开云体育

Peter/Geo/Michael/All,

Take a peek at the regulation curve enclosed..... Simulation is from 800V ramped to 1200V on the UnRegHV input and the RegHV output is a nominal 915V +/-2V over 1000V to 1200V input and -25degC to +75degC. Yes that plot contains three temp curves for regulation! This says nothing about component unit-to-unit tolerance, just that the circuit regulation has been wrung out fine and the circuit errors ameliorated. In simulation....... not breadboard!

Pretty danged good.... Penultimate in fact..... and definitely overkill to the requirement! but fun! I call it the "Penultimate Way-Better Cascoded Shunt Regulator."

Okay, now you can look at the schematic..... I am still using generic ideal JFETs for the cascodes, but everything below the cascodes is a real model of a real transistor. The best actual devices I have seen for the cascodes are either of the two that Michael found at Digikey in this thread below. Used a Rohm 11V Zener.

rogerw

PS - I can also later post the "Way-Better Cascoded Shunt Regulator" in its current final simulated form..... also NOT final breadboarded! That one has about 913Vto 926V over same temp and voltage. Used a Rohm 10V Zener. The Vbe tempco's just behave a little different with the local feedback pairs versus plain BJT's alone. The regulation curve for this one can be improved by using 3 * 6.2V Zeners and adjusting the output voltage appropriately, but the improvement to about 7V change over input voltage instead of 13V change is probably not worth the hassle or cost......

rogerw.

Sounds great!

btw, either of those small signal HV Mosfets that Michael found look good for this.

I have no idea how it might matter to reliability operating so close to the breakdown voltage, at least these currents are miniscule.....

I learned once, a long time ago the hard way, that a breakdown voltage is not a deterministic yes/no number depending on a threshold (or a spread of threshold values), rather it is a time/voltage/temperature dependent stochastic phenomena. When you test a device, say a capacitor, in the lab at room temp for breakdown, you merely found the voltage that raised the probability of breakdown to unity in a second or two of test time....... time, voltage, and temperature. For real high reliability, I learned to stay well below that voltage.......

I am working on a "Penultimate Way-Improved" version of the Shunt Cascode Regulator........ and btw I have yielded to reality and am using a single Zener as you do. However to temperature compensate the 3Vbe drops, either a 10V Rohm Zener (slightly negative tempco of RegHV) or and 11V Rohm Zener (slightly positive tempco of RegHV) works fine. I guarantee what I am doing for "Penultimate" is really overkill..... but fun!

rogerw

On 4/3/2022 6:42 PM, peter via groups.io wrote:

Roger et al:

Spent a few minutes in my junk drawer. Found a 2SK3265( 700V part)
I checked the break down of the body diode, no problem to 1150 V.
Mod'd the cascode to just 1 Mosfet.
Wow, regulation stiffness improved. Set to 900V, raise the input 1150 and may be 15V increase

p

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.