No voltage rating for inductors...

[John Larkin]

On Tue, 8 Mar 2022 09:41:51 +1100, Sylvia Else <sylvia@email.invalid>
wrote:

On 08-Mar-22 1:02 am, legg wrote:
On Mon, 7 Mar 2022 19:03:14 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 2:12 pm, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 12:21:09 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 10:59 am, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

Powdered iron can get lossy if you drive it hard. Then it gets hot and
everything gets worse. KoolMu or Sendust has become affordable.


The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.

I think a few of the Coilcraft parts have voltage ratings, but it\'s
rare. The dual-winding guys, like DRQ127, are probably bifalar and
have more ways to arc.

I usually test them.

How much voltage were you considering? That toroid looks pretty good.

Class D amp?




No. I was pondering why LED light fittings need to have a large
electrolytic capacitor with its limited life.

I have a couple of small LEDs illuminating an otherwise frequently pitch
black corridor. They\'re powered through a transformer, bridge rectifier
and current limiter, with no capacitors at all. They obviously flicker
at 100Hz, but it\'s not visible.

So then I was thinking that one could use a buck converter directly off
the rectified, but not filtered, mains, hence the voltage requirement.

This has not moved past musings at this point, but I was looking at the
voltage specs for inductors, hence my comment.

Sylvia.

Lots of LED bulbs have a bridge and a linear IC current limiter but no
cap. A capless switcher is an interesting idea. The LED could run at
constant current at a very high duty cycle.

An inductor like you showed would be fine at hundreds of volts. Test
one to breakdown!






Seems to work in LTSpice at least:

The switch is a stand-in for a digital isolator.

Apart from V6 which is the incoming mains supply, the various voltage
sources would be small amounts of power derived from the rectified
mains. I haven\'t simulated the fact that this power would not be
available for short periods.

snip

LTspice IV threw a \'singular matrix node error\' here, but only if
the Philips models for 2N2222 and 2N3906 were used. NSC models
allowed the simulation to run.

I\'m not sure that you\'re going to find a \'digital isolator\' that\'s
designed for PWM - they\'re usually pretty slow. You might consider
an integrated gate driver + discrete nmos as a lower-parts-count
solution.

If you check the input current, or the current in the freewheeling
diode (probably not included inside the digital isolator, by the way),
you\'ll see diode current spikes occurring when it turns off and the
switch turns on. The amplitude of these spikes will vary with diode
type and nmos switching speed (slower speeds are generated if a
series gate resistor is present on M1).

These spikes are not simulation artifacts, they are modeling attempts
to simulate schottky capacitance or rectifier reverse recovery.
200V schottkeys are rare beasts, but there are fast recovery
rectifiers offered in the standard LTspice distribution selection
lists.

The Bourns data sheet gives a flux density calculation in gauss
using K * L * dI with a frequency-dependent core loss chart
specific to the SRR1210 series.

In a ripple-regulated circuit dI will vary with frequency - in
the sim it\'s between 200 and 400mAppk.


In a regulated circuit, the peak to peak flux swing in the choke is
determined by Bppk = V * t / ( N * A)
where
Bppk = peak to peak flux density (Teslas)
V = Vout + Vdfreewheel (volts)
t = Tfreewheel (seconds)
N = number of turns on the choke
A = xsectional area of choke

If you\'ve got a table for the choke\'s material, you can work
out core loss (W/m^3 - mW/cm^3) and apply that to the volume
of the core geometry being used.

Voltage is present in that equation. For peak or surge, it\'s
either saturation or turn/terminal breakdown limited. Some
core structures, when impressed with low impedance surges,
will try to turn themselves inside out, in the attempt to
force changes in the magnetic length/area ratio.

Lighting ballast design for the domestic market is not the easiest
road to fame and fortune.

Considering the efficiencies of LED sources and the dominance
of standard hardware formats, it turns into a marketing exercise.

RL

Thanks for taking a look. In the end this was never going to be more
than an academic exercise. I don\'t even intend to attempt to build one.
To be remotely viable commercially, most of the electronics would have
to be bundled into an IC, and I don\'t see that happening.

The real challenge would be to do it with 12 cents worth of discrete
parts.

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

 

[Sylvia Else]

On 08-Mar-22 12:11 pm, legg wrote:

On Tue, 8 Mar 2022 09:41:51 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 08-Mar-22 1:02 am, legg wrote:
On Mon, 7 Mar 2022 19:03:14 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 2:12 pm, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 12:21:09 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 10:59 am, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

Powdered iron can get lossy if you drive it hard. Then it gets hot and
everything gets worse. KoolMu or Sendust has become affordable.


The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.

I think a few of the Coilcraft parts have voltage ratings, but it\'s
rare. The dual-winding guys, like DRQ127, are probably bifalar and
have more ways to arc.

I usually test them.

How much voltage were you considering? That toroid looks pretty good.

Class D amp?




No. I was pondering why LED light fittings need to have a large
electrolytic capacitor with its limited life.

I have a couple of small LEDs illuminating an otherwise frequently pitch
black corridor. They\'re powered through a transformer, bridge rectifier
and current limiter, with no capacitors at all. They obviously flicker
at 100Hz, but it\'s not visible.

So then I was thinking that one could use a buck converter directly off
the rectified, but not filtered, mains, hence the voltage requirement.

This has not moved past musings at this point, but I was looking at the
voltage specs for inductors, hence my comment.

Sylvia.

Lots of LED bulbs have a bridge and a linear IC current limiter but no
cap. A capless switcher is an interesting idea. The LED could run at
constant current at a very high duty cycle.

An inductor like you showed would be fine at hundreds of volts. Test
one to breakdown!






Seems to work in LTSpice at least:

The switch is a stand-in for a digital isolator.

Apart from V6 which is the incoming mains supply, the various voltage
sources would be small amounts of power derived from the rectified
mains. I haven\'t simulated the fact that this power would not be
available for short periods.

snip

LTspice IV threw a \'singular matrix node error\' here, but only if
the Philips models for 2N2222 and 2N3906 were used. NSC models
allowed the simulation to run.

I\'m not sure that you\'re going to find a \'digital isolator\' that\'s
designed for PWM - they\'re usually pretty slow. You might consider
an integrated gate driver + discrete nmos as a lower-parts-count
solution.

If you check the input current, or the current in the freewheeling
diode (probably not included inside the digital isolator, by the way),
you\'ll see diode current spikes occurring when it turns off and the
switch turns on. The amplitude of these spikes will vary with diode
type and nmos switching speed (slower speeds are generated if a
series gate resistor is present on M1).

These spikes are not simulation artifacts, they are modeling attempts
to simulate schottky capacitance or rectifier reverse recovery.
200V schottkeys are rare beasts, but there are fast recovery
rectifiers offered in the standard LTspice distribution selection
lists.

The Bourns data sheet gives a flux density calculation in gauss
using K * L * dI with a frequency-dependent core loss chart
specific to the SRR1210 series.

In a ripple-regulated circuit dI will vary with frequency - in
the sim it\'s between 200 and 400mAppk.


In a regulated circuit, the peak to peak flux swing in the choke is
determined by Bppk = V * t / ( N * A)
where
Bppk = peak to peak flux density (Teslas)
V = Vout + Vdfreewheel (volts)
t = Tfreewheel (seconds)
N = number of turns on the choke
A = xsectional area of choke

If you\'ve got a table for the choke\'s material, you can work
out core loss (W/m^3 - mW/cm^3) and apply that to the volume
of the core geometry being used.

Voltage is present in that equation. For peak or surge, it\'s
either saturation or turn/terminal breakdown limited. Some
core structures, when impressed with low impedance surges,
will try to turn themselves inside out, in the attempt to
force changes in the magnetic length/area ratio.

Lighting ballast design for the domestic market is not the easiest
road to fame and fortune.

Considering the efficiencies of LED sources and the dominance
of standard hardware formats, it turns into a marketing exercise.

RL

Thanks for taking a look. In the end this was never going to be more
than an academic exercise. I don\'t even intend to attempt to build one.
To be remotely viable commercially, most of the electronics would have
to be bundled into an IC, and I don\'t see that happening.

While eliminating the electrolytic capacitor could benefit the consumer
(assuming the high voltages don\'t produce other reductions in life),
it\'s of little interest to manufacturers - even those making the
hypothetical IC.

Sylvia.

Well, they wouldn\'t tell you, if they were, but there ARE application-
specific ICs showing up in SEA - they just don\'t crow about it, or
have any interest in the North American market.

That flat-topped line current waveform isn\'t any easier, w/r to
harmonic distortion on the transformer-coupled grid, than the
capacitive rectifier peak, but it is \'loss-reduced\'. The linear
LED lamps have a similar profile, when caps are removed.

RL

My own home is illuminated almost entirely by low voltage lamps.
Probably someone in the past fell for the notion that low voltage means
low power consumption.

I replaced all the previous incandescent low voltage lamps with low
voltage LEDs with the same fitting. When the caps wear out, which is
within a few years, they start flashing. Next time it happens, I\'ll try
removing the caps entirely. I\'m not that hopeful, but you never know.

Sylvia.

 

[Jan Panteltje]

On a sunny day (Mon, 07 Mar 2022 20:11:59 -0500) it happened legg
<legg@nospam.magma.ca> wrote in <uuad2hpo887r2jdban5a8mghujpq5md96t@4ax.com>:

That flat-topped line current waveform isn\'t any easier, w/r to
harmonic distortion on the transformer-coupled grid, than the
capacitive rectifier peak, but it is \'loss-reduced\'. The linear
LED lamps have a similar profile, when caps are removed.

RL

The Chinese LED lamps from ebay I have use a much smaller electrolytic
capacitor that you could probably leave out:
circuit:
http://panteltje.com/pub/LED_light_circuit_diagram_IMG_6925.JPG

A look inside :
http://panteltje.com/pub/LED_light_fix_IMG_6918.JPG
The current is limited by the 1 uF cap in series with the AC.
The voltage on the eletrolytics is limited by the LEDs working as zeners.

simple...
Some LEDs died
Transients on the mains commom here.

 

[legg]

On Mon, 07 Mar 2022 17:20:22 -0800, John Larkin
<jlarkin@highland_atwork_technology.com> wrote:

On Tue, 8 Mar 2022 09:41:51 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 08-Mar-22 1:02 am, legg wrote:
On Mon, 7 Mar 2022 19:03:14 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 2:12 pm, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 12:21:09 +1100, Sylvia Else <sylvia@email.invalid
wrote:

On 06-Mar-22 10:59 am, jlarkin@highlandsniptechnology.com wrote:
On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

Powdered iron can get lossy if you drive it hard. Then it gets hot and
everything gets worse. KoolMu or Sendust has become affordable.


The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.

I think a few of the Coilcraft parts have voltage ratings, but it\'s
rare. The dual-winding guys, like DRQ127, are probably bifalar and
have more ways to arc.

I usually test them.

How much voltage were you considering? That toroid looks pretty good.

Class D amp?




No. I was pondering why LED light fittings need to have a large
electrolytic capacitor with its limited life.

I have a couple of small LEDs illuminating an otherwise frequently pitch
black corridor. They\'re powered through a transformer, bridge rectifier
and current limiter, with no capacitors at all. They obviously flicker
at 100Hz, but it\'s not visible.

So then I was thinking that one could use a buck converter directly off
the rectified, but not filtered, mains, hence the voltage requirement.

This has not moved past musings at this point, but I was looking at the
voltage specs for inductors, hence my comment.

Sylvia.

Lots of LED bulbs have a bridge and a linear IC current limiter but no
cap. A capless switcher is an interesting idea. The LED could run at
constant current at a very high duty cycle.

An inductor like you showed would be fine at hundreds of volts. Test
one to breakdown!






Seems to work in LTSpice at least:

The switch is a stand-in for a digital isolator.

Apart from V6 which is the incoming mains supply, the various voltage
sources would be small amounts of power derived from the rectified
mains. I haven\'t simulated the fact that this power would not be
available for short periods.

snip

LTspice IV threw a \'singular matrix node error\' here, but only if
the Philips models for 2N2222 and 2N3906 were used. NSC models
allowed the simulation to run.

I\'m not sure that you\'re going to find a \'digital isolator\' that\'s
designed for PWM - they\'re usually pretty slow. You might consider
an integrated gate driver + discrete nmos as a lower-parts-count
solution.

If you check the input current, or the current in the freewheeling
diode (probably not included inside the digital isolator, by the way),
you\'ll see diode current spikes occurring when it turns off and the
switch turns on. The amplitude of these spikes will vary with diode
type and nmos switching speed (slower speeds are generated if a
series gate resistor is present on M1).

These spikes are not simulation artifacts, they are modeling attempts
to simulate schottky capacitance or rectifier reverse recovery.
200V schottkeys are rare beasts, but there are fast recovery
rectifiers offered in the standard LTspice distribution selection
lists.

The Bourns data sheet gives a flux density calculation in gauss
using K * L * dI with a frequency-dependent core loss chart
specific to the SRR1210 series.

In a ripple-regulated circuit dI will vary with frequency - in
the sim it\'s between 200 and 400mAppk.


In a regulated circuit, the peak to peak flux swing in the choke is
determined by Bppk = V * t / ( N * A)
where
Bppk = peak to peak flux density (Teslas)
V = Vout + Vdfreewheel (volts)
t = Tfreewheel (seconds)
N = number of turns on the choke
A = xsectional area of choke

If you\'ve got a table for the choke\'s material, you can work
out core loss (W/m^3 - mW/cm^3) and apply that to the volume
of the core geometry being used.

Voltage is present in that equation. For peak or surge, it\'s
either saturation or turn/terminal breakdown limited. Some
core structures, when impressed with low impedance surges,
will try to turn themselves inside out, in the attempt to
force changes in the magnetic length/area ratio.

Lighting ballast design for the domestic market is not the easiest
road to fame and fortune.

Considering the efficiencies of LED sources and the dominance
of standard hardware formats, it turns into a marketing exercise.

RL

Thanks for taking a look. In the end this was never going to be more
than an academic exercise. I don\'t even intend to attempt to build one.
To be remotely viable commercially, most of the electronics would have
to be bundled into an IC, and I don\'t see that happening.

The real challenge would be to do it with 12 cents worth of discrete
parts.

You might be surprised. I recall being totally demoralized by an
0.08 transformer quote in 2008. If you stick to the right commodity
materials, it\'s a completely different market, or perhaps the
demoralization was intentional. I\'m being paranoid here.

I visualize being paid to use parts, under corporate, tax or political
circumstances, that I\'d really rather not consider . . .

RL

 

[Skittle]

On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid>
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.

If your inductor has a decent Q value (low resistance) then the DC
voltage drop across the inductor is almost zero. Should be no DC
arcing within the inductor itself.

Skittles

 

[Sylvia Else]

On 09-Mar-22 3:07 am, Skittles wrote:

On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.


If your inductor has a decent Q value (low resistance) then the DC
voltage drop across the inductor is almost zero. Should be no DC
arcing within the inductor itself.

Skittles

Yes. The issue related to short pulses.

Sylvia.

 

[Skittle]

On Wed, 9 Mar 2022 09:42:44 +1100, Sylvia Else <sylvia@email.invalid>
wrote:

On 09-Mar-22 3:07 am, Skittles wrote:
On Sun, 6 Mar 2022 10:23:29 +1100, Sylvia Else <sylvia@email.invalid
wrote:

I\'ve looked at a number of data sheets, and they never seem to give a
voltage rating.

Of course, applying any significant DC voltage across an inductor is
unlikely to have a good outcome, but one can certainly put a significant
voltage across one for a short period, and it would be nice to know what
the limits are.

As a random example

https://www.farnell.com/datasheets/1870387.pdf

The obvious concerns here are breakdown of the insulation between the
wire and the core, and between windings at the two ends where they are
close together.

Sylvia.


If your inductor has a decent Q value (low resistance) then the DC
voltage drop across the inductor is almost zero. Should be no DC
arcing within the inductor itself.

Skittles

Yes. The issue related to short pulses.

Sylvia.

https://www.coilcraft.com/getmedia/393e18e1-adbc-45b6-bbe7-5923255e72fc/doc712_Inductor_Voltage_Ratings.pdf


https://www.coilcraft.com/getmedia/1ce5c5e9-3266-4dfa-80a3-e085d659785e/doc1520_operating_voltage_ratings_for_inductors.pdf