backchannel: DC-DC converting charge controllers

[edited and updated 10/26/2019]

I have been thinking lately about a kind of solar charge controller that is starting to pop up more.  In price, it slots in between inexpensive MPPT controllers and inexpensive PWM controllers.

From most expensive to least expensive:

  1. mppt

  2. DC-DC converting charge controllers (DDCCC?) <– we are going to talk about these

  3. pwm

  4. shunt

[519y64wkgl-_ac_us160_](’s specific thoughts are about the Greesonic MPPT1575 15A “MPPT” controller {mentioned in this thread}.

Based on empirical testing by Amazon commenters, the controller does do DC-DC downconversion:

  • 15.2 v @ 5.5 amps –> 13.2 v @ 6.21 amps

  • 15.7v @ 5.8 amps –> 13v @ 6.43 amps

Here’s the thing:  the Vmp of the top panel (HQST amorphous 100w) is 17.7v, and the Vmp of the bottom panel (Renogy mono 100w) is 18.9v.  So it is is not power point tracking (PPT) and definitely not maximum power point tracking (MPPT).

{Edited to add: OP’s panel appears to have a Vmp of ~18.2v (160W/8.8A).  I predict the controller will run the panel between 15-16v as described below.}

My theory

I think a couple of things are going on.  Both are acceptable at this price point, even though they will stick in the craw of folks who need buzzwords/labels to be happy.

As with the newer variants of the CPY-2410 and the CPS 2410 (YT test here), the controller appears to run the panels at a couple volts higher than battery voltage (Vbatt).  This is very rarely the point at which the panel can make the most power (Vmp), but will usually be ~15% better than PWM can do {in cool weather}.

The trickier part comes when the controller needs to reduce current, as when holding absorption voltage (Vabs) or float voltage (Vfloat).  Since it is not true MPPT it cannot manipulate Vpanel arbitrarily along various power points, it cannot move to power points where there is very little current (~Voc).   I would be surprised if they implemented a PWM backend to cycle current off/on.  My instinct is that the controller shunts the power off/on to “hold” a given Vbatt.  {For this reason I would not overpanel a DDCCC the way I would an MPPT controller.}

 My suggestion

I’d like to see a controller manufacturer use shunting to hold setpoints, but approximate (guess) Vmp by calculating it as _VoC x 0.8 _then DC-DC converting to boost amps.  That is a common starting place for MPPT algorithms, and is generally pretty close.  I encourage onlookers to do the math on their own panels to see how close it would be.

This could generate near-MPPT levels of power but wouldn’t require much processing power to operate.

Shunt and PWM would also benefit from this Vmp guessing even without DC-DC conversion.  And it wouldn’t incur the 5% conversion losses.

Admittedly there are drawbacks

  1. Vmp-guessing controllers would only be able to cut power (run panel at Voc) rather than clamp power (run panels closer to Vmp to reduced power coming into the controller).

  2. shunt controllers would experience more intense, faster cycling because power levels would be higher. This might be clicky-click annoying or even harder on the relay.

  3. PWM controllers would run hotter because they’d be switching more of the time to neck down the greater power levels.

The new hierarchy might be:

  1. mppt

  2. mppg+dc - maximum power point guessing + DC conversion

  3. DC-DC converting charge controllers (DDCCC) <– drive  panels at [Vbatt + arbitrary voltage like 2-3v]

  4. mppg+pwm:  maximum power point guessing, drive panel at  ~Vmp but no DC-DC conversion.  Just higher output than pwm.

  5. pwm

  6. shunt