ed , yes , just added the top cylinder to the existing feed as a simple way to expand solar pv heated water storage

ringi, what djh said , i just stole that drawing to show the pipe and vent lay out ignore the thermal bit.

So I just checked whilst running the kids bath, the vent on the top cylinder is the only one sucking .
I turned the tap off. let the levels settle then blocked that vent and ran the hw bath tap.
this solved the problem.
I presume as Ed initially suggested the water is leaving the lower cylinder faster than it can leave the top cylinder and is therefore sucking in air also.
the top cylinder is stainless steel (main presssure though used as open vented as mentioned ) so I presume this cylinder is less likely to collapses.
Wouldn't blocking the top cylinder vent with a one way flapper valve just result in water being sucked in quicker from the header tank rather than cylinder collapse ?
If I remove the top cylinder vent completely wouldn't it just vent and expanded through the lower cylinder vent ?

Steamy , yes , I am suppose to know what i'm doing , but I admit when it comes to the finer points of hydrodynamic I'm a bit of a chancer

The saying "Measure twice and cut once" is especially valuable in what I imagine is an awkward working space, so if you haven't already started, the following info could be useful:

The present height between the header tank water level and the tee on the upper cyl outlet.
The length of the 22mm pipe run from the header to the cyl bottom flange, plus the bit from the cyl top flange to the vent pipe tee.
Does air get sucked in at almost any flow rate, or can you measure/estimate at what fraction of the full flow rate air starts to get pulled in?
Can the depth of water in the header tank be increased ie by tweaking the float valve? Every little helps.

So far as I can tell, 10" or so head should be sufficient, but that is only for a steady flow. Momentary fluctuations due to taps turning on or off quickly are not allowed for!

As you say, increasing the pipe size to 28mm will have limited effect with the cyl flanges at 22mm but could still be worth doing. If there's only 2 elbows in the run, the head reduction should be about 25%, more if there are more upgradeable fittings.

Those figures give a head loss of 235mm for a flow rate of 0.3l/sec, amazingly close to what you have, so maybe not much increase in the head needed to be on the safe side if your flow rate is similar. The head loss increases with the square of the flow rate, so if you can restrict it below the tee just a bit eg. with a gate valve that might be enough. If not and you don't want to do the downward vent pipe thing, then my other suggestions, the easiest first...?