Ripple Problem

In the first place, what observation, prompted you to clamp an AC amp meter, on battery terminal? Please check the clamp meter calibration. What is the ripple frequency of 2 volt, over the battery? Isolate the problem to a particular load, by disconnecting each load, one by one. Once the problem circuit is isolated, it can be throughly checked. We can't visualize your layout. Posting a schematic will surely help.

I have a process, I could refer to an earlier post with many variables:

24 Volts batteries as main storage for 2000 Watts of Solar Panels, working in groups of 34 volts (2 times 17 volts in series) but forced to work at 24 volts, by a group (20 pieces) of inverters that switch in and -out, according a voltage (window) ladder with adjustable comparators.

I underestimated the influence of the inverters on the battery bank. This has to do with the nature of the batteries and of course, the capacity.

While I switch the inverters one by one say, from 24.5V for the first, in steps of 0.3 volts up to the max. 27.2 Volts, I try to keep the batteries in their working range,

It works pretty well, except the ripple makes the set points very noisy and also the power relays. These tend to stick once in a while.

I try to work around 0 Amp loading or unloading current for the battery bank.

The meters and clamp work well. The clamp is a AC/DC type.

When I measure the frequency with a clamp frequency meter, I measure 3 Hz, but I have all reasons to believe this is wrong (due to the duty cycle)

The switch frequency of the inverters is about 10 kHz. This I can see on the scope AC output. (spikes) - The oversweep I brought down with 2.2 microfarad polyester caps.

The ripple is around 60 Hz. It has a sinusoid outlay - with distortion, depending om which inverter is on.

At zero load to or from the battery, the AC current reads 18-20 Amps (Hall element after calibration)

Isolating the problem is stopping the process.

Besides using this (signal) threshold as reference, I also switch the relay out of the output of each comparator through 2 transistors.

I can work out a schematic diagram.

Since you brought down the oversweep with 2.2 caps, did you notice any difference in the ripple then ?

If so, you need more caps !

Routinely switch @ 12kHz and 18kHz on different application, and battery wire length/size as well as capacitor size make big differences in ripple voltages

No, the amplitude and frequency of the "ripple" remain the same. Only the "overshoot spikes" decrease.

I also tried a CLC filter 2500 microfarad/ ferrite coil of computer power supply/ 1200 microfarad. No visible change.

The batteries only have wires coming out #6. Are not supposed to pass more than 10 amps each. Length is 150 cm.

The switching frequency I cannot control. = with fixed crystal.

Since your system I presume was perfectly working and functioning as designed, I still think the noise problem encountered was being caused and related to electrical/chemical changes or imbalanced existing in the battery bank. Some batteries that are in series /parallel combination may well be contributory to this problem. I still think that if all batteries in the battery bank are electrically matched and balance chemically, by itself it will act and behave as a very good filter that will clamp and hold the voltage down to the specified battery terminal level. And limited only by whatever conditions the batteries are presently in. Which generally in my experience was limited by either the total rated battery bank capacity or by even a single battery having the weakest condition in that battery bank, which ever comes first!

Thank you for thinking battery-wise.

Surely, if I had a bank with unlimited capacity so that the solar panels never can fully upload the bank. the problem will disappear, since the Tracking point of the panels will always be held under the battery upper limit. I agree that a battery can be a great filter.

2 other problems however:

1. Too much of an investment to increase the bank.

2. What goes to the bank, doesn't go back to the grid, so efficiency losses in a short

delta time. (unless the inverters do overtime in the dark)

I need to build on a new reference that rids the ripple but follows the load cycle of the battery bank in a reasonable time. Perhaps a parallel battery (small - 4 Ah) coupled to the bank with a small resistor, or through a coil may help. Thank you.

No one stated that you need to increase the capacity of your battery bank! All that is needed is to ascertain that the condition of those batteries in the bank. They need to be all chemically and electrically matched! Even "maintenance free" batteries needs to be maintained!

The batteries are all deep cycle gel batteries - installed 3 months ago. Apart from the stabilizing and forcing down the solar panels, they have no other function. What they do also is providing a more stable power supply for the inverters than a cloudy sky.

Sounds like the grid tie could be the issue. Have you tried charging the batteries from the grid instead using a standard charger?

It won't help.

Your design, I am afraid, appears to be unworkable. We don't bring down, regulate the charging voltage from solar panels, by constantly loadind by switching on a number of invertors.

"I can work out a schematic diagram."

Please do.

In the meantime you may try out on one invertor:

In the 24 V DC from battery, feed line to each invertor, inside the invertor encloser, connect an inductor followed by a capacitor, just before the input power connection of invertor PCB.

This sounds like a lot of ripple (at least in the conventional sense where ripple is thought of as undesirable noise coming from, for example the battery charger, which may have inadequate filtering on its output).

I agree with the first post. Find what is causing the ripple. Perhaps a large commutated motor? An inverter?

If the only problem associated with the ripple is its influence on detecting a voltage to trigger recharging, etc., then filtering the detecting circuit (rather than the source) would seem to make sense. This would only require tiny components.

If you have many loads switching on and off, could the clamp-on ammeter be reading the difference in loads from simple switching? (In other words is the ripple of a constant frequency -- truly AC -- or is it random variations?)

Is there a symptom that is resulting from this ripple?

The what I call ripple is a complete sinus, superimposed on the battery voltage. Thank you- I tried to cover it in one reply.

Most likely an inverter or float charger failure. Swap them out for new/higher quality units.

You could design a choke for this, but the bad parts will still be there.

Tray to add difference values of an inductance between battery and the load (inductance prevents A.C component and passes D.C).

Is it possible that the generators in use are producing the AC? Maybe due to slow rotation and if more than one generator, a mix frequency of two or more (even with fast rotation)?

If the AC is superimposed on a DC level, this may be the source.......

Hi Andy,

I produce the generation out of solar panels. Now about 2500 Watts maximum, to be increased to 5000 Watts.

The system works well in principle. I have been switching on and off inverters by hand for several weeks, until I automated the switching with a ladder of windows.

Per inverter I use a comparator that compares the battery voltage with a reference voltage, that I made with zener diodes. The comparator works on millivolts difference.

I can set the thresholds from 21 volts to about 28 volts with a potentiometer.

Thanks for the correction and update, my comments were not accurate, sorry.

NOTES:

1. Tranzorbs are for lightning protection, in accordance with threats of lightning threats in location of installation of these circuits

2. All component values can be adjusted in accordance with requirements of acceptable ripple levels.

3. The common mode choke values can be 1 uH, 10 uH or 100 uH, select values depending upon your requirements of acceptable ripple levels.

4. The Capacitor values can be adjusted in accordance with requirements of acceptable levels of ripple.

5. The functional maximum current and voltage ratings for these filter components must have > 6 dB margins.

6. Tranzorbs shown in dashed lines will be required for Lightning transients > 250V, 60A

We used to produce transzorbs for a different application: garbage trucks for Henschel Engineering. The electro- hydraulic valves used to send peeks all over the truck's electrical circuitry and e.g blow up all the dashboard lights.

These transzorbs started working at 27 Volts both ways.

The EMI must be suppressed at SOURCE, not at loads, unless loads are also sources of EMI. Once the schematic design has been optimized then PCB layout through auto CAD is very critical, if best high frequency design practices of PCB layout are not used then, all best design approaches tried at schematic level would be futile due to lack of understanding of CM nose and Differential mode noise problems of propagation through PCB structures and how to handle or mitigate through best schematic and PCB layout practices!

The EMI must be suppressed at SOURCE, not at loads, unless loads are also sources of EMI. Once the schematic design has been optimized then PCB layout through auto CAD is very critical, if best high frequency design practices of PCB layout are not used then, all best design approaches tried at schematic level would be futile due to lack of understanding of CM noise and Differential mode noise problems of noise propagation through PCB structures and how to handle or mitigate through best schematic and PCB layout practices!

Both stages of product design i.e., schematic and PCB layout are very critical…..

Do's and Don'ts of Good EMC Practices

Do's

1. High current switching circuits must be shielded inside the enclosure to

mitigate inductive and capacitive coupling to sensitive analog circuits.

2. Run traces for VCC and Ground connections in power supply circuits. Use

flooding of VCC and Ground in power supply circuits sparingly. If flooding

of these planes is to be used, use it in a very localized fashion for

switching transistors circuits areas only, to control high frequency signals loop antenna areas!

3. Do not flood high pass and low pass filters inside power supply circuits

with VCC and Ground planes.

4. All input power and ground pins must have adequate filtering for EMI. All

output power pins must also have adequate EMI filtering. Proper use of

common mode chokes with capacitors for differential and common mode

EMI to form T or π- filters is strongly recommended.

5. VCC and Ground connections are common connections among all circuits

and hence can make common noise appear every where from one circuit

to the others if not connected correctly through filters.

6. Use de-coupling capacitors at the VCC and ground connections of IC's

requiring high instantaneous switching charges (currents). Use 0.01 and

0.1 uF in parallel to provide low frequency and high frequency response.

7. When there are Analog, digital, RF , and power supply circuits on the

same PCB; keep VCC, DGND, AVCC, AGND, and all other VCC and

Ground connections isolated from each other. Use star type connections

within each circuit areas.

8. Use damping resistors as required in high frequency signal transmission

lines.

9. Use pull-up and pull-down resistors to mitigate any false triggering of logic

circuits.

10. All unused logic circuits connections must be pulled up or down

11. Keep high frequency transmission lines as balanced circuits from source

to load.

12. Use software techniques to mitigate RF immunity problems.

Don'ts

1. Do not use X7R type capacitors in lightning protection circuits. They pop

during lightning surges. Use NPO or other high frequency dielectric

material capacitors

2. Do not run clock, read and write traces very long and in zig- zag manner

through out PCBs. These traces ought to be short and direct from

source to load.

3. If all high frequency components such as microprocessors, clocks are

installed on top layer then keep high frequency traces close to top

layer. In other words in case of an eight layer boards, if high frequency

components are installed on layer one then short high frequency traces

should be either on top layer or next first signal layer below those

components not the bottom most eighth layer!

4. Chassis islands, VCC and ground planes should not overlap on PCB

layer stack-up. If they do, there will be capacitive coupling between

them

5. Do not mix sensitive signals with high current and high voltage signals

in cables. Use shields and shielded twisted pair topologies.

6. Do not mix filtered signals with unfiltered signals.

7. Think of highway going around the city and only filtered exit and entry

in and out of cities. Highway being transmission line or cable and city

being individual board on a mother board or backplane or mid plane

etc.

8. Most common problem is common mode noise, which finds its path

through shields. So proper connection or termination of shields is very

critical

9. Capacitors are bidirectional devices. Do not think that they work one

way!

10. Think of filter components from low frequency and high frequency effects

Model point of view.

Hi cbob12.

The systems works better every day, due to also your contributions. I even decided to remake the print to implement your ideas on it. The capacitor was your best input and also the self made transformer coil in the power circuit to the batteries.

I will remember all your Do's and Don't also.

Once the prints are made, I will publish some pictures.

Thank you

Do you think I can use this filter (modified) to connect between the battery bank and the inverters? I am still worried about the AC running through the batteries - up to 85 Amperes (measured with clamp) while the DC ampere meters show zero (system in balance) The wires to the batteries warm up a lot - even with the 0 amps DC.

I will isolate the AC ripple to the battery too with a CT and post the shape of the AC

component. I noticed that it is not always symmetric.

Thank You.

Hi dvmdsc,

I used to work with batteries and chargers as a power system designer. According to my memory, the permissible AC current for a battery was about 1/3 of the 10 hour amp hour capacity - in your case, 400/3 = 133 amps. I would not worry about that, with single phase chargers, a large AC ripple and battery AC current was a fact of life.

The warming of the battery wires is, I guess, because not expecting 85 amps, you never rated the cable for that much. I guess, not knowing your ambient temperature, that AWG #6 is OK for about 50 amps in the usual enclosed location and about 80 amps in free air - at which it will run at about 70 Celsius surface temp, too hot to touch.

If you keep the AC away from the battery with an inductor, you need to give the inverter a similar easy path or you will get a lot of ripple at the inverter input. I estimate your battery impedance as about 10 milliohms - an equivalent capacitor would be about 130,000 microfarad.

Will continue later,

67model

Continuing from post #54,

An inverter designed to run direct off solar cells would have the necessary input voltage range and input capacitors built-in.

An inverter built to run from a battery could take advantage of that low impedance to avoid large and expensive input capacitors. It may not like being fed through an inductor. In any case, 120 Hz inductors with DC current tend to be bulky, expensive, awkward to design and difficult to modify if you got it wrong.

I am concerned by the battery voltage range OP dvmdsc gives in post #4. If you want a 12 cell lead acid battery to get full charge you must keep it at 27 volts for a long time (2 or 3 days). At 25 volts it is discharging.

If you want your battery as a standby, not as a store regularly discharged - say overnight, then it is not a good idea to run it partly charged. At a steady 27 volts it would stay fully charged for years - float charge. If you fluctuate below 27 volts, you really need time above 27 volts to compensate for the periods below 27. The "27" volts is varied a little by the design of the battery and cell temperature.

When the battery is sealed, one cannot use a hydrometer to measure the specific gravity of the electrolyte, in order to judge how well the charge/discharge management procedure is working.

You probably have a practical working system, do not think I am saying it is wrong.

I think my approach would have been to switch on an inverter at 27.5 volts and turn it off at 26.5 volts. If the volts did not drop below 26.5 before rising to 27.5 volts again, the NEXT inverter would be switched ON (two now on) etc. As solar output fell, inverters would be switched off in succession. There would be just one common voltage detector, probably the difference between on and off voltages would be altered in response to light level [because the weaker the solar output, the more stepping the load by one inverter will shift the voltage].

Judging if a system is good or better in real life depends on having records of the really important factors [by hand or automatic means] for "good".

Your objective is probably to optimise battery life by avoiding over or undercharging.

1. What is your main objective?
2. Do you have a monitoring plan?
3. Is it telling you what you need?

Hi 67model,

Your input is highly appreciated. The batteries are gel deep charge sealed types designed for solar. That is why the wires straight out of the batteries are that thin. In all the conditions up to now, I still can hold the wires in my hand. The battery temperature stays below 100 degrees (37 Celsius).

I need to start switching inverters at 24,5 Volts DC ( there are 20 pieces) and I do this progressive. When I start at 27, I have not enough margin and the panels over rule everything: they emit 33-34 to the batteries and at 28 Volts the inverters cease to work.

This condition I can only cure with covering cells of panels, switching on everything and uncover the panels again.

According the manufacturer of the batteries, they can be used between 22 volts and 27.3 (because of the deep cycle and gel)

I use the batteries to get more out the inverters: when directly connected to the panels, the MPPT tracking time is a lot longer and the downtime increases, resulting in less output. I run 2 systems parallel and with cloudy weather, the battery equipped line produces up to 12% more power.

I monitor everything with EKM smart meters (give V,P,kWh,I and cos phi)

The battery set runs always between 0.9 and 1 as power factor, while the other string falls down to .67 with cloudy weather. These are momentary real time results.

The average probably will be better. The design of the inverters is pretty poor: they do not have big capacitors and work at 10.000 Hz to switch.

I did some tests on the reference voltage with the design in a previous post. I want to do a test now with a heavy ferrite coil and 2 sets of windings, to try to eliminate a considerate part of the ripple. (sort of eliminator current transformer)

You can see the principle in the second post.

The ripple model will be posted soon.

Thanks

D.

Hi dvmdsc,

My "rule of thumb" is that 44 Celsius is too hot to keep the hand on - easy number to remember.

Since battery life halves 20 to 30 Celsius, then halves again 30 to 40 degrees (that means that 8 years in the sales brochure could fall to 2 - regardless of depth of discharge and number of cycles which have their own deadly effects) , I would be careful of that temperature. What is your climate/ambient temperature? Day/night variation?

Do you mean that the advantage of the battery is that your sun tracking system always works quickly with continuous power? If that is the case, it might be better to run that tracking system isolated and powered from the battery with a normal AC input charger.

Does your smart monitoring automatically permanently record every value at regular intervals?

It may seem elementary, but does your voltage comparator rig have its own power and sensing connections (+ and -) direct to the battery posts (fuse advisable)? My guess at the supply impedance at the load end of your #6 AWG is about 10 milliohm (of which only 1 mohm is battery). A lot of your ripple could be due to the resistance of the wires N.B. 10 mohm x 80 amps = 800 mV.

Did you note my post #49 about the comparator circuit in your post #29? You have no ripple filtering on the sensing input of the comparator. I cannot read the value of the feedback resistor to give the comparator set-reset voltage hysteresis.

Reducing the ripple current into the battery does not seem a trivial operation. If each inverter is fed by a DC cable with 0.5 volt drop at 10amps (50 mΩ) then a 33000 μF 35V capacitor across each inverter input (40 mΩ at 120Hz) should reduce the current going back to the battery by 6 times. Such capacitors are about 5 euro each, 25 quantity, from Mouser.com. Maybe 6800 μF will halve the current, however the impedance of any capacitor built into the inverter input could mean it is not that good.

A way to find out would be to connect one inverter with 50 mΩ cable. Measure ripple voltage across -ve feed cable [ripple current = voltage/0.025]. Put 6800 μF capacitor across inverter DC input terminals. Measure ripple again and compare.

If you are going to try a double wound toroid power filter - beware - the usual connection polarity for such windings as "common mode filter" [both incoming supply, L & N usually, to winding "start"] cancels the supply current effect on the core but means the effective inductance to supply current ripple - series mode - is negligible. If you reverse one winding polarity, the DC will saturate the core or you can get a situation where the ripple takes the core in and out of saturation causing voltage spikes. Toroid cores have no air gap which is why they saturate easily and draw huge inrush currents (limited only by winding resistance) as power supply transformers [I have measured the 240VAC inrush peak of the toroid transfo in my 30+30 watt audio amp (secondary open circuit) as 5.2 amp, 1000 times the steady state magnetizing current].

67model

Hi 67model,

Today is a cloudy day. The average temperature of solids (garage door- floor- walls) is now 30 Celsius. The batteries have about the same temperature.

But again, today is cloudy. Temperatures here reach easily 45 degrees and 25-30 during the night. We have no heating bills, but enormous AC costs, especially because the heat comes with high moisture content (70-99% relative)

The 5 windings of the toroid measure 33 Celsius. That is 2 degrees lower than a #6 wire I used to connect one of the windings. The toroid transformer is one of 500 Watts, that I brought from Europe, 15 years ago. I used to make first generation inverters (classic 50 Hz with Power transistors) They have 2 X 24 Volts to 230 Vac.

I didn't remove the windings and measure now between what was supposed to be the 230 Volts side. Values there change between 60 to 130 Volts. (depending on how many inverters are on)

With the 5 windings, I use this transformer as choke, one wire to the + and reverse, one wire to the minus and to the batteries. Theoretically, this should eliminate the ripple. I still see phase shifts, probably because of the wiring differences to the transformer. I also tried the saturation way and saw some steep spikes. One inverter killed a mosfet over it. The weakest like always.

I had no clue about the interaction of the inverters towards the batteries when I started the comparator. In the beginning I even didn't put capacitors in the zener circuits. One zener supplies the reference. The other powers the comparator. The inverted input to the comparator is provided by the battery floating voltage through the potmeter array. I must give credit to a friend who helped me over this.

These comparators switched the relays with a few millivolts, fed from a test power supply. I have one dedicated line from the battery, but I also power the circuit with it.

The circuit uses 22 mA and the relay 160 mA - enough to induce some extra variations in this battery follower.(10 relais on = 1.6 amps extra) The comparator input (resistive divider) is about 120kOhms, enough to make it microphonic - it induces even the hum when I touch the potmeter spindle.

I have a new series coming up - MkII - with nice printed circuits and less defects- but need already a MkIII edition. Keeps us busy?

The inverters are in a rack and fed by a 2/0 in almost a plumber array. Solid copper strips to the - and + of the relay power inputs.

I filtered the battery follower line and have 125 mV ripple when the battery has 4.5 Volts ripple. Tcmtech's solution to use a RC filter works well, but before I re-wire the comparators it brings down the voltage out of range of the comparators.

Re- design some resistors is one solution, together with feeding the end stage and relays straight from the panel line. (that shunts the batteries)

However. I have steps of 0.3 volts to play with. (10 comparators =3 volts margin)

I assume that as long the 230 volts equivalent is not reached over the old winding of the transformer, that the core is not close to saturation? I also can make a core with 40 X 0.5mm band iron that I have (strips -leftovers of my roof) I can wind these in a toroid too.

The monitoring system is only on the AC side and measures 2 times 6 parameters per minute. 1. Power rendered, 2 Power used, 3 current - no polarity, 4 line voltage (is between 230 and 248 Volts max. with 40 amps backfeed)

I read your note in your post #49 and you are right. The work -schematic on the picture has been modified afterwards: the feedback resistor (hysteresis) on the comparator is now about 330kOhm, also a 0.1 microfarad polyester cap has been added. The hysteresis now is about 0.18 Volts - workable with the induced 125 mV ripple.

I know there is still a way to go to make this system work flawless. I also do not want to offer too much power in the design, because it brings down the efficiency.

(the watt to the relay doesn't go back in the grid e.g.)

I also figured out that the inverter supplier could have put more work in the input capacitor(s). It runs with an ample 470 microfarad. I assume he thought of the clock frequency of 10kHz instead of the 60 Hz required to deliver.

Why the batteries?

In the beginning I had some questions for the designer of the inverter:

Can you modify the inverter to work, starting at 13.7 instead of 14 volts? My idea was to connect the panels straight to the panel, shunted with the inverter. I would have had a fully loaded 12 Volts bank for back-up power and a GTI system that didn't have to wait for the MPPT.

He told me to use 24 Volts batteries and I followed him, with all the circus stunts I have now. The inverters indeed produce a lot more this way, but keeping the batteries alive, requires the whole set up with shunting inverters over the panels.

The system works great, but needs paint and polishing. More pics are coming I really appreciate the help from all the Techs here. Together we will have learnt something. Thanks. D

Hi dvmdsc,

OK, hot and humid - the least comfortable climate and so the A/C to cut the humidity.

Toroid - when you wrote "ferrite" the alarm bells rang - no good for 50 Hz. My figureings "on an envelope" had come up with 50Hz transfo of a few hundred VA rating as start point for a choke core if you had one on hand - so you beat me to it. As you write, the way to monitor for spikes is to use an oscilloscope - a mean reading on a meter could hide a short but huge spike. Anyhow, since the choke is between a low battery impedance and low impedance inverter inputs, spikes are not likely to give a bad surge.

About the voltage level switches - you have got an effective system by filtering the ripple from the power feed of the whole bank - if you do a re-build, note my suggestion about 10μF bypass from the pot wipers, also there is no need to filter the relay coil power, the high coil inductance does not pass 120Hz when connected to a DC supply with ripple [that will reduce the choke or resistor rating needed].

If you have spare channels on your AC logging monitors, it may be possible to chop the battery DC from a DC shunt/voltage pot. with a 50 Hz transistor switch, filter to a reasonable sine, amplify and feed to your AC monitor as "50 Hz current/voltage" through an isolating transfo. Before high performance transistors came along, a mechanical chopper switch and AC thermionic tube amplifier was one way to get a sensitive DC meter without too much drift, the first good transistor thermocouple amplifiers I remember used choppers because transistor drift as DC amplifiers was not good then.

Regarding the battery, usually they do not mind high charging currents, even 400 amps on a 400 Ah battery so long as the voltage is kept down to 27 (for 12 cell Pb battery or the temperature compensated equivalent). Once the battery gets well charged it limits its own current. The problems occur once the battery starts to make gas, which needs sufficient voltage. Since they got really good 14V nominal voltage regulators and alternators, motor car batteries live this way, with a 35Ah battery having a 45-60 amp alternator.

67model

Again thanks to all the contributors.

Here some more info:

To keep it simple I include a principle diagram. This might be all I am capable of.

There are Relays used that are switched with a threshold voltage detector. Here the prototype:

The ladder array of the inverters is shown here;

This picture needs a 90 rotation to the right

The battery setup is here:

The ripple before it exixts:

then we start one inverter (happens with all of them) and:

The ripple is 1.7 V ptp and has a frequency of 120 Hz. (119-121)

It looks like a sinus but has some distortion.

All comments are more than welcome.

What is your basic purpose of the setup? Do you have no electric power available, remote location?

How many solar panels @ 34 Volts?

What is the power out put of each solar panel?

When inverters are switched on to load, solar panel's excess out put, do you need to use inverter out put or it just goes waste? Can you not disconnect a solar panel instead?

There is a huge mismatch, between battery bank of 24 V & solar panel out put of 34 V

Filter the 24 V input to the most troublesome inverter. Are all the inverters same type, branded or home-brew?

The meter may be measuring DC Amp. on AC range. I doubt that 20 Amps. measured is from 2 V AC ripple.

1.To produce power out of solar panels to the grid. (To lower my bill with $150.00-300.00 per month as is now.)

2.The solar panels are typical 125 Watts, 12 Volts, but as you know depending on the load match the 7.5 Amps@ 17 volts can also change to different values if forced.

Short cut, the amperage is 8 amps. And from 0 to 17 volts, evreything between. I checked my specific panels for max. power and that is @12.7 volts.

3. 2 in series= 34 volts. In use on this array 20 panels Total = 40 panels for next extension.

Except for the losses in the inverters, I use each part of sun. The inverters like a 24 Volts base instead of a fluctuating solar panel output. This shows in the output.

A 17 volts panel has enough voltage left to charge a 12 V battery with inclement weather. (current is lower of course)

In principle e.g. in Europe, they use inverters that even bring a 12 Volts panel down to a few volts, to generate with a little bit of sun. (Fronius- Austria)

4. All inverters are the same brand and capacity.

2 Volts X 20 Ampere is only 80 Watts. The system generates now peak 2400 Watts. So this is 3%? not so uncommon. (in fact RMS is even less- see measurements- now 1.4 PTP, recently reduced again to 0.85 PTP)

5. Measurements with a AC clamp Ammeter show the same. Also with CT. CT doesn't show DC.

Thank You

The problem frequency you mentioned earlier was 3 Hz if I remember correctly, now it is 120 Hz, a very interesting value indeed for anyone generating AC for a 60 Hz environment from batteries with an inverter.

A 60Hz AC inverter will switch on the load to the battery at exactly that frequency....... so its not really a problem as I see it. Its just showing the loaded and unloaded voltage on the battery as (usually) the transistors switch first one side then the other side of a transformer with a center tapped primary to a 120 VAC secondary....

You cannot avoid this effect completely, but with a careful design of extra capacitance/inductance on the battery/low voltage side, it can be reduced....

The reason that the Sine wave is a bit rough and not smooth is that many modern inverters are so to say, "digital" (as a word to use) and instead of making a proper smooth sine wave, it is broken up into approximate segments...it works just fine for most things, except possibly for some really delicate electronics (that were to my mind not properly designed anyway!!!).

So my guess is that its just the effects of the inverters.

When more than one inverter is switched on, the question is are they "synced" or not?

If synced (running in parallel perhaps) then the load on the battery/s would all happen at the same time and accentuate the size of the signal seen on the scope..... I have no idea if they are synced or not at this point....if not synced the load might produce other frequencies/difference signals.

Just a thought or two.....

Hi Andy,

The clamp meter that has also a frequency measurement position shows 3 Hz. I mentioned that I didn't trust this, because of the character of the load.

The oscilloscope gives us right. All the inverters step synchronous. About 10.000 times per second, they compare the grid frequency and magnitude and superpose a block on top of the grid. Behind the first transformer these spikes are not noticeable any more. The inverters work on N and L - one series takes L1 and another N and L2.

I did this because I can keep the cos phi between 0,99 and 1 this way.

You have the right thoughts. The sinus over the batteries increases in amplitude from 0.5 Volts to 1.2 Volts (RMS?) when more inverters are switched on.

It should be clear that the 120 Hz ripple is coming from the multitude of inverters loading and unloading as they generate each half of the 60 Hz sine wave for their gird connected side.

As someone who has designed and built his own grid tie inverter units I just cant see the point of the highly complex design you are using.

If it was me I would simply use the battery bank as an independent system charged through its own MPPT tracking charger and let the rest of the power from the solar panels dump into one correctly sized GTI opposed to having multiple smaller ones switching in and out.

To me this just comes off as an over engineered design that's being nit picked for not working for no reasons other than because its unnecessarily complicated for what it does.

Hi TCMTech,

I have followed your contributions, also on other forums and I recognise you as a inverter authority.

I tried here (in the Bahamas) several grid tie inverters and most have problems following the grid frequency. They switch to islanding @57-58 Hz which seems very common here. The inverters I use just copy the frequency and work between 40 and 79 Hz without compromises. I deal here in total with about 6000 Watts in different circuits:

4 are handling 1200 Watts grid tie, panel to inverter and work on 48 Volts ( actually MPPT'd @ 64 Volts)

The feed line between the panels and inverters is about 400 feet. I used #6 for each group.

I have plenty panels left that are configured @ 24 Volts and have a 2/0 running to the inverter and battery bank.

With the system I use, I run a 4.5 ton AC with scroll compressor. The start current, I get out of the battery that works as a booster - more inverters participate then.

This is the diagram for the comparator:

Works very good, but as you see the dog runs away when the ripple makes the relay(s) sound like buzzers. Right now, I feed the relay transistors also through the reference line. The measure circuit (potmeter circuit) is pretty high resistive and acts a bit sensitive (even when I touch the potmeter).

Getting this circuit lose from all the rest will probably allow me to make a better filter, since the time constant is totally different- almost no current then. (10 mA per relay) against 160 now.

My biggest problem is no components to try it out short term.

The reason for the "overkill" I will explain in my next reply.

There are serious problems with your design, your description, your measurements, or all three. Since you don't seem capable of a schematic diagram, give us a block diagram if you can. I take it that you are generating 120V at up to 50 amps to replace your line power. Is that correct? Is the output of each inverter in parallel? Are the inputs in parallel, in series, or isolated? With 0 amps DC, there would be no load, so no switching would be taking place. Therefore there should be no noticeable ripple.

Probably the problem is me.

When I took the picture of the simplified diagram, part is missing. And now my camera protests.

Imagine an array of solar panels from the left to the + on that picture.

Like this gallery shows:

These panels can easily supply 200A @24 Volts. The 24 Volts is what I want to use about as PPT - Practically, I play a bit between the load state of the battery bank - (from 21 Volts and higher) depending on the sky state - Clouds or no clouds -

But the 17 volts panels deliver well at this "forced voltage". The max. allowable voltage is 27.2 for the batteries. But no guarantee is given that the panels do not force to load the batteries to 34 volts, especially with a up to 200 Amps load current from the panels.

The battery bank can only have 20 Amps to load or unload.

So I decided to take care of the 180 Amps by switching on inverters as load. In automation this on/off state is about the worst regulation, so I decided to switch inverters on/off in steps of 10 Amps getting more a proportional control system.

While in the one switch all, the battery ref. voltage is about uncontrollable, the progressive system takes care of this within a few volts (- or plus on the 24 Volts DC)

I do this with comparators that switch the relay within 0,1 to 0,3 Volts. And give these all a different (ladder like- Vu- meter) set point.

You are right with the partial schema in mind- no current no ripple. But with the panels connected, the battery bank functions as a big capacitor, however with voltage limitations that have to be strictly controlled.

I generate from the main current source, weather controlled, with all the inverters in parallel, as well on the DC side as per group L1 and L2 on the AC side.

I use this system for 3 months now and used to wake up and put my work desk next to a switch panel to switch on the inverters manually. When I thought a comparator could do the job, it misses some intelligence.

I have put some hysteresis in the system but the ripple is pretty embarrassing.

With 0 amps to or from the battery, there can be for 180 Amps or 200 Amps inverter activity with disturbances to the battery bank.

The measurements are sound on the photos.

I still try to use a single inverter but need good references- I am an alien in this strange grid system and most 240 Volts inverters compromise the Neutral and/or Ground connection.

Thank you for sharing your thoughts, D

Wow and huh? Thats about all I got to say out of this whole thing.

If it was me I would just have the battery bank charge up to a specific point then go into a trickle charge mode while the remaining power gets dumped into the inverters and skip the whole power control micro managemnt issue all together.

Right now this all sounds like you just have way too much time on your hands.

But thats just me.

Thanks. Problem is that my battery bank is WAY TOO SMALL to work it out this way. I would need about 5 to 10 times the size. While I really do not like a bank (= too expensive)

One reliable 5 to 10 KVA inverter, say 200-300 VDC to 120-N-120 AC/60Hz is welcome. (and one spare?) should do. Just have no clue which one. I have a 85 kg jewel here but it only works for Europe. When I ask for a 60 Hz print it is a no.

I have a separate bank for 12 Volts emergency LED light and the security systems on a Fronius 500 Watts non- grid tie inverter. Does the job for years already.

Hi, what made me decide also is that I can easily and quickly repair.

The small inverters have cheap MOSFETS (22 cents or free) and are repaired in 10 minutes. They never break down more than one at the time. Now even O for 3 months.

So redundancy on this island here means a lot. Electronics is taxed here more than 50% on the cost price and the same on transportation and shipping costs.

Okay I see what you are trying to do here. Basically you are using the GTI units as a stepable shunt load to limit the battery banks peak charging voltage.

It could work but really you only have a few options to work with.

One is simply living with the fact that your battery bank is going to be working more as a massive capacitor than an actual battery system so with that I would recommend setting up a single point reference for your comparator input reference point by running that as a sub circuit off the main 24 volt buss as basically just a simple resistor connected in series with a capacitor set up to a value that gives a RC time constant well below the 120 Hz ripple. Something with a time constant that gives perhaps a 10 volt/sec rise/fall response time. That should be slow enough to keep the comparators happy but still fast enough to allow for the GTI units to cycle quick enough as the buss voltage goes up and down.

Another option is to change out the analog based comparator design with a basic digital control of sorts. I personally use TECO PLRs for all of my control stuff being they are cheap, well built, reliable, and easy to program plus the software is free as well. With that concept you would be able to set each GTI units individual working range window limits and cycle times plus any nessisary glitch filtering which would stop all the odd oscillation and line noise interference issues.

The other advantage of that is the TECO PLR units have multiple analog and relay or solid state input and output configurations which give a considerably greater range of control over things. The analog inputs they use are standard 0 - 10 VDC with a programmable resolution of something like .01 volts which even when using a resistor divider network would till give you a much better control accuracy over things.

Check them out here. http://www.bb-elec.com/?jadid=12930465846&jk=bb%20electronics&jkId=8a8ae4cd38a702740138dd8072860a5a&jmt=1_p_&jp=&js=1&jsid=31200&jt=1&gclid=CJG3zcLMmLICFYkWMgodeXUAiQ

TECO PLR units. The software package is free to download and try before you ever buy a single PLR unit. I found it rather user friendly and easy to learn myself.

If you read my work elswhere you probibly already know I tend to lean toward recomending using digital for the brains and analog for the muscle for anything but the most basic GTI design.

Yep thats the ones.

Whats too expensive about those? $150 barely buys me one tank of fuel for my pickup or not even one trip to town with the wife.

I like doing all the bits, not just the programming.....

Hi TCMTECH,

I tried your first solution: a RC circuit with adapted time constant. It improves the reference voltage very well. I used a 13 Ohms resistor and a 2500 micrF capacitor.

The problem I had is that I still use more load when more inverters switch on.

The 13 ohm resistor warms up but drops too much voltage then to keep the zeners in the comparators in range. Changing the comparator resistors is quite some job, while they are installed already.

I found a 50 Watts transformer from a burnt down power supply and used the low voltage as self. Luckily the ripple came down from (the ultimate max. 3,5V PTP to 125 mVolts PTP)

This helps a lot. The relay array is pretty quiet now.

I also have a question about the split phase: 120-N-120 VAC/60 Hz.

Is it safe to use 240 Volts systems with no N? What are the dangers? Here they use 240 volts appliances as well as 120 Volts types. Does the system keeps its balance with uneven loads?

Send a PM (private message) from your profile/mailbox to TCMTECH, for early response.

Your question:-

Is it safe to use 240 Volts systems with no N? What are the dangers? Here they use 240 volts appliances as well as 120 Volts types. Does the system keeps its balance with uneven loads?

Of course its OK provided the frequency either matches or is unimportant and the on/off switch is double pole......

The only reason we use Neutral in Europe for the 240 Volts is that phase to phase is 380 VAC or so......

There are no safety problems for you except that some countries may only have single pole switching of the phase, not the Neutral, the UK for example (used to?) allow this. That should be changed to a double pole switch if true.

Due to generally non polarized plugs in many countries of mainland Europe, double pole switches are (must be!!!) fitted as standard.

This question I launched to check on specific US type split phase power.

Here we have transformers on the poles, fed with 7.000 Volts AC/60 Hz and stepped down with a pot transformer that brings out 120 Volts as L1- a centre tap - zero/neuter - and another 120 Volts as L2. This is a residential net where we on the islands have the zero/neuter also connected to the ground.

Lights and small appliances are connected between L1 and the neutral or L2 and the neutral. Between L1 and L2 is 240 VAC/60 Hz.

There is not a lot of choice to connect a 240 Volts grid tie inverter, other than between L1 and L2.

I thought when there are unbalanced loads, the inverter needs its current to run to the transformer on the poles, to get a neutral in play.

My question is: 1. Will the neutral not "float" and divide the 240 Volts unequal? and 2. How does the inverter deal with the ground that is connected to the neutral in our grid?

An inverter like Fronius doesn't come in a "split" phase version, but has a UL certification. Are there dangers involved?

My question is: 1. Will the neutral not "float" and divide the 240 Volts unequal? and 2. How does the inverter deal with the ground that is connected to the neutral in our grid?

A 240V appliance has no reference to the 2 halves; it just needs 240V. The case of the appliance may be connected to the neutral and to earth for safety.

Thank You for the help.

The 240 volts appliances I am not worried about.

What happens with a 3000 Watts load's 120 Volts voltage, if only this is switched on in the configuration where the inverter supplies e.g. 5000 Watts into L1 and L2 (240 Volts) and only the 120 Volts is needed?

How much voltage will it get?

D.

Ah, I think I see the problem. You have 240V inverters with 120V loads, so power is dumped through the other half of the transformer. That looks like bad design to me. A 120V system would make more sense, as your post 20 seemed to indicate you had.

No the inverters are 120 Volts. I just had some thoughts about the functionality of 240 Volts inverters on a US grid. Thank You

Answer:- 120 volts.

Don't forget that the mains in your hose is basically a center tapped 240 volt, single phase connection (I believe phase to phase). The neutral center tap of this is also grounded at some point. (This explanation may be a bit too simple, wait up!)

After this the neutral and ground are always used as two completely separate wires/connections.

I forgot to mention earlier that plugs and sockets for 240 VAC are physically different to ones for 120 VAC for very good safety reasons.....even in the same area/country.

Your question:-

My question is: 1. Will the neutral not "float" and divide the 240 Volts unequal? and 2. How does the inverter deal with the ground that is connected to the neutral in our grid?

Answer:- Your worries are completely unfounded. The neutral will be at some point grounded, usually at a substation, depending upon your locality and the code for that area.

Actually I have already answered most of this in this blog. But if you need more info:-

You should not attempt to connect neutral to anything when using a 240 volt device where you live (a 120 VAC 60 Hz area), just ignore it, though technically speaking, it will be connected via the earth cable (and the path back to the substation to where the earth/neutral link is made) to the case or frame of any device requiring an earth connection....only use the specified earth cable though for safety reasons.

Using the neutral in some situations as an earth in your house can cause REAL problems.....never ever do it.....unless it is allowed for in the code as for example in Germany away from the cities and towns (like me) where no earth is supplied via the mains cable to the house, only 3 phases and the neutral.

So here each house has an earth spear deep in the ground, the neutral is connected to it and so is the earth bar in the fuse box. The earth and neutral are then treated as two completely separate wires in cables within the house.

If it was not so done, ELCBs and the like might have problems to work fully correctly....never tried it myself, so I cannot say exactly what might happen. It probably depends to a degree on how the earth fault is sensed within the ELCB or similar.....slightly out of my knowledge area. There have been many developments over the years with such equipment and someone else here will know far better than I do....

Final note. Earth and neutral, to the best of my knowledge, should only be connected ONCE to each other. Adding extra "parallel" links can cause serious problems of circulating currents in unbalanced single phase loads and the like - not good. Especially if your electricity meter is in circuit with them (I am told!).

It is at least then theoretically possible that you might pay more for electricity than you actually use....(I think!).....can someone who really is up to date say it as it really is please?

Follow code correctly and you cannot go far wrong and you are fully legal as well.....

Thank you Andy,

The industrial net is similar to the one you use in Germany and I used in Belgium.

Here for residential purposes they also have one phase, the poles in the street are similar to the ones used in Europe 50 years ago and the streets here have Can transformers attached on the poles. They supply to a few houses per can with 3 wires, the neutral is a centre tap. The ground is a blank wire along the pole and is connected to the centre tap too. Each house has to supply in a good ground and that is connected to the ground of the pole too, and the transformer can.

The distribution panels have indeed separate ground lugs and also neutral lugs.

In the panel these are jumped together. This way they create 2 times 120 Volts to the neutral from 1 transformer coil.

Thank you very much for your help

I see you solved the ripple problem with a filter external to your voltage level switches.

Starting from scratch, I would note that a 6.2 volt zener at 5 mA current has an impedance of about 10 ohms or less to AC current. So the 10 microfarad across the zener (about 150 ohm at 120 Hz) is doing nothing useful.

On the other hand, the sensing input of the op amp from the potentiometer has no filtering at all.

I would have connected the + of the 10 microfarad to the slider of the pot (disconnect from zener, of course), to make a filter there. I realise that components "in-hand" are precious assets to you, and maybe the 10 microfarad you have is 6.3 volt working rather than 10V - but 10V would be a safe rating for the pot filter while 6.3V is not.

240V appliances should be safe on a 120-0-120 volt system. All user switches and overload circuit breakers must be two pole, else the appliance remains live at 120V while apparently isolated. Double pole fuses would likely blow only in one pole on fault.

The 240V appliance will not unbalance the system, it cannot do other than carry the same current at both its terminals. A normal system would keep an adequate voltage on both halves - so long as you do not exceed the rated current capacity of either half.

67model

Mr.dvmdsc can you publish all the schematic digram of your circuit design ?

with courtesy of Roland Van Leusden: here the basic comparator.The resistor R7 has been lowered to approx.300kOhm fo more hysteresis. Also a 0.1 microF capacitor has been added. Soeey for the late input. D.

In reply to #62,

For the given circuit, hysteresis on/off is about 0.22V : set at 25 volts. Hysteresis would be in proportion to setting.

A capacitor of 0.1μF from pot wiper to ground [not on circuit] reduces 120Hz ripple by 2:1.

As per post #49, moving 10μF from zener to pot wiper would attenuate 120Hz by 150:1. This would cut 1.5V pk-pk at 120Hz to 10 mV. There would be a simple [RC] time lag of about 0.22 seconds.

Regards,

67model

Are you facing problem like an oscillation in relay when the Tr.2N2904 switches on (passes current to relay=R3)?

The relays are switching the inverters on/off on the DC side. Once an inverter has been switched on the current to invert starts at 0 and build up to about 12 amps after 20 seconds. 2 effects occur: 1. more feedback to the batteries, also more current from the batteries, resulting in a lower pilot signal from the batteries (24 Volts + the down filtered ripple -from 3 V PTP to 250 mV PTP) This small ripple makes the relay sing - like a zooming sound before it clicks in.

2. Of course also depending on the average signal voltage the comparators switch according the ladder set points.

The capacitor (I took 1 microfarad) deals with the ripple now. 10 micro made the switch too slow. Inverter load didn't come up on time, resulting in a too high battery voltage (28 V plus) and that is the threshold for the inverters. They do not start up anymore.

Your question:

No, the voltages remain constant enough to keep the relay Transistor saturated.

I used 2 - 2904 in parallel.

The system will be extended next week with another 8 inverters.

I will be working on a print now.

Thank you for thinking with me.

As promised here the picture of the AC through the battery bank. The first one is pretty well, but as you can see, the duty cycle isn't stable when more inverters participate:

The voltage of the measurement isn't correct: It is just what my CT outputs.

I will measure later over the transformer coils.

As suggested by Cbob12 I will now do some tests with a choke transformer.

I use a 500 Watts ferrite ring transformer as base. I will run the test very shortly, with 20 amps breakers between and with few inverters, because the windings are not withstanding all the current.

The idea is just to use one conductor on the plus and the other on the minus.

I will play with the phasing, trying to eliminate the AC to the batteries.

Here is what I brewed for the test: The 3 conductor wire is a leftover that I have.

I will keep the results posted. D