I have a leak valve I have been trying to use with my new chamber setup, and was really struggling to get a uniform flow.
...Then... two breakthroughs.
Firstly, I bought, initially as a spare, another active magnetron gauge. After testing it I didn't really have an immediate use for it so stuck it atop the intake port of the turbo.
Immediately I realised it was giving me a very accurate [relative, perhaps, rather than absolute] reading of flow rates in the system. At its full rated60l/s pumping speed and 1e-5 showing, for example, I then know I have 0.036 sccm flow through the system.
This gave me an immediate insight into the internal workings of the leak valve, and showed it must have a series of volumes, each of which was buffering one way then the other and making control hell.
I had been looking for capillary tubes for a while but, this being UK, I didn't find anything readily available. I found a supplier of 'blunt dispensing needles' [non-medical] and got some a few days back. The smallest, 0.21mm ID, still flowed too much gas but that was rectified by passing a 0.18mm wire down it.
This needle is inserted through two pieces of nitrile rubber I cut out and sandwiched between NW10 flanges.
The 0.21 ID needle (1" long) and the 0.18mm wire gave a flow rate of 0.5sccm from 1 bar to 0. Because it is just one single flow restriction and drops a full bar pressure, the vacuum pressure can vary 'considerably' but the flow rate doesn't - e.g. it could be 0.1 micron or 20 microns, the percentage change from one full bar is negligible, but the flow rate is roughly the same.
To alter the flow rate, I change the pressure on the needle 'intake'. To set the system up, I merely dial in the pressure I want in the chamber on the chamber-turbo vent valve, set the pressure on the needle (I have mapped this out tonight and 1 bar gives 0.5 sccm, while 3 bar gives 5 sccm - not sure why it isn't linear, but that's what the results said. The pressure gauge on my regulator is just a round, dry needle type on a scale and isn't calibrated, so that is the most likely source of error, if it should be linear). There is then a small change on the chamber pressure once the pressure is set and settles after a moment, and that chamber pressure can then be given a final fine adjusted. Or otherwise set the flow rate first, &c...
The whole thing is then dead stable. It will just sit at any pressure I choose, at any flow rate between 0.5 sccm and 5 sccm, with a 0.05 sccm resolution. I can now see why Carl W. has gone for a flow-orifice, as a single, big pressure drop is so much more controllable than a series of 'dead spaces' between valves or restrictions.
[notes; outgassing from the turbo-chamber line was <0.01sccm, and the chamber outgassing rate before and after these measurements was ~0.06sccm, so these are 'error margins' on the above figures.]