Well it has been quite a while since I posted anything on this subject. What with the house move and building a new home, I haven’t had a lot of time, and finding things is rather difficult.
Testing the first prototype showed a couple of straight out errors. However the processor interface to the output section did not work as well as I had hoped, and while I could have had a calibration table, I don’t want to do that, as it adds a lot to calibration.
I received an email from Maxim promoting a new A/D and D/A programmable chip in their PIXI series. Part of the documentation showed an application where a processor is controlling the power supply to an RF amplifier. Their concept is to set up a fixed output boost regulator, then feed a small current into the feedback input to reduce the output voltage.
Depending on the boost regulator the feedback voltage is often in the range from 1 volt to 3 or 4 volts, which is well within the range of a simple micro voltage capability. The way this technique works is if we want say a 25 Volt output maximum, and say the feedback voltage is 2.5 volts, then there has to be a 10 to 1 voltage divider. If we choose the resistor from feedback input to 0V as 100K, then the other resistor will be 900K. If we also assume that the current into the feedback pin is negligible, then the current through the 100K resistor will be 25 microamps. In this case the minimum output voltage would be 2.5 Volts, when the current through the 900K resistor is reduced to 0 microamps. If the micro controller has the capacity to generate a 0 to 4.5 volt output then it must feed 0 microamps through a resistor to the feedback pin (D/A output set to 2.5 Volts) to bring the output down to 2.5 Volts the D/A will be set to say 4.5 volts and to give a current of 25 microamps we would need a resistor of (4.5V – 2.5V)/25uA = 80K. This gives the output range of 2.5V to 25V for an input variation from 2V to 4.5V range.
The astute out there will say that the output could go higher if we reduce the D/A voltage to 0V then effectively we have 100k in parallel with 80K which means there needs to be more than 50uA from the regulator output. This means that in theory the output voltage would rise to 53.125 volts, probably beyond the design capability. The actual value of the 80K resistor nominated needs to be calculated to work with the D/A you have available.
The nice feature of the Maxim Pixi chip is that it has a 12 bit D/A converter with a maximum output of 10 volts. This gives you a great resolution for the output voltage. A word of warning however, the output voltage of the regulator will be indeterminate until the processor has set the D/A. Therefore you need a means to hold the output of the regulator off until the voltage has settled.
I will leave you to workout the fine detail of all of this. I am about to redo the schematic and PCB for the new design.