(1) Well, AWG 2/0 (~64 mm^2)is what I had lying around 
(2) I was trying to keep the cable resistance < 1 milliohm, as the caps have 7 milliohm ESR
Iām seeing about 37A ripple current on each capacitor when doing the 7000W inverter test.
1 Like
OK, thanks!
And an additional question, the same asked to Jeroen above.
Did you looked with an oscilloscope to see if the ripple is the same as the one measured by the system?
Thanks again.
A while ago I did put an o-scope on the battery busbars and the rms values were very close to what was in VRM. However, I should redo that (on the Quattro battery terminals directly).
Ripple is definitely 120 Hz (US power). It would be 100 Hz elsewhere.
Do you have doubts about the measurements made internally and presented in VRM?
Kinda, as it shows ripple only when discharging and almost none when chargingā¦
But this could be because of the frequency used by the boost converters of the MPPTs on my system, which may be over 25kHz and then the ADCs that follow the ripple voltage to not be able to work on such high frequency, probably because of some low pass filters on their input.
I forgot to mention here that I have a Multi RS that is working a little bit different. The MPPTs are pumping on a high voltage bus (over 350V DC) and from there the bi-directional DC/DC converter is charging the battery.
Wait to see the capacitors on the high voltage busā¦ 5 pcs with a total capacity of 2800uF at 500V !
Try to improve that bank too. 
1 Like
Victron should sell something like this as a bolt-on accessory or maybe something to add-in to the lynx - distribution system ectā¦
I thought I went big enough with 300ah 48v LiFePO4 and two 5000va Quattro in split phase, but they constantly give me ripple warning for years, I turned off the warningsā¦
I do believe the manual recommends a 400ah bank, which would help with DC rippleā¦
1 Like
Wouldnāt it be better to have caps internal to each inverter, and add an additional relay (or use one of the 7? relays on the MP) to have a software-selectable way to enable the caps?
These caps are huge! They wonāt fit internally, maybe if they increase the size of the box.
They carry 37A of AC ripple current (in my setup) superimposed on the 52.1 V nominal battery DC voltage. They are part of the pre-charge ritual needed to power up the inverters.
They have less than 7 milliohms of resistance, so they will look like a dead short when fully discharged.
I donāt think having a relay or contactor switching them in is really feasible.
Point taken. Thanks.
But your point is perhaps that Victron should be using larger/more capacitors internally to address the ripple issue, or change the battery specifications to work with the capacitors theyāve used.
I know the Multiplus-II had issues with insufficient capacitors and people were recommending adding these kinds of capacitors externally but close.
Good morning,
@ Rick, so, you mounted 0.1F on each inverter, correct?
That gave you an improvement of 1.4V before and 0.6V now is 0.8V improved, correct?
I would mount these ones , Kemet ALS70A274NT063 together with the ones you already use, that should drop it even more.
Or several 0.1F to increase the current they can handle.
Regards, Jeroen.
Correct, each Quattro has its own 0.1F 100V cap within 300mm of the box with 67.4 mm^2 (AWG 2/0) cabling.The improvement was 1.4V rms to 0.6v rms drawing the same 7000W load.
Iām running out of room and $$ for capacitors 
but yes, adding more will clearly help.
What Iām really trying to achieve is ESS and LOM not dropping out on grid brownouts.
But I have no way of simulating those, unfortunately.
Good evening Rick,
Indeed, when you add more capacitors to your system it will help.
I would like to see the results again when you are going to do that, of course first $$ and there is always space somewhere, hahaha.
Regards, Jeroen.
1 Like
When running a single phase inverter, the DC current has a massive 2nd Harmonic (to power freq) current the RMS value of the DC current will be 1.22 times the average DC current and it is a function of the inverter itself. It is not peculiar to Victron, it happens with every battery inverter i have looked at. It is a direct result of the current delivered to the AC output. At zero crossings, little to no DC current is required while at Maximum (Ā±) AC current the DC current is also maximum. There is little energy buffering inside the inverter and the only way i can see to store large amounts of energy is to use a higher DC voltage or an inverter design incorporating more inductance as they are able to better store energy at low voltage and high current due to Ā½LiĀ². I wrote the attached and recorded the included waveforms a few years ago if anyone is interested.
IES DC Current.pdf (498.6 KB)
3 Likes
Good morning Michael,
Very nice study you made, very clear to me.
Pity you didnāt do that also with a Victron, that island mode SMA is running fine, I would say, isnāt it?
With our field test of about half a year as above shown, we were able to massively improve the ripple by placing capacitors directly on the inverter inputs, and also a CLC between the inverters and battery.
What we are seeing now is that the batteries charge faster and discharge less, till now, preliminary results, but we start to see already a 3% charge and discharge gain due to this ripple improvement.
This we did not anticipated before that this would be so visible, but since september last year we started to notice that the energy is used more āpreciseā, and we run longer on the batteries and so taking less energy from the grid.
We will keep a close eye on that, and after 6 months, then we can be sure about these results.
Regards, Jeroen.
Hi @MichaelHayes and @Jeroen2
Nice study ! 
Thanks for sharing.
Interesting to note that on the SPMC 204AU, the DC current peak is after the AC voltage/current peak, on the time axisā¦
You said ālikely due to minimal capacitor storageā. Isnāt in other way around?
Just like some capacitors inside are quite enough for initial conversion and then, later, are charging back after the conversion depletion? 
For the other inverter, the SMA SI6.0H, the DC current graph is quite in sync with the AC conversion, meaning the capacitors inside are not enough to take the initial load and are drawing, synchronously, from the battery.
So, maybe, as a conclusion, add additional ripple capacitors up until the DC current is starting to lag behind the AC waveform.
Then, you know they are enoughā¦
Hi Jeroen, I wanted to run the test for a Quattro but did not have access to one when i was ready. I do have a Multiplus system i am setting up in a caravan at the moment and i will update with those results when i have a chance. I can see how the input capacitors can improve the situation and when i get a few minutes, i will calculate what the tidal energy flow would be per input Farad as my gut feeling is that it would not give much improvement but your figure of 3% RTE improvement is a good step in the right direction. The other thought i had was that there is thermal and magnetic protection between the battery and the inverter for cable protection. Will large capacitors at the inverter terminals open the door to much higher transient capabilities to the point of harming the inverter??
Regards,
Michael
Good morning Alex and Micheal,
Indeed Alex, at idle, the capacitors inside the inverter are enough, then from idle to 75% load the ripple raises pending on all variable resistances etc in the DC line.
So, as you say, place for 75% load more capacitance (double the inverter capacitors), then for starting a load an extra same amount as the CLC we placed in the busbar or simply also directly to the inverters (a little less effect).
And/or, use the āformulaeā as mentioned by me before.
Regards, Jeroen.
1 Like
Hi Alex,
Well spotted on the DC current phase shift. I had noticed this and also put it down to extra input capacitance. I counted the caps in the sunny island but have not checked the Selectronics. Note that the load i used was a 1000W incandescent light which has very sharp inrush as the Tungsten heats up but should remain very resistive beyond that point so no real need to consider phase shift in the AC current as can be seen by the AC waveforms. Interesting thought on quantifying external input capacitance. Has anyone run the idea of extra caps past the OEMs to make sure there is no problem with it??
The minimal capacitance comment was for ELV battery inverters in general, not as a comparison between these two inverters. The observation comes from designing Transformer Rectifier units in my early days. When you need very low voltage and very high current, you will never achieve it on capacitors alone. Inductors make a huge difference. With just capacitors, the rectifier peak current are massive and given that almost anything with electronics in it these days starts by rectifying the mains, the cheaper units may be what is causing a lot of peak clipping and high harmonics on the incoming grid. The power supplies that are designed correctly do a fantastic job of managing hash back to the grid.
Good morning Michael,
No, it will not harm, its actually a benefit, the DC flow gaps are āfilled inā by these capacitors, and so the inverter MOSFETs get/deliver a very āflatā DC voltage (specially with 3 phase).
Think extreem, on a USB power supply/charger for example, the bigger the output capacitor is less ripple, but, not too big, expensive and not needed.
Then we see even that the inverters stay way lower on temperature, way less fan switching (almost never), way way less overload warnings starting heavy loads, no āhammeringā noises (as many people have in other topics), way more stable 230VAC output voltages during starting a heavy load.
Regards, Jeroen.
