Hysteretic self-oscillating amplifier with constant switching frequency

[sovadk]

Hi

I've made the following simulation model based on IRF's IRAUDAMP3 reference design. They use clock injection to the carrier in order to attain a constant switching frequency in a self-oscillating design. The attached picture shows that switching frequency os rock stable at 500kHz even with a high modulation of around 90%. At the same time the distortion is fairly low and barley present in the spectrum. The modulation actually distorts a lot because of the dirty clock injection, but the clever single opamp second order integrator (which from now on will be my key amplifer component) supresses these errors very well.

IRF uses a more weak clock injection to lock the switching frequency in a multi channel amplifier configuration, but i think this simulation shows that this approach can be used more aggressively and give some interesting results.

[sovadk]

The orcad project files.

By the way, is there some way to get privileges to attach pictures
larger than 1000x1000 and files larger than 100kb?

[sovadk]

Here is another hysteretic self-oscillating amplifier with constant switching freqyency (picture and orcad project is attached). It dosn't use clock injection but instead a discrete circuit that varies the hysteretic window as a nonlinear function of the absolute value of the output voltage.
The discrete circuit does cause some distortion (~0.1%) to the modulation process, but it can be fixed. The modulator only uses a simple pole in it's loop. By exchanging this with an integrator or even better a second order integrator, distortion can be supressed significantly.

The orcad project also contains an analog behaviour modeled version of this constant switching frequency amplifier. It nearly dosn't produce any distortion, therefore it's the discrete circuit that distorts the modulation.

[Jaka Racman]

Hi,

you could also try how PSSR drops with forced synchronisation of hysteresis amps. IMHO this is the main drawback of synchronising.

[sovadk]

PSSR drop, that sounds true. The synchronising causes the equivalent gain of the hysteretic comparator to drop. This decresses the bandwidth of the noise transfer function, where noise could be power supply noise (pumping, ringing, etc)..

[luka]

About those pic that are large...this site will allow you to have pics sizes up to 17", so that should be enough

[koolkid731]

My newer computer can't unpack rar file. Please post a zip file instead. Thanks.

[lucpes]

Originally posted by koolkid731
My newer computer can't unpack rar file. Please post a zip file instead. Thanks.

Download winrar... http://www.rarlab.com/download.htm

[sovadk]

Originally posted by koolkid731
My newer computer can't unpack rar file....

I did start out by makin a zip file, but it became larager than the 100kb limit and the rar version didn't.