# BCA modulator design?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

#### mzzj

I am looking for some hints for BCA modulator, my purpose is to run 2 half-bridges with 90-degree phase shift.

Did I get this 90 degree phase-shift stuff correctly, that triangle waves shoud be generated with 90 degree phase shift to each other?

http://users.rsise.anu.edu.au/~felix/iar/Projects/Thunderball/thunderball_schematic_v0.1.pdf

Like this guy, but instead of inverting that triangle with opamp i would run 2 triangle generators with 90-degree phase shift.

What i had in my mind was running CD4017 from osc, and connecting 1st and 5th output to 4027 to divide it by 2 and obtain 50% duty cycle.
Run these 90deg shifted square waves trough some robust buffer to bang them rail-to-rail and connect the buffer output to opamp integrator(via big cap to make it symmetrical in amplitudes)

Is LM393 comparator toooo slow after all this hassle, even if i use large input signals for it? (something like 17vpp triangle in order to overdrive poor comparator for faster response?)

#### Jaka Racman

Hi,

for one halfbridge you need two PWM signals displaced 180 degrees. You can do that by inverting triangle or easier by inverting input signal.

As I understand your description, you will get two triangles displaced 180 degrees. For 90 degrees you would need 4022 and divide outputs 1 and 3. You were probably thinking about 180 degrees anyway.

But you can use 90 degrees displaced triangles for running each halfbridge and reduce ripple even further. For modulating you would also need both inverted and noninverted input signal. LM393 is to slow, even LM319 is not optimal. Remember you are running open loop.

Maybe this article will explain things a little further.

Best regards,

Jaka racman

#### mzzj

Jaka Racman said:
Hi,

for one halfbridge you need two PWM signals displaced 180 degrees. You can do that by inverting triangle or easier by inverting input signal.

As I understand your description, you will get two triangles displaced 180 degrees. For 90 degrees you would need 4022 and divide outputs 1 and 3. You were probably thinking about 180 degrees anyway.

But you can use 90 degrees displaced triangles for running each halfbridge and reduce ripple even further. For modulating you would also need both inverted and noninverted input signal. LM393 is to slow, even LM319 is not optimal. Remember you are running open loop.

Maybe this article will explain things a little further.

Best regards,

Jaka racman
Yeah, my intention was to use 90 degrees shifted triangles for each half-bridge and invert input signal.

Maybe my circuit description(or enlisgh) was as informative as that crownaudio pdf

#### mzzj

Any ideas for triangle wave generator?

I tried opamp integrator with tl072 and supprise its too slow And besides that I did'nt have any suitable buffers in my dravers to drive square wave before integrator.

4017+4027 works just like a charm, i made mistake in earlier posts, 1st and 6th outputs are the ones to connect to flipflops.(Not 1st and 5th, this gives 72 degree shift)

Nice looking output up to 10Mhz clocks, 5Mhz input clock gives 250khz output. Maybe run this with 2Mhz crystal osc at first, 100Khz should be more than adequate for subwoofer use.

#### mzzj

Jaka Racman said:
Hi,

But you can use 90 degrees displaced triangles for running each halfbridge and reduce ripple even further. For modulating you would also need both inverted and noninverted input signal. LM393 is to slow, even LM319 is not optimal. Remember you are running open loop.

Maybe this article will explain things a little further.

Best regards,

Jaka racman
I just ordered some faster comparators from Elfa, 8ns ones I belive. LM393 seemed to be 10 times slower what I had in my mind. Just finished wrapping first inductor for the power stage , 2x20T center-tapped on EE65 core, 150uH, Isat=25A...

#### Jaka Racman

Hi,

good decision. I had good experience with LT1394. Only problem with fast comparators is limited input range, around +/- 3.5V for the mentioned one. Better is LT1712 dual comparator with +/-6V maximum input range. Your idea to have a large triangle is a good one. There is a problem with analog multiphase designs around 50% duty cycle. Switching transient of one phase tends to influence the other's phase comparator. Be careful with the layout.

You also need decent opamp for triangle integrator. Gain bandwidth product over 100MHz is recommended. I had good results with OPA650 but also with LM359 Norton amplifier.

Best regards,

Jaka Racman

#### mzzj

Jaka Racman said:
Hi,

good decision. I had good experience with LT1394. Only problem with fast comparators is limited input range, around +/- 3.5V for the mentioned one. Better is LT1712 dual comparator with +/-6V maximum input range. Your idea to have a large triangle is a good one. There is a problem with analog multiphase designs around 50% duty cycle. Switching transient of one phase tends to influence the other's phase comparator. Be careful with the layout.

You also need decent opamp for triangle integrator. Gain bandwidth product over 100MHz is recommended. I had good results with OPA650 but also with LM359 Norton amplifier.

Best regards,

Jaka Racman
Thanks for tip, now I have one idea more whats going wrong when it goes to chaos

Yeah, need fast opamp. I have sh*tload of slower opamps at my hands, i can get pretty decent triangle with them if I fiddle around with component values but i dotn like the idea that my integrator needs 20 trimpots

#### mzzj

I ended up using LM361 comparators, not state of the art but i hope it does. Prototype control board is HUGE, point-to-point wired with ground plane and it looks that i have to put ompamps and input level limiter on separate board as I have run out of copper clad

power stage pcb is 2-side etched board, and gate driver assemblies are on two separate veroboards, one for each bridge leg.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.