Non-linear load compensation

[Dave Masters]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Is my hobby driving me crazy, or does such a part exist.

PS: For design reasons, I prefer not to use an LC network, etc.

Any help would be sincerely appreciated.

Dave Masters

 

[[email protected]]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Is my hobby driving me crazy, or does such a part exist.

PS: For design reasons, I prefer not to use an LC network, etc.

Any help would be sincerely appreciated.

Dave Masters

Wouldn't a better driver be the solution?

 

[[email protected]]

Wouldn't a better driver be the solution?

If you only need a couple of volts of voltage swing, a video driver or
a fast op amp intended for use as a video driver might be viable
options.

Tell us more about the amplifier you are using at the moment.

 

[Winfield Hill]

Dave Masters wrote...

I am powering a non-linear load with sine waves in the sub 100Hz
range. Every time (twice during each cycle) the applied voltage
crosses a certain threshold, a glitch appears in the waveform as
monitored across the load. This is due to the load becoming
instantaneously less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel
with the load to take over some of its loading action when the
input voltage goes low, and thus smooth the transition.

The device you are looking for is a resistor.

 

[Mark]

The device you are looking for is a resistor.

or a capacitor..
Mark

 

[Genome]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Is my hobby driving me crazy, or does such a part exist.

PS: For design reasons, I prefer not to use an LC network, etc.

Any help would be sincerely appreciated.

Dave Masters

It's not really going to happen, is it?

I have an apple tree that produces bananas and, every so often, produces
pears but I wanted grapefruits. This is due to the tree graft coming from a
weeping willow.

What I am looking for is a nice entertaining game in plastic, preferably
yellow, that will give me some oranges, but forget about the oranges for now
because that just confuses things.

PS: for design reasons, I would prefer not to use something that will solve
my problem etc.

Thanks, all help really appreciated, glad to be part of the group, perhaps
we can get together for a hug later.

Bestest

DNA

 

[legg]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Is my hobby driving me crazy, or does such a part exist.

PS: For design reasons, I prefer not to use an LC network, etc.

Any help would be sincerely appreciated.

You're worried about the voltage across the load.

Why?

The load doesn't seem to care.

The non-linearity sounds something like a solenoid actuating.

If there are parts of the circuit that need a clean voltage, isolate
the nonlinear portion of the load.

It's possible that what you're just seeing is the load current passing
through zero out of phase with the measured voltage. Some drivers
don't like this.

RL

 

[[email protected]]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Sounds like a woofer.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

Sounds like friction against the voice coil of a woofer, which 'sticks'
when velocity hits
zero. Sticking friction is a higher force than sliding friction.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Huh? What does that mean? Do you want the current to not follow the
demands of the
load? Do you want the driver amplifier to be unaffected by the
transient? Do you want
the load current to be a perfect sinewave, or the load voltage to be a
perfect sinewave?

Or do you want to readjust the voice coil in the woofer so it doesn't
rub the magnet?

 

[jasen]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

What I am looking for is a device that can be wired in parallel with
the load to take over some of its loading action when the input
voltage goes low, and thus smooth the transition.

Is my hobby driving me crazy, or does such a part exist.

a capacitor there would help absorb the hump

PS: For design reasons, I prefer not to use an LC network, etc.

use something that behaves like a capacitor then....

Any help would be sincerely appreciated.

what you don't say is how you're driving it

is it a voltage drive or a current drive

is there room for some feedback to modulate the drive so the glitch
becomes less significant?

 

[joseph2k]

If you only need a couple of volts of voltage swing, a video driver or
a fast op amp intended for use as a video driver might be viable
options.

Tell us more about the amplifier you are using at the moment.

Well gosh bill ReadTheFinePost, OP aleardy ssaid under 100 Hz, hardly a need
for normal video drivers.

 

[[email protected]]

Well gosh bill ReadTheFinePost, OP aleardy ssaid under 100 Hz, hardly a need
for normal video drivers.

Well gosh josephkk, ReadTheFinePost, OP wants to minimise cross-over
distortion, which can be helped by using a driver with a very high
bandwidth and slew rate, such as a nice, cheap, widely available video
driver. This won't get rid of the glitch, but it can feed back the
error signal fast enough to make the glitch very, very small.

Try engaging your brain before descending on your keyboard.

 

[joseph2k]

Well gosh josephkk, ReadTheFinePost, OP wants to minimise cross-over
distortion, which can be helped by using a driver with a very high
bandwidth and slew rate, such as a nice, cheap, widely available video
driver. This won't get rid of the glitch, but it can feed back the
error signal fast enough to make the glitch very, very small.

Try engaging your brain before descending on your keyboard.

Back off and remember that crossover issues are usually bias problems not
bandwidth problems. 100 MHz amplifiers are not necessarily the solution to
100 Hz problems. Fully discrete class "A" amplifiers with appropriate
negative feedback can do the required task easily.

 

[[email protected]]

Back off and remember that crossover issues are usually bias problems not
bandwidth problems. 100 MHz amplifiers are not necessarily the solution to
100 Hz problems. Fully discrete class "A" amplifiers with appropriate
negative feedback can do the required task easily.

Class A amplifiers don't suffer from cross-over distortion. Properly
biassed class B amplifiers are almost as good, but you end up with
quite a lot of bias current, and you have to worry about the junction
temperatures of all the transistors involved in determining the bias
current.

A driver which can slew two Vbe drops in less than a microsecond is a
less elegant solution, but it can work pretty well, and doesn't demand
the same level of skill in thermal design.

In fact, it isn't clear that the OP was having trouble with cross-over
distortion as such - he complained about a non-linear load - and lots
of gain-bandwidth in the feedback loop is probably the more general
solution.

 

[Ken Smith]

Class A amplifiers don't suffer from cross-over distortion. Properly
biassed class B amplifiers are almost as good, but you end up with
quite a lot of bias current, and you have to worry about the junction
temperatures of all the transistors involved in determining the bias
current.

Actually that makes it a class AB not just a class B, but the point is
mostly correct. You can also reduce the needed idle current at the cost
of losing some peak to peak swing by increasing the emitter balasting
resistors.

The idle current for least distortion is just about where the gain for no
signal is equal to the gain where only one transistor is in use.

 

[TuT]

[snip]

Class A amplifiers don't suffer from cross-over distortion. Properly
biassed class B amplifiers are almost as good, but you end up with
quite a lot of bias current, and you have to worry about the junction
temperatures of all the transistors involved in determining the bias
current.

[snip]

If you use Sziklai-connected darlingtons in the output stage, then only the
Vbe of the drivers are in the loop for quiescent current control. Since
these aren't working as hard as the output devices, they don't get as hot
and you can thermally link the transistor used for the Vbe multiplier to the
two drivers and end up with a pretty stable quiescent current over all
output levels.

 

[[email protected]]

[snip]

Class A amplifiers don't suffer from cross-over distortion. Properly
biassed class B amplifiers are almost as good, but you end up with
quite a lot of bias current, and you have to worry about the junction
temperatures of all the transistors involved in determining the bias
current.

[snip]

If you use Sziklai-connected darlingtons in the output stage, then only the
Vbe of the drivers are in the loop for quiescent current control. Since
these aren't working as hard as the output devices, they don't get as hot
and you can thermally link the transistor used for the Vbe multiplier to the
two drivers and end up with a pretty stable quiescent current over all
output levels.

Been there, done that. My home-built amplifier in the basement works
exactly that way, and has done since about 1975.

 

[joseph2k]

Class A amplifiers don't suffer from cross-over distortion. Properly
biassed class B amplifiers are almost as good, but you end up with
quite a lot of bias current, and you have to worry about the junction
temperatures of all the transistors involved in determining the bias
current.

A driver which can slew two Vbe drops in less than a microsecond is a
less elegant solution, but it can work pretty well, and doesn't demand
the same level of skill in thermal design.

In fact, it isn't clear that the OP was having trouble with cross-over
distortion as such - he complained about a non-linear load - and lots
of gain-bandwidth in the feedback loop is probably the more general
solution.

Each to their own, I would use a class AB if i had to go that way.
Crossover glitch is a typical property of class B and class C amplifiers.
It is not difficult to a little class AB and still remain power efficient.

 

[Ancient_Hacker]

I am powering a non-linear load with sine waves in the sub 100Hz
range.

Every time (twice during each cycle) the applied voltage crosses a
certain threshold, a glitch appears in the waveform as monitored
across the load. This is due to the load becoming instantaneously
less, and then more, conductive at that threshold.

Sounds like your load has like back-to-back diodes in series with it,
so below some value the load is disconnected. There's a whole lot of
issues here:

(1) If the load is disconnected and it's not drawing power, that's
probably bad-- you're going to get distortion. The best way to fix
this is not to kludge something, but to find a way to make the load
more reasonable.

(2) You're asking for a resistor that conducts more at lower voltages,
then much less or not at all at higher voltages. That's close to what
a negative resistance does, only a negative resistor would be linear.
Depending on the power level of the load, anytwhere from two to a
tmillion tunnel diodes might be coerced into doing what you need.
(Kidding-- tunnel diodes are getting hard to find).

(3) If you absolutely positively need this, a pair of comparators and
a MOSFET or two will do this.

(4) A better question might be: WHY do you want to smooth out this
glitch? It's not going to affect the signal power delivered to the
load, it's only going to smooth out the voltage waveform.

(5) The best way would be to forget about a changing load, but instead
work on the source. Lowering the source impedance with negative
feedback is the common way to do this.