A
Aubrey McIntosh
- Jan 1, 1970
- 0
Ahhh, the sound of n-MOS singing.
I have been looking at the n-mos switch, M1, in a inductive boost
regulator to light some LEDs. I thought that faster switching speeds
there would result in improvements in the circuit.
See http://www.jump.net/~vima/led/Impulse.pdf
I had read in genome's postings about special drivers for the gate
(Cooper?), which is where I got the idea of using a special driver for
the gate. I came up with my circuit just doodling in SwitcherCAD to
try to practice circuit comprehension.
One view of this is that the total gate energy, (gate charge x switch
voltage) on M1, plus the loss in R1 should equal the total change in
inductor energy in L2. Higher currents and smaller inductors keep the
energy the same but speed things up. R1 is a tax.
At the heart of the circuit, I used Q1 to control the current in an
kick-switch comprised of D3, R1, and L2. When Q1 conducts, current
flows in L2. As the base voltage rises on Q1 it commands "off" thence
the collector switches off slowly. L1 is starved, and the M1 gate
voltage drops, the gate charge flows quickly. The appropriate
simulation traces make nice art. Component values on R1, L2 (should)
limit the M1 gate voltage to -0.6V and D3 provides safety backup on
this limit.
Is "my" circuit published, known, or analyzed? Is it fundamentally
flawed somehow? Is it wondrously novel and my ticket to fame and
fortune?
SwitcherCAD III says that this is a wonderful idea, producing output
voltages following L1 upward of 150V, but hungry for current in L2.
I'll somehow deal with that later, with smaller currents, bigger
inductors and the correct capacitor in the correct place.
There is, however, singing. I looked at the FFT. I am feeding a
square wave in on V2 to simulate the rest of the control circuit, so
there are odd harmonics climbing all up the spectrum. I did an FFT of
the L1 output voltage from 49.01ms to 99.01ms using 256K points. The
harmonic at 220 KHz is 18 db above the harmonic at 230KHz.
Ignoring the FFT, I had assumed that a small signal well below the
linear region on the M1 gate would not matter.
The signal there is 200 mV p-p at 0 DC. Conduction starts at about
1.9V.
I haven't decided what to do next. One thing that comes to mind is to
do a small signal ac analysis on V2 with the other circuit parameters
set as they are at oscillation, and follow the signal through the
circuit. There may be a better injection point, especially if Q1 is
not involved in the oscillation.
I've had fun with this, but my wife thinks I should mow the grass
again this month, so I'll turn it over for comments.
I have been looking at the n-mos switch, M1, in a inductive boost
regulator to light some LEDs. I thought that faster switching speeds
there would result in improvements in the circuit.
See http://www.jump.net/~vima/led/Impulse.pdf
I had read in genome's postings about special drivers for the gate
(Cooper?), which is where I got the idea of using a special driver for
the gate. I came up with my circuit just doodling in SwitcherCAD to
try to practice circuit comprehension.
One view of this is that the total gate energy, (gate charge x switch
voltage) on M1, plus the loss in R1 should equal the total change in
inductor energy in L2. Higher currents and smaller inductors keep the
energy the same but speed things up. R1 is a tax.
At the heart of the circuit, I used Q1 to control the current in an
kick-switch comprised of D3, R1, and L2. When Q1 conducts, current
flows in L2. As the base voltage rises on Q1 it commands "off" thence
the collector switches off slowly. L1 is starved, and the M1 gate
voltage drops, the gate charge flows quickly. The appropriate
simulation traces make nice art. Component values on R1, L2 (should)
limit the M1 gate voltage to -0.6V and D3 provides safety backup on
this limit.
Is "my" circuit published, known, or analyzed? Is it fundamentally
flawed somehow? Is it wondrously novel and my ticket to fame and
fortune?
SwitcherCAD III says that this is a wonderful idea, producing output
voltages following L1 upward of 150V, but hungry for current in L2.
I'll somehow deal with that later, with smaller currents, bigger
inductors and the correct capacitor in the correct place.
There is, however, singing. I looked at the FFT. I am feeding a
square wave in on V2 to simulate the rest of the control circuit, so
there are odd harmonics climbing all up the spectrum. I did an FFT of
the L1 output voltage from 49.01ms to 99.01ms using 256K points. The
harmonic at 220 KHz is 18 db above the harmonic at 230KHz.
Ignoring the FFT, I had assumed that a small signal well below the
linear region on the M1 gate would not matter.
The signal there is 200 mV p-p at 0 DC. Conduction starts at about
1.9V.
I haven't decided what to do next. One thing that comes to mind is to
do a small signal ac analysis on V2 with the other circuit parameters
set as they are at oscillation, and follow the signal through the
circuit. There may be a better injection point, especially if Q1 is
not involved in the oscillation.
I've had fun with this, but my wife thinks I should mow the grass
again this month, so I'll turn it over for comments.
