Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 08:31pm 01 Feb 2019
Copy link to clipboard
Print this post
Why would you want to build a 12v system and then boost the output voltage up to 48v ?
Just use 48 volts worth of solar panels, a 48v battery, and a 48v rectifier. Combine the three outputs through three diodes to drive your load and its as simple as that.
Making things highly complicated just for the sake of doing so makes no practical sense. I just cannot understand why you are so eager to use multiple microprocessors to achieve this ????
Cheers, Tony.
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 08:55pm 01 Feb 2019
Copy link to clipboard
Print this post
The objective of LEARN_MODULE "3INPUTS" is not to build a circuit that can power a 48v LOAD and be done with it.
The objective is to build a project that helps me/us learn how to build something that be used to take in 120V rectified DC from the grid, 120V DC from PV panels and something similar from batteries and seamlessly loadshare between these input sources.
That is a high power design.
But before me/we who do not have a power electronics background like yours can do this, we need to start with smaller voltages. People with no prior exposure to high power could injure themselves because they don't know what the elecronics/concepts are.
Proof of concept and for learning for now.
This is not going to be a product that will be used for anything but for learning and moving to the next step.
The 48 volts worth of solar panels, a 48v battery, and them connected to the load and isolated from each other via 48v rectifiers (is that what you were suggesting?) works for that very specific narrow purpose, but you would never use that design if the panels and grid offered 140V each as the power loss across the diode at that high power would be pretty wasteful, yes?
So we replace the diodes with a set of switches and figure out a logic to synchronize the switches so the inputs share the load according to priority, yes?
So let's do that :D
If I knew how to design the system, I would have. I am not suggesting we do this at all, I am just saying I can handle this if need be.
You know the system. You have designed it before. If this does not make any sense, forget I even said that. I only mentioned that to tell you that as you help me design this, don't be concerned that I will be unable to handle concurrency and message passing. If your inverter design does not require a lot of processing, ignore this bit.
Let us start with a schematic that allows me to loadshare from two or three inputs, each having similar voltages but different power quality/regulation.
The design has to be scalable so that the fundamental design remains unchanged even if the voltage and currents drawn from the design were to go up ten times.
along the way we will learn how to test and check each stage to ensure they work as we expect it to
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 09:15pm 01 Feb 2019
Copy link to clipboard
Print this post
Not that wasteful. A big power diode drops about 1 volt, in a 100v nominal system that is 1% power loss I can live with that.
Never use that design ? That is exactly how my own system works.
I chose to build a 100v system for two reasons. I decided to use thirty 50Ah Winston lithium batteries which fit perfectly into a standard filing cabinet drawer when stacked 6x5. And I already had a suitable three phase transformer which produces around 106 volts dc and capable of very high current.
If I lived in the US and was planning to do the same thing there, I would go to a 150 volt system which would have eliminated the need for a large and expensive transformer for the rectifier.Cheers, Tony.
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 10:11pm 01 Feb 2019
Copy link to clipboard
Print this post
OK, so your inputs are all tied together via power diodes? Each power diode contributes to a 1% power loss, so for a 3 input system, you will have a 3% power loss right at the input.
Did I get that right or you have some tricks to change that?
Understood. Makes sense.
So, your inputs are essentially a "logical OR" of power diodes?
If you don't mind a high level digram of the system, I can get started replicating it :)
I will start with low voltages to being with though. I don't have the confidence yet to start off with 100v+. We will get there eventually!
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 11:09pm 01 Feb 2019
Copy link to clipboard
Print this post
There will only ever be one diode voltage drop, regardless of which source is supplying power so only ever 1% loss.
The diode bridge in the circuit below has four diodes but only three actually do anything. Adding the fourth diode to the schematic only makes it look more confusing.
The rectifier is now optional. It started out with just solar + rectifier. Its now solar + battery.
Cheers, Tony.
mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460
Posted: 01:25am 02 Feb 2019
Copy link to clipboard
Print this post
Hi Tony
I saw this diagram about a month or so ago. It made me think about my next install. The beauty of this implementation for me is the available "Amps" for motor start-up. My cabin install is 10kW of solar and the next install will be at least that. Having the 10+kW array online already will probably start my air compressor - barely touching the battery.
My only hesitation is the dynamic range. For a nominal 96v system, the range for me would be about 104v (low battery) to 160V (solar, unloaded-ish). Sure the design can handle that range but what about the situation where I'm running the hot water at 3.6kW and it turns off , and if there is no other load (which is unusual). Given we would adjust the output at the beginning of every cycle, we could still see a 40+ volt spike on the output during the current cycle? Is the 30,000uF cap sufficient to dampen it enough to allow cycle by cycle adjustment?
What has your experience been?
Cheers Andrew
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 01:53am 02 Feb 2019
Copy link to clipboard
Print this post
You had asked me why I was complicating the system by introducing multiple processors. Right in this diagram I see the need to have 3. 1 for PV MPPT. 1 for managing the charging algorithm for the battery chemistry. 1 for the inverter.
If you think this is not an efficient design, we can talk about it.
Right now, 5 questions:
1. What's the system that represents the dotted line connecting the rectified grid to the inverter? 2. With the heavy ripple from the rectified grid, what life would the large elec. cap have? How big should it be and what ESR? 3. How would those reverse biased diodes behave capacitively with the heavy ripple from the rectified grid? 4. What in this system ensures that power from the PV panels are drawn first and then any slack/difference from the grid?
Depending on the insolation levels, the output from the PV can vary above, at and go below the rectified grid output, but I still want to extract the last possible watt from the PV panels, right?
Let me clarify a point again: my requirement is not a "get power from any one, whichever is higher" source.
It is "always get power from PV FIRST and THEN take the remainder from" the other sources.
AKA Load sharing.
What are your thoughts about the following:
Scenario 1: Let's assume it's bright and cloudless outside and PV panels are in a situation to output their max. power (Pmp)
If I connected a 800Watt resistor as a load to the above circuit, which of the following scenarios would happen (and why):
a. Almost all of the power drawn from the grid and only some from the battery and some from PV b. Almost all of the power drawn from the battery and rest from the grid
Scenario 2: Let's assume it just became cloudy outside and PV panels are in a situation to output half their max. power (0.5 Pmp)
If I connected a 800Watt resistor as a load to the above circuit, which of the following scenarios would happen (and why):
a. Almost all of the power drawn from the grid and only some from the battery and some from PV b. Almost all of the power drawn from the battery and rest from the gridEdited by LadyN 2019-02-03
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 02:21am 02 Feb 2019
Copy link to clipboard
Print this post
Andrew,
I am running 36,000uF across the inverter input and that seems to be sufficient to tame any very rapid dc voltage excursions in either direction. The voltage will definitely jump a fair way on sudden load changes, but slowly enough for the the input voltage correction to deal with it.
I did originally think about the potential for output voltage surges on suddenly dropping off a very heavy load, but have never actually seen that happen.
A possible solution that I never followed through on, was to connect a bunch of fairly low voltage metal oxide varistors (250v ?) directly across the inverter output. Any surge is going to last no longer than 40mS, more likely much less. A bunch of big MOVs should easily be able to swallow that.
The biggest step load I have tried switching on and off was a 1.6Kw heat gun, and there was no visible light flicker from an incandescent bulb. I don't expect there to be any problems, but piling on more microfarads should fix it if there are.
Voltage recovery and correction will be total every 40mS, and the dc input is not going to shift very far in 40mS with 36,000uF and 180 Joules of stored energy it has to shift. Edited by Warpspeed 2019-02-03Cheers, Tony.
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 02:40am 02 Feb 2019
Copy link to clipboard
Print this post
The dotted line is a cable with a plug on the end. If the rectifier is not in use its unplugged and effectively just not there. With a battery providing power at night, its no longer required.
My three phase six diode rectifer has only about 2% rms ripple without requiring any further filtering. A much bigger issue is ripple current caused by the inverter load which is going to be BRUTAL. I am using three 12,000uF 250v electrolytics each of which has a continuous ripple current rating of 18.8 amps at 100Hz, and 14 milliohms ESR.
Capacitance is a non issue at 100Hz.
The highest voltage source will always take up the initial load. The Solar voltage must be higher than both the rectifier voltage and the battery voltage. If its able to do so, it will supply all the required load without the voltage falling very far.
If solar struggles, the solar produced voltage will eventually drop until either the battery or the rectifier will be able to make up any shortfall. But solar will still be producing every bit that it can produce, and its only what extra is needed that comes from the battery or the rectifier.Cheers, Tony.
mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460
Posted: 08:10am 02 Feb 2019
Copy link to clipboard
Print this post
Tony
Thanks for the feedback. My current clock board latches changes every 40ms same as yours but I'll change it to 20ms on my next version. (use an inverter off the 50hz line).
The internal resistance of the solar panels and the Inverter Capacitance will give us the response time. My rough calculations (will be way out) gives me a Rs of 0.6 ohms for a 10kW 120v solar install at 1000w/m2 illumination. So that will give us a Trc of about 22 mSec.
I'm not at the Cabin so I can't measure it. I might buy a used set of panels (1.5-2kW) so I can test this stuff while I'm in Brisbane.
Thanks Andrew
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 04:51pm 02 Feb 2019
Copy link to clipboard
Print this post
The only reason I decided on 25 readings/sec was that the Intersil application notes for the analog to digital converter recommend thirty conversions per second as being maximum.
If you can get yours to work successfully at the higher 50/sec measurement rate, that would certainly be a very worthwhile step forward, its just something I never tried. In retrospect I should have made that an option. It works well enough as it is so I never bothered with any further development along those lines.
Earlier on, I used a voltage controlled oscillator and frequency measurement as a way to average out noise. It certainly did that, but the one bit ambiguity of frequency measurement of a free running VCO caused the least significant bit to jump around randomly and that did not look very nice. There are ways around that problem but I decided to try a completely different approach.
Dual slope analog to digital conversion sampling synchronously with the inverter, and not using the three least significant bits produces a very noise free steady voltage measurement without any jumping around. Very happy with that.
This input voltage correction technique works so very well and corrects much faster for sudden load changes than voltage feedback ever could. I would never now go back to using voltage feedback on an inverter.
Impedance of the solar panels will be all over the place, and infinite at night!
As long as the capacitor bank can store enough energy to see it through each half sinusoid of inverter current demand, with a low resulting ripple voltage, that should also solve any problems with the input voltage correction.
It all becomes a lot easier with higher dc voltages. Some high performance low ESR capacitors right at the inverter could be supplemented by a larger bank of cheaper aluminium electrolytics located further away.
Its rather difficult to visualize all of this, but actual testing produced far fewer problems than I anticipated. The whole project has gone remarkably well.Edited by Warpspeed 2019-02-04Cheers, Tony.
Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Print this post
