UcD / LTSpice help

[fokker]

Greetings to all:

I am trying to simulate a UcD design in LTSpice. The design goal is to produce a class D that has the UcD discrete front-end and an IC gate driver output stage.

The whole thing is based on a small-ucd design I got here. I made sure that it worked with single-ended input, and then I took out the discrete driver stage and replaced with two lever shifters (to make them reference to the negative rail), and then added two voltage-controlled switches, SH and SL, implemented in two .model statements:

.model SH SW(Vt=0.5 Ron=0.01)
.model SL SW(Vt=0.5 Ron=0.01)

Enclosed is the schematic in LTSpice.

I know that the phase may not be correct but that's secondary for now as I cannot get the model to run. LTSpice says that it cannot find model SW.

Any guidance as to how I can get it running?

BTW, this is over stock LTSpice installation. Thanks.

[Genomerics]

Ah Ha....!

What has happened is that in order for you to get the switches the correct way up you have rotated them 180 degrees and then mirrored them.

In doing that the reference and value designators have swapped over. Both of your switches are still of type SW so you only need one model statement to cover both of them.

.model SW SW(Ron=10m Vt=0.5)

You might want to move the labels so they read

SH
SW

SL
SW

DNA

[fokker]

Thanks, genometrics. you are right on.

Here is the model in LTSpice.

[fokker]

and here is the output:

It as you can tell, has a DC offset to it. the high side is conducting a lot more offen than the low side.

Any cure?

also, the shifters are pretty crude and do not turn on and off as quickly. any suggestions to fix it would be greatly appreciated.

[fokker]

I think I found a solution.

the DC offset appears to be dependent on the shifter resistor combination. When used a 1k emitter / 2k collector combination, it produced a nice and DC-offset free sine wave.

It appears that it is possible to produce such an amp with UcD front-end, and IC gate driver back end. The advantage of it is the simplicity in house-keeping items (power supplies to the front-end IC in the IRAudAmp for example), high-speed, flexibility and low cost with the discrete solution.

I would still like to fix the level shifters, tho.

also, is it possible to re-arrange the feedback so that the signal is on the non-converting end? I think so.

[fokker]

here it is. with the feedback re-arranged.

[fokker]

after playing around with the input signal, the amp seems to have non-flat frequency response curve: the gain is 30x at 1k, 27x at 10k and <25x at 20k, and 12x at 30k.

is that normal?

[fokker]

btw, what are those 22p capacitors do?

[Genomerics]

You don't have anything to turn off your level shifters. Their bases are just driven by the cute stuff at the bottom that can turn them on but then there is nothing to take the charge out afterwards.

I've not studied it but it makes my head hurt. Anyway, try adding some resistors from the bases of your level shifters to the top rail... or figure out a way to add an extra two transistors in the emitters of the originals to turn off the opposite one.

UCD is sort of complicated because Bruno (or the agents) wrote his patent without telling someone not skilled in the art how to do it. Obviously if someone skilled in the art (or not so thick to figure it out) does it then they get stuffed anyway if they end up with a result that looks like what's in the patent.

Mind you..... Bruno might have been lucky..... Yeah, I know... I should go suck prunes.

That leaves the rest of us guessing at how to do it properly. Like me....... Interesting, time for a play.

DNA

[Genomerics]

Ooooooooh, I know.

Turn the circuit upside down.

DNA