I am looking into the design of the most simple (yet functional) PWM solar chargers for lead acid batteries. (Numerous examples can be found on Ebay for about 10 bucks.)
I have a hunch that they should be able to work as constant voltage sources without any magnetic filter components since they operate from a current rather than a voltage source. - But I am not sure.
My understanding is that the simplest charge controller is something like this:
My questions are twofold:
Given that the cell and the supply wire have some capacitance which is charged to Voc while the switch is open, does the battery see spikes of Voc when the switch closes or are the capacitances typically small enough and the supply wire LC filter typically low enough to prevent this? - Is there a back-of-the-envelope calculation to convince myself either way?
Assuming a fully charged battery, some chargers (claim to?) realise a constant voltage maintenance mode. Does this require an explicit LC filter at note N2 or can this be realised with the inductance of the wire to the battery combined with a low-value shunt capacitance at N2? - My hunch is as follows: 10 cm of 2mm diam. wire should have about 100nH. Combined with 47µF this gives a cut-off frequency of about 70 kHz. PWM at a faster rate should thus not require an explicit inductance.
Below is an image of the interior of the Juta CMP12 solar controller. (Source: http://solar-nm.blogspot.de/) It looks to me like there are no magnetic components...
Connections from left to right:
- 1&2 pv panel
- 2&3 battery
- 5&6 load
Edit 2: In response to Russel's ammended answer
Two things still have me stumped:
If you connect a PV panel to a battery then current flow will redcue as PWM duty cycle redcues. It will not be a linear reduction as vpanel will rise as load is reduced, tending to act against the current reduction from the PWM.
This is at the core of what I was trying to get at with my question:
- I agree, that average current flow will reduce with PWM duty cycle
- however I would think that it will in fact be a linear reduction even if Vpanel rises during the off-part of the cycle. Vpanel should have no effect on an on/off switch.
- Vpanel should be x Volts during the on-part of the cycle and Vmax during the off-part of the cycle
- Only with a linear regulator (partly-on switch), Vpanel would be somewhere between x Volts and Vmax and here the non-linear effect would occur. Right?
That out of the way, on to how constant voltage PWM of a current source is realised without any filters:
For argument's sake, lets assume the battery was "utterly" full and shall now be kept on float voltage. The current would then have match either the self-discharge rate of the battery. Let's assume it is orders of magnitude smaller than the current produced by the PV panel. - But on/off PWM without filter is not able to produce a smooth instantaneous current. (Only a linear regulator with a partly-on switch could.)
With PWM on/off regulation, I would expect the following should then happen:
If the battery was full and it was (over-)charged with a non-negligible current (during the on-phase), its voltage should increase linearly from 14.4 V upwards similarly to a capacitance. (Otherwise one would not need to reduce current during the absorption phase.)
But if no charge current is applied to a full battery, its voltage is only its open circuit voltage, say 12.8V.
Now if one modulates an on/off switch, should the instantaneous voltage seen by the battery be not as follows:
- A rise during the on-phase starting from some value in excess of 14.4 V upwards.
- Then an instantaneous drop by (14.4V-12.8V) = 1.6V when the switch opens.
- Followed by a slow fall due to self-discharge to the battery open circuit voltage 12.8V.
Knowing that a PWM controller must low-pass filter the voltage seen by the battery over many cycles and regulate it to a temporal average of 13.8V, we can calculate the some value in excess of 14.4 V above:
Since the target voltage around which the PWM controlled voltage shall oscillate is 1V larger than the open circuit voltage (13.8V-12.8V)=1V, the instantaneous value seen by the battery oscillates around 14.4V+1V=15.4V.
- The instantaneous voltage seen by the battery during the off-phase of the switch must be 13.8V+x decreasing to 13.8V-x (since the temporal average is then 13.8V assuming the duty cycle << 1)
- by the same argument, during the on phase the battery should see 15.4V-x increasing to 15.4V+x
So my question: How can this not be very bad for a battery? The effect should be grid corrosion during the on-phase (15.4V overvoltage) and possibly gassing during about half of the off-phase (if the modulation depth x is large enough?).
I suspect that the voltage profile drawn above is inaccurate but I cannot see where my error in thinking lies.