*DC* from PV panels

On 24/11/21 6:47 am, Rich S wrote:
> Hi Don,
> so what exactly do you need the 900W of electricity for?
> 
> If I suppose the goal is to store every bit of ENERGY collected per day,
> then we need a lossless storage medium,
> and a medium whose energy that can be efficiently drawn from at an
> arbitrary rate.

LFP batteries go pretty close to that ideal. Not as cheap as water, but 
anything thermal is going to have much bigger losses.

CH

On 23/11/2021 14:47, Don Y wrote:
> Most panels seem to want to be strung in series to develop a
> higher (potential) output voltage before being fed to an
> inverter -- for AC out.
> 
> What are my options if I'm looking for a *DC* output?
> Ideally, I'm looking for ~40+A @ ~50VDC.  Using PV panels
> in the traditional way to generate AC and then converting
> that back to DC seems like its going to bear more of
> an (in)efficiency cost than just trying to use the nominal
> 48VDC from the panel, directly.
> 
> [I have some leeway with that... maybe 10% on the voltage
> with an inversely proportional change in the amperage]
> 
> OTOH, the output from the panel will vary with cloud cover,
> partial shading, etc.
> 
> So, are the losses trying to redevelop ~48V from each panel
> (at differing current levels based on incident sunlight)
> likely to be more or less than the more traditional approach?
> 
> I should see how folks living "off grid" use DC in their
> applications...
> 
> [I'm wicked busy, lately -- end of year is always rough, this year
> even moreso -- so apologies if I don't reply promptly]

What is the load?

On Tuesday, November 23, 2021 at 11:01:13 PM UTC, Clifford Heath wrote:
> On 24/11/21 6:47 am, Rich S wrote: 
> > Hi Don, 
> > so what exactly do you need the 900W of electricity for? 
> > 
> > If I suppose the goal is to store every bit of ENERGY collected per day, 
> > then we need a lossless storage medium, 
> > and a medium whose energy that can be efficiently drawn from at an 
> > arbitrary rate.
> LFP batteries go pretty close to that ideal. Not as cheap as water, but 
> anything thermal is going to have much bigger losses. 
> 
> CH

I was being a bit cheeky, but yes, you gave a sane answer.
Indeed, for off the shelf products, the home energy storage
marketplace is using Lithium chemistries (LFP, LNMC, LTO).
I found this article to be very informative:
https://www.cleanenergyreviews.info/blog/home-solar-battery-cost-guide
regards, RS

On 11/23/2021 2:12 AM, Martin Brown wrote:
> On 23/11/2021 03:47, Don Y wrote:

>> What are my options if I'm looking for a *DC* output?
>> Ideally, I'm looking for ~40+A @ ~50VDC.  Using PV panels
>> in the traditional way to generate AC and then converting
>> that back to DC seems like its going to bear more of
>> an (in)efficiency cost than just trying to use the nominal
>> 48VDC from the panel, directly.
> 
> I suspect that your losses (~2%) with the fattest Schottky diode you can find 
> to wire or them together will be somewhat lower than the losses in a typical 
> active DC to DC converter. This is especially true when compared to panels in 
> series with for example one partially shaded.

I dunno.  Some of these converters claim really high efficiencies -- 95-97%.
There's just a shitload of competitive pressure on the industry to deliver
the most power per incident solar watt.

Unfortunately (for me), that's in line with generating AC for domestic use.

In my case, the "naive" approach -- a "typical residential solar installation"
with a 48V 40A power supply "bolted onto its back" -- starts out at a lower
efficiency over the solar solution as the power supply won't be 100% efficient.

Just tying the panels together leaves you with the same old PV problem;
getting the most *power* out of the panels.

My thinking is that a controller on each panel driving into a 48V output
setpoint with the outputs of these tied together is likely the best
compromise.  If a panel is shaded/degraded, the controller will still
try to maintain that 48V -- albeit at a reduced current.  So, the panel
is still contributing to the output capacity.

And, if a converter fails, you just lose the output from that panel
(it's as if the panel is permanently shaded).  Contrast that with
the single inverter (or "power supply") failing in a "naive implementation"
(AC inverter feeding DC power supply) and rendering the entire facility
useless.

Likewise, if you ADD a panel, you don't have to upgrade the inverter
(or, buy it with excess capacity to start with) or the power supply.

And, you can harvest (digital) information from each controller as to
its instantaneous performance so you know if the system's output is
falling, how much "untapped surplus" is available, etc.

Finally, converters can sit on the panels, outdoors, so I don't have to
move that waste heat out of the household interior.  (OTOH, they will
have to operate in 120F+ ambients)

The question is whether the same sorts of overall efficiencies are
available when tackling it in smaller pieces.

> Wired OR you get the current from all the ones in full light but in series your 
> maximum current is limited by the weakest link in the chain. Also when not in 
> direct sunlight the diodes drop will be less.

They have solutions that allow underperforming panels to pass the full
current of the rest of the chain.  The voltage is lowered to the point
where the required current can be supplied (or, shunted)

[The sorts of gizmos available are dizzying as each attempts to address
a different aspect of the issue and different deployment topologies.]

> I guess it depends a lot on whether your load can tolerate the variation in 
> supply voltage that will come from the solar panels directly.

I want 48V +-10% (5% would be nicer but not worth much in recurring costs).

And, ideally, if the source can't supply the required *power*, the voltage is
maintained but available current drops.   E.g., shed load or supplement supply.

I started looking into a converter design (as I already have something similar
for a different design) but have had to discipline myself NOT to pursue it
until I get these short term deadlines met (I committed to releasing six
designs by year end -- but, then went and "played" for a while, travelling
and, in general, "having fun".  So, now I'm under the gun to get those off
my list as promised... thankfully, release engineering is relatively
uneventful, just time consuming.  Five more to go!)

It's a "sufficiently interesting" problem that I've managed to interest
my power supply guru friend to "come out of retirement" to review my
design.  I'm sure he'll spot efficiencies (and potential fault modes)
that I'd have overlooked!