designed to do.