Post Go back to editing

LTM4653 as a variable output voltage power supply


I’m currently designing a variable output voltage buck convert for driving a baffle heater (my company works on EGSE for space instruments).

I need to generate a variable voltage between ~5V to ~50V for closing thermal loops on resistive elements (25 Ohm minimum resistance).

My idea is to use the LTM4653 buck module which have good specification on paper (good Vin and Vout window).

To control the output voltage I was thinking about two solutions:

  • Using a 5V DAC along with a current mirror that would drain the output current from the Iset pin by removing it from the resistor that sets the output voltage. So when the DAC out is max (5V) the output voltage is minimum and vice versa.
  • Connect 5V DAC directly to Iset pin (through buffer) and connect the Vosns pin not directly to Vout (as I see in all circuit implementation of LTM4653) but at a midpoit of a 1/10 voltage divider.

Would you recommend this kind of approach and this component for this application? Do you have any suggestion? Has anyone tried to use the component in this way?

For now LTSpice simulation shown quite good results. I think the tricky thing is finding the right switching frequency to get the best compromise between ripple and efficiency across the entire output voltage range. Its also importat to mantain good stabilty margin aswell.

Many thanks and happy new year to all of you.

Francesco - Saitec s.r.l.

  • The ISET pin can be driven by an external supply capable of sinking 50µA, so I would suggest the 2nd option.

  • Hi Francesco,

    Can you tell us more info about the reason you want 5V-50V output and the meaning of closing thermal loop? If you can, please provide project info and estimated annual usage.

  • The ultimate goal of this application is to adjust the temperature of the baffles and focal planes of space optical sensors. As we know, such sensors perform best in certain temperature ranges away from absolute zero. The thermal loop is closed by the reading of a temperature sensor. The data, properly processed, modify the output of the DC/DC with a rate ranging from 100ms to 1s. I want to specify that the environment of use of this solution is the laboratory; only the sensor will be inside the thermal vacuum chamber.

    As you may imagine, given the final application, the annual production rate will be no more than a few dozen.

    Thanks for the answer. I thought the module was designed with the constraint that the VOSNS pin be connected to VOUT. So introducing an attenuation of a factor of 10 on the feedback will not affect the stability of the loop. Right?
    Would you recommend the ability to configure the switching frequency while the DC/DC is running? Maybe using an RDAC along with a resistive network. Thanks.

  • I don't anticipate a problem with the feedback divider, but it is not the normal configuration. External loop compensation may be needed. I think programming the frequency on-the-fly should be ok; I suggest avoiding sudden changes.