Diversion ...

[gridloose]

... consider separate bank for water heater/other DC loads.

PV is cheap, but not yet free.  The electronics to store otherwise wasted power as hot water should be considerably cheaper than an extra 5 kW of panels.

A Black box processor could easily minimum interface to the CC yet provide more effective measurement of available PV power and adjust the DC load to match.

Perhaps you can glean enough information from the register interface of the CC through a com port to make this work.  But it would be easy to disturb the CC's MPPT scan if you are switching kilowatts of power from the PV panels to an opportunity load without carefully considering what is going on inside the CC.   Not obvious to me how the black box provides a more effective measurement than the CC's MPPT scan, or how you plan to adjust the DC load.

...  Can the CC actually be 'aware' of what the max power available from the pv array is, at any moment? ...

The CC does an MPPT scan every few minutes if it is not getting all the power it wants.  But you're right, there is a complication here if the PV array is larger than the CC can handle, or if the battery/inverter can't load the CC enough to bring the panels down to Vmp.   I was thinking the water heater could be sucking most of that current when the CC is doing the MPPT scan.  But it should do a full scan of the entire power curve for the panels to avoid getting stuck on a local maxima.  The CC could measure the average water heater current with an external shunt, and control that water heater current with the AUX signal to scan the entire power curve of the PV panels.  That's getting a bit messier than I had first imagined.

Perhaps you're right, and we should simply have a dedicated array for the water heater.  With perhaps a switch to allow us to manually send all panels into the CC when the weather gets gloomy for a few days, and a relatively small waste-not load on the CC for when the weather is sunny.  I think you still need the PWM controller to send a varying amount of power to the water heater, such that the dedicated array panel voltage stays near Vmp.  And eventually something monitoring time-of-day, insolation, and battery state to automate that switch, only allowed to flip it once every hour or so to avoid messing with the MPPT scans?  I guess that could be your blackbox.

[dgd]

Perhaps you can glean enough information from the register interface of the CC through a com port to make this work.  But it would be easy to disturb the CC's MPPT scan if you are switching kilowatts of power from the PV panels to an opportunity load without carefully considering what is going on inside the CC.   Not obvious to me how the black box provides a more effective measurement than the CC's MPPT scan, or how you plan to adjust the DC load.

Therefore to eliminate risk of disturbing CCs mppt scan it's best to only consider diversion to heater when CC is in float mode.
The BB can use various methods/sensors not available to the CC to measure insolation, at any time, which can then be used to evaluate pv power capability. The BB software then uses this data and loads the pv array appropriately - likely using pwm.

..  I was thinking the water heater could be sucking most of that current when the CC is doing the MPPT scan.  But it should do a full scan of the entire power curve for the panels to avoid getting stuck on a local maxima.  The CC could measure the average water heater current with an external shunt, and control that water heater current with the AUX signal to scan the entire power curve of the PV panels.  That's getting a bit messier than I had first imagined.

And with the Classic it would need significant firmware changes to achieve this. Since this is not open source we would depend on MN to implement such changes. I won't comment on the likelihood of this happening.

Thanks for your suggestions and ideas on this diversion issue. It's been interesting for me to read about other control options and possibilities.

Dgd

[gridloose]

...
Therefore to eliminate risk of disturbing CCs mppt scan it's best to only consider diversion to heater when CC is in float mode.
The BB can use various methods/sensors not available to the CC to measure insolation, at any time, which can then be used to evaluate pv power capability. The BB software then uses this data and loads the pv array appropriately - likely using pwm.
...
And with the Classic it would need significant firmware changes to achieve this.

Yup, we're pretty much on the same page here.  I think the CC firmware changes could be quite trivial if the CC uses a temperature compensated Vmp as the target panel voltage based on a user entered value from the panel data sheet and a temp sensor on the panels, so the CC does not use MPPT.   But that's not a plug and play solution, and probably not something midnite would want to put their name on.  Depending on user configured loads to implement MPPT at power points greater than that which can be handled by the CC is most likely not in the cards as well.

That calculated Vmp approach might work better for you than measuring insolation.

I think a common problem here is that of implementing PWM to a water heater, with power coming directly from PV panels.  You need big enough caps (with high enough surge current capabilities) near the PWM switch such that the panel voltage remains pretty much constant, otherwise the panels will be quite inefficient with the panel voltage a square wave popping far above and below Vmp.  So either a whole bunch of large high voltage caps, or fast PWM.  And with fast PWM you get FET switching losses.  I'm figuring on getting to ~30 kHz, perhaps with four FET's each operating at 7.5 kHz and each handling about a kilowatt.  Avoid long wires from the PWM FET's to the water heater to reduce the inductance of the load, perhaps mounting the FET's on the water heater.  Drive the FET's very hard with a gate driver capable of several Amps to reduce switching losses.  Have large schottky diodes across each FET to allow any inductive current to continue to flow as the FET turns off.  Snubbers at the FET's might be needed, but I'd guess the heater element inductance is low enough to get by without them.

[gridloose]

I think the CC firmware changes could be quite trivial if the CC uses a temperature compensated Vmp as the target panel voltage based on a user entered value from the panel data sheet and a temp sensor on the panels, so the CC does not use MPPT.   But that's not a plug and play solution, and probably not something midnite would want to put their name on.

Does the CC register interface allow us to know when the CC is in MPPT mode, taking all available power from the panels?
Or perhaps midnight could provide a testpoint which is high when the CC is in MPPT mode (preferred, as that is a much simpler interface).

If so, then we can do everything else in this PWM controller for the water heater.

Whenever the CC indicates it is in MPPT mode, the PWM controller ramps power to the water heater down to zero, waits a few minutes for the CC to perform an MPPT scan, then if the CC is still in MPPT mode the PWM controller records the panel voltage for future use as the target Vmp.  When the CC leaves MPPT mode for more than say 30 minutes, the PWM controller ramps up power to the water heater until the panel voltage falls to the Vmp value previously recorded.  The PWM controller then regulates power to the water heater in such a way as to maintain the panel voltage at Vmp.  The PWM controller may or may not include a temp sensor at the panels to allow it to adjust the recorded Vmp value to meet current conditions.

(edit:)  The PWM controller could probably detect when the CC is doing an MPPT scan by monitoring panel voltage, checking for the characteristic pattern.  That assumes the PWM controller is slewing power to the water heater slowly enough.  But having that information come from the CC as previously suggested would be quicker and more robust.

[boB]

I think the CC firmware changes could be quite trivial if the CC uses a temperature compensated Vmp as the target panel voltage based on a user entered value from the panel data sheet and a temp sensor on the panels, so the CC does not use MPPT.   But that's not a plug and play solution, and probably not something midnite would want to put their name on.

Does the CC register interface allow us to know when the CC is in MPPT mode, taking all available power from the panels?
Or perhaps midnight could provide a testpoint which is high when the CC is in MPPT mode (preferred, as that is a much simpler interface).

If so, then we can do everything else in this PWM controller for the water heater.

Whenever the CC indicates it is in MPPT mode, the PWM controller ramps power to the water heater down to zero, waits a few minutes for the CC to perform an MPPT scan, then if the CC is still in MPPT mode the PWM controller records the panel voltage for future use as the target Vmp.  When the CC leaves MPPT mode for more than say 30 minutes, the PWM controller ramps up power to the water heater until the panel voltage falls to the Vmp value previously recorded.  The PWM controller then regulates power to the water heater in such a way as to maintain the panel voltage at Vmp.  The PWM controller may or may not include a temp sensor at the panels to allow it to adjust the recorded Vmp value to meet current conditions.

(edit:)  The PWM controller could probably detect when the CC is doing an MPPT scan by monitoring panel voltage, checking for the characteristic pattern.  That assumes the PWM controller is slewing power to the water heater slowly enough.  But having that information come from the CC as previously suggested would be quicker and more robust.

What you are talking about is "Waste-Not" mode.   Waste not allows the water heater or other load to be the diversion load that
regulates the battery voltage at Absorb or Float or EQ so that the CC does not have to reduce its power output to keep
the voltage from going higher than the regulation set point.

Usually the relative voltage (relative to the Absorb or Float or EQ set point) is somewhat lower  and so those Absorb, Float and EQ voltages only come in to reduce the power when the Waste-Not load can't do the job.  That might happen if the water heater
turns off because the water is already hot.

When the Classic is in MPPT mode, the battery voltage is below the Waste-Not set point and the Aux output is off so the
diversion load is not drawing any power at that time.

boB

[gridloose]

What you are talking about is "Waste-Not" mode.

Yup.  I dropped the discussion for a month, and now I'm circling back around to what I was thinking then.  I'm still a bit hung up on wanting an oversized array to be useful during bulk, though perhaps that's not really much of an issue.  Another possible issue is that if we have an oversized array, the CC's MPPT can only scan part of the power curve on a sunny day since much of the power is getting shipped off directly to that water heater from the panels.  Perhaps that's not an issue unless the panels get to 5x oversized or so.  A third is that even fast PWM direct from panels to water heater could cause enough voltage ripple to disrupt the MPPT algorithm.  So I was trying to figure out how to only send power to the water heater when the CC was not in MPPT.  I'm probably over thinking this, and could do well to just try using the existing waste-not AUX signal to ramp up that 30 kHz PWM of power from panels to water heater.

[boB]

What you are talking about is "Waste-Not" mode.

Yup.  I dropped the discussion for a month, and now I'm circling back around to what I was thinking then.  I'm still a bit hung up on wanting an oversized array to be useful during bulk, though perhaps that's not really much of an issue.  Another possible issue is that if we have an oversized array, the CC's MPPT can only scan part of the power curve on a sunny day since much of the power is getting shipped off directly to that water heater from the panels.  Perhaps that's not an issue unless the panels get to 5x oversized or so.  A third is that even fast PWM direct from panels to water heater could cause enough voltage ripple to disrupt the MPPT algorithm.  So I was trying to figure out how to only send power to the water heater when the CC was not in MPPT.  I'm probably over thinking this, and could do well to just try using the existing waste-not AUX signal to ramp up that 30 kHz PWM of power from panels to water heater.

Remember that Waste-Not is intended to be diverted from battery and not PV input voltage.

I wasn't sure which side you were thinking of diverting from.

boB

[gridloose]

I may be driving off into the weeds here, but there seem to be a few others with the same goal of putting up a 2x sized array to avoid foul weather generator use, and are then too cheap to just let that power go to waste when the sun's out.  I'd like to store that power in a large insulated basement cistern, use it for hydronic floor heat.  A few have figured out ways to more or less successfully send surplus power direct from the panels to an electric water heater, but all the schemes I've seen thus far look to be either inefficient or overly complicated to get going, often both.  The outline in posts #32 and #33 still seems reasonable the next morning, which for me at least is quite remarkable.  Hard to say where my head will be in a week.  Standard AUX Waste-Not is a potential plan B if the objections of post #35 don't get in the way (would need fast PWM with large caps, ramp the power very slowly to avoid disturbing the CC's MPPT).

[TomW]

I may be driving off into the weeds here, ...

Well, if you are off in the weeds, this is out of left field:

Maybe the easiest and simplest thing to do is grab an el cheapo Chinese inverter in an European configuration of 240? volts 50 Hz for your battery bank. Use a relay to toggle it on and off by whatever means and criteria you choose. Drive whatever loads you desire from AC like common everyday water heater elements  or AC resistance heaters for heating. A few losses but..

Personally, I would heat water and the rest would go to space heating in cooler weather. Or just let it go to waste.

Sadly, I don't have this problem of excess power much except in mid summer and long days.

Just thinking into the keyboard.

Tom

[gridloose]

Maybe the easiest and simplest thing to do is grab an el cheapo Chinese inverter in an European configuration of 240? volts 50 Hz for your battery bank. Use a relay to toggle it on and off by whatever means and criteria you choose. Drive whatever loads you desire from AC like common everyday water heater elements  or AC resistance heaters for heating. A few losses but..

That's hardly left field thinking.  Most successful attempts at heating water from PV I see posted here involve sending it through the CC and battery and inverter as you suggest.  But if I have a 4 kW water heater, that's probably twice as big as any other loads I plan to burden the system with, so everything would have to be sized much bigger.  If I can go direct from panels to water heater with $50 in electronics (and 6 months of screwing around, it's a hobby!) that's a big win.  If you do use the el cheapo inverters unattended, be sure to fuse them well and put them in a fireproof box.  I've seen them blow for no apparent reason in most spectacular fashion, amazing how quickly and thoroughly a bank of lead-acid batteries can destroy a few FET's.

[Westbranch]

  I've seen them blow for no apparent reason in most spectacular fashion, amazing how quickly and thoroughly a bank of lead-acid batteries can destroy a few FET's.

I hope they (inverters) were well protected by CB or fuse at the battery,  as you comment about using a fuse implies...??

[TomW]

amazing how quickly and thoroughly a bank of lead-acid batteries can destroy a few FET's.

I have witnessed a few spectacular failures involving Fire Emmitting Transistors over the years.

T

[dgd]

If I can go direct from panels to water heater with $50 in electronics (and 6 months of screwing around, it's a hobby!) that's a big win.

And just to say how easy that is...
Use dc SSR, snubber diode and decent blocking diode. SSR control to aux1 set to Float High. Use SSR to switch in heater to output from pv array - or just part of array that has block diode between array and CC. Leave enough of array direct to CC  to maintain float with normal loads.
Aux1 has a few seconds off delay that stops SSR chattering.

[zoneblue]

> I may be driving off into the weeds here, but there seem to be a few others with the
> same goal of putting up a 2x sized array to avoid foul weather generator use, and
> are then too cheap to just let that power go to waste when the sun's out.

Thats about an apt a summary of what we are trying to do as well, as youll ever get.

And I also am suprised at how hard it is to do. I refuse to tie up that much inverter capacity. But we have short distances and lots of fat wire laying around, so the plan here is to use DC elements, 100A DC SSR, and waste not.

But ill be interested on hearing how your pv diversion method works out.

[dgd]

...and using aux1 float high for heater controlmeans aux2 could be used as input for WBjr so ending amps will probably terminate Absorb sooner with over sized array. Hence reducing potential pv power waste during Absorb.