Solar to DC-DC Converter Spike

[Burned Traces]

Hello,

This is my first post, apologies if this has already been covered elsewhere, but my quick search didn't yield any results.
I would like some advice on the correct approach to the following:

I have a 24V solar panel that I would like to use to power a 5V radio board. This board is very sensitive to over voltages, which I found out the hard way.
I have a small no-name eBay Special DC-DC buck-boost converter that I am hoping to use to regulate the voltage swing of the panel throughout the day.
The converter works well until the input voltage drops down to around 3V and then its output spikes up to about 30V. When this occurs, I seem to be measuring about 10mA from the output.
The voltage in this raised condition will change if I adjust the value of the trimpot used to set the nominal output voltage of the DC-DC converter.
This doesn't seem like alot of power, but I would rather not have to replace the diodes on my radio board again.
I am currently testing this with a variable DC power supply and handheld multimeter.

I know I could buy a 5V panel and be done with it but couldn't resist the chance to learn something for once.

My questions:
Is this a little surge from the inductors in the converter collapsing? Would a standard TVS or Zener Diode be appropriate on the output?
Is this caused by some leakage within the cheap DC-DC converter that could be remedied with some filtering?
Since I am going to the trouble of sourcing discrete components, what would be an eloquent way of clamping this? It would be more interesting to set up a voltage-controlled switch rather than just dump the power through a zener.
Should I instead focus on clamping the input voltage? I am not seeing this rise in voltage when I turn off my DC Supply, only when I set it to hold at about 3V.


I don't need a complete schematic for any solutions, but if anyone has any suggestions on what a good approach would be or keywords to search for, I would very much appreciate it.

Thanks

-BT

[paulca]

Is the panel connected directly to the 5V Buck regulator?

What are the input voltage limits on the converter?  Often they have a minima.  3V sounds like it's well below a 5V regulator minima.  A "cheap" buck module will just go into "undefined" behaviour.  A "good" one might shut the output off.

It might help to put a charge controller and a ballast battery before the 5V regulator.  A basic charge controller will often has "Panel", "Battery", "Load" connections and will automatically shut off when the battery is flat.

[Burned Traces]

Hey Paulca, thanks for the reply!

Is the panel connected directly to the 5V Buck regulator?

Currently none of it is connected, mostly random bits I have laying around the garage.

What are the input voltage limits on the converter?  Often they have a minima.  3V sounds like it's well below a 5V regulator minima.  A "cheap" buck module will just go into "undefined" behaviour.  A "good" one might shut the output off.

I am not sure about this. I ordered it for very cheap years ago on eBay, so I don't have any of the original documentation. However, I am pretty certain this is not a Good module.

It might help to put a charge controller and a ballast battery before the 5V regulator.  A basic charge controller will often has "Panel", "Battery", "Load" connections and will automatically shut off when the battery is flat.

This is an interesting idea. I do have an unused MPPT of similar dubious quality, but I was hoping I could augment the DC converter with a few discrete parts, as a learning exercise.


As I research more, I am leaning towards using a 5V zener and a transistor to disconnect the output of the DC converter when the input voltage from the panel drops below 5V. Does that seem reasonable? Any features you would recommend beyond this?

[shapirus]

How much current does your radio board draw (at its normal 5V supply)?

[Burned Traces]

Shapirus,

According to the Datasheet https://docs.rakwireless.com/product-categories/wisblock/rak19007/datasheet/, it will draw up to 130mA when transmitting, but it also has an onboard charger for 18650 batteries that can draw 350mA.
So I think I would think about 500mA at maximum draw.

[shapirus]

Shapirus,

According to the Datasheet https://docs.rakwireless.com/product-categories/wisblock/rak19007/datasheet/, it will draw up to 130mA when transmitting, but it also has an onboard charger for 18650 batteries that can draw 350mA.
So I think I would think about 500mA at maximum draw.

I was going to suggest a linear regulator, but no, at this current it won't be a viable solution.

I'd look for a better buck-boost converter with a wide input voltage range and UVLO (input undervoltage protection) functionality. I've heard that some converters can go nuts when the input voltage drops too low, and it seems that yours is one of those.

Additionally, an input overvoltage protection before your radio board (using a fuse and a powerful TVS diode, for example) might also be desirable.

[Burned Traces]

I will keep an eye out for a different DC converter, it may be cheaper to buy a new one rather than paying shipping for parts

[mtwieg]

It's certainly not normal for a DCDC to increase its output voltage suddenly as its input voltage decreases. It's likely that the PWM controller circuitry simply can't operate properly at 3V, and its behavior is undefined. A good DCDC will have a well-defined undervoltage lockout (UVLO) on its input which forces it to shut down before it malfunctions.

You might also want to get a DCDC module which is only capable of stepping down voltage (a buck) rather than a buck-boost.

[Seekonk]

"The converter works well until the input voltage drops down to around 3V"

Not sure I understand this whole discussion.  With a 24V panel that has to be an extremely low light level.  Most converter boards have just enough input and output capacitance to keep them from oscillating.  A large value input capacitor will likely make it more stable and do a controlled shutdown. 

[BCRADIO]

Hello All:
I may have missed something somewhere, but how about an old LM7805 VR? The supporting circuitry is minimal and you can get an amp out of it.

BR