Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Input supply voltage regulation loop for solar peak power tracking

Status
Not open for further replies.

eddie_matos

New Member
Hi all,

This is my first "non-Arduino" electronics project. I'm a mechanical engineer and this is my first go at trying to build a more efficient, economic and compact circuit then what you end up with when using an Arduino.

I'm trying to build a DC-DC step-down converter circuit that will output 5V (0.5A) to a USB socket (for connecting low-power appliances such as mobile phones). The power supply is a 15V solar panel. I know there are tons of example circuits all over the place, and I have already racked up a few hours playing around with a step-down converter using the MC34063(ACN).

My main difficulty is when the available power from the solar panel dips below the load power demand (e.g. when a cloud comes along). When this happens the step-down converter forces the solar panel operating voltage way below the maximum power point voltage (into the variable-voltage zone), and it gets stuck there even after the cloud has passed. I know this is only a problem because I'm choosing to spec my panel operating voltage at 15V instead of 7V. I tried using a 7V panel very successfully: when a cloud came along the resulting decrease in the panel voltage also decreased the output voltage, thereby decreasing the power demand and preventing the input voltage from falling any lower. Unfortunately however, the 15V panel is a must.

I've looked into several maximum power point tracking solutions but none of them seem cheap enough for my needs. My favourite so far is the LT3652 from Linear Tech. You can use it to set a minimum input voltage (e.g. set at the maximum power voltage of the solar panel). So when a cloud comes along and the input voltage drops and hits this set minimum, the output voltage is decreased to lower the power demand and prevent the input voltage from dropping lower.

The problem is that the LT3652 is still pretty expensive and overkill for my needs (it does stuff like temperature compensation and is rated at 2A output), and I really want to try and keep the cost down. So I was basically hoping that someone knew of a cheap and cheerful way to trick the MC34063 into doing something similar. It doesn't need to track the optimum panel voltage to the nearest mV, it just has to prevent the panel voltage from going so low that it gets stuck over there and doesn't come back!

I'd be very grateful for any links to another post, or a website, that describes such a solution. Though if my problem is more than simply not knowing the right search terms, then feel free to tell me go back to my sodding Arduino or something ;)

Cheers,

Ed
 
Further to my post above...

I used a 555 timer for the first time the other day to build a low-voltage disconnect circuit to protect a lead acid battery from over-discharge (http://www.gorum.ca/lvdisc.html). Perhaps I could implement it here in a similar way to disconnect the charger when the solar panel voltage gets too low? Not very elegant, especially as the zener diode/shunt regulator used in the link above wastes too much power so perhaps I'd have to give the 555 its own linear voltage regulator, but it would probably do the trick, right? It would probably involve playing with the R-C values on the "THRESH" and "TRIG" inputs to make sure it switches off quickly enough, but not so fast that it causes lots of nasty noise. Perhaps a low R-C value for the switch-off and a high R-C value for the switch-on?

Any recommendations appreciated.
 
Maybe you could add a battery? If I understand your problem correctly your USB 5v output will be used to charge a phone etc, which requires a fairly high constant power, so when a cloud comes along the solar panel can't supply the SMPS input power (or load power).

I think your only options are to make sure the panel is so large that even with clouds etc it can still ALWAYS supply the minimum power, or add some type of battery etc to the system to cover the power shortfall during the cloud times.
 
As your SMPS IC you're using will have a feedback pin, you can just inject a little current there to increase the feedback voltage and reduce the output voltage. The attached circuit shows a way of doing this.

The zener holds the left transistor on (and right transistor off) while the input voltage is at a sufficient level. When the input drops, the left trans. turns off, allowing the right to turn on, thus increasing the feedback voltage and reducing the output voltage (and allowing the input voltage to creep back up).
 
Last edited:
I don't have any 12V zener diodes handy but I had some 7.5V ones from another project so I was able to test the concept using a benchtop power supply... and it worked a treat! Thank you very much dougy63 - this is exactly the kind of cheap solution I was looking for. I just have to order some 12V zeners now so I check out the transient response when powered by a solar panel. Am I right in suspecting that it will switch-on-and-off a lot as opposed to finding a stable point? But perhaps that's what the capacitor is for...
 
I'm glad it works. Depending on the values of the components, especially the capacitor and resistor on the right NPN, the SMPS can be made to turn off slowly. Yes, the capacitor was meant for that purpose, though it could be rearranged to have more effect.

I've redrawn the circuit to show where a resistor and capacitor may be located to cause the response to have a gradual onset; this should stop the SMPS turning on and off sporadically, though it may cause it to oscillate somewhat... you'll have to experiment (or calculate the appropriate values and response) and you should be able to find a range of values that will be stable.

As for tracking the maximum power point, that is only possible if the load will readily accept all the power that you can give it -- e.g. some types of batteries & loads. I'm not sure how your phone charger will handle a drooping voltage. If the load is amenable to accepting whatever current you give it, the maximum power point can be set by changing the zener voltage, or by feeding the left transistor via a potentiometer (with a series base zener for added accuracy).
 
Thanks for the follow up. I got my zener diodes today. Actually I decided to go for a 10V zener in the end, and tried it with my solar panel. The result was exactly as I needed. It sets a minimum voltage on the panel (at around 10.5V) and stops the regulator from getting "stuck" in a low-voltage stable point. So hooray! It works.

Unfortunately the way it does it is by switching the regulator on-and-off very quickly. The output voltage (initially set at 5V) doesn't go all the way down to zero, but it does fluctuate at a high frequency by about +/-0.75V. It'll be hard to tell how this affects my 5V appliances, but I guess the only way to find out is by trying! As long as it doesn't break the appliance (which is presumably very unlikely as the fluctuations are happening way below 5V), then your solution still solves my problem - finding a stable operating point was just a nice bonus!

The reason why the capacitors in your circuit couldn't reduce the frequency of the fluctuations was because I couldn't put them in at all! The solar panel would just go straight to its low-voltage stable point where it would get stuck - as if your circuit wasn't there at all. I suppose this must be because the capacitors inserted a delay, long enough for the system to move to its low-voltage stable point. Once it got stuck there, it appears the regulator could not pull itself out. Or maybe the input voltage was so low (around 1V) that the transistor logic wouldn't work properly. I'm afraid I don't understand enough about how the transistors and the SMPS circuitry works to tell you.

Because of this, I suppose the only way to stabilise the output (instead of having it switching on and off) would be to have some sort of proportional influence on the feedback pin voltage - rather than the discreet high/low influence carried out by the transistors. Any tips on how I might achieve this? I can just about imagine some sort of arrangement using a carefully calibrated potential divider working with the zener, but I can't quite figure out what the circuit would look like.

Ultimately however, this is exactly the cheap and cheerful solution I was looking for - anything else is just a bonus - so many thanks again for that!
 
I'm really glad that you have something that works for you :)

Can you post a quick sketch of the circuit & values you're using? There may be a simple fix to get the proportional control you're after.

The other way could be to increase the output capacitor value -- this will reduce the output ripple and although the SMPS will still work in burst mode, the larger cap will smooth out the voltage fluctuations.
 
Sorry for the late reply, its not been very sunny lately so I haven't had a chance to try increase the value of the capacitor on the output yet. I can how this would work, but I imagine the capacitor would have to be a bit of a whopper! I promise to try as soon as the sun comes out again.

I've attached your sketch (sans capacitor) with the values of the resistors I'm using. I played with the value of the 5k resistor a bit. At 10k, the output voltage would not go all the way down to zero during a low-input-voltage condition but finds a stable point at around 2V. This is probably as good as zero for me, given that the load is a 3.7V lithium battery, but I didn't want to risk this so I stuck to 5k - which does take it almost all the way down to 0V. I also tried a much lower resistance of 1k, but this increased the amplitude of the "switch-on-switch-off" fluctuations on the feedback pin to an alarming degree (purely a non-technical subjective opinion there!) so I concluded 5k would do better. Actually the amplitude of the fluctuations on the feedback pin don't appear to affect the fluctuations on the output voltage much - presumably something to do with there being a trade-off between switch-on-switch-off frequency, and amplitude - but I didn't look into this any deeper then that.

Untitled.png

Finally, a quick note on this controllers functionality, though I imagine all these things work pretty similarly: The controller tries to keep the voltage on the feedback pin to 1.25V. So the potential divider I'm using here keeps the output voltage to 5V, because 1.25*(1+3.6/1.2) = 5.
 
If you change the resistor values, you should be able to slow the transition from ON to OFF. From the left, try changing the second and third resistors to 100k and 1k respectively. This will increase the voltage that the circuit comes into effect, so you may need to play with the zener value (perhaps 6V2 or 5V6) to get the voltage back down.

A simulation is shown, but without the SMPS - it should give a basic guide to how the circuit is operating. The multiple traces in each colour are because the simulation was run for temperatures between 10-40 degrees Celsius (lower temperatures = higher voltages). Note that the left circuit "SlowResponse" has a 6V2 zener fitted.

You'll notice that the range of cutoff voltage has increased with the slow response - this is because it is more dependant on the current gain of the transistors (whereas the original circuit was basically just ON or OFF).
 

Attachments

  • inputreg.png
    inputreg.png
    11.5 KB · Views: 226
I've just replaced the resistors as suggested and did a high->low sweep on the voltage of the input power with a benchtop supply. As you predicted it appears to have created a far more "gradual" impact on the output voltage which should bode well as soon as I can test this with a solar panel (still waiting for that sun!).

I'm going to get a bunch of zener diodes to see which zener voltage works best. As you predicted, the output power does indeed start to get "throttled" at a much higher input voltage than in the "on-off" circuit we were trying earlier. My only concern is that, with a low zener voltage of around 5V, the minimum panel voltage set by the system might then be too low for the circuit to carry out its primary goal - to avoid the panel getting stuck in a low-voltage stable point. But I'm feeling optimistic that this won't be an issue and I'll be back in touch as soon as I've got those new zeners and the sun finally comes out!
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top