Simpler inverter/charger

[johntaves]

I am trying to understand if it possible to create a simpler pure sine wave inverter/charger. This is your typical inverter/charger where if the mains AC is present the unit turns off it's own invert function and passes AC mains to the AC output, and supplies DC power to charge the battery. If no mains AC is present, then it is an inverter.

However, the typical inverter/charger on the market has a bunch of additional features for protection, like low volt cutoff, high volt cut off. They also have to have a user interface (LCD and buttons, or bluetooth, or serial port) to be able to specify the type and size of the battery (e.g. flooded lead acid, AGM, etc) and manage the charge profile.

The one I might want to bring to market would have no user interface, no buttons, no lcd, just a few inputs and outputs as follows:

1) mains input
2) AC output
3) DC
4) control wires
a) inverter on/off (closed circuit = on, open curcuit off)
b) charger on/off
c) max charge amps (maybe an analog voltage 0-5v say that means 0 to 100% of rated charge current from mains)
d) charge volts (analog signal, where voltage determines the output charge voltage)

Is there a relatively easy dividing line between the brains of the unit that supplies the control signals 4a, b, c, d to the analog components? (I say analog, but I think there is a pwm controller that generates the pure sine somehow, so there might be a brain in the "analog" side). In other words is there a separate communications/UI processor that can be ditched to save money?

Is it possible to hack apart an inverter and find that dividing line and make a prototype?

 

[rjenkinsgb]

You are describing a "UPS" - an uninterruptible power supply, as mass produced by many companies.


The display part is trivial compared to the cost of the other components; a typical small LCD is less than $1 in mass production.

They need a microprocessor (or microcontroller) to do various functions so the extra monitoring is pretty much just software and again has no real effect on manufacturing cost. The same part would link to the display, I cannot see why there would be a different part to the main control CPU. A serial or USB link to the control CPU also has trivial extra cost.

The expensive parts are the power electronics, which do need protection and monitoring circuitry to avoid damage from overload, plus the transformers etc. for voltage isolation and conversion, power factor control (required by regulations) plus PWM filtering for noise.

There are also a lot! of legal requirements for anything electronic produced for consumers, plus more when mains voltages and electrical safety come in to the picture. Things have to pass UL / CE approval or you can be prosecuted and the product banned from sale.

Such regulations do not apply to prototypes or personal projects, but you need to do a lot of studying and research - and be able to afford the approval testing, which is done by an independent service - before you can sell such a product.


The simple fact is that you can buy such things ready built cheaper from the far east than you can buy the parts to build one in the west...
eg.

This product is no longer available.

This product is no longer available.

www.alibaba.com

You can certainly build one yourself for educational purposes, there are a few designs online for inverters etc., but unless you are going to buy parts for tens of thousands in one go, you cannot get the unit price low enough to be competitive.

 

[johntaves]

Thanks, for the info.

Can you explain why "They need a microprocessor (or microcontroller) to do various functions"? I get that the lcd, buttons, serial are cheap, but can I get rid of the microprocessor/controller? What functions require the controller?

I'm not interested in making one of these. I would have it built in China. I am trying to understand if there is an opportunity of simplifying them if charging lead acid is removed from the goals.

Thanks for your help.

 

[Nigel Goodwin]

Thanks, for the info.

Can you explain why "They need a microprocessor (or microcontroller) to do various functions"? I get that the lcd, buttons, serial are cheap, but can I get rid of the microprocessor/controller? What functions require the controller?

Everything you've just described, LCD, buttons, serial, and everything else - microcontrollers make things possible and cheap - using a microcontroller drastically reduces your costs in almost all cases.

I'm not interested in making one of these. I would have it built in China. I am trying to understand if there is an opportunity of simplifying them if charging lead acid is removed from the goals.

If you're not charging the battery, how is a UPS supposed to work?

I don't see at all what you're trying to do?, it sounds exactly like current UPS's?, most of which don't have displays or many buttons - mine has two buttons, power ON/OFF and Master Enable (which allows printers etc. to be powered down when you turn your PC off.

 

[rjenkinsgb]

Can you explain why "They need a microprocessor (or microcontroller) to do various functions"?

That's usually what synthesises the sine modulated PWM. Such as a PIC MCU with PWM suitable for driving a push-pull or H bridge for the inverter is cheaper and simpler than trying to make one from discrete components

As that takes little CPU time, it has plenty of resources to do all sorts of other things; voltage monitoring, overload protection etc. etc. which have trivial cost but add to the features and prevent the device getting a reputation for being easily damaged,

 

[johntaves]

Everything you've just described, LCD, buttons, serial, and everything else - microcontrollers make things possible and cheap - using a microcontroller drastically reduces your costs in almost all cases.

But, I am trying to get rid of LCD, buttons and serial and the microprocessor, so what exactly is "everything else"?

If you're not charging the battery, how is a UPS supposed to work?

I want to charge lithium, which requires it's own microprocessor to monitor individual cells. Consider a typical solar lithium build.

1) microprocessor in the solar controller
2) microprocessor in the lithium BMS
3) microprocessor in the inverter/charger

Look at how this came about. Years ago we have an inverter/charger (UPS) and a lead acid battery. So, they put a microprocessor on the charger to control the charging of the lead acid. Great, then along comes the solar charger, and it too needs to have a microprocessor to control the charging. Now we plop a lithium battery in there, and because the individual cells are more sensitive to over/under volts it has a microprocessor too. This begs the question, why do we need microprocessors in the chargers?

Are we in this "of course it needs a microprocessor" thinking only because of the path that got us here?

 

[johntaves]

That's usually what synthesises the sine modulated PWM. Such as a PIC MCU with PWM suitable for driving a push-pull or H bridge for the inverter is cheaper and simpler than trying to make one from discrete components

As that takes little CPU time, it has plenty of resources to do all sorts of other things; voltage monitoring, overload protection etc. etc. which have trivial cost but add to the features and prevent the device getting a reputation for being easily damaged,

Ok, that makes sense. But do they do that? I mean do they have just the one CPU?

See the EGS002, I thought this would be the core of the inverter. To finish the inverter we add the transformer and other analog components. Is that correct? The EGS002 does not provide any way to monitor and control the lead acid charge profile, much less drive a user interface, right? Is this thing even involved with the charge functionality?

Again, thanks for helping me understand.

 

[rjenkinsgb]

In that module, it looks like the "EG8010" IC may be a pre-programmed MCU, purely for a single-function inverter.

For a more generalised device you would program your own MCU with the appropriate functions.

 

[Nigel Goodwin]

Are we in this "of course it needs a microprocessor" thinking only because of the path that got us here?

No, we're in the "if you don't use a microcontroller your device will be far more complicated and expensive than all others, and have a much poorer performance" mode. As already mentioned, it creates the sine wave, it does all the control functioning, charging, switching modes, safety functions - everything - it may quite easily have more than one micro, they are so cheap and versatile.

Li-Ion BMS's don't usually use a micro, they use custom chips designed and manufactured for that exact purpose - although you can certainly do it with a micro if need be.

Why Li-Ion instead of lead acid?, don't you think there are good reasons why UPS's seem to universally use lead acid? - car batteries are lead acid, again same reasoning.

You seem to be wanting to wanting to 'simplify' it by making it more complicated and more expensive, and probably less effective.

 

[johntaves]

As already mentioned, it creates the sine wave, it does all the control functioning, charging, switching modes, safety functions - everything - it may quite easily have more than one micro, they are so cheap and versatile.

My goal is to understand what functions need a microcontroller.

1) "it creates the sine wave" -- OK, I get that. And I also agree that maybe the charger/inverters on the market all have just one controller.
2) "charging" -- Sorry why does it need a microcontroller, unless you want to charge lead acid, which I don't?
3) "switching modes" -- Why? Do we need software to decide if the mains power is OK?
4) "safety functions" -- What safety functions require a micro? Isn't putting a micro in charge of safety just adding another layer of complexity?
5) "everything" -- This is the problem. As soon as you say "everything" you are now telling me the best way to make a typical inverter/charger for a lead-acid world. I don't want to know that.

they are so cheap and versatile.

I am sure that the margins are nil on these inverter/chargers because there are so many choices that have very little differences. Any savings is beneficial, so even if you are really trying to say that the transformer cost overwhelms the cost of a micro, the additional micro is still a cost. You can argue that the design cost is simplified by buying the inversion board and slapping on an additional micro to handle the UI stuff, and that's exactly what I am wondering. Can that additional complication and minor expense be eliminated if we drop the need for lead-acid and the need to fit into the existing complementary products based on lead-acid.

Li-Ion BMS's don't usually use a micro, they use custom chips designed and manufactured for that exact purpose - although you can certainly do it with a micro if need be.

I don't know of any that don't use a micro for any significant storage size. But regardless, the BMS I need for the market does need a micro.

Why Li-Ion instead of lead acid?, don't you think there are good reasons why UPS's seem to universally use lead acid? - car batteries are lead acid, again same reasoning.

Yes, of course there are good reasons, lead acid was cheaper. Emphasis on "was".

 

[ChrisP58]

snip

Yes, of course there are good reasons, lead acid was cheaper. Emphasis on "was".

I think they still are. Can you post links to Li-Ion batteries that are cheaper than lead acid of the same, or similar, capacity?

 

[johntaves]

I just bought 8 cells of 280ah for $1000 from Xuba in China delivered by slow boat. This will be 6.7kwh capacity. I figure that's similar to 8 trojan 6v 225ah for $160ea., assuming we have an effective usage of say 85% of nominal from lithium and 50% of nominal from lead acid.

 

[ChrisP58]

I just bought 8 cells of 280ah for $1000 from Xuba in China delivered by slow boat. This will be 6.7kwh capacity. I figure that's similar to 8 trojan 6v 225ah for $160ea., assuming we have an effective usage of say 85% of nominal from lithium and 50% of nominal from lead acid.

Eight 280 AHr LiFePO4 cells at 3 volts is 24 volts, for a WattHour capacity of 6.7K WattHours. But 8 cells at $330.68 each is $2,645.44, not the $1000 you state that you paid.

As for the Trojan batteries, 8 225 AHr batteries at 6 volts is 48 Volts (24V in 4x2 series-parallel) with a WHr capacity of 10.8K WattHours. That's about 60% higher than your LiFePO4 pack. At a cost of $1,280, about half the cost.

But, to be fair in your comparison, you need to use Chinese pricing for both. This is a 2v 300ah Lead-acid cell. 12 cells is 24 Volts, with a total WHr rating of 7.2K WHrs, at a cost of $900.

Unless I'm missing something, Lead-acid is still cheaper than LiFePO4.

What are you basing your "effective usage" numbers on?

Notes:
- Shipping cost are not factored in. Lighter LiFePO4 cells will be cheaper to ship than Lead-acid.
- LiFePO4 (Lithium iron phosphate) have less energy density, lower operating voltage, and are less expensive than LiIon.
- There are other factors that affect the comparisons of LiFePO4 and Lead-acid. The info in this post is based only on initial cost.

 

[johntaves]

But 8 cells at $330.68 each is $2,645.44, not the $1000 you state that you paid.

That's $330.68 for a set of 4. It was $985 total to my door for 8 pieces.

What are you basing your "effective usage" numbers on?

My understanding is that generally you want to charge your lead acid to 100% and avoid going lower than say 50% too often. With lithium, generally stay away from fully charged and zero by say 10%.

I was hoping to get information about inverter/chargers, and not reasons why an inverter/charger should handle lead acid.

 

[rjenkinsgb]

My goal is to understand what functions need a microcontroller.

1) "it creates the sine wave" -- OK, I get that. And I also agree that maybe the charger/inverters on the market all have just one controller.
2) "charging" -- Sorry why does it need a microcontroller, unless you want to charge lead acid, which I don't?
3) "switching modes" -- Why? Do we need software to decide if the mains power is OK?
4) "safety functions" -- What safety functions require a micro? Isn't putting a micro in charge of safety just adding another layer of complexity?
5) "everything" -- This is the problem. As soon as you say "everything" you are now telling me the best way to make a typical inverter/charger for a lead-acid world. I don't want to know that.

1 - OK.

2 - It does not "need" one, it is just simpler with one, if there is already one somewhere in the device.
Connect a couple of ADC inputs to sense voltage and across a current shunt resistor, plus an output to control charge current and you have limiting and charge calculation capability, with a far lower component count and cost than using separate components for charge control.

3 - Same again. Changing modes is quite complex, several different things need to happen for optimal control. Having one part that can "see" all information and control everything is a lot simpler than trying to use several "dumb" controllers and get them to work together.

4 - And again. It can monitor every voltage and current, in / out / battery etc. so can calculate safe limits and shut things down or sound an alarm if they go wrong.

5:
The battery type is irrelevant to all that so far, but with a multi-cell lithium pack a micro is far better as it can also monitor individual cell voltages and control balancing or limit charging.
Lithium charging is far more complex and critical, with multi-cell batteries, than lead acid - they are a doddle by comparison, as they can stand permanent trickle charge, all that's needed is overall voltage control and some current limiting.

You cannot use an cheap chinese BMS with eg. 60mA max balance current with 280AH cells! You need something that can switch a fairly high current load across any cell as needed to maintain accurate balance.

 

[Nigel Goodwin]

That's $330.68 for a set of 4. It was $985 total to my door for 8 pieces.

My understanding is that generally you want to charge your lead acid to 100% and avoid going lower than say 50% too often. With lithium, generally stay away from fully charged and zero by say 10%.

I suggest you do some more research, as Li-Ion are easily permanently damaged by over discharging, FAR, FAR more so than Lead Acid.

 

[johntaves]

1 - OK.
...
5: The battery type is irrelevant to all that so far

OK, thanks for your patience and willingness to break it down.

...with a multi-cell lithium pack a micro is far better as it can also monitor individual cell voltages and control balancing or limit charging.
Lithium charging is far more complex and critical, with multi-cell batteries, than lead acid

I know. That's why I was exploring the idea of having the BMS provide the charge instructions to the inverter. It has to shut off loads and shut off charging, so instead of having high current circuits for that in the BMS, why not send signals to the inverter/charger and hopefully lower the cost of the inverter/charger.

You cannot use an cheap chinese BMS with eg. 60mA max balance current with 280AH cells! You need something that can switch a fairly high current load across any cell as needed to maintain accurate balance.

That may be true for high C charge/discharge, but the off grid/RV market generally does not have that situation. At most I will be doing .2C.

Thanks again for your information.

 

[johntaves]

I suggest you do some more research, as Li-Ion are easily permanently damaged by over discharging, FAR, FAR more so than Lead Acid.

I'm not sure why you concluded I needed to do more research from this statement: "With lithium, generally stay away from fully charged and zero by say 10%."

The whole point of this thread was for me to understand why the inverter/charger cannot be made less expensively given that the BMS has to have smarts to keep the lithium OK.

 

[rjenkinsgb]

At most I will be doing .2C

0.2 x 280 = 56A charge.
Than needs some serious balance and monitoring to prevent imbalance!

BMS boards as typically sold are last-ditch safety devices only. they are not charge controllers.
The charge section of the system must have both voltage and current control for optimum charging and battery life.

It's another situation where trying to stitch together random parts intended for different single purposes it more complex and expensive than using a CPU to do the overall control.

 

[schmitt trigger]

I worked for 15 years at a company that built UPS, power inverters and large power supplies. Up to 150 kW. I designed several commercial products.

I can tell you that the cost of such products is never, never, never, in the microcontroller.
The power semiconductors, the large heatsinks and/or fans, the huge bypass capacitors, the magnetic components (inductors and transformers), the protection and monitoring devices like fuses, current sensing circuitry, large capacity transient suppressors....... that is where all the cost is. Easily 95% plus.

Actually, because of all the necessary monitoring and protection circuitry, and I mean necessary and not optional, a microcontroller is actually a cost reducer.
And since the microcontroller is already gathering and manipulating so much internal state information, it only requires a few lines of code and an inexpensive LCD display to provide all sorts of good information to the user.

Oh, did I forget that additional lines of code can also provide very useful diagnostics? Like how many charge/discharge cycles the battery has had, the actual run time, overload events, and many more.