Ecoflow failure

[piranha32]

@EEVblog, you have EcoFlow batteries at home, right?

[electr_peter]

A bit dramatic video. Ecoflow does not seem to be bullet proof, but then you do not put in the line of fire. What was that comment about connecting to neutral line?

Not sure of the exact issue. Failure mode, if correctly depicted, is worrying.
Their marketing about X-boost says it can handle overload for resistive loads only (X-boost basically lowers AC voltage so that max power draw is limited - something similar to CC control on DC power supplies). Maybe this implies do not try using big inductive loads.

[IanB]

Not sure of the exact issue.

The issue seems to be that if you connect a load like a brushed power tool (or vacuum cleaner?) at the same time as sensitive electronics (phone charger, maybe?) then it might be possible to damage the sensitive electronics. Under these circumstances the output from the inverter can have a short period of instability with a high frequency, high voltage component. This can interfere with capacitive dropper power supplies since the high frequency causes the capacitive dropper to have a much lower impedance and overload the downstream voltage regulator.

This might be a rare situation, but if you were the one whose equipment got damaged, I could see being pissed off about it. I think it would be reasonable for Ecoflow to provide a warning, rather than keeping quiet about it.

[piranha32]

After publication of the first video Matthias noticed a mistake, and published an updated one. But the problem of destroying sensitive electronic devices connected in parallel with inductive loads remains

[IanB]

So if inverters like this can have instability and produce high frequency on the output, couldn't this be mitigated by putting some kind of low pass output filter in place? If you are paying a lot of money for a true sinewave inverter, you have reason to expect it will stay within limits on the output frequency and voltage.

[piranha32]

I would hate to pay a boatload of money for a product which destroys my electronics and the manufacturer saying me that this is normal behavior 

[electr_peter]

In video it was mentioned that analog scope ground was connected to Ecoflow output (floating neutral) ... while Ecoflow was in bypass mode No wonder mains fuses blew and device was damaged. While not in bypass mode doing the same if fine, as inverter is floating.

High frequency oscillations stuff is dangerous, I wonder what safety margin these devices have before going crazy. Not sure if external protection device can be added to mitigate this effect.

[thm_w]

Quoting myself from the other thread:

TLDR: it can blow up capacitive dropper supplies. Sounds like other inverters have similar issues.
As he points out, with the small box, its an edge case. But with the larger battery boxes, that claim to do whole home power, it could be a concern.

So if inverters like this can have instability and produce high frequency on the output, couldn't this be mitigated by putting some kind of low pass output filter in place? If you are paying a lot of money for a true sinewave inverter, you have reason to expect it will stay within limits on the output frequency and voltage.

Somewhat, but, a 100Hz lowpass filter at 10's of amps is a massive device.
Maybe someone can try a RC snubber.

[robzy]

EcoFlow has posted an analysis of the failure:

https://medium.com/@TimDolidze/everything-you-need-to-know-about-inverter-instability-bfd1ce0471a5

Are there other techniques to manage over-current while not introducing a bunch of high-frequency?

[IanB]

If the problem is cycle-by-cycle overcurrent protection on the IGBTs, then I am inclined to wonder if the problem can be reduced by enforcing a time delay on the reset between each overcurrent event, so that the IGBT has to wait at least half a cycle time before being turned on again. Then, instead of getting a bunch of closely spaced cycles, you would get short cycles spaced out at the normal AC frequency? There would still be high frequency content in the output, but the total energy would be lower.

[thm_w]

If the problem is cycle-by-cycle overcurrent protection on the IGBTs, then I am inclined to wonder if the problem can be reduced by enforcing a time delay on the reset between each overcurrent event, so that the IGBT has to wait at least half a cycle time before being turned on again. Then, instead of getting a bunch of closely spaced cycles, you would get short cycles spaced out at the normal AC frequency? There would still be high frequency content in the output, but the total energy would be lower.

But then the device would probably not spin up, or be incredibly slow to start.

They could just shut off the output, and offer the user a "high inrush mode" and warn them about only connecting power tools. IF its an actual issue that people care about.

[uer166]

If the problem is cycle-by-cycle overcurrent protection on the IGBTs, then I am inclined to wonder if the problem can be reduced by enforcing a time delay on the reset between each overcurrent event, so that the IGBT has to wait at least half a cycle time before being turned on again. Then, instead of getting a bunch of closely spaced cycles, you would get short cycles spaced out at the normal AC frequency? There would still be high frequency content in the output, but the total energy would be lower.

But then the device would probably not spin up, or be incredibly slow to start.

They could just shut off the output, and offer the user a "high inrush mode" and warn them about only connecting power tools. IF its an actual issue that people care about.

Pretty cool, looks like they don't have much filter capacitance at the output. Ideally even in a cycle-by-cycle limit scenario, you shouldn't generate so much HF voltage at the output, but it's a massive tradeoff for circulating reactive current at low-load. I'm personally okay with this design, it's a trade-off. Capacitor droppers suck and blow up regularly on the grid as well, nothing new.

[f4eru]

filter at the output could be counterproductive, since the load impedance is a big unknown (the whole point here), you could easily get to an unstable regime, or oscillations due to the added filter, especially if there is a feedback loop acting on voltage.

The thing is, having  low impedance load on motor startup is a common situation to happen, and it should make for a more graceful failure, like limiting the output current to a sine wave.
The point is valid, this failure of output control should be adressed.

[Geoff-AU]

There's already a filter on the inverter output, it's just filtering the 24kHz (in this case) switching waveform rather than 50Hz.  A 50Hz filter would be bigger than the inverter. 

You can't "limit" the output current to a sine wave with a device that switches rapidly.  It's like having an on/off toggle switch and saying you want it to be "half on".  However you slice it, it's an approximation.  In this case you can generate a smooth sine wave but only when the circuit operates within its design parameters.  During overload, you're not within the design parameters any more so what do you do?

How it behaves when overloaded falls into 3 camps:
1. The inverter's output devices die
2. The inverter cuts off, protecting itself and potentially sensitive devices but making it unable to power inductive loads
3. The inverter tries to push through a brief overload, which allows inductive loads to start up

Option 2 is the conservative one but when someone plugs their fridge in they find they need a 3,000W inverter to run a 300W fridge.  Needing that amount of headroom just for startup surges makes the product look terrible compared to others who fly closer to the wind.  Option 3 is arguably the better one, if your capacitive dropper regulator can't handle a handful of cycles of overload then it almost deserves to die in my opinion.  It's an ugly waveform but mains can be ugly too.

[NiHaoMike]

A particularly smart inverter can dynamically reduce the output voltage and frequency to ease starting of motors.

[IanB]

Nice follow-up video here:

[langwadt]

A particularly smart inverter can dynamically reduce the output voltage and frequency to ease starting of motors.

but that could have other side effects, some devices might not like the reduced voltage or frequency

if you computers reset when you start a motor it is not much of an UPS

[uer166]

massive tradeoff for circulating reactive current at low-load. I'm personally okay with this design, it's a trade-off

I like that he did an actual test for the above and demonstrated it. Adding a X-cap across the outputs reduces the ringing substantially, but he gets 1.5A of circulating current, not fun for standby!

[NiHaoMike]

Adding a X-cap across the outputs reduces the ringing substantially, but he gets 1.5A of circulating current, not fun for standby!

So then add a triac/SSR to only switch it in when needed.