Radio WC6W wrote:
> The shunt regulator consists of a venerable uA723 regulator IC driving
>a FET. The FET is connected to the screen through a series R which drops
>most of the voltage and therefore dissipates most of the heat. The '723
>is powered from a +15V supply which is independent of the plate supply.
>
> In the event that the plate supply is incongruously enabled without
>the +15V supply or the +15V supply fails, I have a 1M resistor connected
>between the drain-gate of the FET. There is an opto-isolator, with its
>input diode connected through a 10V zener and a small R to the +15V, that
>couples the 723 output to the gate of the FET . So, in the absence of
>+15V the opto' is off, the FET is turned on hard by the 1M R and the
>screen voltage is clamped to a low value.
>
> There is addtional positive and negative screen current sensing beyond
>the regulator which provides overload fault protection.
>
> Long, long ago (about 20 years), I designed a series-shunt regulator
>for a friend who used it with a pair of 4CX250F's. It used the same
>shunt regulator as detailed above. The series pass was a FET operated as
>a current source. He powered this regulator from a stiff 400V supply so,
>a small R inserted in series with the shunt leg reduced the drive to the
>gate of the series FET and therefore the current sourced, as the shunt
>regulator current increased to limit the total dissipation. I don't
>recall from what source what he powered the '723 in this implementation.
>Incidently, this regulator replaced a string of three 100V/50W zeners...
>I forget how much idle current he was running in this previous
>arrangement.
>
> I have recently been playing with a screen regulator design, for a pair
>of 4CX600's, using a TL431 in a cascode arrangement with a FET for the
>shunt leg plus the series input as noted in the paragraph just above. It
>is also foldback limited at about 10mA into a dead short. This will be
>sourced from a 400V switching supply and requires no other power source.
>There is an SCR crowbar on the output, in case this scheme departs from
>the accepted laws of semiconductor physics.
>
> Note in context the thread this continues, the two designs just
>described require parts in the series leg that can accomodate the full
>source voltage, in the event the output is overloaded, shorted or
>crowbar'ed.
>
> Incidently, the test I use to check stability (and regulation) of
>these critters is to feed a square wave (at a KHz or so from a generator)
>back into the regulator. I typically run 40V pp through a suitable
>(600V) coupling cap and a 600 ohm R. This is +/- 30 mA test.
I usually trim my quotes, but couldn't leave out any part of that good
stuff.
It shows that an idea that I'd though un-promising can be made to work
well and reliably, but that's because Marv has the engineering know-how
and has thought through all the problems.
He also identified the potential failure modes and took precautions
against them. He also knew he couldn't think of every possible failure
mode ("in case this scheme departs from the accepted laws of
semiconductor physics") so he defended against that too. That's good
engineering.
However, it does show that the more you think about screen supplies, the
more potential failure modes you see, and the more complex the design
becomes to defend against them. Marv's is complex; Acom's is complex;
Alpha's is complex; and G4JZQ's and mine is complex. We all did it
differently, but we all agree that a handful of cheap zeners won't cut
it.
>73 and probably good night by you,
You guessed that right, and a good morning to you, Marv. Thanks for
taking the time to line all that out.
--
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.com/g3sek
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|