Maker Pro
Maker Pro

Cute little circuit

T

Tim Williams

Jan 1, 1970
0
http://webpages.charter.net/dawill/Images/High_Voltage_Generator.png
(That feedback winding is part of the main transformer...)

Great regulation, as TL431 always provides. Efficiency probably around 50%
at full output, not great but reasonable given the control method. PSRR
isn't all that great, simply because the feedback network is VCC-referenced.
Output ripple is in the mV.

Hmm, occurs to me Sloman will probably appreciate this. Though the
distortion is obviously not very good, which isn't the point.

Tim
 
T

Tim Williams

Jan 1, 1970
0
tm said:
Yeah, and lose the series 914s and use single UF400x diodes.

Waffor? 914s are low capacitance, low leakage and very short recovery time.
Keeps switching noise down (as does the series inductor). The additional
voltage drop is negligible.

Tim
 
T

tm

Jan 1, 1970
0
Tim Williams said:
Waffor? 914s are low capacitance, low leakage and very short recovery
time. Keeps switching noise down (as does the series inductor). The
additional voltage drop is negligible.

Tim

Hey, let's optimize it. Cost, efficiency, parts count, reliability, ...

List the design requirements:

9 volts in, 100 volts out @ 3 ma., regulation, noise, ...


Anyway, two parts vs. four? Board space? Holes? Both are fast, both are low
C, Both about the same cost though the 914 might beat it out with quantity.
 
L

legg

Jan 1, 1970
0
http://webpages.charter.net/dawill/Images/High_Voltage_Generator.png
(That feedback winding is part of the main transformer...)

Great regulation, as TL431 always provides. Efficiency probably around 50%
at full output, not great but reasonable given the control method. PSRR
isn't all that great, simply because the feedback network is VCC-referenced.
Output ripple is in the mV.

Hmm, occurs to me Sloman will probably appreciate this. Though the
distortion is obviously not very good, which isn't the point.

Tim

At low power levels, you get away with murder. Actually, at 500mW, the
50% efficiency is pretty good, considering the large number of parts.

Couldn't figure out if you were getting regulation by increasing the
cross-conduction - a boost converter - or through base current
starvation. I guess it has to be the latter because the frequency
would normally be set by core saturation in this topology, and because
you need it to start up (a handy feature in any supply).

Did you model it first? Will it model?

RL
 
T

Tim Williams

Jan 1, 1970
0
Bill Sloman said:
It looks rather like a Baxandall class-D oscillator

http://home.planet.nl/~sloma000/0344_001_Baxandal.pdf

but the two 2N4401 switching transistors won't be getting enough
base-drive to saturate, and will presumably run hot.

They run linear for light loads, which isn't terrible as it's a light load.
Efficiency vs. load and vs. supply voltage will vary more than otherwise.
It might make more sense to PWM the voltage applied to the 270uH
inductor - switching it between 0V and 9V - which is a configuration I've
simulated, but the feedback would be a bit of a nightmare.

I've tried it before on other occasions; you need a huge inductance if the
switching frequency is not considerably higher than the oscillating
frequency. I know of (relatively small) induction heaters which do this
(replacing the transistors with IGBTs and a PLL); they use a 15kg choke to
keep ripple low enough.

Also, the extra switching may introduce undesirable noise, say if it's used
for a photodiode or avalanche supply.

Tim
 
T

Tim Williams

Jan 1, 1970
0
legg said:
Couldn't figure out if you were getting regulation by increasing the
cross-conduction - a boost converter - or through base current
starvation. I guess it has to be the latter because the frequency
would normally be set by core saturation in this topology, and because
you need it to start up (a handy feature in any supply).

Assuming low bias and small signal conditions, "cross conduction" will look
more like a differential amplifier with lots of feedback. It never really
runs out of voltage, so it'll tend to look more like an alternating
beta-dependent current source.

Timing is provided by the 220pF on the secondary; surprisingly, frequency
doesn't change much under load, despite the squashed waveform that would
tend to suggest inductive or saturation induced commutation.

The frequency is on the low side for an FT50-43 at this number of turns, but
the full-output waveform looks clean (a sine wave, somewhat skewed due to
core nonlinearity and hysteresis losses I think).
Did you model it first? Will it model?

Of course I did, on the best simulator available... Nature 1.0 ;-)

If it's switching due to saturation, it may still switch due to inductance
alone, without using a nonlinear core model. I've seen that before with
blocking oscillators. Saturation will commutate quicker, though.

One thing the simulation is almost guaranteed to miss is the squirrelies.
At low power, the waveforms look funky due to myriad resonances (it doesn't
maintain a clean/clamped sine appearance).

Tim
 
L

legg

Jan 1, 1970
0
Assuming low bias and small signal conditions, "cross conduction" will look
more like a differential amplifier with lots of feedback. It never really
runs out of voltage, so it'll tend to look more like an alternating
beta-dependent current source.

Timing is provided by the 220pF on the secondary; surprisingly, frequency
doesn't change much under load, despite the squashed waveform that would
tend to suggest inductive or saturation induced commutation.

The frequency is on the low side for an FT50-43 at this number of turns, but
the full-output waveform looks clean (a sine wave, somewhat skewed due to
core nonlinearity and hysteresis losses I think).


Of course I did, on the best simulator available... Nature 1.0 ;-)

If it's switching due to saturation, it may still switch due to inductance
alone, without using a nonlinear core model. I've seen that before with
blocking oscillators. Saturation will commutate quicker, though.

One thing the simulation is almost guaranteed to miss is the squirrelies.
At low power, the waveforms look funky due to myriad resonances (it doesn't
maintain a clean/clamped sine appearance).

Tim

So your model was close. How did it model start up?

Oh. I get it. A little slow today. NatureSpice.....

RL
 
L

legg

Jan 1, 1970
0
Not knowing the exact currents, I can't say anything definitive, but my inclination would have been to PWM at twice the resonant frequency of the transformer-capacitor LC - use a phase-locked loop to lock that frequency to the VCO in a 4046 running at close to 10MHz, then divide down to get a waveform or two which could be used to switch the 2N401s, and separately to get a PWM waveform synch'd to twice that frequency.

Feed-back would control the on/off ratio on the voltage applied to the inductor.

That would basically fiddle with the second harmonic current you've got flowing through the inductor anyway, and if you got the phase right you could probably minimise that. Switching back and forth between the two almost right PWM on/off ratio's to get the output voltage right would introduce some ripple, but not a great deal.

The digital logic would all fit into a smallish programmable logic device - something bigger than a 22V10, but not all that much bigger.


They survive. You've got to filter like hell after any kind of inverter. You can play trick to make the switching noise easier to filter out, but there's always some ripple to get rid of.

This is a 500mW circuit.

RL
 
J

John S

Jan 1, 1970
0
I've now cleaned up my original model, by adding the ferrite beads L9
and L10 to stop the high voltage generator messing up the inverter so
that the inverter now oscillates at a reasonable 135kHz, rather than
8MHz. The rest of the extra components stop all the other oscillations
(more or less). John Larkin's idea might be rather easier to put into
practice.

Apparently, you did not run the simulation out to 100ms. Pretty crappy
waveform out there.
 
L

legg

Jan 1, 1970
0
What might be awkward is that each of the 2N4401 switching transistors is going be be running as base-current-controlled current sink when it's on. There's no guarantee that they'll have the same current gain, so they'll pull down one side of the transformer harder than the other. Presumably the circuit sorts itself out, but it isn't exactly surprising that it gets "squirrelly" under light loads.

<snip>

Sometimes it's useful to separate the regulator from the power train -
just to sort of get an indea of what the power train wants to do over
the intended control range.

http://www.magma.ca/~legg/TVS/cute_circuit.zip

These show the basic requirements and also the limits of the thing
you're trying to control.

I always get suspicious when I see control node voltage or current
ripple at multiples of the conversion frequency, at the last point of
control for a power train that is itself incapable of significant slew
rates, and may require 'balance' for at least a full conversion
period.

RL
 
Top