Ripple in Double Conversion UPS

[SWSL]

I've been warned about "ripple" charge/discharge in my (discreet component) double conversion UPS setup. But I can't find anything here or on the web about this.

Does anybody here know what it is ?
How bad it is for LFP packs?
Would adjusting settings on my Victron charger help avoid?

Thanks!

 

[DIYrich]

Search the web for:
double conversion ups "ripple"

 

[SWSL]

Search the web for:
double conversion ups "ripple"

Exactly what I did for an hour or two before asking. I am usually pretty good at scouring for info. I also searched this forum. I dropped ripple and just read anything I could find on double conversion. This "thread" is now the top result.
...........

Hey, I just did a different search (50 posts into this thread) and got lots of hits. The trick was to leave "double conversion UPS" out. I searched "ripple effect Victron" (I was looking for the Victron paper on this) and lots of other info showed up. So there. Curious minds will find plenty.

 

[zanydroid]

Did you go to the author websites to get the PDFs of the dozen or so academic papers that pop up on Google? At least one of them should be an Ok start if you go through the intro, they have to provide some background to support their narrative.

 

[HarryN]

Just for my own clarity, is your system more or less:

- 120 vac or solar power coming in - chargers - battery pack - dedicated inverter -- 120 vac out?

If you are talking about (battery DC voltage fluctuation ) - This will be no worse than any other system.

If you are talking about 120 vac power coming out that is coming only from the inverter - then the quality of this power will entirely depend on your inverter quality - which is a good thing.

Personally I like this type of setup.

 

[zanydroid]

For more intuition one could also do mental "cross training" with reading up on ripple in capacitor ratings. It is different chemistry and physics but maybe 50% of the concepts are similar. So if you understand one it will help you understand the other.

I'm pretty confident we can conclude, based on the fact that power capacitors almost always have ripple ratings, and batteries do not, that it's a more significant issue for capacitors than batteries.

And, there appears to be two types of ripple (at least) people are talking about. Ripple when net power goes into the battery (or always out of the battery), probably doesn't matter. This is likely not what people are worried about with double conversion. Ripple around zero point, where it goes in and out, is the concern.

I was able to find more information on quick Google of ripple for power capacitors a few months ago, compared to Google for ripple in Lithium ion batteries earlier today.

I doubt changing settings on the charger/inverter will help because this is a lower abstraction level than the charger/inverter is specified at. But there could be some operating regimes where the ripple is better. If it's important to worry about. Need to buy an oscilloscope to test this kind of heavily time-domain hypothesis, you will get nowhere with this otherwise. You can likely understand the theory without buying stuff.

 

[Vigo]

Im sure ive got some of it wrong but from what i’m seeing, ripple is just talking about “when you convert ac to dc, its hard to get a perfectly flat output from a wavy input without adding a bunch of extra electronics”. Is that right?

Is there even evidence of this being a problem for lifepo4?

I check for ‘ripple’ on alternator charging systems pretty often, and at least in a car its considered acceptable to have hundreds of millivolts of ripple. Granted this would result in less current on a lead acid system than a lifepo4 system, but i would be surprised if the kind of ripple that comes out of SMPS that switch at super high frequencies, was significant or harmful to the battery?

 

[SWSL]

Just for my own clarity, is your system more or less:

- 120 vac or solar power coming in - chargers - battery pack - dedicated inverter -- 120 vac out?

If you are talking about (battery DC voltage fluctuation ) - This will be no worse than any other system.

If you are talking about 120 vac power coming out that is coming only from the inverter - then the quality of this power will entirely depend on your inverter quality - which is a good thing.

Personally I like this type of setup.

Good to hear that you like this approach! I don't really know what I'm asking about in reference to ripple since I am only pursuing this after being warned that this "ripple" is hard on batteries.

In this particular setup I am (currently) only charging with AC using a Victron charger. No solar online, though I do have a tiny bit (100 watts!) of panel and a spare SCC on hand to connect in case of a long term disaster. As it is, I can weather a more than a day of outage with my battery.

I understand that the process of converting AC to DC will have ripple from the sine wave that needs to be smoothed out. Any good DC power supply must address this. I can imagine that a cheap UPS system will have cut more corners than Victron did with their electronics.

Not concerned about the inverter - it's fine. I'm only trying to make sure that I don't (heavily) abuse my LFP pack with poor charge settings. In that context I don't know if the concern is 60hz ripple that makes it through the charger or perhaps some other oscillation at a much lower frequency within the charge/deplete action at the battery. Clearly the battery is acting as a big capacitor and will absorb any draws that exceed the charger output and then make it up from the charger, depending on the voltage settings. Is that also ripple?

 

[SWSL]

OK, thanks mucho folks. I'm concluding that ripple - whatever it means to an LFP battery - is not an issue for me.

If somebody has a suggestion for best V settings on the charger that would be great. I have control over Bulk/absorb and "storage" Still not sure why or when storage mode is effected with the Victron charging algorithm but it's mostly sitting there. I guess it's a kind of "float" that can handle minor drains but clearly not intended to keep battery at full charge.

But that's for another thread. I don't want to wear out my welcome!

Cheers

 

[Hedges]

My math says that if you draw 100A (mean) from a battery to create 60 Hz single-phase AC, the current will be 111A (RMS), and that is what will cause heating. So don't be surprised if the power you can draw appears to be 90% of what BMS is rated for.

I had measured battery ripple current for my inverter/AGM system with 40% load, and assumed that at full load ripple would increase to about 100%

 

[zanydroid]

Im sure ive got some of it wrong but from what i’m seeing, ripple is just talking about “when you convert ac to dc, its hard to get a perfectly flat output from a wavy input without adding a bunch of extra electronics”. Is that right?

I think there are two notions.

If it's the output of a charger, then that is fine, it's always going to be charging, just might be slightly faster with the higher voltage. Battery doesn't care, BMS probably won't be upset if the ripple is somewhat sane.

If it's the combined effect of charger + inverter load. Then there is the ripple the charger wants to put in as a source, and the ripple that the inverter wants to put in as a load. This ripple may be enough to continuously charge and discharge the battery which will cause extra charge/discharge chemical reactions to occur. I was too lazy to download the PDFs but from the abstracts it seemed like this is what papers are concerned about. Now what would be interesting is, is the effect of this different from counting these as regular cycles? Or, is there a concern that it's not possible to count this correctly (BMS not sensitive enough? frequency of ripple higher than the sampling frequency of the ADC? especially if the ripple is at the PWM frequency, if it's at the AC frequency then the BMS can probably see it), or the rapid cycling has a different chemical consequence. Too lazy to read and not a material scientist :laugh:

And this is also what is a problem for power supply capacitors - they're being exercised continuously from charge moving in and out, and that results in heat / chemical degradation.

 

[SWSL]

My math says that if you draw 100A (mean) from a battery to create 60 Hz single-phase AC, the current will be 111A (RMS), and that is what will cause heating. So don't be surprised if the power you can draw appears to be 90% of what BMS is rated for.

I had measured battery ripple current for my inverter/AGM system with 40% load, and assumed that at full load ripple would increase to about 100%

Thanks. I've got a 200A continuous rated JBD bms and am pulling around 10A constant. Probably more when fridge is in defrost but not gonna be 200A

 

[SWSL]

I think there are two notions.

If it's the output of a charger, then that is fine, it's always going to be charging, just might be slightly faster with the higher voltage. Battery doesn't care, BMS probably won't be upset if the ripple is somewhat sane.

If it's the combined effect of charger + inverter load. Then there is the ripple the charger wants to put in as a source, and the ripple that the inverter wants to put in as a load. This ripple may be enough to continuously charge and discharge the battery which will cause extra charge/discharge chemical reactions to occur. I was too lazy to download the PDFs but from the abstracts it seemed like this is what papers are concerned about. Now what would be interesting is, is the effect of this different from counting these as regular cycles? Or, is there a concern that it's not possible to count this correctly (BMS not sensitive enough? frequency of ripple higher than the sampling frequency of the ADC), or the rapid cycling has a different chemical consequence. Too lazy to read and not a material scientist :laugh:

And this is also what is a problem for power supply capacitors - they're being exercised continuously from charge moving in and out, and that results in heat / chemical degradation.

Nice! That's the closest to understanding any of us have found so far I think. Didn't think about how the inverter would have a ripple type effect on the draw.
The crux of the question with this double conversion is - "how does constantly drawing and re-charging right around a given point - say 70% charge - affect the chemistry?" And "is it better to have some more dramatic swings to keep the chemistry..um.. "exercised"?" The Victron charger will do that every 7 days automatically and frequent one hour power outages will too. Adding some solar may be helpful for the same reason. A bit more "exercise' of charge/discharge and less "ripple" which I'm now taking to mean tiny variations around a charge level.

 

[Hedges]

If it's the combined effect of charger + inverter load. Then there is the ripple the charger wants to put in as a source, and the ripple that the inverter wants to put in as a load. This ripple may be enough to continuously charge and discharge the battery which will cause extra charge/discharge chemical reactions to occur.

A HF inverter, fed by LV battery, may not have this ripple.

A LV "LF" inverter draws current from the battery bus, and any capacitors in parallel can only supply current by rippling in voltage, so battery supplies ripple current too. That is what I measured (with CT and scope). The caps can only supply HF ripple of the PWM that generates sine wave, not the 60 Hz LF ripple.

If inverter architecture boosts 48V battery to 170V or 340V HV rail and makes AC from that, capacitors on the HV rail could ripple several volts to supply single phase AC current while boost converter draws constant current from battery. Or, boost converter could draw ripple. Have to measure with AC ammeter or CT to observe.

I measured (with clamp ammeter) PV input of GT PV inverter while it delivered a couple kW. Voltage rippled about 3.5V out of 350V.

 

[zanydroid]

Nice! That's the closest to understanding any of us have found so far I think. Didn't think about how the inverter would have an effect on the draw. But the crux of the question I originally had is "how does constantly drawing and charging close to a given point - say 75% charge - affect the chemistry.?" I think that with lead acid, you keep it charged right up at the top and it's just different. Even so, they say that the UPS' with lead acid go bad way sooner than same chemistry in a solar system.

Right.

LFP should last a lot longer already. So let's say both are affected by 30% degradation in cycles or calendar aging, that's a lot more meaningful on lead acid.

So if there is NO chemistry difference, and it's just regular heating and cycles, then you can do some math based on how much ripple you think is happening, and guesstimate the impact. Maybe do some sensitivity or scenario analysis if not sure what the ripple amount is, in that case you can pick some numbers that look way bigger than likely to happen. (The ripple should also be measurable with normal tier lab equipment)

For maybe some more intuition. There were some other threads on ripple specs for capacitors where a power engineer was quite concerned that an off-label use of that MPPT (feeding it from a SMPS rather than a clean source like battery or panels with no high frequency switching) would overload the MPPT input capacitor's ripple rating. In the approved use, the only high frequency switching is from the MPPT's own actions, which presumably the engineers already factored in. You might guesstimate the effect of the additional SMPS stage as "doubling" the ripple, which is a lot.

 

[zanydroid]

A LV "LF" inverter draws current from the battery bus, and any capacitors in parallel can only supply current by rippling in voltage, so battery supplies ripple current too. That is what I measured (with CT and scope). The caps can only supply HF ripple of the PWM that generates sine wave, not the 60 Hz LF ripple.

Does either the HF or LF architecture send instantaneous power back into the battery (IE does the ripple current go negative from the battery's perspective and cause charging chemical reactions)? Seems unlikely. I guess inductors in a buck stage (which the HF approach needs) may have some kick back that is partly buffered by the battery. Not sure what the LF transformer will do, but I think anything that makes it through will be intrinsically low pass filtered.

 

[Hedges]

People have said HF architecture has trouble with AC coupling, slow to reverse between charge and discharge.

LF, at least true LF like early Trace and Warpverter, simply have FET turned on between battery and transformer. That swings both ways.

Most "LF" inverters have LF transformer but a buck converter between battery and H-bridge that turns pulsed DC into AC for the transformer. I don't understand why those wouldn't have same issue with changing directions.

But anyway, driving a "real" resistive load, current drawn from battery would be a rectified sine wave, never switch to charging.

Driving a purely reactive load, like a capacitor on an inductor (e.g. transformer primary with no load on secondary), sine wave current is 90 degrees out of phase from voltage, would alternately charge and discharge battery.

Or so my theory says. Some day I should measure more, make sure I'm not all wet.

The key is that capacitors in LF inverter can't store 1/120th of a second worth of inverter's rated wattage, at least not without massive voltage swing. Which battery won't allow. Nor can you get an inductor able to store that much. Figure something like 250# Line Tamer ferro-resonant transformer to deliver lost AC cycles to a 250W load.

I have measured AC voltage & current feeding a transformer (VA) and compared DC current draw from battery (W). In that case, a 9000VA rated transformer drew about 15 VA no-load and battery current draw increase (vs. no transformer connected) reflected 7W actually dissipated. The difference of 8W reactive power was cycling somewhere.

Could be partly in capacitors, given voltage variations allowed by AGM before much current shows, not sure. If it was a massive amount of reactive power, I think battery would have been charging and discharging.

Some of the GT PV inverters like TriPower offer "Q on demand", reactive power compensation by means of their capacitors and switch-mode supplies (could multiply apparent capacitor size, or even simulate an inductor.) Sunny island also has some reactive power features, and I presume those use battery for the storage across AC phases.

 

[krby]

I'm going back to the OP's statement about "double conversion UPS" having to worry more about ripple. I can only guess this is because the AC->DC converter/charger is always on, vs AC going directly to the load through a transfer switch like a standby UPS.

How is this any different from a float-stage charger on an SLA battery? I have trouble believing mV of ripple (at whatever frequency) actually affect a battery in any real way.

Related but not quite the same problem, Lithium chemistry batteries don't "like" to be kept at 100% full forever, which is exactly the pattern for a UPS. It's one reason (well, mostly cost, if I'm being honest) I went with SLA AGM batts when I cobbled together my DIY UPS from discrete components.

 

[zanydroid]

How is this any different from a float-stage charger on an SLA battery? I have trouble believing mV of ripple (at whatever frequency) actually affect a battery in any real way.

It's interesting/real enough for scientists people to get grants to investigate and write research papers on this. Remember the ripple will cause charging and discharging chemical reactions if current alternatingly goes into and out of the battery. I don't have the background/time right now to look into whether this is just a jobs program or very high social value research.

I would imagine for any new (or major update to the geometry/recipe/whatever) chemistry someone will want to create test cells and see what happens when subjected to this workload.

Lithium chemistry batteries don't "like" to be kept at 100% full forever, which is exactly the pattern for a UPS

You don't need to make it sit at 100%, you can use current shunt to stop at 80% and periodically charge to 100% to calibrate without compromising UPS endurance.

 

[zanydroid]

People have said HF architecture has trouble with AC coupling, slow to reverse between charge and discharge.

There is no reversing happening with double conversion. Dedicated separate charger and inverter stages.

You do make a good point about the power factor being something sometimes overlooked. Inverter rating usually assumes a minimal power factor, if the load deviates beyond that then the VA and W that the inverter can push out will both go down.

However I think the ripple problem will still happen with a 100% resistive load, it just won't be an inherent issue with the load type (IE, a 60Hz ripple). Fundamentally the charger/power supply + battery and inverter are pushing/pulling at the same time.

If hypothetically you disconnected the battery, at least the battery isn't subjected to extra load from this. But in double conversion UPS the simplest approach is to keep the battery connected all the time to minimize transfer time.