Is this a safe way to build a solar inverter?

[rritoch]

Hi,

I have a grid-tie inverter but it is useless during power outages. During a recent power outage I wanted to get a 150W fan running. We have 3 18V 100W panels connected in series though they only produce about 120W total. We also have an old, broken, UPS, so what I did was disconnected the broken battery from the UPS and plugged the 50+ V solar input directly into the UPS. The fan ran for a few seconds and the UPS shut-down, it was probably overloaded since it only expects 12V input. I have about 8 115W transistors (4 NPN, 4PNP), a 12V zener diode, some electrolytic capacitors, and some resistors laying around. I am thinking of trying out the following circuit and plugging the power output (powering a light in this circuit) into the UPS. Would this be safe, and is there a good chance that this would work?

http://www.falstad.com/circuit/circu...+12+0.4+0.4%0A


Just in case this link doesn't work for you. I would attach the 50V+ of power input to a 5k ohm resistor, and from there to a 12 volt zener to ground and also to the input of a transistor, taking the output of that transistor into the bases of the two other transistors to increase the power from what the zener diode can handle to what the UPS input would be.

The zener diode I believe can handle 1W and the transistors are MJ2955 & 2N3055

[inetdog]

Although that circuit might work it would waste an enormous amount of power and would be strictly limited in the amount of power available because of the 5K resistor.
I would not recommend it.
A very small battery and a normal pure sine wave inverter (or even a Modified Square Wave inverter) would, IMHO, be a better solution.

[rritoch]

Although that circuit might work it would waste an enormous amount of power and would be strictly limited in the amount of power available because of the 5K resistor.
I would not recommend it.
A very small battery and a normal pure sine wave inverter (or even a Modified Square Wave inverter) would, IMHO, be a better solution.

Thanks for the feedback. As of now efficiency isn't an issue since anything is better than nothing. For now I'm just going to go with what I have as the local stores don't have inverters for sale. I've constructed a test circuit using a 3.9k ohm resistor, and a 47uF capacitor as that's what I have available. If it works I'll probably need to construct a more permanent circuit as this one is monkeyed together with loose wires and alligator clips. No sun until tomorrow so I won't know anything for sure until then.

[SunEagle]

IMO it is always best to Keep It Simple Simon. Have fun with your experiment and please don't get an electrical shock.

If you really need to run your 150w fan during the next power outage you can add a battery and battery & inverter to those 100 watt panels. Or get a generator to run even bigger loads.

[bcroe]

Hi,

I have a grid-tie inverter but it is useless during power outages. During a recent power outage I wanted to get a 150W fan running. We have 3 18V 100W panels connected in series though they only produce about 120W total. We also have an old, broken, UPS, so what I did was disconnected the broken battery from the UPS and plugged the 50+ V solar input directly into the UPS. The fan ran for a few seconds and the UPS shut-down, it was probably overloaded since it only expects 12V input. I have about 8 115W transistors (4 NPN, 4PNP), a 12V zener diode, some electrolytic capacitors, and some resistors laying around. I am thinking of trying out the following circuit and plugging the power output (powering a light in this circuit) into the UPS. Would this be safe, and is there a good chance that this would work?

Connecting 50V to a 12V circuit is a good way to smoke everything. Anything you connect to the
panels is going to quit every time there isn't enough sun.

That linear regulator you showed will just burn up most most of the panel power. And it needs to
be modernized a bit with MOSFETs and an OP AMP or 2.

To get some 12V efficiently, try one of those switching DC-DC converters. There are some really
cheap ones on E+ay, but maybe not powerful enough for your setup. Or, go with a real MPPT and
a modest battery. Bruce Roe

[rritoch]

Thank you everyone for your feedback. Efficiency was apparently the killer of this circuit. In my first test, while it did produce 12 volt output as needed, it simply didn't have enough power and the UPS didn't respond at all. In my second test I reduced the resistance to 440ohm using two 220ohm resistors in series. With this second setup the UPS made a small beep and never activated, the only thing this second configuration achieved was burning one of the resistors. I will keep my eye out for some DC to DC converters since I can't find an inverter that can handle the 50 volt input.

[max2k]

Thank you everyone for your feedback. Efficiency was apparently the killer of this circuit. In my first test, while it did produce 12 volt output as needed, it simply didn't have enough power and the UPS didn't respond at all. In my second test I reduced the resistance to 440ohm using two 220ohm resistors in series. With this second setup the UPS made a small beep and never activated, the only thing this second configuration achieved was burning one of the resistors. I will keep my eye out for some DC to DC converters since I can't find an inverter that can handle the 50 volt input.

Can you connect your 3 18 V solar panels in parallel? This way you'd only lose about 30% of the total power produced by the panels. In your current arrangement you're losing about 75%- basically 2 out of your 3 panels will simply heat transistors and resistor of the linear regulator. Your parallel arrangement of the transistors won't work well too as one of them would end up driving the load due to exponential dependency of the input voltage to output current and negative temperature coefficient of the input voltage: hotter transistor will flow most of the current making it hotter and so on. Small 0.1-0.2 Ohm resistors in series with each of the output transistor bases would solve this problem.

You'd need about 200 mA current into base of the first transistor so with 18V arrangement the resistor in series with zener should be 30 Ohm and rated to at least 2W. With your current setup (50V in) I won't even bother to calculate as it is so inefficient. Your transistors and zener would need to be on a heat sink or this project won't last even 1 minute. Depending what you're after buying standard MPPT controller is probably much better alternative but also much less fun .

[rritoch]

IMO it is always best to Keep It Simple Simon. Have fun with your experiment and please don't get an electrical shock.

If you really need to run your 150w fan during the next power outage you can add a battery and battery & inverter to those 100 watt panels. Or get a generator to run even bigger loads.

We actually have a 5kw gas powered generator. It is very loud, and usually more than we need. There are some 25v inverters for sale in Bacolod, a nearby city, but the 600W grid tie converter we have won't activate at the 18v these panels produce so we need them connected in series, producing too much voltage for the inverters I've found so far. The DC-DC converter seems to be the best route. Since this is only meant for back-up power I'm not sure that batteries would be the best way to go since this system isn't going to be used often and the batteries would likely die quickly since they won't be used (charged) often. The big complexity is that the grid tie needs 22v to 60v, the inverters need 25v, and the panels produce 18v. The only batery-powered solution I could think of would be to setup two charge controllers and feed the DC output of that, in series, to the grid-tie (12v X 2). I'm also not sure if that would work. For now the simple solution would be a DC to DC converter, which is basically what I unsuccessfully tried to make.

[rritoch]

Can you connect your 3 18 V solar panels in parallel? This way you'd only lose about 30% of the total power produced by the panels. In your current arrangement you're losing about 75%- basically 2 out of your 3 panels will simply heat transistors and resistor of the linear regulator. Your parallel arrangement of the transistors won't work well too as one of them would end up driving the load due to exponential dependency of the input voltage to output current and negative temperature coefficient of the input voltage: hotter transistor will flow most of the current making it hotter and so on. Small 0.1-0.2 Ohm resistors in series with each of the output transistor bases would solve this problem.

You'd need about 200 mA current into base of the first transistor so with 18V arrangement the resistor in series with zener should be 30 Ohm and rated to at least 2W. With your current setup (50V in) I won't even bother to calculate as it is so inefficient. Your transistors and zener would need to be on a heat sink or this project won't last even 1 minute. Depending what you're after buying standard MPPT controller is probably much better alternative but also much less fun .

This wasn't just about having fun, I was also, somewhat unsuccessfully trying to impress "the wife", lol. She was rather amused that I was able to get the fan to turn on at all. I really can't go for the parallel connection because it isn't enough to power the grid tie. The small resistors added to the circuit seems to be a great addition. Sadly these 100W X 3 panels only produce 120W which is only 5W above what the transistors are designed to handle so the second transistor was probably not needed for more than the possible future addition of more panels. The 1W zener diode was the best I could find at the radio supply store. This was a fun experiment and luckily no one got hurt. The next time I'm in the city I'll see if I can find some resistors that can handle the power load to see if I can continue this experiment to a functional model.

If I use 2 1W zener diodes in parallel would that solve the 2W requirement?

I may be missing something but to produce 200ma wouldn't I need to use a 300 ohm resistor? (50/300 ~ 167ma) This would be in-line with the reality that I had some response with 440 ohm resistance, excluding the reality that one of the resistors burned.


I updated the circuit with the new specs

http://www.falstad.com/circuit/circu...4+400+224+0%0A

[rritoch]

Connecting 50V to a 12V circuit is a good way to smoke everything. Anything you connect to the
panels is going to quit every time there isn't enough sun.

That linear regulator you showed will just burn up most most of the panel power. And it needs to
be modernized a bit with MOSFETs and an OP AMP or 2.

To get some 12V efficiently, try one of those switching DC-DC converters. There are some really
cheap ones on E+ay, but maybe not powerful enough for your setup. Or, go with a real MPPT and
a modest battery. Bruce Roe

I really like the idea of using a MOSFET but I don't have much experience with them. What I do know about them is that they're sensitive to static electricity and if I go in that direction I'll need a more "sterile" (static free) environment to work in. With the alligator clips I don't expect big problems with the MOSFET but when it comes to making a permanent circuit I would need a higher quality solidering iron than I have now. I'll start researching the MOSFET solution, the grid tie uses MOSFET's so it probably is the more ideal solution.

As for using an OP AMP circuit, I'm not aware of an OP AMP that can handle 10A+ current, and it seems unlikely that I could find them in a local store. I'm trying to avoid importing equipment myself since it typically takes over a month for delivery.

edit: After review of MOSFET based voltage regulators, and testing out on the circuit emulator it seems that the MOSFET circuits are quite simply too complex for what I'm trying to do. I have a hard enough time finding basic components, it seems almost impossible that the controller IC's needed for MOSFET circuits are going to be available. I live in a village in central Philippines, the availability of electronic parts is limited at best.

[max2k]

I really like the idea of using a MOSFET .... I live in a village in central Philippines, the availability of electronic parts is limited at best.

MOSFETs would require special IC to be used in power supply regulator even linear one. They're not that sensitive to static but are more complex in application. They can be connected in parallel, even to the same IC but again- driving them requires some circuitry and nowadays that part is handled by specialized IC chips. MOSFETs shine in switchable DC-DC converters and I'm sure there are plenty of IC chips to control them in those applications as well. The problem is if you're not going to make electronics as your permanent hobby it's a little too involved.

For your original setup with 50V feed your calculation is about right but you have to keep in mind the resistor would have increased voltage drop of 50 - 12 = 38V and at the current ~0.2 A it would need to dissipate 38 x 0.2 = 7.6 W so you need at least 10W resistor there. Zener would need to dissipate the 12 x 0.2 = 2.4 W. BTW, zener gets it hardest when there's no load as in that case it has to conduct to the ground all the current that otherwise would flow into the first transistor base.

Since your wife is already impressed I guess other more standard ways already suggested in this thread would solve the problem more reliably. BTW, I thought Philippines is the place where a lot of semiconductors are made or has it been moved out of there since 1990s ?

[bcroe]

This wasn't just about having fun, I was also, somewhat unsuccessfully
trying to impress "the wife", lol. She was rather amused that I was able to get the fan to turn on at all.

If I use 2 1W zener diodes in parallel would that solve the 2W requirement?

Impressing the wife is always a worthwhile thing.

I found a DC-DC on Eb*y that ought to work, if the panels don't exceed 60V. Search
"DC Converter 48V to 12V 20A 240W Step-down Buck Power Supply"
I can get about anything (not already stocked) off the internet.

Don't ever connect zeners in parallel to increase power. One will be slightly lower in
voltage and take all the power while the other does nothing. Instead use 2 in series,
each rated half the desired voltage .

MOSFETS are actually a lot easier to use than bipolar, driven by voltage instead of current.
There are ICs especially designed to take advantage of them; not required for this ap.
A problem with the bipolar circuit you showed, is in order to increase the max output current,
you have to increase the currents of all the other driving circuitry. The current gain of each
stage must be carefully controlled. And those stages end up using a significant amount
of power. Since a MOSFET is voltage driven with essentially INFINITE current gain, the
drivers can operate at microwatt power and then let a big fat FET (or several) handle
big output current.

You wouldn't need a 10A OP AMP. Most of these do their work with voltages and
practically no current. Really an OP AMP is a shortcut to avoid building up a complicated
and limited discrete circuit. Add a voltage reference bridge to the OP AMP/MOSFET
and you have a linear power supply. Bruce Roe

[SunEagle]

We actually have a 5kw gas powered generator. It is very loud, and usually more than we need. There are some 25v inverters for sale in Bacolod, a nearby city, but the 600W grid tie converter we have won't activate at the 18v these panels produce so we need them connected in series, producing too much voltage for the inverters I've found so far. The DC-DC converter seems to be the best route. Since this is only meant for back-up power I'm not sure that batteries would be the best way to go since this system isn't going to be used often and the batteries would likely die quickly since they won't be used (charged) often. The big complexity is that the grid tie needs 22v to 60v, the inverters need 25v, and the panels produce 18v. The only batery-powered solution I could think of would be to setup two charge controllers and feed the DC output of that, in series, to the grid-tie (12v X 2). I'm also not sure if that would work. For now the simple solution would be a DC to DC converter, which is basically what I unsuccessfully tried to make.

I agree that using a very large gen set is noisy and very inefficient. They do make smaller quite inverter style generators that provide solid emergency power on demand.
I "bolded" something you said above. What type of 600w grid tie inverter are you talking about? I do not know of any legal grid tie inverters that small.

[rritoch]

Just a small update...

I was able to buy 12, 10W 25 ohm resistors to provide the 300 ohm resistance. I also added the second 12V zener diode, and in series with the zener diodes added two parallel led's to both increase the output voltage to 12.7 to 13.4V, and provide an indicator that the circuit was receiving power. This was able to activate the UPS for a short amount of time, though it was beeping as if it was facing a low-battery situation. I ran the test as the sun was going down so this was potentially just bad timing. The zener diodes ran cold which was good. The resistor's were a little warm, but nothing significant. The only serious problem was with the power transistors. One of them didn't have any connection at all, which I didn't realize until I tested the continuity afterwards, so the full load was on one transistor. The one transistor ran VERY hot but as far as I can tell wasn't damaged. I'll try again during better sunlight conditions, but this was certainly an improvement since the voltage was correct and the UPS activated (charging a small wall-mount flashlight) but only for a short time. It wouldn't turn on at all with the fan attached but there are a number of reasons that could have caused that, from inefficiency, to not enough sunlight, or even the single transistor not being able to handle the load. Either way, it looks like this is very close to a functional design.

[rritoch]

What type of 600w grid tie inverter are you talking about? I do not know of any legal grid tie inverters that small.

Here's a link to a comparable grid tie inverter http://www.amazon.com/SUN-600G-Micro.../dp/B0089UA9KO, the only difference is the one I use is designed for 220V AC.