Trying to deeply investigate LED pulsing flickering with HF off-grid inverters

[fmeili1]

I just caught up on this thread.... Gotta toss something in here.

I run Samlex EVO LF Inverter/Charger. LF because it handles my well pump, compressor, MIG Welder & more and never any flickering or noise on the 120VAC line. If Charging from Genset, Pure Sine Inverter type, again No issues, no flicker or noise. If Charging from my Big Construction genny (not inverter) the EVO passes the sinewave through and there is some flicker & noise. You can actually hear motors (fridge, freezer etc) "grumble" and LED's (cheapo's not the Phillips or EcoSmarts) do flicker terribly. The cheapo "Walmart"ones actually BUZZ, tossed them fast.

Finished a new offgrid solar install for a neighbour on the 23rd. I installed a different LF Inverter for him, a Value Grade 3024 (identical to an AIMS 3000W/24V LF model) and he has a "filthy" 15 year old Basic Genset... Oi, it floats from 100V to 125V and frequency floats from 52Hz to 67hz... okay, it is quite fugly, the Pure Sine inverter still passes that mess through and I have Never Ever seen such horrible flicker, even with low watt incandescent.

Inverters/AIO's when taking "non-grid" power are subject to whatever the AC Source is putting out, they will NOT fix the sine wave or clean it up. HF Inversion is Hyper Sensitive and in a sense delicate - it is really NOT intended for motors & heavy devices unless way over built and that would push their costs above LF Inverters. LF Inverters with the large capacitors & Torodial Coils are designed to handle light to extreme & even "hard" loads like motors, this is also why they can handle 3X surge vs HF's 2X surge.

HF systems are attractive for ONE Reason - COST ! They are Cheaper & Lighter than LF.
Sadly, this is where "Dollar Wise & Penny Foolish" comes into play!. Spend the $ and get LF and suffer No Issues with all gear in your home or save a few pesos and suffer issues, especially if "motor heavy" with Big AC, Fridges, Freezers & Pumps, let alone Commercial/Industrial applications. FYI, if you look at Industrial Solar Systems, they are ALL Low Frequency.


Good article that sums it up.

Inversion Methods Explained: High Frequency vs Low Frequency | Magnum Dimensions

Understand the difference between high frequency and low frequency inverters with this quick article.

www.magnum-dimensions.com

I understand the difference and especially the different behavior about surge load and large inductive load capabilities.

In principle, both limitations could be somewhat reduced by oversizing the HF inverters and maybe by implementing a soft start circuit in front of some large inductive loads (I've done both oversizing and soft start for the central heat pump). The very short voltage drop when a large load kicks in, is a principle problem of the HF inverters (not enough iron to store energy for short times), even if they are oversized - but most people could live with just ONE short light flickering if these loads start.

But I don't really understand why the LED pulsing will happen permanently while a "dirty" load is running - not just at the moment when it turned on. Do you think this could also be explained because of the different amount of "iron" in the transformers between HF and LF inverters? Maybe it's because of the higher switching frequency of an HF inverter which may result in a more sensitive feedback loop of the PWM controller - just an idea. If the flickering would NOT be because of the less "iron" in the transformer compared to an LF inverter, I still have hope to find a type of filter for the HF situation to stop this oscillating problem.

 

[fmeili1]

You are looking to fix this problem from wrong end. This problem is caused by poor voltage feedback loop design inside the inverter. No amount of tinkering on the outside will fix that. You can try bruteforce way of rectifying to DC like I suggested earlier but that requires adding those rectifiers in front of every LED light which is a lot of work. Much easier to add small 800VA Victron LF transformer inverter and a sub panel.

Got it!

 

[Steve_S]

HF are all FET Based no torodial coils.

Yes, you are largely correct, many compensate by Upsizing the HF Inverters - the Sledge Hammer solution for finishing nails. There is a HUGE DOWNSIDE to that as well, which is not realized till it's too late... Bigger Unit = More Consumption to operate. HF are already heavier when idling or low power demands are being served. Again, using the Samlex as the example, see below...
Running an extra LF Inverter for lights etc... well... seriously, falls far outside of the KISS Keep It Sweet & Simple rules and over complications means troubles...

Samlex EVO Inverter Efficiency (Peak) 94%
No Load Power Consumption in Standby Mode (Inverting/ charging are suspended) < 5 W
No Load Power Consumption in Power Saving Mode < 8 W
No Load Power Consumption in Normal Mode (120 VAC Output, Typical) 25 watts

The guts of my Samlex EVO... look at that 65 Pound Coil. And this is just a 4000W-12,000W model.

 

[AntronX]

Samlex EVO Inverter Efficiency (Peak) 94%

Peak efficiency is only achievable at 15% rated load. Dirty little secret they don't say. HF inverter wins in that regard. The problem here is not LF vs. HF. The problem is cheap engineering vs. quality engineering. Buy 1st tier inverter brand and this problem goes away.

 

[Steve_S]

Peak efficiency is only achievable at 15% rated load. Dirty little secret they don't say. HF inverter wins in that regard. The problem here is not LF vs. HF. The problem is cheap engineering vs. quality engineering. Buy 1st tier inverter brand and this problem goes away.

Big part of it and the the $$ Factor attached to stepping up to Tier-1. My Samlex is around $2500 USD - that would be a Really hefty EG4 HF system.

For just a couple of pesos more: ($400)
EG4 12kW Off-Grid Split Phase Inverter Bundle
| 2 x 6000XP| 12000W Output | 48V 120/240V Split Phase | All in One Solar Inverter System [BNDL-E6000-2]
What's Included With Your BNDL-E6000-2? ✔ Ships from Texas! ✔ 2 x EG4 6000XP Off
List Price: $3,137.96
Your Price: $2,937.96

 

[NC_hydro]

Curious how the Midnight Rosie HF inverter would do in this scenario. Probably going to be the best built HF out there.

 

[webbbn]

This not going to work. It's a low pass filter. I don't think high pass filter will work either.

If my theory is correct, and I think you stated essentially the same, that it's the response time of the voltage feedback loop (essentially a lower limit on the bandwidth of the control loop), it requires a high-pass filter. The impulse response of the PWM is mostly high-frequencies.

I'm pretty sure the low frequency affect is simply the higher-level control loop that is trying to maintain 120/240V across a very noisy waveform. In fact, I'm pretty sure someone posted a picture in another thread of the waveform, and it had a large amount of high frequency (much higher than 60Hz) noise.

The fact that adding some load smooths it out, tells me that it's due to the high-frequency PWM overshooting due to being underdamped.

 

[AntronX]

Oh i see what you trying to do. Better to place LPF on the voltage sense line inside the inverter. Placing sufficient LPF on the power carrying circuit is very inefficient as it will have to pass inverter's full output current. But then you may be messing up the impulse response of the inverter. This should be done with better software control algorithm.

 

[webbbn]

Oh i see what you trying to do. Better to place LPF on the voltage sense line inside the inverter. Placing sufficient LPF on the power carrying circuit is very inefficient as it will have to pass inverter's full output current. But then you may be messing up the impulse response of the inverter. This should be done with better software control algorithm.

Maybe, but ultimately hardware determines limits for how fast the FETs can switch, and if they can't switch fast enough, that can't be fixed in software. Still a fundamental problem with the inverter, but a different kind of problem.

Personally, I would just figure out some sort of constant load to put on the system. Even if you have to add a bit more battery capacity to cover it, it's probably the cheapest solution overall. You could probably even automate it so it only kicks in when the total load gets too low.

 

[AntronX]

Maybe, but ultimately hardware determines limits for how fast the FETs can switch, and if they can't switch fast enough

Usually AIO inverters with IGBTs switch at 19 kHz PWM and that gets filtered by output LC filter. The problem has to be choppy non linear loads presenting non sinusoidal current load which confuses inverter's voltage feedback circuit.

 

[SparkyJJO]

This is one reason why solid LF inverters are superior.

 

[jmjack]

I still have hope to find a simpler solution.

What do you think about putting an isolating transformer in front of the dirty load which may reduce the interference between the inverter and the "dirty" load?

My "dirty" loads have a max. of 100W (old slot machine) and 300W (my 3D printer) so these would cost not very much. Maybe I should give my "diry" 100W load a try.

I tried using a medical isolation transformer for the issue you are describing/experiencing, no solution. I am now using a pair of 6000XPs, solved all LED flickering (strobing) issues. Can’t be happier!

 

[rhino]

This is one reason why solid LF inverters are superior.

As I mentioned earlier my Victron Quattro 5kVA still has lights flicker when a heat gun is heating up and when the fridge turns on there is also a noticeable blip in the lights.. so not sure what the issue is.

 

[ricardocello]

As I mentioned earlier my Victron Quattro 5kVA still has lights flicker when a heat gun is heating up and when the fridge turns on there is also a noticeable blip in the lights.. so not sure what the issue is.

Presumably you mean flickering when inverting, but no flickering when in grid passthru (or maybe you are off grid).

I get some flickering with grid passthru normally anyway, same when I bypass my quattros altogether.
Grid power isn’t all that clean at the end of the line where I live.

I actually think there is less flickering when inverting, with the sole exception of the well pump kicking in.

 

[Hedges]

Filters have been tried by many, the result was never as hoped. Many here tried filters & conditioners & failed - some of it is in threads.
I am sorry BUT do not shoot the messenger... You are free to explore it and chase the proverbial tail while spending cash but it will never be what you expect.

You would not BELIEVE the scale of program where I have seen that attempted. Filtering of a practical scale can't clean up a bad inverter waveform or output characteristic. Solution was to fall back on the tried & true.

Btw. I will use the wording "dirty" load for loads which interact in a negative way with the HF inverter to make some LED loads pulsing.

Now I ask myself if it's possible to put a filter in front of the "dirty" Load2 like shown in the following picture to stop the pulsing of the "LED Load" (to stop the 3-7Hz interference oscillation between Load2 and the HF inverters which causes the "LED Load" to pulse).

I don't think that is a dirty load. As AntronX says, it is instability of regulation/feedback in the inverter.
Maybe a purely resistive load would not do that, and capacitance causes the problem. Or could be just some level of load causes inverter to hunt.

I've actually chased interference of similar frequency range (in a completely different sort of system.) Using oscilloscope with 100 second sweep time and FFT, to see the frequencies present. We're working with DC, so other guys are looking at large capacitors, while I'm designing linear regulators. Either way, we don't care about efficiency/losses.

Maybe if you had the "knobs" of the control loop (PID or whatever), you could tune it to stop the flickering. Likely the original developers just sought to hit voltage specs, didn't have any measurement of what causes your lights to flicker.

You can buy inverters that work better, and you can buy lights that don't flicker so much. Maybe. It is one thing to not flicker on chopped 60 Hz sine wave from dimmer, quite another to ride to 300 millisecond ripple. Maybe a good non-dimmable LED or fluorescent lamp with current regulator fed by capacitor.

 

[Checkthisout]

As I mentioned earlier my Victron Quattro 5kVA still has lights flicker when a heat gun is heating up and when the fridge turns on there is also a noticeable blip in the lights.. so not sure what the issue is.

This to me is not flickering. A single voltage droop when a high draw appliance is turned on is different to me than flickering which is rhythmic or random oscillation in light output from the bulbs.

 

[Steve_S]

You would not BELIEVE the scale of program where I have seen that attempted. Filtering of a practical scale can't clean up a bad inverter waveform or output characteristic. Solution was to fall back on the tried & true.

Falling back to Tried & True for some means accepting their choices were either a mistake or made in error. It's the 21st Century...

It appears this is a GENERALLY - ALL HF DO THIS type of Thread. SORRY But NOT all HF inverters flicker LEDS or have these issues... The GOTCHA is the Price Point where folks went out and bought the "Proverbial Clone" because it was cheaper than the Name Brand... DEJA-VU with Personal Computers in MANY ways !!! The Parallels are almost Spooky.

IF one of our member EE's with a lot of spare time, would "Tear Into" a Value HF AIO (like an EG4) and a Tier-1 HF AIO and compared the chips & quality, there would more than likely be a few Eye Openers... I don't know if anyone has compared the "guts" of the various devices and made it available.

 

[fmeili1]

Thanks a lot for all your feedback and hints!

A short interim review of your experiences and comments about this problem and possible solutions.

Where it happens

• Usually HF inverters are much more prone to LED pulsing problems compared with LF inverters, but even with LF inverters it was seen by some
• Depending on the brand and models of HF inverters the issue is more or less visible, but even with high price / high quality HF inverter models the problems are known but usually much less common compared to low-cost inverters
• There are users out there which have no issues with pulsing LED's even with low-cost HF inverters
• When it comes to LED pulsing, this is not the case with all brands LED brands, cheaper LED's usually are more problematic compared to high quality LED's

When it happens

• Some users have LED pulsing always
• Others have LED pulsing only if a specific additional load is active (e.g. laptop charger, fridge, 3D-printer, etc.)
• If the inverters have low overall load, the problem is more visible. Sometimes the pulsing went completely away if there is a high load active - sometimes with pure resistive load sometimes with inductive load also.

Possible solutions

• From what I've learned so far, it's at least problematic (if not at all impossible) to reduce the problem by adding a "typical" filter in front of the "dirty" loads, which causing the LED's to pulse (to reduce the "dirty" feedback between the HF inverter and the problematic load which causes the HF inverter to produce "dirty" output voltage). I've tried 4 different passive filters and none of them worked.
• Replace cheap LED's with better quality LED's. In case of lamps with integrated LED's, it's possible to add additional and rectifier and capacitor to reduce the pulsing
• Replace the HF inverter with an LF inverter (not an option for me)
• Add an small additional LF inverter just for the lights - this would require an additional "light" sub-panel and a lot of re-wiring.
• Add a "dummy" load in case of low inverter load (e.g. heating water or drive a chargeverter to charge the battery by itself ) to reduce inverter voltage fluctuations.

The problem really doesn't seem to be simple and easy solvable, or if at all, it's difficult to solve.

I still have hope and not yet given up to find a "sort of filter" to block my "dirty" loads (which causing the HF inverter to fluctuate the voltage). It would be much easier to have such a solution compared to modding all my pulsing LED's in the house which are not possible to replace with high quality LED's.

Idea:
Even if such a "sort of filter" would be a "small online, double conversion UPS", it would be the better solution to place just ONE of these "sort of filter" in front of the "dirty" load to totally cut the interaction between the inverter and the "dirty" load to stop the inverter from voltage fluctuations.

A "small online, double conversion UPS" is just a small off-grid inverter with a small battery and an AC charger. I've already built one for my solar black-start situation. It's funny, but I've used a small Victron Phoenix with 500VA which is in fact an LF inverter (together with a 8Ah battery).

If I'll find time over the next couple of weeks, I will try to add such an DIY online double conversion UPS in front of my 3D printer and hope that this will stop the flickering... hope dies last.

 

[Hedges]

• Add a "dummy" load in case of low inverter load (e.g. heating water or drive a chargeverter to charge the battery by itself ) to reduce inverter voltage fluctuations.

Hey, at least the seeming "free energy device" would be more efficient than a dump load which wasn't needed!

Similarly, a grid-tie inverter would be a good load for testing a DC power supply.

 

[johnnysolar]

Add a "dummy" load in case of low inverter load (e.g. heating water or drive a chargeverter to charge the battery by itself ) to reduce inverter voltage fluctuations.

Wouldn't power going from batt, to inverter and back to battery thru chargeverter make it a physical circuit, literally? Like circle all the way back, connected? Sounds potentially hazardous. Man I gotta learn more ?