SPV1040 - using LX to generate a boost supply

[Andrew Lohmann]

This is to turn on MOSFETs that operate above 5.2V supply of the Boost regulator.  I have followed a forward converter guidance roughly in one of the application notes, but I have not used a transformer but placed a high value resistor in the voltage doubler output.  How much can I reduce the resistor values R6 and R69 without disrupt the operating of the SPV1040?  Reducing them to 1 to 10K would be good.  The input is limited to 4.5V so that a diode is not required between LX and Vout. 

The application is a generator start-up circuit for vintage/veteran car dynamo.

An Engineers Perspective: Vintage and Veteran Car dynamo power management

Thanks

[Peter BENSCH · Accepted solution]

By application note, you probably mean the DN0012 Boost converter with MPPT drives isolated solar panel loads?

If the voltage doubler works together with the SPV1040, you should be able to reduce the resistors to 2k. However, the time constant resulting from C123 (100nF) and the resistors R69+R6 is then somewhat small, so I expect at least 10% less output voltage behind the doubler. If necessary, you can compensate for this by increasing C123 to 220nF or even 330nF.

But instead of increasing the current through the resistors, you might want to consider reducing the demand for it. The LM431 typically requires at least 400µA and a maximum of 1mA for stable operation. Since you are only controlling MOSFETs with it, you can use the TLVH431 instead, which can only be used up to the acceptable maximum of 18V, but only requires a maximum of 200µA, which allows for resistors of (in total) 50k without any problems.

If you have selected R69 and R6 with C124 and C47 due to the different time constants, you could also make both resistors the same value (e.g. 100k) and adjust the capacitors accordingly (10nF and 1µF respectively for 100k).

Hope that helps?

Regards
/Peter

[Andrew Lohmann]

Thanks Peter,

LMV431 is better than the LM431 at 55uA minimum.  Otherwise, your answer 2K minimum works out very well and 100nF still filters the 100KHz at LX well enough.