Is there any variable DC-DC converter suitable for pre-regulation?

[bloguetronica]

Hi prasimix,

Why so many resistors in that feedback circuitry? Any particular reason?

Kind regards, Samuel Lourenço

[prasimix]

There is just one voltage divider (R3 + R9) that define max. output voltage (here about 60 V) that is well over max. working range (43 V). Max. voltage can be set much lower but it's intentionally in my case much higher then Vin (i.e. 48 Vdc) to push controller into 100% duty cycle by disconnecting PNP tracker to bypass pre-regulator. That is not applicable in your case. You can eventually remove R9 or decrease its value.

[bloguetronica]

Thanks! And what about R8 in series with R7, is it to set the voltage difference higher?

Kind regards, Samuel Lourenço

[prasimix]

Hm, it's not really in series with R7: R8 is connected on Q2's base and set it's working point. Try to change its value in simulation and measure current thru it and check how it affect pre-regulator's output voltage (i.e. offset will increase for lower R8 or decrease for higher value).
You can also use this simulation to see how such kind of tracker cannot provide the same offset over the whole range, e.g. for Vout_sim=0 V offset is above 4.5 V while for Vout_sim=40 V offset drops below 2.5 V. Therefore you can expect higher dissipation on post-regulator's pass element for lower voltages.

[bloguetronica]

Hi prasimix,

Finally had some time to run some simulations. Without R9, the circuit behaves linearly with the offset pretty much constant. I've then adapted the circuit to my needs (also changed the PNP transistor to the one I'm using). I'll not use the same DC-DC converter, but I hope the behavior is the same. The offset is set to around 3.6V, which I think is the ideal compromise, as it give some headroom so the pass element can regulate.

Anyway, thanks for your excellent help. It was very valuable.

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

Finally I've assembled a board based on the TPS54233 from Texas Instruments. The offset measures 3.711V when the control is set to 16V, and measures 3.744 when the control is set to 0V. Unfortunately, I only have a board layout and not the schematic. But the control circuit is the same as used in the simulations, only with a different converter.

Unfortunately, I've damaged the board when assembling it, because I've used a 2N3906 in place of the MMBT3906. Lets say that the "water tower" fell and took two pads with it. Now I'm waiting for a bunch of MMBT3906s to arrive. I'll post pics when I have another board soldered.

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

Managed to recreate the schematic from the board layout. See the attached files.

Once I've a board assembled as it should, I'll post some pictures. Anyways, thanks for all the help.

Kind regards, Samuel Lourenço

[prasimix]

Hm, your PCB layout doesn't look optimal at all. Did you study all that app. notes and recommendations and check other people design? I'd try to reduce distance between parts as much as possible, reorganize Cin and Cout to have gnd connections much much closer, move R4 very close to IC1, increase power lines width, connect R3 to output terminal instead of  C6, etc.

[bloguetronica]

Hi prasimix,

Indeed the layout is not optimal at all. This was done quickly (I remember that it took half a day) and there are many things to improve (the diode could be a little closer to the IC, the inductor closer too, and many other things). Anyway, this is just to prove the concept, and actually the board works really well. The ESR imposed by the PCB seems not to be very relevant.

Anyway, the capacitors should have priority over R3 and the remaining control circuitry, which in this case wasn't done. Ideally, C6 and C7 should be right next to the inductor, and R3 should be after the second capacitor, and not in between them. R4 should not be close to IC1, as it doesn't have priority over C2, C34, C4 and R1.

Note that, in the TPS54233 datasheet, the feedback is taken from Vout after the filter capacitor, and loses priority over the remainder of the components. They even use a long feedback trace. The feedback network is closer to GND and near the IC, but de-prioritized over the compensation filter components. In a sense, these components should be near the IC, but as close as possible without moving any other components.

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

This is the schematic for the regulator module itself. The "CSENSE_UP" signal is fed back to the pre-regulator, so that its output is set to 3,7V above the voltage of that signal.

As you can see, the regulator module has voltage and current regulation, and over-temperature protection as well.

Anyway, I think it is better to leave this subject for another topic, should problems arise.

Kind regards, Samuel Lourenço