Water Heating Controlled by a Classic 150

[Treehouse]

I have to agree with OperaHouse I don't think it is working exactly like you think. According to the classic manual "The difference in Aux 2 is the use of PWM running at a hundreds of Hz rate" The spec sheet of the ssr relay response time is 10 ms on and 10 ms off. That means the ssr has a max switching time of 50 Hz.. Unless I am missing something I think what you are seeing is the ssr grabbing random pulses. It would be interesting for you to connect a regular light bulb to the ssr output and see if it is a smooth dim to bright (as more loading is needed) or just flashing at random like I would expect. 

[OperaHouse]

I think the PWM is allowed to free run and that time period could be anywhere from seconds up to that several hundred hz.  Remember the system is dynamic.  It tells the relay to turn on and that likely doesn't change till the controller sees the battery voltage drop.  A long time ago my physics teacher taught me that "Order emerges out of chaos."  Chris has seen a quantom change over using the relay that is seems smoother thn it really is.    Any time you mix two frequencies strange things will happen.  Midnight even says that its inductors will sing under certain loading conditions.  If the SSR was not a zero crossing type the results would likely be very bizare.  With zero crossing the relay is likely to just stay on as the PWM rate increases.  I agree that the fineness of control is a lot less then if this signal controlled a FET and the water heater operated directly off the battery.

There has been a lot of interest in this.  I have only 800W of panels and just started diverting the excess to a 10 gallon water heater.  I can take hot showers with only 140W being diverted.  This was just crude testing and I can easily divert 250W on most days.  That is not a lot of power. Even with a 30 gallon tank these numbers are reasonable.  My gas water heater sprung a leak two years ago.  A buddy works at Lowes's and was going to get me a dented one cheap.  I had a 30 gal electric one handy  and jusp plumbed that one in tempotarily.  I plugged it into a 120V outlet.  That gave me 1/4 power or about 750 watt element.  Two years and haven't got a ROUND TUIT.  Just can't shower, dish washr, and laundry at the same time.

This winter I will be writing a program for the UNO that mimicks that of the Midnight AUX DUMP and uses a zero crossing SSR.  This can be used with an existing all the time on inverter or a separate inverter that turns on when diversion is approaching.  The SSR will then turn on from 1 to 120 times a second depending on diversion need.  A 2000W element at 240V is only 500W at 120V  Connecting that to a cheap 1000W inverter is far more practical than buying a low voltage heater element where half the power is lost in the wires going to the water heater.  Also included in the UNO will be a GHURD like output to control a FET for a resistive dump when dump power needs exceed what the water heater can take or the thermostat has shut off.  That is my idea and it seems practical.

[Treehouse]

Are you going to include a zero crossing detector? If not how are you gong to know when to fire the scr? I too have been thinking about a similar design, just can't seem to find all the right parts to fit together. Most ssr I have found (Less than a few hundred dollars) are to slow or  to low of voltage. 

[ChrisOlson]

It would be interesting for you to connect a regular light bulb to the ssr output and see if it is a smooth dim to bright (as more loading is needed) or just flashing at random like I would expect.

Treehouse, I would suspect it is not "firing" the SSR with true infinite range, but the end result makes it appear that it does.

The whole idea with this was to vary the power to the 2,000 watt water heater element so that the proper bank charging voltage would be maintained.  It has achieved three things for me:
- the controller is now able to maintain proper charging voltage at all times during absorb
- the solar panels remain at 100% possible power output at all times - the controller never "throttles them back" to maintain absorb voltage
- I am getting as much hot water as I got before with the mechanical relay

So even though it may not work as it appears to work on the theoretical level, on the practical level it is working perfectly.
--
Chris

[DamonHD]

"In theory there is no difference between theory and practice. In practice there is."

    -- YogiBerra

[OperaHouse]

Treehouse, The device I will be building is a stand alone dump controller for those who don't have a Midnight controller.  Functionally it will act just like a Midnight in that you enter the absorb voltage and what voltage below that it will start drawing power to a water heater.  The output will be proportional between those two voltages.  Output will simply be a timed period.   That timed cycle period may be 1/2 second or maybe a second.  Increments of 2% (50-60 potential half cycles) may be fine enough and create less pulsing on the inverter.  This on/off time period controls an external zero crossing type solid state relay (SCR or TRIAC).  There is no need to sync the output.  If it misses the first cycle there will just be another added on at the end.  If one time there are 31 half cycles in a row and the nest time there are 32, that is no big deal.  I will be looking to see how the cheap inverters handle this wham bam output.  I may have to mix things up so no more than a couple half cycles are on at a time. 

[Eric L]

It would be interesting for you to connect a regular light bulb to the ssr output and see if it is a smooth dim to bright (as more loading is needed) or just flashing at random like I would expect. 

FWIW I tried exactly this with my Classic and the same SSR before wiring it into the hot water heater. The bulb went gradually from dim to bright but there was some flickering. At a certain point the bulb reached full brightness and stayed on without flickering.

I can take hot showers with only 140W being diverted.

Since installing this, my hot water heater thermostat is shutting the lower element off (the one tied to the SSR and pv) after only around four hours of operation (55 gallon hot water tank). The measured water temperature at the tap at that point is about 162 degrees F. Of course prior to being heated, the water is still warm from the day before. But yeah, it's surprising how hot it's getting from the surplus off of my pv array. It's certainly not running the element all day, but there is  plenty of hot water for two people. So far I haven't needed to turn the upper, grid-connected element on at all.

It's working very well, IOW, but not needing to use the whole surplus pv capacity.

[ChrisOlson]

I haven't tried the lightbulb trick.  But I think based on the voltage I see going to the element that it's doing the same thing as you describe, Eric.

Sounds like your setup is working like ours, with the exception that we have two water heaters and the bottom element in the primary, and the next two in the pre-heater are daisy chained so the power goes to the preheater when the bottom in the primary gets up to where the stat kicks it off.

One thing about this system - as you noted it doesn't take long to get 55 gallons at 160+ degrees.  We put in one of those mixing valves on the heater that mixes cold water with the hot water coming out so the temp of the water going to the house is about 125 degrees.  That prevents bad burns in the event you get guests or something that don't know about the extremely hot water coming out of the heater.  The mixing valve was only about 100 bucks and it has a thermostatically controlled thing in it that changes the amount of cold water being blended with the hot as necessary to get the water out at 125 degrees.
--
Chris

[Eric L]

Thanks Chris for the mixing valve idea. It has been less than two weeks since I installed the SSR, but our meter is showing that our draw from the grid has nearly been cut in half (down to about 3 kwh/day). When there's sun, what's left to power is just an electric dryer and electric range that my inverter can't run.

By the way I just saw your edit to the Oct. 9 post above. It may be that the battery-inverter wiring is undersized in my case (I sized it for the maximum rated draw only). I don't have an analog meter but I'm going to look for one to try your test.

I do have a small battery bank relative to my array size (18 KWh for a 4.1 KW array). This is because I designed the system only around a 'dusk-to-dawn' battery draw to 80% SOC, since I didn't want or need to oversize the bank. So it's also possible that the bank is struggling with the load. However it seems to accommodate other surge loads (well pump at 4.2KW start surge, e.g.) without the huge voltage drop, although there is one there too. So I'm a bit puzzled; maybe it is a wiring issue as you say.

[ChrisOlson]

When there's sun, what's left to power is just an electric dryer and electric range that my inverter can't run.

We power my wife's electric range (Electrolux with induction cooktop) and electric clothes dryer with inverter power.  It does require an inverter with generator load support to run them, though (Magnum doesn't have one yet, but they're coming out with one shortly called the MSH4024RE).  The inverters will handle the load from either one long enough to start the generator on load amps, then the gen supplies roughly 4 kW of the load while the inverter supplies the balance.  The inverter shuts the gen off when the load goes away (below the load start threshold).

Our clothes dryer pulls dead on 20 amps (4,800 watts).  The range can pull up to 7.5 kW if my wife turns on both the oven and a couple of the induction surface units.  I have the load start set so it starts the generator within 6 seconds of the load going over 25 amps per inverter (dual Trace/Xantrex SW+ 4024's).  It takes about 15 seconds to warm up the generator and sync with it, and then it goes to full load to take over the lion's share of the load.

We've never had the inverters kick out due to overload even when my wife really pushes things - they can put out 14 kW for about five minutes - and usually if she turns on the oven in the range it takes about 10 minutes or so to get it up to the temperature she has set, and then it cycles on and off to maintain the temp.  It's only the preheat time that draws a lot of continuous power.

Because we don't have grid power, I changed all the heating elements in our water heater from 4,500 watt (what they came with) to 2,000 watt.  The top element in the primary heater is on all the time with the stat set at 125 and the inverters will handle that, on demand, any time its required.  If we get good solar and wind that top element never comes on - that would be the element that you currently have on grid power.
--
Chris

[ghurd]

Uh,
http://www.fieldlines.com/index.php/topic,145256.26.html

[ChrisOlson]

We didn't get any solar power all week - only 0.8 kWh today.  But the wind turbines have been running pretty much balls out for several days.  We get WAAY more power from wind than our little 3.5 kW solar array can produce on its best day.

Tonight I snapped this photo of the solar controller to show folks that even though the controller is resting with no incoming power on it, the AUX function remains active.  When the turbines get the bank up to the temp compensated absorb voltage it activates the SSR and starts heating water (AUX 2 is on).

So it works not only to keep the solar panels at full power when the bank doesn't require all the amps - it also works as a "dump load" controller for the wind turbines to keep them at full power instead of their controllers activating the voltage clippers to control charging voltage.


--
Chris

[JW]

Wow Chris, that is a very lightweight picture 18.5k and the resolution looks great...


JW

[Treehouse]

After a lot of time... with a scope and a zero crossing ssr it does look like one can indeed control it with pwm at a higher frequency. There is some flicker but over all without syncing it to the output of the inverter via a zero crossing detector, it's very close. I'm now working on a stand alone controller that does what the xantrex c-series controllers do but using your inverter as the dump load control. Not to hijack this tread, but if there is some interest  in this I will start a new thread. I would like some input as to what features and functions people would like, maybe at some point offer a kit.

[Eric L]

Here's an update on my AC voltage issue using Chris' setup for those that might have a similar issue (the problem was 10%+ AC voltage sags and surges). It has now improved considerably (down to about 2 v +-).

I got ahold of an analog voltmeter and performed the test between the battery terminals and the inverter lugs that Chris suggested and applied a very heavy load. I could not see any difference in DC voltage between the two points. I general, I haven't had this level of voltage drop with any other load, so I figured it was likely something with the PWM function and the SSR.

Just as an experiment, I tried increasing the width setting on the Classic from 1.0 v to 3.0, v and noticed an immediate improvement in the degree of the voltage sag (this was immediately audible in the character of the transformer hum on the inverter). I didn't think of trying this originally because I figured it would make the swings worse, but it made them better. After experimenting with different settings of width and offset, I settled on a .7 offset and 3.0 width; this produced a much more manageable 2 volt swing on the AC side. I basically tuned it by ear to get the steadiest hum from the inverter.

So for whatever reason, I guess the Magnum inverter likes the 'slower' frequency of the wider pwm width.

The minor drawback is that this setting is not maximizing the draw from the panels, but I'm still getting plenty of hot water, so I'm not going to fuss over it.

Incidentally, I'm still using the 4KW element on my heater since I appear to be stuck with it; literally. My heater is 21 years old and no amount of cranking with a 6 point socket and breaker bar would budge it. As it happens, the pwm setting is not drawing more than 2kW even under full sun, so I'm inclined to leave the element alone until it fails.

[ChrisOlson]

Feel free to hijack it.  I showed folks how to do it with a Classic 150.  It's good to see other methods too.

I wish I could say something intelligent about the zero cross and high frequency PWM.  But I don't know enough about it.  I only know that it's working fine with the Classic.  When it's modulating the power to the heating elements I see the load amps on the inverters constantly vary from like 2 amps to 8 amps.  8 amps (@240 volt) is full power to the element.  Since I have SW+ inverters with the triple noisy transformers in there, I can also hear the transformers changing their buzzing as the load varies when the SSR is modulating the element.

I never did try the lightbulb trick.  I guess the results were definitive enough for me - the controller always maintains proper absorb and float voltages, the panels stay at full possible output when they would otherwise be only at partial output because the bank wouldn't need the current, and I have been getting enough free hot water from it for the last three weeks in the preheater, so that the primary water heater has not gotten below 160 degrees for the entire three weeks.  All of the power didn't come from solar - some came from wind.  But it still works.
--
Chris

[ChrisOlson]

Just as an experiment, I tried increasing the width setting on the Classic from 1.0 v to 3.0, v and noticed an immediate improvement in the degree of the voltage sag (this was immediately audible in the character of the transformer hum on the inverter).

I don't get no voltage sag on my inverters.  But more voltage width lowering the PWM frequency is something I'm going to revisit too, to see if I can make it smoother.  If I understand this correctly, the PWM switches the relay way faster than the frequency of the current.  So if you lower the PWM freq to match the current freq, in theory it should work better (maybe).

Thanks for pointing that out.
--
Chris

[OperaHouse]

Treehouse....what was that higher frequency and did you try multiple frequencies?

[Treehouse]

I tried every frequency between 60 Hz to 400Hz I wrote a pic program to step through each one watching the wave forms on the scope and found that certain frequency's work better than others. Try watching a scope for that  long...lol  You can try 156Hz, you will have to experiment with it because the internal resonator of the pic drifts a little, choosing the wrong frequency will result in what seems to be perfectly synced output, but leave it for a bit and it will drift and then never line up, giving you no output. Also some frequency's will not give you 0 -100% only 0 - 50% Anyway.. Having lots of fun..   My next step is moving from two x 100w flood lights to a 1500W heating element. It will be interesting to see if my 1000W msw inverter will run it. If I have the frequency right with the pwm it should work up to at least to 1000w, will be fused for 900W to be safe. If not I will get a 1500W- 240v and try it at 120V for  750W.

 

[ChrisOlson]

We had a bad week this past week and never got no water heating at all.  But finally got some today.  I've been wanting to play with more different settings.  I found that by changing the width to 2.0 that it seems to better match and is not "surging" on the inverter all the time.  I also decreased the offset to -0.2V.

By increasing the width to 2.0 it appears to switch the SSR on and off at full voltage somehow and wasn't really telling me much.  So I went to reading amps instead.  What it appears to do now is vary the amps to the load by turning it on and off really fast with voltage almost up to 240.  So I assume that means that it's firing the SSR at the right speed to get a full sine wave out of it?

--
Chris

[gww]

Chris
Thanks for the update.
gww

[ChrisOlson]

Well, I"m just playing with that voltage width, according to what Eric experimented with on his Magnum inverter.  I have to agree that it does seem to be smoother by using the wider width.  But I can't figure out now why it isn't modulating the voltage very much, and only amps.  How can it do that?  I'm seeing from 220-240 at the element all the time, just that it's not drawing much amps.

I never got a chance to take another video this afternoon because I was welding on a wind turbine that I'm building.  But it was up to about 6.5 amps @ 240 volts when I checked it.  That's 1,560 watts to the element.  The voltage was varying a little - like from 230-240.  But not much.  If I turn the port manually on it draws 8 amps @ 240, which is 1920 watts.  The element is 2,000 watt, but those elements draw a little less power after they get heated up.
--
Chris

[OperaHouse]

Treehouse.... I read with great interest your post on multiple frequency and duty cycles being generally proportional. I finally got a chance to test a Crydom D4875 zero crossing relay on a 120V lamp using frequency generator with variable duty cycle.  My results are at total odds with yours.  It was a real light show with very little proportionality above 200 hz mostly just being on or off. At lower frequencies there is a vague proportionality but a lot of beat flashing.

[ChrisOlson]

operahouse, can you explain why mine, after I increased the voltage width to 2.0, is showing 220-240 at the element (that varies some within that range), but the amps will go from 1.2 or so up to 7-8 amps?

The higher the amps goes, then the voltage will get up closer to 240.  The lower the amps, then the voltage is closer to 220.  But I never seen it drop below 220 yesterday if the SSR was active.
--
Chris

[OperaHouse]

I found that "showing 220-240"  interesting in the previous post.  Many voltmeters just rectify the AC and then just calculate the RMS AC value from an averaged DC value.  This is not a TRUE RMS.  You may be seeing a long peak hold value from a cycle thaqt occurs very often.  I still find it very strange as a cheap Harbor Freight DVM will give a wide range of values even though they nay not be accurate.  This is likely just a meter issue.  The light bulb test always works.

There wasn't much I could draw from the solid state relay test as frequency dominated pulsing action. One thing was apparant.  As the frequency of the PWM increases the duty cycle for full on decreases.  In the higher frequencies a duty cycle of only 15-20% will turn the relay fully on.  As i mentioned earlir the Midnight likely has a variable PWM frequency.  As you increase the voltage width the upper range will result in the relay just being on because the PWM will be in the higher frequencies. 

[Treehouse]

Did you try 156Hz? There may be some diff. from your signal generator to my pic.. I would assume there is a lot less drift than with a pic. Maybe the more random output of the pic is what is keeping it a little more linear? I too do see a fair amount of beat and flicker, looking at it with different wattage loads my controller seems to "step"  10% at a time. Not as linear as I would have hoped.

[ghurd]

I found that "showing 220-240"  interesting in the previous post.  Many voltmeters just rectify the AC and then just calculate the RMS AC value from an averaged DC value.  This is not a TRUE RMS.  You may be seeing a long peak hold value from a cycle thaqt occurs very often.  I still find it very strange as a cheap Harbor Freight DVM will give a wide range of values even though they nay not be accurate.  This is likely just a meter issue.

Yes, strange.
The typical $1.99/free w/coupon HFs read within a couple~few % of my RMS meters, even over a wide frequency range.
The cheapies are so close I never bother with a good meter anymore, unless the test result numbers are overly anal, which was the situation when I learned that the cheapies are pretty darn good.
G-

[ChrisOlson]

We went to 48 volts on our system with a different inverter and a Outback PSX-240 6.0 kVA auto transformer to supply the 180° out-of-phase leg for our 240V split phase power.

It took some different settings in the Classic with the 48V system - had to increase the voltage offset to -1.0 and the PWM width to 3.0 volts.  With the new system I suddenly had a problem with the transformer.  I made a video of what it does when the PWM width is at 1.0 vs 3.0 vs 5.0:

The total load of the water heating is 8.3 amps @ 240 volts.  This draws 17 amps from the inverter at 120 volts.  With the PWM width at 5.0 the inverter was surging to 40 amps and back to zero every time the transformer made that loud "groan".  So since nobody gave me a better explanation I figured that pulsing the SSR at too high or too low of a frequency bleeds the core down in the transformer somehow and it draws a lot of inrush to re-magnetize the core.

That Outback transformer is one heavy duty unit and it weighs about 50 lbs.  If the core is totally de-magnetized I measured the inrush to magnetize the core at 102 amps with my Fluke meter set to capture the "spike".  And that's with no load on the transformer at all - just turning the power on to it.

This is probably why Eric was having problems with his Magnum PAE split phase inverter too.  He was getting a low voltage problem at certain PWM widths.  But the Magnum don't have the raw surge power of my SW+5548.  So I don't get a voltage drop problem, just big inrush surges to the transformer.

Why it does this to transformers (whether it's in the inverter or an external one like this) I have no clue.
--
Chris

[ghurd]

You misunderstand what they meant.

[OperaHouse]

I use HF meters for everything.  I'm very guarded abou using my good meters for anything.
It is still very disconcerting that very good meters disagee on AC.  I selectivly use
different meters for different waveforms.  The problem Chris has must be due to the meter
using peak values with a long average or possibly a DC offset. I'll discuss that later.
Chris, you should try meqasuring a 12V battery on the AC range and post the results.

Nice this thread is back.  I had done more testing with the treehouse 156hz and other
frequencies again.  Couldn't really come to any conclusion. Going a couple hz in either
direction could change thing a lot at times.  Certainly torn between randomness and
bursts of on and off.  A while back I was reading in a PWM discussion that only complete
cycles should be used to prevent transformer saturation.  In the case of a dedicated incerter,
that wouldn't be a problem. A MSW inverter would likely give much different results than a
sine wave because of the shorter on time.

In performing the test I used a CRYDOM D4875 zero crossing SSR and a 70W lamp on the 120V
line.  To isolate the AC line from the scope I used the first transformer I found, a 277V
to 6V from an emergency exit light. This seemed like a good match since I wanted the inductive
load to be dwarfed by the lamp currents.  I had a couple of AAA batteries on the desk.  Once
I had everything fired up those batteries started dancing around.  At different PWM levels
The scope clearly showed that groups of all positive or negative pulses were grouping together
turning the transformer into a big magnet attracting the batteries.  Even on the 277 tap the
transformer was overheating.  This is likely what is happening with your transformer. The
core develops a DC bias resulting in inrush currents.  I also saw another odd thing happening.
When the turn on pulse didn't last long enough at the start of a line cycle, a partial turn on
occured that only lasted about a quarter cycle.  This was a very strange peaky looking waveform
repersenting maybe just the driver current to the SCR.  I wonder what will happen to some SSRY
if operated in this mode for long periods of time.   

[Eric L]

Chris,

How do more sensitive loads like CFL bulbs look while your transformer is humming like that with the PWM load on? Mine still flicker a bit.

Transformer saturation makes sense to me; I suspect there are still a lot of excessive harmonics on my AC while the heating element is on, even though I've neutralized the transformer hum and voltage drops by adjusting the pulse width. I don't have a scope or even a true RMS meter though so I can't directly check the AC power quality.

[ChrisOlson]

To answer one question, my Fluke meter is AC only.  But I tested a 12V battery with it and it shows 3-4 volts very briefly (about a second) and then goes to zero.

Eric, the CFL's in the house flicker a tiny amount when the PWM load is active.  But, and this is weird - they only flicker on the inverter output, not the transformer output.  Since we put in our new inverter the 240V is gotten between the inverter "hot" and the transformer output.  From either of those to neutral is 120V.  Some of the 120V lights and outlets in the kitchen, and the microwave, are on the split phase leg supplied by the PSX-240 transformer.  Those lights don't flicker when the PWM load is on - only the other lights that are on the split phase leg supplied by the inverter's "hot" output.

I don't really understand that either.  All that transformer does is take the inverter's output and generate a 120V sine wave that's 180° out-of-phase with the inverter's output so you get 240V between them.

One other thing that's strange about this slight flickering of the CFL's when the PWM load is on, is that they don't do that when the gen is running on Load Start.  Our system was absorbing here just before noon and the PWM was operating the water heater.  My wife decided to do laundry and with the clothes dryer running the total load was 7-8 kW so the inverter started the generator to help out:



With gen support the system continues to absorb and operate that water heater element but the CFL's stop flickering.
--
Chris

[ChrisOlson]

One further thing I want to experiment with is using Waste Not to divert high voltage DC direct from the panels to the water heater.  I ordered a 40A DC output SSR so I can try this:
http://www.lightobject.com/40A-Solid-State-Relay-SSR-DC-In-DC-Out-P315.aspx

I have some 7.2 ohm 2,000 watt 120V heater elements that I can swap out in the heaters.  I figure at 90 VDC these elements should draw up to 1,100 watts or so.  The switchover will be fairly easy - simply replace the AC SSR with the DC one (it bolts to the same heat sink), run the DC system ground to one leg, and run 90 VDC from the high voltage DC bus that feeds the Classics thru the SSR to the other leg.

The only thing I'm unsure of is if the thermostatic relays in the water heaters can handle switching 90 VDC power.
--
Chris