LCS708 variable output voltage and current

4 posts / 0 new

We need to build a LLC power supply with variable output voltage and/or current.

The converter has 2 operating modes: Constant voltage, Constant current mode.

We plan to use LCS708 @400W

Our solution is control Feedback voltage through an external device (microcontroller)

We may use digital potentiometers or digital programmable current source through a microcontroller.

The control loop of the power supply is / can be very slow. We do not need fast regulation.

What may be your recommendation?

We also need to paralel (combine the outputs of ) 2 LCS708 to reach 800W. (Each controller has its own current shunt and it is regulated (by regulating the resonant tank frequency) seperately. A host controller combines the total current to the set point. by communicating the slaves) Would you recommend it? Any issues we should take care of?

Best regards.

Hi rxpu,

Thank you for your interest in using Power Integrations' part.
I don't see any problem with your approach as long as your tank design(should be low Kratio Lmag/Lres) can cater the variable voltage range of your output. Then on output sharing it should also be okay since you have master and slave configuration and not passive sharing which probably go into output current hunting. My only concern is start-up with 800W load. If they are not synchronized enough, one might go to overload protection before the second one can catch up but since you have master-slave with microcontroller, I think you can address it easily. Thank you.

Thank you for the recommendations.
Each supply has its own current shunt resistor and own closed loop control.
Even at voltage control mode there will be a constant current/power limitation

I plan to design the power/current limitation by 2 factors:

- Small and limited power tranformer for each converter. Effective area of the core is limited to 500W for a 400W supply.(Even at secondary short there will be a limitation by the isolating transformer. )

- Current upper limit control even in voltage mode

- Very slow current starting point and long ramp to set current. (up to 500ms).

According to your recommendation:

if Kratio=smaller

1- Lmag should be as small as possible. (Would you recommend using planar E cores to lower Lmag?)

2- Lres should be chosen as big as possible. (this is a tradeoff between swithing frequency and Lres. Is it true?)

Would you recommend 250khz or 180khz (higher Lres) or 120kHZ (much higher Lres) nominal resonance frequency to satisfy the above requirements and reach optimum robustness for current control/limitation. The key point is the current limitation capability and robustness of the converter.

What would you recommend for the interface between the feedback pin of LCS708 and the microcontroller? A current source or a digital potantiometer?. The microcontroller measures the secondary output shunt resistance and output voltage of the correspanding converter and closes the loop and sets the feedback setpoint.

Best regards.

Hi rxpu,

According to your recommendation:
if Kratio=smaller
1- Lmag should be as small as possible. (Would you recommend using planar E cores to lower Lmag?)
> E cores is fine. To lower Lmag, just grind the center leg to meet your desired inductance.

2- Lres should be chosen as big as possible. (this is a tradeoff between swithing frequency and Lres. Is it true?)
> Yes Lres is choosen with Cres for your desired resonant and switching frequency. For a given desired resonant frequency you can have different combinations of Cres and Lres. There's always pros and cons. Increasing Cres will increase your operating range/higher gain but Lres have to reduce together with Lmag to maintain Kratio but higher Lmag means higher circulating current which is less efficient.

In frequency selection, it's up to your requirement. There's always pros and cons. Increasing the frequency will make your magnetics smaller, less ripple/less output cap but less efficient. So my recommendation is chose the biggest core that can meet your form factor then design with that core at lowest frequency you can possibly operate with 250mT flux density.

A current source feedback is typically used like an TL431 feedback but it's up to which is easier for you.