Design review / comment

[jharvey]

@krzysztof9nowak
I figured I'd start a conversation here. The charge controller repo is just a copy of the FUGU's hardware. This repo is very similar, but also includes more features.

A short version of the below.
-- Do you have suggestions for the transformer? I need to find a solution for this. I'm hoping it's around 60Hz of 120VAC at 20A. However I also expect each panel can't really handle the full load, so I could probably get by with something like 5A instead.
-- Do you have suggestions for BMS? I'm starting to look at BMS options.

Longer description of the above is below.

My goal here is to make a portable solar power useable source of energy generator. I want to mount the electronics behind a solar panel and making it portable. Basically I want to be able to put the panel in the yard, prop it up at an angle, and have it generate VAC, preferably with a battery which can capture energy during the day and allow it to keep powering during the night. One panel in the 45 degrees of latitude area can probably provide enough energy to recharge a 50AH 12V battery during the day. Then the battery allows consumption of that energy during the night. I'll attempt to make the panel system hold a 100AH battery as that can commonly power a refrigerator overnight. I understand a normal lead acid car battery is approx. 100AH.

My current goals, at the moment for this repo are below. Feel free to add feature requests. I'll incrementally pick manageable bites and make incremental progress. However I like to have the full feature list so I can avoid making obstacles for myself relative to future work.
-- Portable all in one self contained useable solar power generator. MPPT, Battery, BMS, pure sine inverter, etc. All in one unit. It may power a refrigerator or it may power a ham radio. So pure sine wave is desirable. I want it to be physically capable of powering 120VAC 60Hz at 20A. However battery capacity and the panels ability to generate the energy are not going to obtain that with a single panel in a sustainable fashion. So I'm thinking one panel with the ability to generate useable power, which includes a remote storage connection that allows other panels or other storage options.
-- We have a distributed neutral in the US, so I was planning for 220V to be possible by using 2 panels. However I tend to flip flop on if I should make the transformer with an optional neutral. I understand many areas like the EU do not distribute the neutral.
-- I think it should be able to sync with other sources of VAC. So if I have a generator, wind, car battery / inverter, or other source of AC, I want to plug in this panel and have it sync with the existing AC then contribute power.
-- Sync with other nearby solar generators. If I put multiple of these in the yard and plug them all into a power strip, they learn about each other and manage each other. For example, the load is 500 watts, and I know that panel 1 has an empty battery and 300 watts of solar energy, but panel 2 has full battery and 300 watts of potential energy, then the panels should figure out that panel 2 should provide 300 watts of the load energy, and panel 1 should provide 200 watts of load energy while banking the other 100 watts goes into charging the battery.
-- If mains voltage is detected, then use it to charge the solar battery. If you have another source of intermittent energy like a generator or wind, then capture that energy and store it.
-- Hot water generator. Basically put tubing on the panel, and capture the heat energy. I plan to include some components to power a pump and allow it to circulate water through the tubing. This can be as simple as liquid cooling, or it could be used as a source of hot water.

I've got 7 of these panels, so to some extent I'm designing this project accommodate these panels. For example, I know I've got 36V to work with so my battery max voltage will be 36V. I'll desire to use a 24V battery, which allows the MPPT buck some head room to work with as it attempts to track.
https://store.santansolar.com/wp-content/uploads/2022/03/document.pdf

BMS I'm considering things like this. https://www.beyondlogic.org/review-li-ion-lipo-lifepo4-lithium-battery-active-equalizer-balancer-energy-transfer-board/ I've also found this https://github.com/stuartpittaway/diyBMSv4 I think I want active cell balancing instead of the resistor balancing.

However I'm thinking of using it with lead acid, not lithium based tech. Also about batteries, I have some odd plans. I'm thinking that I'll likely experiment with a bucket of acid, then if the cell voltage drops, use a peristaltic pump to replace the acid. If the voltage of the battery get's to high, then reverse the pump and fill up the acid. Basically I don't need high amps from the battery, but I do need high energy storage capacity. I'm thinking lead acid as I could re-claim the lead and re-cast plates. This would allow the battery cells to last a very long time for low $. I've also heard that Epsom salts can be used as an electrolyte. So I may experiment with different chemistry.

At the moment I'm drafting the pure sign inverter and I'm starting to look at BMS. A key issue I'm trying to figure out is how to get my transformer for the sine inverter. I'm thinking of winding a toroid. Do you have thoughts or suggestions on how to get a transformer? Remember I'm aiming to keep is low profile so it can fit behind the solar panel.

Also I'm starting to look at BMS. For now I'm thinking of a pile of lead acid car batteries. But longer term I'll want mount the battery behind the panel. The panel has large X and Y dimensions with a fairly short Z dimension. A car battery does not fit in this physical area, and would make the overall panel system less portable. I'd like to incorporate the battery in the the back of the panel. So I may look to balance the individual cells but for now I'm thinking of balancing multiple 12V cells. With only 2 or 3 cells, the balancing may not be strictly required.

On a side note, this is an interesting method for boosting your panels power for low $. https://www.youtube.com/watch?v=FKhszB4E1_M

It's a rainy day, I go start chipping away at your design suggestions.

[jharvey]

I took a stab at doing a simulation to get a better idea about the heat generated by the MOSFET's. I put that simulation here https://github.com/jharvey/Solar_Kit/tree/main/Simulations/MOS_Heat_Test2

It shows some odd things. I'm not sure if it is very helpful yet. I used PSPICE for TI, as it has a SPICE model for the MOSFET's. I then found a semi similar driver, and I modeled the circuit topology. The timing of the MOSFET's is critical, the overlap is critical and I'm having trouble getting the simulation to match what is probably done in the real world for these driver timings.

[krzysztof9nowak]

That's a lot of interesting ideas. Some of them, like syncing multiple off-grid inverters seem quiet ambitious, but could be achievable.

Regarding the transformer, at the moment I am working with 250VA toroidal 18-0-18V to 0-230V transformer. I was planing for max power of 800W, which is a power of a typical microwave transformer, which can be sourced cheaply. I planned on keeping the primary wingdings and replacing the HV with a ratio more appropriate for my battery voltage. In Europe we have a three phase grid with neutral, but the voltage L1 - N is 230Vrms and L1 - L2 is 380Vrms. The three phase power is used by very power intensive devices and would require more complex transformers, so I am not looking into it. In your case, it's just a matter of winding a 120-0-120V transformer. The only downside would be increasing the transformer size and cost. I would go for it. However I don't know how to make it flat enough to fit in the panel frame. You could use "transformerless" design i.e. based around high frequency transformer, but that gets complicated and pricey very quickly.

I am driving the inverter with STM32F103C8T6 (basicly any STM F1 or better will do) as they have very versatile timers and PWM output, a good 12bit ADC and plenty of computational power.

With this topology I can output off-grid, synchronize on-grid and charge batteries using boost converter and rectifier (not tested yet).

I haven't research BMS' a lot because I've scored a few 1.5kWh 13S LiIon modules with builtin BMS. I've read about the batteries with pumped electrolyte, but haven't seen any DIY implementations. Let me know if you achieve something.

[krzysztof9nowak]

I've requested access to PSPICE for TI, once I get it I will have a look at your simulation.

[krzysztof9nowak]

The core you have selected for L1 has a very high magnetic permeability and it will saturate much sooner than at 36A.
The magnetic field at 36A and 6 turns is:
$$B = \frac{\mu_r \mu_0 I n}{L} = \frac{2000 \cdot 4 \pi \cdot 10^{-7} \frac{H}{m} \cdot 36A \cdot 6}{0.043m}=12.6T$$
It's way above $B_{sat} \approx 0.55T$ of Epcos N27. It's a material well suited for transformers, but not inductors. I have used Micrometals core T130 with material 26. You might need a bigger one, but I would use the same material (or find something similar from Epcos).

[jharvey]

About the Hz sync, I agree ambitious. I may not try this for the first one. I'll probably just leave an option to connect the analog control pin to the DAC of the ESP32. Then if I can find time, then try adding the feature. Seems the day job and kids consume most of my time. 10 minutes a day I'll keep chipping away at it.

Hmmm, saturation core issues. I've attempted to predict saturation before, but did not find any good references to help with the predictions. Do you happen to have a reference doc or app note that you can recommend? At 60Hz there really isn't any way to keep the transformer low weight. I understand increasing the frequency allows decreasing the size of the transformer. Perhaps it's worth while to make a high frequency DC to DC and bump the 48V up to something 200V or 300V then use that with no transformer to power the VAC.

Any luck with PSpice for TI? I couldn't get beyond the apparent typo in the SPICE file for the transistor. So far the best simulation I've got is with LTSpice, which uses a similar but different MOSFET.

[jharvey]

SUMMARY: I'm looking into designing a LCC / ZVS topology circuit to boost battery voltage to around 390VDC. This would be used for the pure sine wave circuit. As well I'm looking for most of this to be JLCPCB source-able parts. Longer description is below. If I can find a suitable purchased DC to DC converter, I'd likely just use a purchasable option.

MORE DETAILED: For pure sine wave 220VAC non distributed neutral OR 120VAC with a distributed neutral, it appears the EG8010 can be configured to operate from a 390VDC supply. If you have this kind of high voltage to work with you can negate heavy transformers, like those found in microwaves. I'm looking for reasonably light weight and physically compact, so I'm planning to negate the heavy transformer. This means I'm looking to convert an input voltage of 10.4VDC to 38VDC battery storage to around 390VDC. I'm looking to have enough power to run loads like a refrigerator, so I'm looking for around 1kW, or around 10A at 120VAC. To make this happen, I need to boost around 1kW of output energy. Efficiency at these power levels becomes a concern.

The below link shows that a LCC controller or at higher power levels a full bridge topology is the optimal efficiency topology.
https://www.ti.com/power-management/acdc-isolated-dcdc-switching-regulators/overview.html

This video provides a brief description of the LCC transformer topology.
https://www.youtube.com/watch?v=TVQuPWtxN34

So I'm in part looking for a transformer. However this means I need to select the driving components first. I need to know if those driving components can drive at 100kHz or if we are some other frequency. The frequency will determine many aspects of the transformer. Keep in mind this transformer is probably very light weight, and small, and is not big and heavy like a microwave transformer. By increasing the frequency, you decrease the size and weight.

Finding MOSFET's which can handle these switching frequencies and amps from JLCPCB is fairly easy. However finding a chip which can drive a LCC circuit while monitoring the output voltage is a harder task to accomplish. I looked for chips suggested by the above TI page like the UCC25600 which might be winner winner chicken dinner. I see the below is at both JLC and on the TI suggested list.
https://jlcpcb.com/partdetail/TexasInstruments-UCC25600DR/C130223

I also found this ZVS circuit on eBay. It can be found with this search term. "DC-DC Boost Converter Board 8-32V to 45~390V"
https://www.ebay.com/itm/293851632650

Other DC to DC converters exist on eBay. However they are around 80% efficient. These LCC / ZVS topologies appear like 90% efficient is achievable. So I'm looking for a ZVS or LCC design. If you dig deep enough, you can find the core chip at the heart of this circuit is probably a UC3855A
https://www.ti.com/lit/ds/symlink/uc3855a.pdf

However that chip is not found at JLC.

So I'm looking for a JLC source-able ZVS or LCC topology circuit. Then I'll be looking or making a transformer to match what ever that frequency is that the components can drive. I'm expecting the transformer will be around 100kHz to 400kHz.

[jharvey]

I've made some progress in getting a DC to DC 400Vout LC transformer topology circuit simulated. It needs more tuning to step up 300W of power. Currently it only works with around a 10k ohm load. It was harder then desired to get this rough model working. So I wanted to toss this out there for now.

I plan to tune this a bit more manually, then eventually I plan to use the parameter sweep to find the peak tuning for each parameter.

[krzysztof9nowak]

The LLC probably is a good option, but I tried to avoid it, as the frequency control seemed a bit complicated. You have mentioned a very wide input voltage range (10.4-38V) do you expect your converter to work over this range or you will choose a more specific value? The LLC can achieve it's best efficiency only when operating only in a quiet narrow voltage range.

Here is an interesting document comparing all topologies for application in solar devices: https://www.ti.com/lit/an/slla498/slla498.pdf

[jharvey]

I'm OK with using a more specific value. I'm going to plan for a 3 cell flooded acid battery, so V low of 10.4*3=31.2V and V high of around 38V, nominal at 36.6V.

Thanks for the link to the topology review. That got me to this article https://www.ti.com/lit/ug/tiduct9a/tiduct9a.pdf which includes some calc's for determining the LLC's resonant frequencies. Once the LLC caps and inductors aligned with the calc in that PDF, the simulation started to comply and started to produce useful results. I'm still attempting to tune this circuit, it shows problems. Vtrans is 1kvp-p I can't help but wonder if the 6.2nf is a realistic value, the output voltage is still to low, and it's not producing the watts I need. So it needs work, but it's getting better. This is a snap shot of how it's currently doing.

I need to find a core and make predictions about realistic uH on the primary and secondary windings. Once I have those estimates, I can then start to tune the LLC to match those predictions. So far I have not found an off the shelf transformer which I expect to work for this design. I'm thinking I really need to wind a toroid. So I guess toroid transformer winding is the next round of predictive calcs.