Baxendall Class-D oscillator with a big source inductor

That's just Sloman's interpretation of what he thinks he remembers. But at least he's doing some electronics, at last. Well, simulated.

John

hether your

you talked

which means

above the

t you are

the

The

ing

twice

line

eed

ope it

e t g

My Ph.D. thesis includes a couple of simulations of the chemical reaction I was studying - I used a non-linear least squares minimisation routine to curve fit the simulations to my experimental data to allow me to pull out best fit rate constants, so I do know how

- and why - simulations fail.

In this particular case, unrealistic starting conditions lead to an irretrievable build-up of rounding errors.

If John Larkin knew anything about the nuts and bolts of simulation, he'd be less likely to believe that a perfectly trivial failure of a Spice model was persuasive evidence that climatologist's computer models could be ignored, but he suffers - like Eeyore - from delusions of copetence in areas that he really doesn't know anything about.

-- Bill Sloman, Nijmegen

t g

I've done enough work with inductors that I automatically trhow in a parallel capacitance to get the self-resonant frequency into a sensible ball park.

The Toko 8RB 3.3mH part has a self-resonant frequency of 700kHz which equates to a parallel capacitance of 16pF. The series resistance is

13R. The sort of inductors I wound on R6 and RM8 ferrite cores would have had higher capacitances and lower resistances at that inductance.

Correcting the 0.1F capacitors to 0.1uF did the trick. That is one of the nice things about simulation - it does help to clean out the mindless errors - the the circuit designer's equivalents of typos - that plague every real design.

-- Bill Sloman, Nijmegen

whether your

you talked

, which means

s above the

nt you are

the

=A0The

sing

d twice

c line

feed

hope it

he

nt

ng

s n I

You must be one of those picky people who use the proper Greek alphabet "u" for "micro".

-- Bill Sloman, Nijmegen

-D

in

wn

eg.

ut

cuit.

a

any

You two really are don't know much about anything outside your little ponds. Note the following sentence from my web-site - Rich would have read it if he ever bothered reading anything.

"The circuit is probably best known from Jim Williams=92 series of application notes for Linear Technology, on high frequency inverters for driving cold cathode back-lights used in laptop computers (application notes AN45, AN49, AN51, AN55, AN61, AN65)."

Those application notes were extraordinarily popular, so it seems likely that quite a few other people think they remember the same circuit.

-- Bill Sloman, Nijmegen

s-D

u

r in

own

ueg.

out

rcuit.

a

any

nant%20Class...

If you do it right, the sine wave coming out of the transformer is pretty good - around 1% third harmonic content and very little else - until the load degrades the Q of the tank circuit below about 5 - and the efficiency is usually better than 90%

There are ways of getting the distortion down to the 0.1% level - my favourite degraded the efficiency down to around 50% which didn't matter in my application, while distortion did.

-- Bill Sloman, Nijmegen

On Dec 30, 5:00=A0am, snipped-for-privacy@ieee.org wrote: [... saving bits for future generations ...]

No, somehow my copy of LTSpice got set to change uF into (mu)F. I use uF.

In the package I use for doing my real schematics I always use the caps lock to make everything upper case. This is because the lower case letters don't make good looking prints.

e

ent

ing

cs

rn

n

=A0I

The output side diodes are way the wrong sort of thing to be using for this task but they were ones that LTSpice had.

I think I have the K1 value a little large for what would be between the primary and secondary and a little small for the primary to primary. If we assume that it is a transformer of the "any idiot can make it" class, the secondary would be in a different winding bay on the former. It would be easy to call out a bifilar primary. It is only going to be about 10 turns. The inter-winding capacitance on the secondary would be moderate in value. High voltage windings tend to have lower values.

I didn't put any output side filtering because that makes the spice run take a lot longer and this isn't a real circuit.

Q6 needs a resistor or two on it. Bad things could happen if it turns on all the way.

ass-D

you

tor in

down

nd

he

squeg.

thout

circuit.

as a

nd any

sonant%20Class...

That wouldn't be into the nasty nonlinear voltage multiplier circuit.

At 50% you could use a hifi amp to drive the transformer. I think you could get that low with a multiple tuned transformer section. To keep the capacitively coupled stuff from being a problem, you could use a linked transformer pair.

:

class-D

f you

uctor in

ng down

ound

the

o squeg.

without

e circuit.

has a

find any

resonant%20Class...

.

on

I was using the oscillator to excite an LVDT in a precision weighing system that monitored the weight of a single crystal of GaAs as it was pulled from the melt in a Metals Research Czochralski crystal puller.

The original circuit that I was replacing had used a couple of parts that had gone obsolete, so I had the fun of designing the replacement into the few cubic inches of space directly above the LVDT (and below the slip ring that carried the electrical signals in and out of the rotating pulling head) that had accomodated the original electronics. The replacement circuit was more complicated than the circuit it replaced, but there really wasn't room for a hifi amplifier,

-- Bill Sloman, Nijmegen