VDC ripple improvements

[Jeroen2 (Jeroen Kuijf)]

Good afternoon,

VDC ripple improvements.

As discussed on the previous forum, the results are in about the VDC ripple improvements we made since this past half year on a 3-phase 24V 8K Quattro system with 4200AH bats.

2 graphs are captured, the first is in normal view and the second in detailed view.
Till about the 24th of June the installation was running bog standard, from then till 30 september we added on each inverter a 0.27F capacitor over the busbar, from then till now we placed 2 0.75F capacitors with a choke in between.
The result is very clear, ripple can be improved enormously.

Regards, Jeroen.
VRM Portal - Victron Energy, no password needed.

[ricardocello (Rick)]

Great job!

I’ve got dual 48V/5k Quattros (USA split phase), and decided to add a 0.1F low-ESR TDK electrolytic cap across my busbar. This reduced the rms ripple from 1.4 Vrms to 1.0 Vrms when inverting 7000W. Looks like additional caps would really knock it down.

I’m measuring peak AC ripple current through the capacitor at almost 60A.
Did you check to see that your caps can handle the large ripple current or take any measurements?

[Netrange (Steffen)]

I use 3 parallel capacitors in my 48V system with ‘only’ 2 Mp2 3000 but another 2 3kw BASBA for hot water tank and air conditioning …

Elko Kondensator radial 47000µF 100V | Schraub Elektrolytkondensator bis zu 85°C...

Works perfectly when large loads are switched on

[Jeroen2 (Jeroen Kuijf)]

Hi Rick,

Thanks for the compliment.
Yes we did/do measure a lot, as this test is still running.
Over the busbar before dividing to the 2 batteries we have placed 2 Licap 18V 62F Ultracapacitors in series (24V system, 36V capacity, 32V max charging voltage) already years ago, these “boost” up to about 80A when we start heavy equipment.
This was our bog standard system for years but we noticed secondary ripples (3 phase), and a supercapacitor is not really good for that.
So, we placed first these 3 Kemet ALS71A274NT063-ND capacitors right on the busbar of each inverter with silver cables (speaker cables), these produce without load about 6-8A and when we start heavy equipment about 43-45A just for a flash second, just over the limit of 34A.
Then we mounted 2 Kemet ALS71754QT040 capacitors on the busbar to the batteries and the positive side of this busbar is cut and a IHXL1500VZEBR68M5A
mounted in between.
This since its an inverter-charger so this ripple is “handled” while inverting but also while charging, a so called CLC-circuit for 2 directions in this case.
These also use about 5-6A without load and with heavy equipment about 35-40A.
That created that the 3 other capacitors are reacting slightly less and are now also at about 35-40A when starting heavy equipment.
Under load, after start, all capacitors do 12-18A and the ripple is never higher than 0.2V.

Your TDK capacitor is from the B41xxxx-series 63V, correct, these can handle about 23-25A max, so watch out to run long time on full power as these can get very hot, also, it reduces lifetime by 5-fold or so…
I would recommend to place Kemet ALS70A334QS063 directly over the busbar connections of your inverters, this, as you have a way bigger ripple to solve than our system, these can handle 40A with ease and covering your issue.
Also, you have 48V, ours is 24V, meaning, the choice of big capacitors is limited to 63V and/or above, leaving you with the above mentioned capacitor (in stock) as I write.
But, again, 1.4Vrpp is quite high for a low amperage system, I would check cabling connections once more really good, meaning, take these loose, clean with P400 sandpaper (for prepping cars before paint), then scotch brite and then reconnect, you will be surprised what that does before investing in capacitors and chokes.

What is your VDC-loss?, between the quattro voltage sense and quattro voltage, this we monitor as well.

Here you can see the 3 inverter voltages and the voltage sense of the master, that’s our safety to monitor cable connections, etc, before things get out of hand.

Jeroen.
VRM Portal - Victron Energy, no password needed.

[Jeroen2 (Jeroen Kuijf)]

Hi Steffen,

Yeah, it does, but the only thing I can say is watch out with the starting amps with single capacitors connected to solve such issue.
Victron places multiple capacitors instead a single one in their inverters, that’s not done for the capacity nor for the space (yeah, a little it is), but mainly for the amperage, a 24V 8K quattro has 26 capacitors with a total of 0.213F built in, these can handle 83.2A with ease and a similar single one just about 40A, that’s why multiple small capacitors are in an inverter.
Mostly of the time a ripple occurs is due to the battery is to small (specially Lithium), and the cable connections, and the start load to big so starting takes longer.

Jeroen.

[ricardocello (Rick)]

Jeroen,
I did look at ultracaps, but putting enough of them in series to get past 51.2V was cost prohibitive for me. Also, their ESR is too high for meaningful ripple reduction. But for starting big equipment that makes total sense.

I’ve built an experimental holdup-box that I put in my rackmount setup very close to the busbars, fused with a MEGA 80A fuse using 4 AWG (21.1 mm^2) welding cable. I regret not making the box bigger to handle more capacitors in parallel, but it was just an experiment. Seeing what you’ve done, I realize getting much closer to the busbar would greatly improve the results.

I’ve only got 300 Ah total LiFePO4 rack mount batteries. In general, I find the charging ripple to be less than the inverting because I only charge at 100A maximum. If I had a bigger battery I would expect the ripple to be similar to inverting. Sounds like you’ve got sufficient capacitors in parallel to keep the ripple current down.

Yes TDK B41458B8100M000, and I worry about cutting the voltage so close, but it has a surge voltage rating of 72.5V unlike normal caps. The next step is 100V caps (no 80A) and have a lower capacitance of course. The data sheet says they can do 57A without cooling. With cooling significantly more.

When I built the setup I attempted to measure the actual resistance in milliohms through the system by cranking 150A through and measuring the voltage drop. I’ll revisit that and see if things have gone downhill.

[ricardocello (Rick)]

150A drawn in these plots. Voltage drop from battery to quattros about 0.7V.
3.3 millohms from battery to quattros.

[Jeroen2 (Jeroen Kuijf)]

Hi Rick.

Yep, you explained your issue dead on, you are correct, due to that you only tried to “smoothen” the busbar to the battery you get these high inrush voltages as this busbar from the battery to the inverter is always “full” or resistances (very small for each single connection).
Therefore its always better to “start” with capacitors at the inverter itself and at the “entrance” of the busbar extra capacitors.
Yes, supercaps pricing went up 20% since this year.
The supercapacitor or ultracapacitor is opposite mounted, meaning, this one you put bang on the battery terminals, for extra “boost” from the battery.
In your case I would use 3 of these Licap 18V 62F Ultracapacitors (SM0062-018-P-1) in series, (we will mount these this year at another installation), with 2 200A fuses directly in between and then 2 silver cables to the terminals to connect them directly instead through fuse holders and cables (a lot less resistance again).

As you can see, these 3 can then be put right besides each other and linked.

Jeroen.

[Jeroen2 (Jeroen Kuijf)]

Hi Rick.

I see your graphs now, that’s way to much resistance, example, a 500A victron shunt is 500A@50mV=0.1milli-Ohm only, you see?, I calculate that your system is almost 5milli-Ohm.
When you run for an hour just below maximum power, nothing gets 3-5 degrees higher than ambient?, or more?, must be…, these connections must be paid attention to (clean very well).
Paint the bare steel or bronze parts, red cables, transparent plastics, etc, black, as that does not reflect light so much and so you can measure more precise with a temperature gun (easiest way).
Can you show me a “detailed setting” ripple graph of when you start your heaviest load, that might give even more information.

Jeroen.

[ricardocello (Rick)]

I’ll recheck and clean the connections, but I suspect the extra Class T fuses and MEGA fuses are adding resistance I cannot avoid. I’ve put a FLIR camera on the busbars after running for a while at full load and didn’'t see any obvious hotspots, but that was over 6 months ago. I’ll give everything a thorough review sometime next week and let you know.

Thanks for the great advice on reducing resistance and proving that the ripple reduction is feasible!

[Jeroen2 (Jeroen Kuijf)]

Hi Rick,

You are welcome.
Yes, that can very well be the case as these 200A fuses have 0.24milli-Ohm for the 70V version.
The 32V line of Mega has 0.26milli-Ohm for 200A, so no option there.
At 200A (100%) these fuses will blow after 14400 seconds and at 200% in about 1-15 seconds.
So to reduce your ripple drastically you can try to mount temporarily 2 400A fuses, these are only 0.13milli-Ohm, this to test the other parts/connections which might have high resistance and to see if the reduction is confirm the “new” resistance.
Then, you put a blank in the fuse holders, again to test, again you get results, and these might now be very close to 0.15-0.2V ripple, then you know for sure where the resistance is/was.

Jeroen.

[Jeroen2 (Jeroen Kuijf)]

Hi Rick,

Any improvements with your installation?
Did you try some blank fuses, just to see the difference?
This is my Ripple till so far, we monitored it now for a long time and we see that more improvements are not anymore needed.

Again 2 captured pictures, one in VRM normal and the other in detail.
Just the 24th of June is visible when we started our testing, and from the 30th of September nothing else was done anymore.

Jeroen.

[ricardocello (Rick)]

Your ripple is outstanding!

I’ve purchased a YR1035 4-wire meter to check resistances.
I’m going to measure the internal battery resistance, because I suspect it isn’t great.

It’s hard to bring things down when it is producing power,
but I should be bringing everything down and disassembling/testing/reassembling sometime this week or next, with Winter solstice being a good excuse.

I’ll post any ripple improvements here, thanks for checking in.

[Ludo]

@ricardocello

I’m going to measure the internal battery resistance, because I suspect it isn’t great.

I must have missed something in school, for me battery resistance is more a calculated value then something that can be measured with an ohm meter.

Keep us informed please.

[lxonline (LX)]

Meter probably does something similar to a normal dmm for rms readings - it is calculated.

[ricardocello (Rick)]

YR-1035+ Meter uses four wires, can measure milliohm resistances and battery internal resistance. Uses 1 kHz AC signal, and properly excludes inductance from the resistance measurement.

I’ve purchased a YR-1035+ 4-wire meter, and I’m measuring everything I can get my hands on. First thing I did was measure a 15 mΩ 1% wire-wound resistor to check the calibration.
It read 15.01 mΩ, so I’m happy already.

What I’m trying to do is reduce resistive losses causing DC ripple in the Quattro split-phase rack mount system.

The Layout
[– Battery Rack] → [AWG 4/0] → [SmartShunt] → [AWG 4/0] → [– Busbar] → [2x AWG 2/0] → [L1 Quattro, L2 Quattro]
[+ Battery Rack] → [AWG 4/0] → [+ Busbar] → [2x MEGA 70V 200A] → [2x AWG 2/0] → [L1 Quattro, L2 Quattro]

[Ludo]

Ok, you’re checking the wiring, not the battery itself.

THAT makes sense

[ricardocello (Rick)]

I also measured inside the battery rack and batteries.

This is a standard EG4 3-slot battery rack, with customized 360A Amphenol connectors mounted on the side to make the rack portable.

[Jeroen2 (Jeroen Kuijf)]

Good morning Rick,

Give me some days to “draw” a picture of your measurements.
This as the total resistance you measured does not add up with the ripple you have.
Actually you measured way more resistance than expected, the cabling and fuses I agree fully, as was proven already more or less, before your measurements.
But the internal resistances of the batteries are giving me a bit of a doubt.
Well 3 batteries in parallel of 12mOhm is of course only 4mOhm, but still a lot.
Resistance of a battery is not so important, because as soon as it starts delivering power, this resistance drops dramatically, its only when you charge that there is this resistance you measured, with Lithium not so much, with lead a lot of course.
So, I like to propose one test, easily done, can you charge at max power and then make the voltage loss graph once more in VRM, with the voltage sense from the master Quattro right on the master battery terminal, this to see the real voltage loss over all the cabling, fuses, etc.
If that graph is “almost” the same as the discharge graph (see above), then it is mainly the resistance outside of your batteries, see my point?

Regards, Jeroen.