pv dump load to water heater on midnite solar classic?

13»

Comments

  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?

    > Through a bunch of cloudy days, certainly not.

    The whole point of excess panels is to cover us on moderately cloudy days.
    But when the weather is heavy with rain clouds, panel power can go down
    towards 1% of full sun.

    The Classic's MPPT scans for max power only once every few minutes.
    So a simple minded tracking mechanism employing an "MPPT-Active" signal
    won't work unless we ramp the PWM current from min to max over a period
    of an hour or so. One of the other schemes mentioned in these last few posts
    where we measure battery voltage or panel power would probably work better,
    not require any change to the CC, and should work with any MPPT CC.

    If it's the Classic, I'd probably just use the Aux1 waste-not and write off any excess
    power on current limit in Bulk. If drawing power directly form the PV panels,
    the PWM duty cycle for the heater should ramp up over a second or two
    when Aux1 tells us to, holding the duty cycle steady when the panel voltage gets down to
    approximately Vmp. Heater PWM for use directly on the panels must cycle fast
    enough and the caps into the PWM's FET switch be big enough that there is not
    more than a volt or two of ripple on the panel voltage, or the panels will not
    be operating efficiently. I figure around 30 kHz and 1000 uF.

    Aux2 with PWM is probably best used only if drawing power from the battery
    (perhaps through the inverter) since it at 500 Hz it cycles too slow to be able to avoid
    monstrous ripple on the panel voltage when drawing power directly from the panels.

    Edit 9-17-2013: No, do not use Aux2 with 500 Hz PWM through a 60 Hz inverter!
    That would be very silly. Only when drawing power directly from the battery.

    Well, that's what I think today.
    We'll see about tomorrow.
  • NorthGuy
    NorthGuy Solar Expert Posts: 1,913 ✭✭
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    The whole point of excess panels is to cover us on moderately cloudy days.
    But when the weather is heavy with rain clouds, panel power can go down
    towards 1% of full sun.

    My worst day (which was very snowy) was 0.11kWh/kW. However, most of the cloudy days you can get 0.5-0.7kWh/kW even here in the North.

    I also want to have more panels for cloudy weather, and it would be already installed if I had spare $5K. This would increase my panel/battery ratio, but I don't think it would be a problem, exacept that I might need to limit current a little.

    However, the panal/battery ratio doesn't mean much by itself, only in relation to loads.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?

    > My worst day (which was very snowy) was 0.11kWh/kW.

    Assuming 5 hours of usable sunlight, I'd figure 0.11kWh/5h = 2%.
    You were well on your way towards my off-the-cuff 1% figure.

    The PV sales pitch around Portland OR often includes something like:
    "With cloudy days the light on the panels is well dispersed throughout the day rather
    than concentrated in just a few hours of direct light on the panels"
    Which may be true, but it's still bunk.
    Five minutes of sunshine gives more than an hour of mucky weather.
    Thankfully, NE Oregon gets considerably more sunshine than Portland.
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    The PV sales pitch around Portland OR often includes something like:
    "With cloudy days the light on the panels is well dispersed throughout the day rather
    than concentrated in just a few hours of direct light on the panels"

    That is indeed true. The few Watt hours you get will be a more consistent flow of meager current for a longer time. :p

    Not much good for charging batteries though.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    There were a few sharp curves, but I think I figured out all he was trying to say.
    Except just what the heck's going on regarding Aux2's PWM duty cycle and frequency.
    But I'm no longer planning to use Aux2 PWM anyways.
    I'll post a response to the Midnite thread cited, see if there's any interest in
    finding a better way to do waste-not. In particular, see if there's any interest
    in having the Classic emit a signal to show when it is using MPPT.
    Will probably have it up within a day or so, once my registration gets through.

    Registration never came through, perhaps Midnite discriminates against people with a history of obnoxious posts? Oh well.
    (Edit: Now see that my registration does work, just never got the promised second email stating my registration was approved.)
    Here's how I interpret boB's description of waste-not as found in post #5 at http://www.midniteforum.com/index.php?topic=890.0
    I found it very confusing on first read, so I'm filling in what I think are the missing details.
    Any corrections to this would be most welcome.

    ##########
    Waste-not only works in Absorb, Float and EQ, where the Classic is trying to regulate the voltage across the battery. It does not work in Bulk, where we assume the Classic is trying to dump all available power into the battery. (Most installations have the battery and charge controller sized to be able to accept all available power from the panels when in Bulk.)

    If the panels cannot supply enough power to reach the target battery voltage, (or the current limit setpoint has not been reached, in the case of Bulk), the MPPT algorithm in the Classic charge controller is actively looking for the ideal combination of voltage and current from the panels for the most possible power to the battery.

    If the panels can supply enough power then the MPPT algorithm is not active. The charge controller draws only the needed current from the panels and the panel voltage naturally rises above what would be the Vmp for the current conditions of temperature and insolation.

    Waste-not only turns on when the panels have an excess of power to do the job at hand.

    Edit 9-22-2013: This statement is wrong: "That is, it only turns on when MPPT is not active."
    The battery voltage when waste-not turns on is slightly below the target voltage, so the MPPT is still trying to send all the power it can to the battery.

    Waste-not is configured with two voltage setpoints which are relative to the target battery voltage for the mode (Absorb, Float, or EQ) that the Classic is in. We configure the waste-not Aux output to turn on slightly below the target battery voltage, and turn off at a still lower voltage. The Classic's voltage regulator algorithm never quite reaches its target voltage, it is the time averaged power to the waste-not load that dissipates the extra power from the panels that would otherwise cause the battery voltage to exceed the target voltage.

    Assume we have 1000 watts from the panels, the battery needs 100 watts to reach the Float target voltage, and the waste-not load is 2700 watts. The waste-not Aux1 port might be on for 1 second and off for 2, giving an average load of 2700/3 = 900 watts, equal to the otherwise wasted 900 watt difference between the panel capacity and the battery's need. We're counting on the battery to average out the waste-not load, so battery voltage remains about the same.

    The discussion of temperature compensation, Aux1 and Aux2 in boB's post is fairly clear except for the details on the Aux2 PWM duty cycle and frequency. I'd assume that the Aux2 PWM duty cycle varies linearly from 100% on at the hi voltage setpoint down to 0% on at the low voltage setpoint, operating at a constant frequency. Diagram 7 on page 44 of rev C of the Classic Manual shows a more or less constant duty cycle that increases in frequency with increasing voltage. The notes from boB in the Midnite thread suggest it is the "voltage width" that determines the PWM frequency, nothing much said about duty cycle. If the "voltage width" really does affect frequency, I would expect battery size and waste-not load impedance to affect frequency as well.
    ########

    Most posts reporting success using the Classic's waste-not to heat water are switching on a water heater powered from a 240 Volt inverter using Aux1.
    Some use a low voltage DC water heater powered from the battery, avoiding the additional load to the inverter.
    For waste-not direct from PV, posts 13 and 14 of http://midniteforum.com/index.php?topic=727.0 might be a good jumping off point. No custom electronics to design and build. Note that he's only drawing 780 watts from 4*5*140=2800 watts worth of panels, so the panel voltage will remain fairly high in decent sunlight even when the heater is switched on. I'd start with Aux1. Using the 500 Hz PWM of Aux2 plus the very large caps of post 14 might allow higher power heating elements to be used without disturbing the MPPT in the Classic too much.

    I'll probably be buying a Classic in the next few months and try some of this out.
  • boB
    boB Solar Expert Posts: 1,030 ✭✭✭✭
    Re: pv dump load to water heater on midnite solar classic?

    Hi Gridloose (et al. Whoever "Al" is ?)....

    The PWM on Aux 2 is normally set for a 1 volt width. The frequency of the PWM at that width is
    about 500 Hz. The reason for that is that the Classic's software that runs this is at a fixed
    timing of 200 microseconds. If you take 200 microseconds and multiply that by 10, you get
    2 milliseconds which is 500 Hz.

    OK, so having divided up a 1.0 volt width by 10, you get 0.1 volt divisions or increments.

    When the Classic encounters the low end of that 1.0 volt, it starts to pulse the Aux 2 output
    at one tenth of its 2 millisecond (1/500 Hz) time. That would be 10% duty cycle.
    As the battery voltage rises every tenth of a volt, the Classic increases the duty cycle
    by 10%. At the top of that 1.0 volt width, the duty cycle would be at full on or, 100%.

    When the width is widened to, say, 5.0 volts, the 5 volts is divided up into 50 tenths
    of a volt. The frequency of the PWM in this case is 5 times slower, or 100 Hz.

    The tenth of a volt division is because that is the minimum voltage change the Classic
    sees and has to work with.

    boB
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?

    boB,

    Great answer! I've now got Aux2 PWM waste-not understood.
    Will continue the discussion up on the Midnight forum at http://midniteforum.com/index.php?topic=890.0

    One correction to my description of waste-not:

    > Waste-not only turns on when the panels have an excess of power to do the job at hand. That is, it only turns on when MPPT is not active.

    Waste-not is on when the battery voltage is goes above the waste-not setpoint, but is still below the target voltage for Absorb, Float or Equalize.
    Since the Charge Controller has not reached it's target voltage, it is still using MPPT and trying to send every bit of power it can to the battery,
    which is what makes waste-not efficient.
  • Grey
    Grey Registered Users Posts: 6
    Re: pv dump load to water heater on midnite solar classic?

    Have been reading and following with interest the opportunity and diversion debate and some very creative ideas for how
    to better harness and manage opportunity.
    This thread has gathered a lot of data on CC operation and managing "opportunity" loads ..
    After reading as many posts as I can find, I would summarise that much of the effort is focussed on using opportunity
    for "variable load heating" of one form or another - with various implementations using PV, DC or AC inverter loads.
    Which no doubt works great for those in cooler climates ...

    However my scenario is somewhat the reverse, and with some other less common characteristics.
    I have no reqt for "heating water or similar" and hence can't use this variable load method.
    But I have numerous "opportunity" loads that can be used in a combination over time to to simulate "variable"
    (but over a longer averaging period)

    Perhaps this is incorrect thread for this ... however over time this thread has canvassed various similar requirements from users

    Not wishing to provoke here a debate about Definitions of "opportunity" "dump" "diversion"...
    However I have defined my own sub categories of "opportunity" for "fixed" loads (Details of loads in each category deleted for brevity)
    Cat 1 Opportunity : time shifted fixed loads
    Cat 2 opportunity : discretionary fixed loads

    CC has capability for Opportunity-Hi (waste not hi) which uses PWM style output to be able to vary the power to the load whilst
    minimizing or avoiding voltage drop measured at the battery ... And thus maintains Battery charge cycle/status and avoids drawing
    current away from maintaining or charging batteries
    Information avail confirms the CC uses voltage offset points, combo of input voltage control and output voltage droop and
    Then intelligently Reduces PWM accordingly the CC is doing its absolute best, using the information available to it, to
    Regulate and enable waste-not for opportunity loads.
    With the PWM method more efficient at keeping load near the optimum and minimizing impact to battery charging/state.

    However, the CC doesn't have all the possible information, for human like decision making available to us.
    As has been noted by many, with a knowledge of available loads, and an observation of "the sun is shining" ...
    And monitoring battery state, charge state and Power flows... We can make a decision to intervene and manually switch
    On/off various loads. The CC does this - but with far less information available to it.

    1) It's unclear how the recently added current sense module will enable CC to adjust more intelligently this algorithm.
    Noted that midnite have confirmed it coming soon (so lets not put extra Pressure on them, as has been noted by bob)
    Regardless, it's yet another "input" (good) into the decision making algorithm.
    Change in current flowing Into battery is presumably a more precise way to ensure that opportunity loading does not impact
    battery charging status.

    2) the CC knows MPPT state, charge cycle, Input and output voltage, current flows,
    But doesn't have a way to determine the potential available power from the solar array?
    .... Except by the "suck it and see method" and in particular, when in latter stages of charge cycle,
    it needs a heavier load applied ... To be able to determine quanta of "opportunity availability"
    It's basically a feedback loop - that's not predictive, it's reactive and adjusts accordingly.
    (Note this is not a criticism of the CC, merely a reality of what information it's working with )

    ... But what if we could more intelligently forecast or estimate the available potential power and then match it to
    our available "opportunity" loads....

    Anyway to my point .... I've been speculating about a method to measure the solar potential energy (at a given
    point in time) and then use it as an "input" to a decision making process for "opportunity" load control
    (particularly For loads that cannot use the PWM methodology .... Hence the need for more informed decision Making)

    Which brings me to the "blackboxproject" (BBP) in midnite forum
    Here we have a fantastic opportunity to bring various inputs into a box that will already have access to much
    of the data needed, ... And use it to make (or assist) informed decisions that can be used for load control ...
    Note this is not in anyway a suggestion to modify, interfere with or substitute for the core CC functionality
    it would simply "wrap around" the Existing CC/inverter and Other components.
    .. And leverage the work and inputs already underway in BBP.

    I will separately post into the midnite forum BBP thread my suggestions on how this might be achieved,
    and to ascertain interest to expand BBP capabilities.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?
    Grey wrote: »
    ...
    Not wishing to provoke here a debate about Definitions of "opportunity" "dump" "diversion"...
    However I have defined my own sub categories of "opportunity" for "fixed" loads (Details of loads in each category deleted for brevity)
    Cat 1 Opportunity : time shifted fixed loads
    Cat 2 opportunity : discretionary fixed loads
    ...
    However, the CC doesn't have all the possible information, for human like decision making available to us.
    As has been noted by many, with a knowledge of available loads, and an observation of "the sun is shining" ...
    And monitoring battery state, charge state and Power flows... We can make a decision to intervene and manually switch
    On/off various loads. The CC does this - but with far less information available to it.
    ...

    2) the CC knows MPPT state, charge cycle, Input and output voltage, current flows,
    But doesn't have a way to determine the potential available power from the solar array?
    .... Except by the "suck it and see method" and in particular, when in latter stages of charge cycle,
    it needs a heavier load applied ... To be able to determine quanta of "opportunity availability"
    It's basically a feedback loop - that's not predictive, it's reactive and adjusts accordingly.
    ...
    Which brings me to the "blackboxproject" (BBP) in midnite forum
    ...


    I'll definitely read through the BBP thread when I have the time and maybe respond further there.
    But in response to what you've presented here:

    Seems time shifting an electrical load will need battery capacity, and enough charge controller to run it. So you may just need to build a bigger system. The thing about water heating is that it's a cheap way of storing usable power, an insulated tank of water can be much cheaper than enough golf cart batteries to store the equivalent power.

    The proposed battery current monitor (edited) will allow the charge controller to know how much of the power is going to the battery. And since it already knows how much total power it is sending out, it can also deduce how much is going to the load (typically a 120/240 VAC power inverter). The "suck it and see" method is a fine way to determine how much power can be had from the array, and can be applied to diversion/opportunity loads as well. I'm not sure the charge controller needs a lot more information at hand than that in most cases.

    When the MPPT algorithm in the charge controller does a scan, it knows precisely how much power is available from the panels. My thoughts on the ideal solution vary by the hour (as a quick read of this thread will show), but perhaps the ideal would be for the charge controller to somehow communicate to our opportunity load controller exactly how much current the opportunity load should be pulling from the panels. Occasionally the charge controller will shut down the opportunity load (by telling it to pull zero Amps (edit: or some appropriate number of Amps if the array is bigger than the charge controller can handle)) so it can cleanly do a scan of the panel array. The charge controller can measure power going out to the battery, and tell the opportunity load how many Amps of additional current from the panels are available, perhaps updating that figure at 100 Hz or so. That could be a much better solution than having two competing algorithms in two different boxes trying to maximize power from a single array of panels.

    There is plenty of room for an external controller box to do things that have not been implemented in the charge controller. If such additional features generate interest, it may make business sense to incorporate them as firmware updates to the currently available product, or perhaps as hardware changes to a future charge controller.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?

    My thoughts are still evolving on this.
    The continuing discussion is on the Midnight forum at http://midniteforum.com/index.php?topic=890.0
  • NorthGuy
    NorthGuy Solar Expert Posts: 1,913 ✭✭
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    My thoughts are still evolving on this.
    The continuing discussion is on the Midnight forum at http://midniteforum.com/index.php?topic=890.0

    Using a heating element as a dump load is easy. Too bad I don't have such loads.

    But what would you do with a load that cannot be PWMed, such as an air conditioner, or a load located hundreds of feet away?
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?
    NorthGuy wrote: »
    Using a heating element as a dump load is easy. Too bad I don't have such loads.
    But what would you do with a load that cannot be PWMed, such as an air conditioner, or a load located hundreds of feet away?

    For an air conditioner, I think you're stuck with a bigger CC and bigger battery and bigger inverter. Though a water heater could be hundreds of feet away if the PWM FET's and input caps are mounted at the water heater and power to the water heater is ramped up and down slowly. Those long wires to the water heater are then seeing pretty much constant current.
  • NorthGuy
    NorthGuy Solar Expert Posts: 1,913 ✭✭
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    For an air conditioner, I think you're stuck with a bigger CC and bigger battery and bigger inverter.

    Yes, but how do you know when to start it and stop it? You prabably need to wait for at least a minute or so between switching it on and off.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?
    NorthGuy wrote: »
    Yes, but how do you know when to start it and stop it? You prabably need to wait for at least a minute or so between switching it on and off.
    I suggest you start it when it's too hot, and stop it when it's too cold. Once you have upsized the entire system, it no longer counts as a dump load.
  • NorthGuy
    NorthGuy Solar Expert Posts: 1,913 ✭✭
    Re: pv dump load to water heater on midnite solar classic?
    gridloose wrote: »
    I suggest you start it when it's too hot, and stop it when it's too cold.

    Your logic is undeniable.
  • gridloose
    gridloose Solar Expert Posts: 40
    Re: pv dump load to water heater on midnite solar classic?
    NorthGuy wrote: »
    Your logic is undeniable.

    Well, the part about it not being a dump load (opportunity load for some) probably is deniable.

    And yes, you're right, a timer is a good idea, or that AC will just cycle on and off every few seconds.

    Probably good enough to turn it on when the Aux Waste-Not has been true for 30 minutes or so,
    in the hope that the panels are then providing more than just marginally over what the battery needs.
    And a 30 minute delay after Waste-Not goes false before turning it off, to give the AC a chance to run efficiently.
    When you turn the AC off, maybe somehow keep the fan running for another 5 minutes to properly distribute all that cool.

    edit: Could make the 30 minute AC turn off timer start counting when the AC turns on, so it can turn off immediately when Waste-Not goes false. That would tend to deplete the batteries less. Probably want to drive some sort of control line to the AC (perhaps hacking into the thermostat?) rather than just switch off power to it, most AC units should then know enough to continue to run the fan for a while.