I am buliding a very small and low power "intermittent" sensor, powered by a tiny solar panel and an ultracapacitor. It works with a voltage from 1.9V to 3.3V, but the transmitted radio range greatly depends on it.

When I used an LDO voltage regulator I had to "target" a lower-than-necessary voltage for it to work properly. But having a constant voltage is not required in my case: I would better benefit from a higher power when the sun is shining a lot.

What are the tradeoff if I ditch the LDO regulator and I simply cap the maximum tension with a Zener diode as shown below?

I do not want the zener to draw energy while the circuit is idle, due to its leakage current. So I moved it "before" the usual schottky diode.

Also, I use a slightly higher zener value to compensate for the schottly voltage drop, so as to charge the cap to the maximum voltage that the circuit withstands.

Still, I have the weird feeling there is a drawback. Am I just too cautious?

Also, the tiny solar panel will probably not reach its dangerous 4V level when loaded. Actually, I could even wake the MCU up to keep a load when the tension is "too" high, but this is a riskier business I guess.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Looks ok. You can probably fiddle around with low current zeners (ON has some excellent), perhaps even two in series if you can't find an exact voltage and ditch the shottky to gain some voltage. \$\endgroup\$ – winny Jul 17 '16 at 10:26
  • \$\begingroup\$ Also, how much current are we taking about? Can you afford some quiescent current for a regulator? \$\endgroup\$ – winny Jul 17 '16 at 10:28
  • 2
    \$\begingroup\$ Set the LDO to 3.3V. If it can't meet that, it should just drop a small and constant voltage (maybe 0.2V) so giving 1.9V with 2.1V in. NOTE 3.6V zeners are notoriously imprecise. If you're using one, highball its value (often 5% tolerance), look at the added voltage from its (not very flat) slope at the solar cell's max current, subtract the minimum Vf of D1 at very low current (may be 0.1V) and check you're not exceeding your sensor's rating. \$\endgroup\$ – Brian Drummond Jul 17 '16 at 10:28
  • \$\begingroup\$ @winny I measured 7.7mA across a 98 ohm load and (under a harsh sun), 7V empty load. \$\endgroup\$ – MoonCactus Jul 17 '16 at 11:20
  • \$\begingroup\$ Watch the leakage on the ultracapacitor, btw. \$\endgroup\$ – pjc50 Jul 17 '16 at 11:23

You can use a precision shunt regulator to create a "zener " with a sharp voltage regulation commencement point. The TL431 (>= 2.5V ) or TLV431 (>= 1.25V) are low cost and will do what you want well.

Be aware that the voltage drop in the Schottky isolation diode D1 decreases with decreasing current and may be much less than the typically 0.3V seen at currents of 10's of mA. This can mean that a 4V PV panel voltage clamp might deliver almost 4V on the supercap. Allow enough tolerance to avoid damaging the capacitor.

Be aware that Schottky diodes can have very greatly increased leakage currents at elevated temperatures. If the diode was adjacent to the PV panel and in full sun this may be of relevance in some cases.

The use of diode D1 is wise in this case, but in some systems may not be strictly necessary. PV panels that are illuminated at very low light levels have low reverse currents and will not discharge a directly connected capacitor as rapidly as may be expected.

  • \$\begingroup\$ The bit about the panels not discharging a directly connected cap sounds interesting Russell - any links to any studies? \$\endgroup\$ – Andy aka Jul 17 '16 at 16:21
  • \$\begingroup\$ @Andyaka - carefully note the " ... as rapidly ..." :-). The only link I'll gives is the sign of Jonas ... - whoops, no, wrong question :-). While it's "semi obvious in retrospect" it's based on my own measurements. I've been involved in the production of semi-vast quantities of small portable devices and always used a series Schottky as of right. But I was doing some work with a 250W, 30 Volt panel and made specific measurements without the series diode. At very low light levels back current is in the same order as forward current optimally loaded. In the wee small hours in a dark yard ... \$\endgroup\$ – Russell McMahon Jul 17 '16 at 16:36
  • \$\begingroup\$ ... that's 'very small' . I actually used LED lanterns and CFL bulb lamps to test low level illumination effects. YMMV. However, as Vpanel remains close to Vmp if unloaded until light level falls to say 5% of max then I reverse will be small if Vmp is usefully more than Vbat in normal use (eg Std "12V" arrangement of 18V panels and 12V battery). | If battery is only load then as light falls Vpanel is clamped by battery if Voc > vbattery and only allows reverse flow at very low light levels. \$\endgroup\$ – Russell McMahon Jul 17 '16 at 16:39

The solar panel being small cannot probably produce amps and probably not even hundreds of mA so, do some reasearch on the panel and estimate how much current it might generate into a zener of 3V3 and forget about all that extra voltage dropped by D1, the series schottky.

If you can guarantee the panel won't produce more than say 70 mA you could use a precision and adjustable zener like the TL431.

If you are intent on using D1 to prevent back charge into the panel then you could supplement this with a MOSFET transistor controlled by the MCU to "short" out the diode and give less volt drop.

  • \$\begingroup\$ Yes, this panel certainly cannot produce more than a dozen of mA (I read 7.7mA with 1-100 ohms). D1 prevents back charge indeed. Are you suggesting to add an MCU-controlled MOSFET to avoid the voltage drop of D1, but only once the MCU runs and knows about VCC? Smart idea, too bad I run short on GPIOs already. \$\endgroup\$ – MoonCactus Jul 17 '16 at 11:29
  • \$\begingroup\$ There are ways of dual functioning IO pins - all you need is one that toggles continually during normal operation and you can steal that signal to make a DC signal that can drive the gate. \$\endgroup\$ – Andy aka Jul 17 '16 at 11:33
  • \$\begingroup\$ There are also some very low quiescent current shunt regulators from TI I believe but how much is "low" depends on what you think. \$\endgroup\$ – Andy aka Jul 17 '16 at 11:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.