Commanda's solar mppt take 2

[commanda]

Following on from discussions in this thread
http://fieldlines.com/board/index.php/topic,144675.0.html
which followed on from earlier threads started by Madscientist267 (Steve),
and some valuable input from Opera House, I now present a new design.

This pretty much ticks all the boxes on my wishlist.
High Efficiency (synchronous buck converter)
Holds the panel voltage at a fixed point
Shuts down once the batteries have reached a certain voltage
Main parts readily available in through hole packages
Works as expected in the simulator



I have a more pixels version that I can email if you pm me.
I will be building this in the coming weeks.

Amanda

[Madscientist267]

Oh schnap... breakin out the synchro version...

This should be interesting...  I'll take one of those, please. 

Steve

[hayfarmer]

looks great commanda,that's the trouble with the 50 kb limit on pic's ,the resolution gets all fuzzy on detailed schematic and hard to see details on many projects seems like an added 25 k.b. would help.

hayfarmer

[commanda]

As much for my own reference as anybody elses, I'll detail the inductor I'm using.

EPCOS - B66325GX187 - FERRITE E CORE, E42X21X15, N87 ferrite core.
Farnell part number 1190596. You need two of these for one inductor. These are not gapped, and have to be gapped manually.

EPCOS - B66242J1000R1 - FORMER, E42
Farnell part number 1190598. You need one of these for one inductor.
This has 5 solder terminals down each side, which is why I wound it 5 in hand.

5 strands of 1mm wire. Each strand was 1.4 metres. These were twisted together using a cordless drill with the far end clamped in a bench vise.
Managed to get 15 or 16 turns on.
Watch the termination order. You want to terminate them in the same order as you started, so each winding is the same length.

Set the gap to 0.8 mm to give 50uH using the inductance meter.
Fixed the gap using a cut piece of Press Pahn (elephantide). This is the cardboard stuff you sometimes see used to isolate mains wiring.
This was applied to the center leg of the ferrite core.
The finished winding was then wrapped in the proper ultrathin yellow tape you see commercial coils wrapped in.
The outside of the ferrite core was also wrapped with the same yellow tape to hold it all together.

Calculated resistance should be about 6 milliohms.

Haven't built a saturation current tester yet, but I expect this inductor to far exceed what I really need.
I'm aiming for about 10 amps out into a nominal 12 volt battery, or a single 120 watt solar panel.

Amanda

[Madscientist267]

Ok, now that I've had a good look at this, the differences are a little more clear.

To make sure I understand:

1 - You've gone back to the more 'traditional' use for pin 4, with duty cycle able to go from 0 - ~88% (instead of ~12-100%) with Q1 forming your inversion to track panel voltage rather than output...

2 - R2 brings the output voltage to the chip for full battery shutdown...

3 - Dedicated MOSFET driver to run synchronized...

That's about what I see for the changes... am I close?

V2 I am presuming will be identical to the Prereg-Pass/7805 combo used in the non-synchro version...

My only reservation about this is in the battery voltage feedback. Been playing with a few concepts, not sure anymore if withdrawing pulse width is the way to handle a full battery or not... still got some more experiments to run... 

Looks good though. I can't wait to see what you get out of it (in terms of efficiency).

Steve

[commanda]

Steve,

Your understanding is pretty much correct.
Note that V2 is 12 volts. The IR2184 won't run off 5 volts, so it has to be a 7812. I've used a TO92 package 78L12 with a TO220 pre-regulator.

I also did away with the pot on the input voltage sense. Used a series string of resistors connected to a 2x8 pin header. I can set it using a jumper shunt, or later will fit a rotary switch wired up with ribbon cable and an idc connector. Should give about 1/2 volt per step.

I finished assembling the new prototype on Friday at work. Ran a quick test with the bench supply and a single 12 volt 40AH SLA, only ran it to about 1 1/2 amps output, and used the power supplies inbuilt meters to measure the input current, and a clamp meter to measure the output current, so the setup was a bit rough, but the efficiency numbers were 96 point something. So ballpark figure where one would expect.

And, of course, now the weekends here and I've got all the time in the world to play, sunny Sydney just ain't so sunny.

Amanda.

[Madscientist267]

Excellent, Smithers.

Looks like in terms of difficulty, it ranks with the first version, meaning not that much more complex. You never cease to amaze me. 

I still have to get my hands on some of these ideal diodes too... I've got one Schottky in play as a blocker that I really want to replace with an ideal. I bypass it manually during the day, but this can bite me if I forget about it and leave it jumped out overnight.

Can they handle the tests and trials of high speed switching too, or are they just better suited as almost-zero insertion loss blocking diodes? Wasn't entirely clear on the site (unless I missed something...?)

Steve

[commanda]

Ideal diodes. They are too slow to use as rectifiers. Only good (excellent, really) as blocking diodes.
They are surface mount, and tiny at that. Definitely have to go on a pcb.

I thought  I gave you a link to a guy that was making and selling them. Don't really want to start a business out of it myself.

And don't get too excited yet. I've yet to see the new mppt work with a solar panel.

Amanda

[Madscientist267]

It's all good...

I did get the link to the page, but couldn't recall if they were fast enough to be used for rectification.

I'll probably order me several of them when the finances straighten out a little around here. Right now everything that wasn't already pretty much rolling has been put on pause until I can locate another suitable means of income. Friggin economy.

As far as working with the new mppt, I'm sure it will be fine, it benched ok, right? The biggie between bench and panel performance that I found is the caps! Caps caps caps! Don't forget!

There's a couple of other things that I've been playing with that may be worthy of mention for use in version 2 -

I've been dinking with the use of a battery as a kind of cap. Currently I only have it on the output, looking to put something similar on the input in the near future as well. Re-borrowed the concept from the attempt with the LiPo cells (which failed miserably; I don't think the control circuitry in each cell particularly cares for the voltages that get seen when that many are put together in series... ?). Went with lead-acid this time.

The output side has a 'sacrificial' 12V 5AH SLA that runs pretty high most of the time when the system is in operation (14.7-15V, thanks to the incorporation of a Schottky diode afterward, hence 'sacrificial'), but it is close (electrically) to the output of the buck, and smooths out the pulses rather well. Afterward, the mentioned Schottky diode provides main battery discharge protection, but of course also gives a 0.3-0.5V drop depending on current and temperature. This 'battery-cap' will of course burn all it's water up eventually, but ATM it seems like the only real viable solution to an annoying problem.

Worked out the theory on paper, and even with the drops from the Schottky, it does in fact appear to be driving the overall system efficiency up a couple points by significantly reducing the high current pulses that are still coming out of the buck. Still plan on shortening the leads going in and out of the control modules (buck and dump) to further increase efficiency. Noticed a significant drop in efficiency when I went from the very first smoke tests with this to the currently 'installed' version. There's a big difference between a battery being 6 inches away from the output of the buck and 6 feet away. Losing just over half an amp from this!

I've tried to keep the lengthy runs on the higher (wild) voltage coming from the panels, and then do all the conversion and processing as close to the target battery as possible, but even with that, I was still getting losses from lack of filtering.

Going to also try something similar on the input; have done it as a very basic conceptual test, not quite sure of all of the potential gains just yet. Pros vs Cons still need to be weighed in this part. The battery is going to need to need to be ~30V for my particular layout. Was considering two 12V and one 6V @ 3AH, but finding that the peak power from the panels is typically at a slightly lower voltage than originally designed with in mind, so considering just two 12V 3AH in series for 24V nominal. They of course are going to boil dry rather quickly too, so again, this is not ideal, but the use of a battery very close on the input side significantly reinforces the 'stiffness' of the buck in general; with both input and output, efficiency is much better than without either.

Simply here FWIW, you may want to consider playing around with this as an idea. Caps would be better in terms of 'proper' obviously, but at the power levels we're dealing with (particularly as they increase), filtering and hardening are going to be more and more important. Unfortunately it's difficult to get the stiffness that these things demand with just capacitors alone.

... Unless you want to try an entire bank of them haha But even then, while each one added helps, and I stiffened both the input and output on mine with as many smaller caps as I could fit under the board when I tweaked the last revision, no amount of capacitor seems to have as great an effect as a battery.

Steve

[jeffbirkle]

Hello,
Nice work.
What are the input-output voltage min/max, and max amps?

[Bruce S]

Hello,
Nice work.
What are the input-output voltage min/max, and max amps?

Depends on what you are asking about, the last post is almost a full year old.
Bruce S

[jeffbirkle]

Yah Wow, didn't even notice that!

[MadderScientist]

Ahem... well I WAS replying to this... as me, but then some horrific events took place, and so now, well, I'm ME... I digress... here's the original reply...


I can't comment directly on this particular version, but there are some similarities between this and the original version (asynchronous) that can give some idea:

IIRC, for mine, the max input is about 40V (limitation of the pass transistor in the MC34063 chip supply). Keep in mind, that this only happens under no-load (or fault) condition.

The output will vary wildly with the availability of power at the input; remember that it targets a very narrow voltage window at the input, and then "cranks down" on it to squeeze everything it can into the output without falling below the set input voltage.

The overall efficiency of my version is about 90% or so, squeezing a fair amount of extra power out of the panels that otherwise would have been lost to mismatch.

My best output was about 54W, into a 12V nom battery with ~36V (24 nom) input from the panels, targeting a Vpeak of 30V.

The efficiency goes up slightly as you go to lower output voltages, but much above 50W or so, and the MOSFET in mine begins to waste more and more heat. I attribute this to design flaws more than anything, as theory suggests it should be able to push much more before liberating significant heat. It's my understanding that Amanda's version (which mine was based from) does exactly that, although I'm not entirely sure to what degree.

It's unclear whether the synchronous version (from this thread) was ever built, and I haven't seen Amanda in quite some time... In theory however, it should be capable of much more efficiency (in the 95-97% range as opposed to the 90 I get with the async version).

Hope that helps...

Steve