HV divider for James

Pretty. Are you overcoating? Otherwise that Dremel'd gate-drain clearance looks a bit sparky (~30V/mil?). Some Krylon might help.

I'm not allowed to d.c. load my node, so I was using 2.2M resistors.

---------------- ////////////// The other end of the coax will be a 20 dB attenuator and a 50 ohm/500

50 ohms is classy. Not feasible for me. Even 1000:1 only lets me use 50k for my divider, which ain't allowed. My final divider is a compromise: simple and cheap, but still reasonably accurate.

Cheers, James Arthur (downloading a new copy of OrCAD -- 6GB!!!!!)

What's that inductor doing on the drain? And is that a cap to ground (a HV one?)

--
 Thanks, 
    - Win

If you can load things, the Caddock MG series of resistors are mega-wideband

and can stand lots of spike voltage. They are essentially all that's in the 6 GHz HP scope probe.

You can get a pretty good 100:1 scope probe from Amazon cheap:

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

That's just a hacked breadboard, mostly to characterize the Cree part model... which is not entirely to be trusted.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I've been making great progress with them using analytical design calculations, rather than SPICE modeling.

--
 Thanks, 
    - Win

But I don't know anything about the internals of the Cree parts, so I measure things. I already learned that the body diode isn't modeled very well.

When I design my real product, I want to get it right the first time, so I simulate, or breadboard if there is any doubt about the part models. Things like the Miller effect of nonlinear d-g capacitances would be a serious pain to approach analytically.

I do very little classic EE math these days. My style has evolved towards instinct and simulation. I even simulate resistive voltage dividers, which is more fun than punching keys on a calculator, especially when trying to make permutations of in-stock values work.

What's especially bad is RF parts that have s-parameters, but no decent DC, capacitance, or nonlinear characterization. People just don't specify Rds-on for RF fets, or pulse behavior of MMICS.

Breadboarding, dremeling and soldering, is fun now and then.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Den tirsdag den 23. januar 2018 kl. 16.29.34 UTC+1 skrev John Larkin:

this a bit more specs