film cap test...

[John Larkin]

We want some biggish film caps to use in some LC filters for a couple
of switching power things, so we got some to test.

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe, but if I scrooch it
into the test rig with longer leads the ESL spikes go way up,
consistent with the loop area, so the low ESL seems real. It will be
even lower soldered on a PCB, into some hunky copper planes.

Of course ESL is not a lumped property of a wound film cap, but this
is close.

We were considering some big ceramic caps, but they lose 80% of their
c at our 48 volts DC.

--

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

 

[Phil Hobbs]

John Larkin wrote:

We want some biggish film caps to use in some LC filters for a couple
of switching power things, so we got some to test.

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe, but if I scrooch it
into the test rig with longer leads the ESL spikes go way up,
consistent with the loop area, so the low ESL seems real. It will be
even lower soldered on a PCB, into some hunky copper planes.

Of course ESL is not a lumped property of a wound film cap, but this
is close.

We were considering some big ceramic caps, but they lose 80% of their
c at our 48 volts DC.

Fun! That\'s quite a lot lower than I would have expected, and a useful
wrench for the toolkit.

Of course the null position just demonstrates that some series-LC
someplace in a more complicated model has that inductance--the external
leads, banana jack, and so on will contribute as normal.

That 8 nH is reasonably believable because the conduction current in
adjacent foils goes in opposite directions and the capacitive
displacement current will be more or less radially symmetric, so that
the B field tends to cancel there too. (Inductance goes as the volume
integral of |B/I|**2.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[whit3rd]

On Wednesday, July 6, 2022 at 2:35:09 PM UTC-7, John Larkin wrote:

We want some biggish film caps to use in some LC filters for a couple
of switching power things, so we got some to test.

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe, but if I scrooch it
into the test rig with longer leads the ESL spikes go way up,
consistent with the loop area, so the low ESL seems real. It will be
even lower soldered on a PCB, into some hunky copper planes.

Of course ESL is not a lumped property of a wound film cap, but this
is close.

Could it be a stacked-foil item, instead? Mechanically, those have
windings, but the leadwires contact in two spots on each layer.
They\'re always radial, not axial (I think...).

 

[server]

On Wed, 6 Jul 2022 18:01:22 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Wednesday, July 6, 2022 at 2:35:09 PM UTC-7, John Larkin wrote:
We want some biggish film caps to use in some LC filters for a couple
of switching power things, so we got some to test.

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe, but if I scrooch it
into the test rig with longer leads the ESL spikes go way up,
consistent with the loop area, so the low ESL seems real. It will be
even lower soldered on a PCB, into some hunky copper planes.

Of course ESL is not a lumped property of a wound film cap, but this
is close.

Could it be a stacked-foil item, instead? Mechanically, those have
windings, but the leadwires contact in two spots on each layer.
They\'re always radial, not axial (I think...).

I really want a high-res xray machine. I guess I could whack this cap
with a hammer and get an idea of the construction.

I just tested one of these

https://www.mouser.com/ProductDetail/XP-Power/SRH05S12?qs=w%2Fv1CP2dgqoB4ok%252BZU%252BVXg%3D%3D

and it was remarkable. It would be cool to see what\'s inside.

 

[Phil Allison]

John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?
Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf

..... Phil

 

[Phil Hobbs]

Phil Allison wrote:

John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?
Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf

The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[server]

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?

Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

The conductive path, end to end, is probably many concentric cylinders
of foil, which I guess could have very low inductance. In my
measurement, the leads themselves seem to dominate L.

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

 

[John Larkin]

On Thu, 07 Jul 2022 07:51:24 -0700, jlarkin@highlandsniptechnology.com
wrote:

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?


Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

The conductive path, end to end, is probably many concentric cylinders
of foil, which I guess could have very low inductance. In my
measurement, the leads themselves seem to dominate L.

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

We\'ll use lots of these so it\'s good to understand them.


--

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

 

[Cydrome Leader]

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Thu, 07 Jul 2022 07:51:24 -0700, jlarkin@highlandsniptechnology.com
wrote:

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?


Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

The conductive path, end to end, is probably many concentric cylinders
of foil, which I guess could have very low inductance. In my
measurement, the leads themselves seem to dominate L.

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

We\'ll use lots of these so it\'s good to understand them.

 

[legg]

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?
Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

I think the rule of thumb takes into account a return
path for a straight wire separated by a PCB thickness.

The test jig illustrated has terminals mounted on
0.75 inch centers.

If that is being \'nulled\', then you\'d get a different
measurement, not the full loop.

The rolled film cap construction has a larger width
than a straight wire. This gives the path a reduced
L/W ration, reducing both R and L.

RL

 

[legg]

On Thu, 07 Jul 2022 10:07:49 -0700, John Larkin
<jlarkin@highland_atwork_technology.com> wrote:

<snip>

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

.. . . . . called \'schoopage\'.

We\'ll use lots of these so it\'s good to understand them.

RL

 

[John Larkin]

On Thu, 07 Jul 2022 15:55:04 -0400, legg <legg@nospam.magma.ca> wrote:

On Thu, 07 Jul 2022 10:07:49 -0700, John Larkin
jlarkin@highland_atwork_technology.com> wrote:

snip

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

. . . . . called \'schoopage\'.


We\'ll use lots of these so it\'s good to understand them.

RL

Maybe they manufacture them shorted and use the self-clearing effect
to remove the shorts.

And maybe they test them to arc-over, ditto.

--

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

 

[Phil Hobbs]

legg wrote:

On Thu, 07 Jul 2022 10:07:49 -0700, John Larkin
jlarkin@highland_atwork_technology.com> wrote:

snip

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

. . . . . called \'schoopage\'.


We\'ll use lots of these so it\'s good to understand them.

RL

Fun--named after a Dutch guy named Schoop.

Inquiring minds and all--is that \"scoopadge\" or \"shoopadge\" or
\"showpadge\" or some *tertium quid*?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Phil Hobbs]

legg wrote:

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?
Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.


I think the rule of thumb takes into account a return
path for a straight wire separated by a PCB thickness.

Nah, the 20 nH per inch number is for wire flapping in the breeze, per
the ARRL Handbook et al. It does depend slightly on the gauge.

The test jig illustrated has terminals mounted on
0.75 inch centers.

If that is being \'nulled\', then you\'d get a different
measurement, not the full loop.

The rolled film cap construction has a larger width
than a straight wire. This gives the path a reduced
L/W ration, reducing both R and L.

Yup. Also, while the field outside doesn\'t depend on the radial
distribution of current, the field inside does, and the spiral
construction forces it to be pretty well uniform.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Phil Hobbs]

Phil Hobbs wrote:

legg wrote:
On Thu, 07 Jul 2022 10:07:49 -0700, John Larkin
jlarkin@highland_atwork_technology.com> wrote:

snip

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

. . . . . called \'schoopage\'.


We\'ll use lots of these so it\'s good to understand them.

RL

Fun--named after a Dutch guy named Schoop.

Inquiring minds and all--is that \"scoopadge\" or \"shoopadge\" or
\"showpadge\" or some *tertium quid*?

Never mind, he was Swiss, so I expect that means it\'s roughly
\"scope-age\", with a bit of \"soup\" in the \'o\' sound.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Dimiter_Popoff]

On 7/7/2022 20:07, John Larkin wrote:

On Thu, 07 Jul 2022 07:51:24 -0700, jlarkin@highlandsniptechnology.com
wrote:

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?


Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

The conductive path, end to end, is probably many concentric cylinders
of foil, which I guess could have very low inductance. In my
measurement, the leads themselves seem to dominate L.

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

We\'ll use lots of these so it\'s good to understand them.

I am puzzled where the huge inductance, so nicely visible on your
scope shots, comes from as the layers are all shorted on each
side.

 

[Phil Allison]

John Larkin wrote:
==================

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

** Bet you will, polyester film caps are not rated for tolerating high frequencies with high applied voltages.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?

** I have actually done that.

250VAC* at 10 to 20kHz will smoke nearly any such cap in seconds.

* 50watt audio amp driving cap and 1 to 5mH air core choke in series at resonance - big fun.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm.

** Polyester caps are quite fragile, even at 50/60Hz they will fail over time below rated AC voltage.

The issue is internal trapped air pockets and corona discharge across same.
Class X1 and X2 types are \"double wound \" so act as 2 in series to stop this.

Seems winding in vaccuo is just too damn hard.


...... Phil

 

[John Larkin]

On Fri, 8 Jul 2022 01:41:08 +0300, Dimiter_Popoff <dp@tgi-sci.com>
wrote:

On 7/7/2022 20:07, John Larkin wrote:
On Thu, 07 Jul 2022 07:51:24 -0700, jlarkin@highlandsniptechnology.com
wrote:

On Thu, 7 Jul 2022 10:39:39 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Phil Allison wrote:
John Larkin wrote:
===============

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

I like to measure (and explode) parts. I wonder what will happen if I
put kilovolts into this one. Any guesses?


Says <1nH per mm of body and lead length.

It\'s a bog standard, Ilinois Capacitor: MMR series, metallised polyester.

https://www.farnell.com/datasheets/2900812.pdf


The usual rule-of-thumb for wire is 20 nH per inch (800 pH/mm for the
imperially challenged). However, that cap is 26 mm long, not even
counting the banana connectors, which ought to be above 20 nH.

Cheers

Phil Hobbs

The conductive path, end to end, is probably many concentric cylinders
of foil, which I guess could have very low inductance. In my
measurement, the leads themselves seem to dominate L.

The frequency null computes to 8 nH against the 6.8 uF capacitance.
Things may not be that simple, but the tiny net impedance seems real.

I have a cool LCF program if anyone is interested.

I connected it to my trusty ole (it has one tube!) Kepco HV supply. I
was expecting a kilovolt at least, but at 250 volts it made popping
frying noises and got warm. But IR is still off the scale on my Fluke
DVM and it\'s still a capacitor around its original value.

https://www.dropbox.com/s/n1gp3h7h1v4bzw6/CDE_Guts.jpg?raw=1

It appears to be a metalized film spiral that has big metal (flame
sprayed?) electrodes on both ends.

We\'ll use lots of these so it\'s good to understand them.



I am puzzled where the huge inductance, so nicely visible on your
scope shots, comes from as the layers are all shorted on each
side.

8 nH isn\'t huge. Extending the wire leads up an extra half inch
triples the inductive spikes.

The leads aren\'t even zero length on those Pomona adapters. Inductance
would be even lower if the cap were soldered onto a pcb\'s power pours.

I should try that, just for fun. The caps are plenty good enough for
our power supply uses.

At our voltages, we need films because ceramic caps will lose all
their C.


--

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

 

[John Larkin]

On Thu, 7 Jul 2022 17:28:56 -0700 (PDT), Phil Allison
<pallison49@gmail.com> wrote:

John Larkin wrote:
==================

The best one is this ugly brown radial-lead CDE thing.

https://www.dropbox.com/sh/d2mwz6i6yr0j9sp/AACJZ9IfJEGzdk0WGwsFb0MCa?dl=0

The 8 nH ESL number is shocking, hard to believe,

** Did you read the maker\'s spec?

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

** Bet you will, polyester film caps are not rated for tolerating high frequencies with high applied voltages.

We did some testing today. One of these caps runs cold at 48 VDC and
half an amp RMS ripple curent at 250 KHz. And we\'ll run four in
parallel.

--

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

 

[Phil Allison]

John Larkin wrote:
==============

No, I tested a few sample caps for another engineer, to check his
power supply Spice model and see if we\'ll have a problem with heating
from ripple current at 250 KHz. I\'m better equipped to measure these
things.

** Bet you will, polyester film caps are not rated for tolerating high frequencies with high applied voltages.

We did some testing today. One of these caps runs cold at 48 VDC and
half an amp RMS ripple curent at 250 KHz.

** So the ripple voltage is TINY !!
It\'s the di-electric loss in the film that heats.


....... Phil