Parametric voltage stabilizer circuit

[Alper özel]

Hello,

I (am supposed to) design mainboards for TV units. In our mainboard we need to add a parametric voltage stabilizer before panel supply.

Here is the situation. Our power supply may be varying between 11.8V and 14.2V and our Panel accepts maximum 13.2V (Absolute maximum). So, the ideal function we need is something like this:



The output PANEL_PWR must follow the input while input is below threshold 13.2V but if input exceeds 13.2V it should regulate it.

To do that I added this circuit:



The circuit in the red line is fixed standart power shutdown circuit of us and I can not make any changes on that. So I added the parametric stabilizer circuit in green line. The problem with this stabilizer is that it manages to trigger when the threshold is exceeded but it has enormous ripples:



between 10KHz to 23KHz, up to 4V ripple when input voltage hits to 14V and above.

So can you suggest me anything here? Ripple must be below 300 Vpp and threshold must be below 13.2V. At this point I have to warn you that we do not use opamps! This circuit can be formed by only transistors, references etc. simple components.

Thanks per advance.

 

[mtwieg]

Seems you've made an unstable circuit. The 4.7uF cap on the FET's gate will make response time much slower than your threshold detector. Varying the transfer function of the threshold detector may improve your results. Is adding hysteresis an option?

In the worst case you may have to make a completely independent circuit rather than using the FET in the red box.

 

[Alper özel]

Seems you've made an unstable circuit. The 4.7uF cap on the FET's gate will make response time much slower than your threshold detector. Varying the transfer function of the threshold detector may improve your results. Is adding hysteresis an option?

In the worst case you may have to make a completely independent circuit rather than using the FET in the red box.

Thanks for the advice. I also tested it without 4.7uF input capacitor and it hardly made a difference.

I also tried a shunt 1 nF capacitor with R537: it did not do much, just made so little.

Also I put 22 uF large output capacitor at output: did not make a difference.

 

[FvM]

I also thought the 4.7 µF capacitor might be the primary reason for slow reaction. But more generally speaking, you need a sufficient fast control loop and also a defined roll-off characteristic with sufficient gain margin for stability.

I would normally use an industry standard LDO with very low dropout voltage.

- - - Updated - - -

I guess APA432 is a TL432 clone. It has too high gain to work in series with two additional gain stages T1002B and T1001 without dedicated compensation. Usually RC compensation is connected between TL432 cathode and ref pin.