Constant off time controller for flyback control?

Hello,

I wish to do a flyback (non isolated) with vout = 50V, vin = 4v8, and 3W power.

I don’t want to need slope compensation, so I will just use the HV9910B pwm controller from supertex (in constant off time mode)

HV9910B DATASHEET
**broken link removed**

I intend to regulate the Vout to 50V by using an external error amplifier and making it drive the “LD” analog dimming pin of the HV9910B.

…..do you think that this sounds feasible?

Sounds like it should work fine, but why not use a normal PWM controller with an error amplifier built in?

Your 4.8V supply must be able to provide 625mA average, in order for you to obtain 3W of power.

If your duty cycle is 50 percent that comes to 1.25 A instantaneous. Add 15 percent to make up for losses.

Furthermore if you want 4.8V to push 1.25 A through the primary circuit, then my simulation shows you must trim resistance down to a fraction of an ohm. There will also be added impedance due to transformer reactance.

Thanks Mtwieg

why not use a normal PWM controller with an error amplifier built in?

Click to expand...

.....no constant off time controllers exist with error amplifiers built in.

BradTheRad:
Thanks, yes , i also want the current sense reference voltage to be low so that the primary side sense resistor is not eating up too much voltage on the primary side

Only problem with HV9910B is the gate drive is only 160mA peak

treez said:

.....no constant off time controllers exist with error amplifiers built in.

Click to expand...

Oh yes they do, though most of them are aimed at step down applications. But many general purpose constant frequency controllers can be made to operate in constant off time mode with a bit of external circuitry. Check near the end of this white paper: https://www.venable.biz/tp-05.pdf

Thanks Mtwieg,

yes, the great UC384X range is the one that can be made constant off time with the addition of the comparator as you point out in fig14.

...but it illustrates the point that you need an extra component to do it (the comparator)......chips just dont exist with constant off time inbuilt and an error amp.

im not knocking your article...that really is good in fig14...and maybe the comparator is better than the bjt and resistor that you need for slope compensation.

...but wouldnt that comparator need some hysteresis?....and then thats more components needed.

the blessed UC38C4X range unfortunatley have a 1V current sense reference voltage, which means losses and is not good when your vin is low.



I am amazed that constant off time is not more common.....after all, who wants slope compensation?

treez said:

...but it illustrates the point that you need an extra component to do it (the comparator)......

Click to expand...

I've implemented the same thing using a tiny FET in place of the comparator as well.
[quotes]chips just dont exist with constant off time inbuilt and an error amp.[/quote]Again, they do exist, but they're all specialized, most for LED drive (error amp is just a comparator) or they're for fancy step down drivers (like controllers for CPU power supplies).

the blessed UC38C4X range unfortunatley have a 1V current sense reference voltage, which means losses and is not good when your vin is low.

Click to expand...

The current sense limit is controlled by the error amp, and the absolute current limit can be configured externally, it's not fixed to 1V?

I am amazed that constant off time is not more common.....after all, who wants slope compensation?

Click to expand...

I feel the same. I suppose most engineers are just more comfortable with constant f converters. Though it's not like the math is any more difficult or anything...

Mtwieg,

The current sense limit is controlled by the error amp, and the absolute current limit can be configured externally, it's not fixed to 1V?

Click to expand...

...yes , forgive my slackness of description here...the endstop is 1V, but as you rightly say, ultimately the error amp controls it.

.....but the fact is that making the endstop lower than 1V needs a couple of extra components, and its not good when you get component crowding round a pwm controller.