DC-voltage doubler reaches 96% power efficiency
Analog switches let you charge a capacitor from two directions
Marián Štofka, Slovak University of Technology, Bratislava, Slovakia; Edited by Martin Rowe and Fran Granville -- EDN, July 15, 2010
| View as PDF |
The voltage-doubler circuit in Figure 1 can convert 2.5V dc to 5V dc or 1.8V to 3.3V. Most voltage doublers use an inductor, but this circuit doesn’t need one. The circuit uses a capacitor, C, by charging it through serially connected switches. The charge switches let capacitor C charge, and the discharge switches are open. In the subsequent discharging phase, the charge switches are off, and the discharge switches close. The two discharge switches now connect capacitor C between the source of the input voltage, VS, and the output capacitor, COUT. This connection scheme lets the applied voltages combine. Thus, the voltage at the output terminal has a value close to 2VS.
The two phases of operation repeat periodically at frequency f, which clock generator IC2 determines. The duty cycle is about 50%, but the value isn’t all that critical. One half of the Analog Devices highperformance ADG888 analog multiswitch provides the switching. The IC’s two halves have independent control, so the other half occasionally shorts Rp, the 10Ω inrush-current-limiting resistor, which protects the charge switches from an initial overcurrent. That current occurs after power-on, before the output voltage reaches the predetermined percentage of the output’s full voltage.
| Read More Design Ideas |
During soft start, the output voltage can’t reach the threshold level for loads below a certain value. Consequently, the circuit remains in soft-start mode. The minimum value of RL, which activates the protective subcircuit, is RL≤m2×(α/(1−α))×Rp, where the multiplication factor m=(VOUT/VIN) and α is a fraction of VOUT, at which the soft start turns off. For m=2, α=0.8, and Rp=10Ω, RL is 160Ω. Thus, loads of 160Ω or less will overload the circuit if you connect them to the circuit’s output before power-on. IC2 and IC3 get their power from the input supply. IC1, however, switches voltages of as much as 2VIN, and its VDD1 supply-voltage pin must remain at the same level. An analog OR switch comprising Schottky barrier diodes D1 and D2 provides that voltage. The higher of the input or output voltages appears at the VDD1 pin of IC1. The high levels of output voltages for both IC2 and IC3 suffice for control of IC1 because the ADG888’s data sheet allows a 0.36VDD1 value for the high value at the control inputs. The circuit has been tested at an input voltage of 2.386V, RL of 178.46Ω, a frequency of 200 kHz, a supply voltage of 2.377V, an input supply current of 51.285 mA, and an output voltage of 4.588V. Evaluating these data gives a multiplication factor of 1.929 and power efficiency of 96.39%.
This power efficiency remains more than 96% for frequencies of 150 to 350 kHz. The 9-mV drop at the switchshorted Rp at the given input current indicates that the on-resistance of the paralleled switches has a value of approximately 0.175Ω.
Talkback
-
It should be noted that the TL7660 or ICL7660 are very old work-horses of the industry. I have some designes dating back over 10 years using this or equivalent technology chips.
So even though a new part comes along with improved technolgy, that doesn't mean it is better.
The NE555 may be old and there are plenty of CMOS versions now. But we can't argue with something that is simple and works well enough.
Price is often the most important part of the design parameters when selecting a component!
Stanley Hirsh - 2010-27-7 10:26:59 PDT -
I would like to add the following:
I was so impressed by the SC1462 IC, that I did not compare it with the DI thoroughly.
Today I did it for input voltage value of 1.5 V.
The doubler diagrammed in the DI exhibits a power efficiency (eta)of 96.54% at Io=16mA, fosc=405 kHz.
For the SC1462 can be found in the Datasheet: eta-95.6% @ Io=6mA; eta=92.1% @ Io=12mA; eta=91.2% @ Io=16mA.
Marian Stofka - 2010-25-7 07:31:50 PDT -
Thanks for pointing out the Semtech's SC1462.
This IC is an "other coffee", as we say here; as compared to TL7660. Let the TI people forgive me this expression.
If I needed to apply a DC voltage doubler, I bought the SC1462.
If I wanted to learn, what is inside, I bought the EDN, issue July 15, 2010.
Marian Stofka - 2010-24-7 08:04:28 PDT -
Follow-up: Here's an example of a single-chip solution that costs less and performs about as well: the Semtech SC1462. Typical efficiency is 96% at 2.5V in and about 25mA out, according to the data sheet. It comes in a SOT-23-6 package, which is only about 3mm square. External components required: three ceramic capacitors. Cost: $1.30 in single pieces.
Marc Stewart - 2010-24-7 02:49:08 PDT -
My point is not that the TI part can do the same thing as this circuit. My point is that there are plenty of readily-available monolithic devices that perform the same charge-pumping voltage doubling, for less than the cost of the ADG888 switch used in this circuit. Just go to Digi-Key's website and do a search for "charge pump" for over 1,600 examples. If getting 96% efficiency instead of 90% is worth the extra complexity to someone, then fine. But it would have been better if the article had at least mentioned that monolithic solutions were available, instead of implying that the only other option was an an inductor-based boost regulator.
Marc Stewart - 2010-24-7 02:32:33 PDT
