Following on from post #37, where I attempted to get close to the original design specs, I will now describe what we actually produced.
Quite a bit of time was spent playing with high voltage solid-state regulators, and discrete regulators, because I was determined to try and produce a supply that performed as well as the original, but was a good bit lighter - and smaller! A few of the semiconductor regulator solutions were knocked up and tested on the bench. The trouble with them is that they produced a very stable voltage (tick!) with very low ripple (tick!) - BUT they were extremely easy to destroy, sometimes with utters stupidity, like shorting the HT line to ground, and sometimes it just died without any apparent help.
Given that the supply was being made for a radio user, not necessarily very technical, I decided in the end that semiconductor cleverness was probably a bad idea. So I returned to the ultra-simple solution of a bridge rectifier, followed by a single L-C filter. Same principle as the original, but with high value C available these days (they weren't in 1940), and also the ability of a colleague to wind any size choke I wanted, it sounded promising.
After a few simulations with PSUD2, I came up with this schematic:
Simulation results as follows. First up, the start-up voltage with the 8.8k load:
Then the same, steady state after a couple of seconds:
Then with a load of around 75mA, 8.8k ohms, representing the R1155 on normal reception mode with no DF use:
Then finally at maximum (spec'd) current of 110mA:
A summary of the results under these three conditions are:
Results - Condition 1:
Load value = 8.8kΩ
Load current = 25mA
HT voltage = 220V
Ripple (Vp-p) =0.35
Results - Condition 2:
Load value = 2.8kΩ
Load current = 76mA
HT voltage = 212V
Ripple (Vp-p) =0.35
Results - Condition 3:
Load value = 1.85kΩ
Load current = 112mA
HT voltage = 207V
Ripple (Vp-p) =0.35
So the ripple is now a lot lower than the original psu managed. The voltage stability is also better dropping only 5V (compared to 11V) with the load ranging over 75 to 110mA.
And the actual circuit is about as simple as its possible to make a linear psu. The only drawback is the cost of the choke - I decided the increased weight compared to a SS regulator was not that important, given it was "lost" when put in with the T1154 transformer etc.
Chokes are available - often in junk boxes, sometimes at rallies, and of course you can do what we did - wind a new one from scratch - or get the helpful people on this forum to make one for you.
Richard
P.S. I forgot to add that the simulations showed the 1N4007 PIV rating of 1000V being exceeded in the first 0.5 seconds after switch on. That doesn't matter for a simulation - but it's not a good idea to go with that in real life! I think we actually used some 1500V diodes in the end - BY228s as I recall.