AC300 to Transfer Switch

So I was showing a friend of mine who is an Electrical Engineer what was happening, and he has a pretty good theory what the problem is…

When the power is restored to the AC300 after an outage, it switches to “bypass”, sending power directly to AC outputs. This is probably done with a relay or something similar. The process of sending power anew through there will usually cause a voltage spike (just like if you watch an outlet spark when you plug something in). There has to be some kind of surge suppressor to prevent dangerous levels of spikes. It is probably tuned for a typical range of voltage spikes that you would expect for devices plugged directly into the AC300.

However, a transfer switch is capable of generating a much larger voltage spike when connected, even when no current is flowing, because it has much higher capacity wires running directly to the panel. He suspects their surge suppressor is not tuned for these larger voltages. This would obviously be consistent with the error we get (high bus voltage), and also the fact it sometimes works (sometimes spikes are small).

I asked him if he thought the suppressor might be software adjustable. His answer was “something along those lines would be 10x or more the cost of a simple suppressor, so unless there was some other need for it, highly unlikely”.

The arc you see when you plug in a plug tot he wall is called arcing. The voltage should be constant (or even dropping due to inrush current) though so I’m very doubtful your friend is correct (My dad is an electrician and my formal degree is computer science with some electrical engineering work).

You have to remember as well that this works until you simulate a power failure (I.e. the switch from off to grid power works fine, the switch from grid to UPS power works fine when the power failure occurs, it just fails when the power is restored.

I’m done running tests after one of my AC300’s blowing up due to this but an interesting test is probably running the AC300’s in split phase with a generator that is not ground/neutral bonded to the AC300’s and then simulate the power failure with that method.

Without really knowing the root cause of the problem I do believe bluetti could solve this. Clearly the case works fine where we switch the unit on when grid is on and clearly the swap to battery power works fine as well. It seems that by increasing the software delay between power restorations, detecting it occurred that they could delay the relay switch longer. First swap the inverter off and return this to being the base case of powering on when the grid is on. Yes, you would see the power blip and some devices would probably reboot but it’s less bad than what is occurring now.

Bluetti RMA’ed my ac300 back to Las Vegas for analysis. what ever that means. Hopefully there are internal logs and stats that are recorded for them to understand the problem. I made another video with different tests and outcomes and asked them to view the video carefully. I am hopeful it can be resolved by tuning the sensitivity of their software/firmware. The ac300 was delivered by UPS and received by bluetti yesterday.

I also made a request for bluetti engineering to improve the app allowing the powering on and off of the ac300 through the app. Also to enable the viewing and clearing of alarms through the app. The request was passed on for app improvement. This will hopefully at least enable remotely reactivating the ac inverter.

I have been connecting my transfer switch for actual use periodic exercise of the ac300 and the battery with really no issues other than the ac inverter turning off during these tests.

Voltage spikes from relays is a well-known issue in electronics. Since the issue only happens when the relay is activated with an AC-load and the transfer switch attached, it seems a reasonable theory.

A software “work-around” might be the most practical solution. However if the theory is correct, there is potential to cause electrical damage each time it happens. I would suggest not testing the relay any more – it might be why some of you have damaged units.

Voltage spikes with relays occur when relays are disengaged and the relay then acts as an inductor sending voltage to the wrong place. In the case where people are running into issues it’s the opposite, the relay should have already been disengaged and is reengaging to allow grid power to power the transfer switch. I don’t think this is would be a reverse flow problem.

It’s been almost six weeks since this issue was first reported and no word from @BLUETTI.

@BLUETTI, can you tell us if you have been able to reproduce the problem? If so, have you figured out a cause? Is it something that may be fixed with a firmware update?

If none of these are the case, can you provide a software mitigation? Three ideas:

  • Have an option to automatically restart the AC Output once, after the alert triggers. This would cause a short-power outage but otherwise alleviate the issue. (Restarting only once would prevent endless failures if the error is caused by some other problem.)

  • Allow the app to notify and clear alerts so we could manually restart the AC Output remotely.

  • Add Custom Online-UPS to the firmware, so we could avoid the bypass circuit but still take advantage of the other UPS settings.

Using a transfer switch is really an important feature for many of us, especially when we already have one installed.

I have not read this entire thread. But I can confirm that the AC300 should not be grid-tied to typical transfer switch.

I have not sign any NDA, but have worked with some of the engineers over the past few weeks. The input and outputs current bearing wires (Hot and Neutral) need to be isolated completely. The ground should be connected (I have not verified this, but it is code).

An isolation transformer on the input or using external switching mode PSU that outputs DC directly to AC300, or using something like the T500 directly into the B300 modules should work. I am waiting for my replacement AC300 which should show up later today before I can firm ways to isolate input charging methods.


Thank you Doug, that is really interesting. Nice to hear Bluetti is aware of and working on it too.

Any idea if Bluetti’s Fusion Box has some sort of isolation?

I did hook up a T500 into both the AC300 and the B300 and it does work successfully…

…but, the problem is none of the UPS functions works with the T500, which means you are stuck constantly charging and discharging the battery at ~100%. From what I’ve read, that is about the worst thing you can do to the battery in terms of longevity.

If @BLUETTI could modify the firmware to offer full UPS functionality when using a T500, I think that would be good enough for me.

My thoughts for isolation is to get a hold of a DC PSU that meets the required demand (somewhere between 1000 and 2000 watts) and then using the DC input into AC300 and that should still maintain a SOC of ~70% or whatever you desire.

The DC inputs can be selected between PV and “other”. I have not used the DC input in PV or other mode yet. Suspect they could probably offer firmware updates to expand how “other” mode can operate given certain edge cases.

How did you connect the T500 directly to the AC300? Did you make some adapter that goes from the barrel to MC4 ? I’ll have to check, but don’t remember having something like that on hand.


Additionally AC isolation transformers can be costly for high power:

(this should work with the AC300 input in factory 15A mode)

However being able to lower the charge rate for smaller transformers will be required, but it is “locked” until one gets permission from whoever at Bluetti .


The T500 came with a DC input adapter that plugs directly into the AC300 through one of the DC inputs. In addition, you can plug the T500 directly into the B300 without an adapter.

If you’re looking for a cheaper alternative for experimentation, in an earlier post in this thread I mentioned buying a 120V to cigarette lighter adapter off Amazon for $25. I connected it to the AC300 car adapter and it pumped in 96watts.

I looked at those AC isolation transformers – pretty expensive for what is kind of a hack. I was looking into the Echoflow Delta Pro because I heard they work with transfer switches, when I ran across this video. It sounds like the Delta uses an unbounded neutral when running the inverter, but when it switches to bypass mode, it essentially becomes bonded. I suspect the same thing is true with the AC300 – the only difference being the AC300’s firmware catches the fault, and Delta doesn’t (willfully or not).

They make transfer switches that switch the neutral wire but they are hard to find. I think I see why Bluetti and Echoflow sell sub panels for these systems and not transfer switches.

I ordered one of these – it allows you to switch the neutral on one circuit. I am going to install it and see if that works.

I just went back and “Explored” the T500 offerings. The T500’s didn’t offer a AC300/AC500 pigtail (DC7909 to MC4) when I got my T500. Will be sourcing parts to make that pigtail. My T500’s only came with out DC7909 and I have pigtails that go to XT90.(Note purchased them for my AC200 , NON-P crowdsourced model and the B230 I got for it a while afterwards. Original AC200 came with the “T400” 400 watt adapters that have no ventilation and get really hot.

Making XT90 → MC4 is what I am planning so I can mix and match.


Doug, I would just contact Bluetti support – I’m sure they’ll send you the adapter. When I ordered my T500, the adapter was mailed separately from a different location, so they have them available individually.

Well, I hooked-up the EZ Generator Transfer Switch, tested it out, and… it works. When you restore power to the AC300 with it attached, no errors, AC Output stays on.

The difference between this switch and most transfer switches is that it switches the neutral, so the circuit is totally isolated. That seems to make the difference. Most transfer switches don’t do this, probably because it greatly increases the wires going into the panel: four instead of two for each circuit.

This particular switch is for a single circuit and costs $100, so I’m not sure you would want to pull out a 10-circuit Transfer switch and replace it with 10 of these. I’m trying to find such a transfer switch, but it seems like a sub-panel may have to be used instead for practical reasons.

I finally got my T500 and attempted to charge B300 directly with it (so I don’t lose out on a PV input)… but it does not work. Seems the B300 will not charge via DC directly when it’s hooked up to the AC300. This means my plan of unhooking the AC300 from the grid and using the transfer switch will not work. It sucks, but I guess until @BLUETTI listens to all our requests about enabling the clearing of the error code via mobile app, I’ll just have to use the unit w/out my transfer switch.

Brian-- it definitely works, I have done it. The display on the AC300 does not acknowledge the charging input, but the battery charges. If you watch the five green LED’s on the battery, they will indicate charging. You may need to run down the battery a bit to see it.

It’s not a great solution though because you can’t control the charging with the UPS limiting controls… so it will always charge the battery to full when attached. Some people have suggested using a timer to turn the T500 on and off at intervals, but then you are sort of shooting in the dark as to the charge levels…

Interesting, I tried it yesterday running for two hours (battery was at 29%) and the B300 would not charge. In fact, it actually dropped in % according to the AC300. I had the two PV inputs coming in from my roof panels, then ran the T500 directly into the B300. It was night so had 0 coming from panels, but B300 didn’t show charging at all.

That is interesting… I don’t have any PV panels connected yet, maybe that has something to do with it… Does the LED indicator on the T500 stay green or does it ever turn red indicating charging?

It stayed green (despite the T500’s fan sounding like it was taking off). I only was able to get the T500 to actually indicate charge when I went T500->one of the PV inputs->AC300

I know what you mean about the fan noise – that was one thing that turned me off to using it!

I would try disconnecting everything and reconnecting one at a time to see if charging works: e.g. see if you can charge the B300 when it’s disconnected entirely, then connected to the AC300 when no solar, etc. Like I said, it definitely charged for me. Not sure why your’s would be different. I have only one B300, don’t know if you have more than one…