Switcher boost circuit

A study note on an almost analogue switching boost power supply with a flip-flop clock and biasing via a flip-flop R arm ) with a possibly custom error definition and a compensator proposal that should make the response transient load independent

analog_switcherDownload
Standard

Water maker

watermaker

Image

This thread starts with inverters, the massive humming things which appear between the rows of solar panels up atop the Carrizo plane where FirstSolar and SunPower are operating 250 and 550 MW solar power plants. The humming things in the case of the SunPower plant are the 760cp SMA inverters which handle up to 0.76 MW of power, take a large DC voltage from a series linked panels and turn it into an oscillating signal at the frequency of the grid ( 60 Hz ). The need for an AC conversion is actually not all that obvious, as recent ABB transmission lines show. For long distance transmission one would use HVDC as they do to get hydropower from the Amazon basin to Sao Paulo. The Carrizo plane power, I believe, is carries to the PNM consumer without any HVDC segments.

SMA inverter discussion SMA 760cp

The 760cp SMA is an ‘H5’ bridge, but for this thread I picked a circuit that self-oscillates: a strange thing. It is both a controller and a driver, runs off a battery, or a DC supply and looks simple: two transistors, two caps, and four resistors. Essentially mirror halfs for a 50 percent duty cycle signal, or suitably assymetrized to produce any value of the duty cycle. This comes in handy in speed control application. Birds eye view: the half the flows current turns the other half on when it goes off. How? The low valued resistor biases the BJT ( sets the current flowing through it ), the high value resistor combined with the capacitor set a 1st order transient voltage which when low keeps the mirror half shut off and when just at the right voltage value turns the mirror BJT on. The rushing current in the mirror brange drops the collector voltage and the voltage at the cap, high resistor node ( because the charge in that capacitor will not drain out immediately ) and this in turn flips the switch. And so it goes.

Now a bit more quantitatively:

V R4-Q1 V R3-C2
3. 0.8
0.2 -2.0

Table 1: Voltages at nodes R4-Q1 and R3-C2 as Q1 switches from OFF to ON

RC transient duration for the Q1 side control R3-C2 is as follows:

\displaystyle \begin{array}{rcl} -2.0 + ( 3 - (-2.0) ) ( 1 - \exp{ \bigg ( -\frac{t_{ON}}{\tau} \bigg ) } ) = 0.8 \\ t_{ON} = -R_3 C_2 \ln \bigg ( 1 - \frac{ 2.0 + 0.8 }{ 3 - (-2.0) } \bigg ) \end{array}

This give 82 ms for the Q2 time. The measure value is more like 73 ms. When ON Q2 appears to operate just barely in reverse bias to let the current flow. The R1-Q2 voltage steadies out at 0.2 V ( and not say 1.4 as it would in forward bias)

A visit to the TechShop on Howard in San Francisco gets me access to a National Instruments workbench: an all-in-one scope/power supply/multi-meter controlled via a Virtual Workbench on a laptop. End up buying wrong caps at Leishman’s on University in Berk and have to scratch my head as to how to get anything to go. The circuit that actually begins to oscillate on the table top is the one in the figure. Not symmetric, not 50 percent duty cycle. LTSpice ( free to download here: http: ) does a good job reproducing the control and the collector voltage relative to ground. Control in this case is the 100 micro F, 1kOhm side of the circuit and the collector voltage is on the other side.

So, is this circuit in the large humming inverters on the Carrizo plane: not sure. Can it handle large currents? Yes, if biased well. Can one bring the line voltage up? Yes, possibly. More thinking / searching required. As a fountain that keeps on giving, the circuit is it. Is there a way to drive an LED ribbon using its variant? Collect N half circuits as above and daisy chain in a ring. This actually would work as a ring oscillator …

Standard