Workaround for Cheap MPPT Controller?

[MagicBlueSmoke]

Hi,

I have an urgent need to get a better solar power system going. I have a chance to get a 280W panel for dirt cheap, and I'll need a better controller than the Harbor Freight one I have.

While searching for one, I've run accross the MPT-7210A controller. It seems to be a real mppt device, the catch being that it's a boost converter based controller. Which makes it suitable for those almost unheard of situations where the storage banks are higher voltage than the panels used for charging them.

This poses a problem for me: The panel I have this chance to get produces 35.2Vpmax, 44.8Voc, 7.95Apmax, 8.33Asc. I currently have a single 35AH AGM-SLA battery from Harbor Freight. Getting a higher voltage battery bank isn't an option for me in the immediate future. With the higher power panel, I could at least run my laptop during daylight while still keeping my battery charged for lighting at night.

Than I had the "bright" idea (solar pun intended) of sticking a cheap buck DC-DC converter in between the panel and the MPT-7210A, reducing the voltage to around 12V to let the boost controller do its mppt thing to charge the battery. Would this be feasible? Would the buck converter let the variability of the panel's output just pass through to be handled efficiently by the boost converter?

(And if my idea is absurd... Please be gentle!    )

[NiHaoMike]

I think the use case is for 48V systems. That's a common voltage used by those who have expanded their systems beyond what a 12V system can practically do.

For your case, maybe build your own buck MPPT controller? Start with an existing buck converter design and add a signal to limit the duty cycle. (Consult the block diagram of the switching controller used to find out how.) Then find a way to use a microcontroller to perform MPPT.

Or go all the way to a digital power design with a dsPIC or similar directly controlling the buck converter. I'm actually working on one for a 65W, 18V panel I recently managed to get for cheap.

[MagicBlueSmoke]

Thanks for the suggestions. I'd love to take on a project like that. But my resources & situation don't allow for it any time soon.

I did have a facepalm moment though: Maybe I could just put a buck converter between the cheap mppt controller and my battery? The controller's output should be fairly stable, enough for a buck converter to handle?

I understand there's the split-pi converter topology, which is a boost-buck converter. So maybe I'm not completely bonkers? I know this wouldn't be a proper split-pi device, but do I have a chance of getting it to work without blowing up what little equipment I have without losing too much power? I'm just looking for something to get me by until I can put together the resources to do a proper job of it. I really do hate resorting to cobbling things together like this, but I'm in total bootstrap mode right now out of sheer necessity. I hope I'm not being a total pest.

[NiHaoMike]

If you try to cascade the two converters as described, it would probably just end up being unstable.

A buck converter with the duty cycle limited to a certain value (can generally be done by clamping the compensation pin voltage with a potentiometer to ground or similar) would likely work fairly well (would need manual adjustment with the help of a multimeter) and be easily upgraded to a proper MPPT setup later on. No need for the boost part.

[Someone]

Than I had the "bright" idea (solar pun intended) of sticking a cheap buck DC-DC converter in between the panel and the MPT-7210A, reducing the voltage to around 12V to let the boost controller do its mppt thing to charge the battery. Would this be feasible?

Battery charging/management relies on measuring the battery voltage (among other parameters) so you would need to do quite some work to have the output of the MPT-7210A (or any other battery charging/management device) "see" the battery load it is expecting. If you want a system with separate MPPT and battery management you'll need modules designed for this.

[tatus1969]

Than I had the "bright" idea (solar pun intended) of sticking a cheap buck DC-DC converter in between the panel and the MPT-7210A, reducing the voltage to around 12V to let the boost controller do its mppt thing to charge the battery. Would this be feasible?

Battery charging/management relies on measuring the battery voltage (among other parameters) so you would need to do quite some work to have the output of the MPT-7210A (or any other battery charging/management device) "see" the battery load it is expecting. If you want a system with separate MPPT and battery management you'll need modules designed for this.

second that, exactly. No way to make this work with an addon buck, this will hide the battery state from the MPPT charger.

[David Hess]

If the buck regulator has a fixed duty cycle, then the input to output voltage ratio will be fixed and the MPPT controller which is also a battery charge controller will be able to "see" the battery.  Since the buck regulator will have a non-zero input to output resistance, use feedback in reverse to adjust the duty cycle so the input voltage will be a fixed multiple of the output voltage.

It will not be pretty but it will work.  It would be safer to get the proper MPPT controller.

[tatus1969]

If the buck regulator has a fixed duty cycle, then the input to output voltage ratio will be fixed and the MPPT controller which is also a battery charge controller will be able to "see" the battery.  Since the buck regulator will have a non-zero input to output resistance, use feedback in reverse to adjust the duty cycle so the input voltage will be a fixed multiple of the output voltage.

It will not be pretty but it will work.  It would be safer to get the proper MPPT controller.

agree on that. but you would need to prevent the buck from regulating at all, and that will not be possible with most integrated buck controllers. That will likely end up in a custom built fix-pwm buck design. P.S. I have something equivalent at home. It is a full bridge forced 100% duty AC/DC that powers my home audio amp. This makes it 2-quadrant capable, effectively "translating" the primary HV cap to the secondary, providing the necessary bulk cap for the amps. Pretty clever, but the credit must go to Bruno Putzeys.

[David Hess]

If the buck regulator has a fixed duty cycle, then the input to output voltage ratio will be fixed and the MPPT controller which is also a battery charge controller will be able to "see" the battery.  Since the buck regulator will have a non-zero input to output resistance, use feedback in reverse to adjust the duty cycle so the input voltage will be a fixed multiple of the output voltage.

It will not be pretty but it will work.  It would be safer to get the proper MPPT controller.

agree on that. but you would need to prevent the buck from regulating at all, and that will not be possible with most integrated buck controllers. That will likely end up in a custom built fix-pwm buck design. P.S. I have something equivalent at home. It is a full bridge forced 100% duty AC/DC that powers my home audio amp. This makes it 2-quadrant capable, effectively "translating" the primary HV cap to the secondary, providing the necessary bulk cap for the amps. Pretty clever, but the credit must go to Bruno Putzeys.

It is not that the buck regulator has to be prevented from regulating; a fixed duty cycle will do that easily enough so for instance driving the switch with the output from a toggle flip-flop (50% duty cycle) could be used to create a fixed 2:1 voltage ratio.  The problem with this is various tolerances like on/off switching times, forward voltage drop, and inductor series resistance conspire to increase the effective input to output resistance preventing the output from tracking the input precisely.

Instead use the switching regulator's error amplifier to set the output voltage at exactly 1/2 of the input voltage; then it will slightly adjust the 50% duty cycle to make up for everything else.  To do this, the input voltage (divided down) is used as the reference.  Not all integrated regulators will easily allow this because they use a built in reference however many provide access to the transconductance output node used for frequency compensation which makes it easy to add an external error amplifier.  Or a regulating pulse width modulator like the 3524 could be used.  Or build the regulator control circuit from discrete parts; it is not that difficult.

[Red Squirrel]

I was toying with this idea for a small 12v project then never ended up doing it and sticked with a cheap pwm controller.   My idea was to build a buck converter that monitors the input and output amperage.  Once in a while it would sweep the voltage range until it's generating the most watts and hold it for a few minutes, then repeat.  Basically a really really crude mppt.  It would never go higher than 13.5 though.  Now if you wanted you could add smart charging capability too but that gets more complex.  Need to know how much current is going into the battery, separate from the load.

I've also never had good luck measuring current with shunts, I'm probably doing something wrong, but it's always super unstable. 

I definitely will revisit this at some point though.  Need to read up more on SMPS design in general, I want to learn more about it.

[NiHaoMike]

I've also never had good luck measuring current with shunts, I'm probably doing something wrong, but it's always super unstable. 

You need to use a low offset opamp, preferably one designed for that use. Ordinary opamps have too much offset voltage to get reliable readings at DC. If you're doing high side sensing, you'll also need very good CMRR.

[Red Squirrel]

I've also never had good luck measuring current with shunts, I'm probably doing something wrong, but it's always super unstable. 

You need to use a low offset opamp, preferably one designed for that use. Ordinary opamps have too much offset voltage to get reliable readings at DC. If you're doing high side sensing, you'll also need very good CMRR.

Yeah, I was just using a fairly general purpose instrumentation op amp. Was trying to low pass filter and all sorts of stuff and no go.   I also read more on it after the fact, and they also make op amps designed for high side current sensing which is probably what I'd use. (they can handle the higher voltages, like say, up 6o 60v).  Problem with low side is messing with the ground path. Even more problematic if you have two shunts in series. (Ex: one measuring current going into the SMPS and one measuring current coming out, or something similar)

[David Hess]

The usual problem I see with constant current feedback loops in power supplies is a slow current sensing circuit in series with the error amplifier leading to poor phase margin causing oscillation.  My first bench power supply design was a real dog in constant current mode because of this.

Solutions include using differential video amplifiers for current sensing or a discrete transistor circuit with offset compensation provided by an external slow operational amplifier.  Better yet is to level shift the current reference (not the current sense!) so that the current mode error amplifier can directly see both signals and have it take over from the voltage error amplifier as required.