Just in case this is useful...
I had the same problem once, and it turned out to be a meniscus had formed between the bell and the breather so the breather was sucking water up even though it was 1 cm above the water. (it was a very narrow breather)
And...
if there are not enough holes in the BOTTOM of the media screen it can reduce the flow when the tide is almost out, to the point where the siphon hickups then starts again but with just a trickle. When that happens out-flow can match in-flow and it can stay that way forever. As the siphon almost stops, the local tide within the media screen can rise up from the flow still entering through the screen, and the small amount falling back out of the bell and restart the siphon again.
The rate at which your siphon flows still depends on the head. ie it depends on how high the grow bed's water is above the outlet at the bottom of the standpipe. In my small test system that's only 5 cm or so, so there is very little pressure.
The height of water at the end of the cycle ie when the tide is almost right out is lower, even though the water actually goes up to a set height in the siphon, its the head that sets the exit flow rate. So as the tide goes out, the flow reduces. If the reduced flow matches the input of the pump then the siphon reaches an equilibrium and the level inside the media screen will settle down to a constant.
Look inside the media screen and see if you can see water squirting through the holes. If you can it might mean there is a greater water level outside (in the GB) than inside the screen. That might mean there are not enough holes in the screen. To test it just lift it up a tiny amount to let some water flow under it. Don't over do it and let in any media, or better still put a tea strainer under your standpipe when you do it to catch any media and protect your pump.
If the flow out of the standpipe is reduced in any way, especially at the low pressure (tide is almost out) stage, you can get a situation where equilibrium is reached. Make sure there is no way air can be trapped in your system after the bottom of your standpipes. I find the best way to be certain of this is to have your standpipes dump into fresh air rather than a sealed drain pipe. When the siphon is stopping, air gets sucked into it. if the first tiny bit of air can collect in your pipes, and it comes in at just the right rate, it can reduce the flow just enough to make the system settle into a situation where the flow in equals the flow out. This is easier than it might seem, because if the air is coming in gradually enough, the incoming air makes the siphon suck less, reducing the amount of air being sucked in. If the air can be trapped, the system self regulates the amount of air it takes in, to the point where equilibrium is reached.
For a siphon to work it needs to have a flow in and a flow out that are within the siphons ability to start and stop. Each siphon, depending on its size and shape will have different maximum and minimum flow rates that it can operate within. If you have 2 almost identical siphons, and one is working and the other isn't, it could just be the one that is working is only just working. ie its working right at the extreme of that bracket that allows success. The tiniest variation in flow to the second one, or the flow through the media screen, or a bit of air trapped in the system could be just enough to take it out of that successful area.
Depending on your lengths of pipes and height of head, its possible to have a siphon trigger properly even though there is some air in the drain. But if it's sitting right at the extreme of the allowable flow when it comes to the stopping stage, its possible for that tiny bit of air to make it fail.
Even though its often helpful to use raised sections in your drain pipes to trap water etc, if you make them fall straight down into fresh air, then control the flow into your garden bed, you will find it very easy to get siphons to work reliably.
You might need to adjust the flow to the growbed to get a siphon to work within its maximum and minimum required flow rates.
I should stop now
