I adapted this schematic from a UPS no-break external bypass board so I’m fairly certain it will work with a Multiplus, but I thought I’d double check with the community and Victron staff before actually implementing this.
The components:
To the left is the power from the main Grid breaker board, going to breaker F4 that is wired to the AC IN
In the center is the Multiplus, in a 3 phase system there will be 3 of those
Below is breaker with 3x NO and 1x NC contacts
Each of the NO contacts is wired to a Multi input that is configured to put the Multi in Passthrough if the input is activated
The NC contact is wired to a Disconnect Shunt attached to breaker F2 so it can not be closed if the Multi is not in Passthrough mode
At the top is breaker F2 that connects the AC IN to the AC OUT if closed
At the right is breaker F3 that connects the AC OUT to the load breaker board.
Operation to enter bypass:
Close F1 to enter Passthrough mode
Close F2 to enable the external bypass
This should be safe since in Passthrough mode, the Multi’s AC IN and AC OUT should be at the exact same voltage
Open F3 to disconnect the loads from AC OUT, power now flows through the F2 external bypass
Open F4 to disconnect AC IN from the grid
The Multi can now be physically disconnected from the installation or reprogrammed
Operation to exit bypass:
Close breaker F4 to reconnect the AC IN to the grid
Close breaker F3 to reconnect AC OUT to the load
Since the Multi is in bypass, there should be no voltage difference between AC IN and AC OUT so this is a safe operation
Open breaker F1 to bring the Multi out of Passthrough mode
The system is now back in normal operation.
When entering or exiting bypass mode, any manipulation in the wrong sequence will either not be possible (due to the shunt on F2) or result in power loss to the loads.
In the UPS bypass board I have here, breakers are installed (left to right) F1, F2, F3, F4 with F1 and F2 installed upside down.
This way, in normal operation, all breaker handles are pointing up and pointing down in bypass mode.
I’m not sure I completely understood what you’re trying to achieve, or how, but what happens when the multi does not go into passthru mode, for example because AC In is off? Then, AC out is probably out of phase with the grid and a lot of sparks will happen.
I’m trying to find a way to bypass the Multi, without losing power to the critical loads behind it.
You have a valid point when saying “What if the Multi doesn’t go in bypass”.
This should of course be checked before continuing the bypass sequence.
And if AC IN is off (and you’re running on batteries), it wouldn’t make much sense to try and take the Multi out of service
In case there not that high put a ups in between multi output and critical loads and switch the whole system with a normal bypass switch on ac in and out
Far less risks of doing it wrong and creating magic smoke…
That’s certainly an option but kind of defeats the purpose of having a Multiplus to power critical loads, if a UPS would do the exact same thing.
I’m replacing a big UPS with a more intelligent Multi install to have less components, not more
Not very often. Probably only when there’s a new firmware for the Multi.
Or when a Multi breaks down
In my assessment, the risk of using my bypass is limited but, as with any setup that’s a tad more complex than a toaster, prudence is always a good thing.
I’m considering an extra contact to the Cerbo, to force ESS off in passthrough mode as well, so the Multi would just be 3 closed relays in series between grid and load.
update only when absolutly necassary, or if you really need new functionality, you dont have to update with every new version ( victrons official recommandation)
Most devices will handle the brief interruption of power when switching is break before make.
The Multi is not a true UPS in that the output voltage and frequency are the same as the input when it is passing through grid power to the output. That means there will be no compensation for a brown-out (until the output voltage drops to the point the Multi rejects it).
If your loads are that critical then they deserve to run on an on-line UPS (the kind that runs the charger and inverter at all times so there can never be even a brief interruption of power.
If the interruption is short enough they’ll survive, yes.
But you need very fast relays for that, and those get tricky with bigger loads.
And yes, very good point that a Multi behaves like an “off line” UPS in that in normal mode it passes AC straight through.
To detect a dirty grid I’ll probably install some extra grid monitoring devices anyway.
I have an 8kVA online UPS (and in storage also a 30kVA and a 40kVA with some vague error, but those don’t matter now) that I could use but those have an idle load of +400W and that tends to tick on the electricity consumption in the long run.
But if the grid is kind of dirty over here, the 8kVA will be put in service again.
Which would mean extra breaker boards and I already have too many of those.
There’s plenty options but I’d rather keep it simple.
And if my proposed bypass board would actually be a good way to implement this, it could be beneficial for others as well.
I’m going to install a bypass setup to my Multis anyway, the question will just be if it will be this “no-break” bypass or a plain I-O-II transfer switch that cuts loads halfway.
FWIW, I have a 1-0-2 bypass switch across my Multis and found that if I turn the knob quickly (i.e. one decisive motion, not stopping on the “0”), not a single one of the way too many Raspberry Pis that populate my house crashes. I haven’t tried it with a desktop PC or server, tho.
The bypass switch is a Kraus & Naimer KA40B.T904.VE2.
Late to the party, but just thought to mention anyway that there are products that already do this. Not that i am discouraging you from making your own.
Those are valid options, but with a few drawbacks:
They are all single phase, no option for three phase
They are all “break before make” contacts with, granted, very short interruptions
In the “break before make” near-hitless transfer section, there are many options.
My hope was to find a validated solution to achieve a completely hitless transfer
Hmm, your drawing shows a transfer on only one phase? The filafax is fast and designed to switch computers.
Possibly then update the drawing.
Break and make is mush safer.
For ease of drawing, I only drew the Multi from one out of three phases.
Readers’ discretion to imagine the other two
Except a minor detail*, it’s a 1:1 copy of a no-break bypass board used in UPS systems, so I’m 99% confident that it will work with Victron.
Was hoping a Victron tech would pick this up but not sure who to tag. @mpvader ?
* : a UPS is typically switched to ‘internal bypass’ through the front panel, then to ‘external bypass’ using the above switchbox.
To achieve a comparable state from a Multiplus I’ve added F1 in my drawing that puts the Multi in Passthrough mode. All the rest is pretty much identical.
Some UPS bypass boards even omit the shunt preventing F2 to be closed too soon but rely on internal UPS electronics to just ‘deal’ with a dirty transfer.
Yes, I admit it’s more complicated and sequence-depending than a fast break-before-make transfer switch and should only be manipulated by skilled and trained personnel.
Sometimes that’s just what it takes.
