Maker Pro
Maker Pro

Capacitor ESR ??

J

Jim Thompson

Jan 1, 1970
0
I'm having trouble tracking down typical Capacitor ESR values for
Aluminum and Tantalum electrolytics.

Can someone point me to a page?

Or some rules of thumb I can use in simulations?

Thanks!

...Jim Thompson
 
J

James Beck

Jan 1, 1970
0
I'm having trouble tracking down typical Capacitor ESR values for
Aluminum and Tantalum electrolytics.

Can someone point me to a page?

Or some rules of thumb I can use in simulations?

Thanks!

...Jim Thompson

I did a quick look at some of our suppliers websites and I think this
has some references to Al Cap ESR in it.

http://www.nichicon.com/english/lib/alminium.pdf

Jim
 
R

Rene Tschaggelar

Jan 1, 1970
0
Jim said:
I'm having trouble tracking down typical Capacitor ESR values for
Aluminum and Tantalum electrolytics.

Can someone point me to a page?

Or some rules of thumb I can use in simulations?

The ESR is a matter of technology.
Have a look at the manufacturer's pages, eg
http://www.epcos.com

Rene
 
A

Andreas Hadler

Jan 1, 1970
0
Jim Thompson said:
I'm having trouble tracking down typical Capacitor ESR values for
Aluminum and Tantalum electrolytics.

Can someone point me to a page?

Just look up in LTspice's capacitor list to see what
voltage/capacity/ESR ranges are viable?

aha
 
J

Jim Thompson

Jan 1, 1970
0
Jim,

Capacitor manufacturers no longer like to state ESR, because it depends
on the frequency of interest. As you may have noticed, they do state
Dissipation Factor (DF) instead. I had to do some poking around a few
weeks ago to find the secret formula to convert DF into ESR, and here's
what I learned.

DF/(2*pi*frequency*capacitance) = R

Assume a data sheet DF of .05% And assume you are doing some work at,
say, 1kHz, and the capacitance is, say, 20uF.

So in this instance:

0.0005/(2*3.1416*1000*.000020 = 3.97 milliohms

Wouldn't that mean you need DF given at a specific frequency so that
you can work back to ESR (which should(?) be fairly constant)?

...Jim Thompson
 
M

Mike Rocket J. Squirrel Elliott

Jan 1, 1970
0
Jim said:
Wouldn't that mean you need DF given at a specific frequency so that
you can work back to ESR (which should(?) be fairly constant)?

At 0Hz, the ESR of a capacitor will be darn high. It drops as one goes
up in frequency until inductance takes over.

When you do find specs for ESR in capacitor data sheets, they specify at
which frequency (usually 120Hz in the US) they are providing the ESR for.

According to
http://www.cooltron.com/component/technical/library_of_capacitor.shtml

7. Equivalent Series Resistance (ESR)

It's the sum of all the internal resistances of a capacitor measured in
Ohms. It includes:

- Resistance due to aluminum oxide thickness
- Resistance due to electrolyte / spacer combination
- Resistance due to materials (Foil length; Tabbing; Lead wires; Ohmic
contact resistance)

The lower the ESR the higher the current carrying ability the capacitor
will have. The amount of heat generated by ripple current depends upon
the ESR of the capacitor.

ESR is both frequency and temperature dependent, increasing either will
cause a reduction in ESR. The ESR is an important parameter in
calculating life expectancy as the power dissipation (internally
generated heat) is directly proportional to its value.
 
J

John Popelish

Jan 1, 1970
0
Mike Rocket J. Squirrel Elliott said:
Jim,

Capacitor manufacturers no longer like to state ESR, because it depends
on the frequency of interest. As you may have noticed, they do state
Dissipation Factor (DF) instead. I had to do some poking around a few
weeks ago to find the secret formula to convert DF into ESR, and here's
what I learned.

DF/(2*pi*frequency*capacitance) = R

Assume a data sheet DF of .05% And assume you are doing some work at,
say, 1kHz, and the capacitance is, say, 20uF.

So in this instance:

0.0005/(2*3.1416*1000*.000020 = 3.97 milliohms

I think dissipation factor includes several factors, one of which is
ESR. The bigger factor at higher frequencies is a fairly constant per
cycle loss. Your formula looks like it is interpreting those per
cycle losses as ESR. Not very useful, I suspect.
 
M

Mike Rocket J. Squirrel Elliott

Jan 1, 1970
0
John said:
I think dissipation factor includes several factors, one of which is
ESR. The bigger factor at higher frequencies is a fairly constant per
cycle loss. Your formula looks like it is interpreting those per
cycle losses as ESR. Not very useful, I suspect.

You may be right. My formula was provided by the engineer at ASC, and
I've also seen it on some online sites as well. I'll bet that the
formula is useful for line-frequency based power supplies, though.
 
Q

qrk

Jan 1, 1970
0
Wouldn't that mean you need DF given at a specific frequency so that
you can work back to ESR (which should(?) be fairly constant)?

...Jim Thompson

Jim, if you tell me what style capacitor (el cheapo, low-esr, ...), I
might be able to measure the Z on our network analyzer. Z data varies
quite a bit depending on capacitor construction. ESR can vary quite a
bit over frequency, perhaps over 100 to 1 for some ceramic caps (as
reported by AVX's SpiCap program.

FYI, AVX <http://www.avxcorp.com/ProdInfo_Listing.asp> has good data
on their ceramic and tantalum caps.
Aloha, Mark
 
L

Leeper

Jan 1, 1970
0
I've tried some of these AO caps, the A700 series, and one
nice thing, unlike alot of tantalums and some niobium, is they
don't go up in flames....ran a 4V version to 14V and held it
there, and then back down, it nearly recovered all its original
characteristics.
 
D

DarkMatter

Jan 1, 1970
0
John Popelish wrote:

Au contraire- if you look at that so-called formula it is the definition
of DF which is the cotangent of the impedance angle of a simple series
R-C equivalent circuit. The DF being small allows the identity between
ratio and tangent function. The small in-phase component of voltage with
current is exactly that fraction of the VA producing dissipated versus
stored energy per cycle and is the equivalent resistance at the
measurement conditions, where resistance is abstracted from being just a
chunk material of finite conductivity to an energy-to-heat conversion
element. It will be non-linear, a functional dependence on temperature,
frequency, and signal level.

Nice catch!
 
J

John Popelish

Jan 1, 1970
0
Fred said:
Au contraire- if you look at that so-called formula it is the definition
of DF which is the cotangent of the impedance angle of a simple series
R-C equivalent circuit. The DF being small allows the identity between
ratio and tangent function. The small in-phase component of voltage with
current is exactly that fraction of the VA producing dissipated versus
stored energy per cycle and is the equivalent resistance at the
measurement conditions, where resistance is abstracted from being just a
chunk material of finite conductivity to an energy-to-heat conversion
element. It will be non-linear, a functional dependence on temperature,
frequency, and signal level.

If you want to interpret all losses as if they were a result of an
actual series resistance, and the total losses are low, the formula is
fine. If you want to know what the parallel losses, the series losses
and the hysterisis losses are, it is not much use. I think the best
way to measure the actual series resistance is to subject the cap to
its series resonant frequency and and measure its impedance.
 
J

Jim Thompson

Jan 1, 1970
0
[snip]
If you want to interpret all losses as if they were a result of an
actual series resistance, and the total losses are low, the formula is
fine. If you want to know what the parallel losses, the series losses
and the hysterisis losses are, it is not much use. I think the best
way to measure the actual series resistance is to subject the cap to
its series resonant frequency and and measure its impedance.

Are you saying that the best model would be a series R-L-C evaluated
at resonance?

...Jim Thompson
 
J

John Popelish

Jan 1, 1970
0
Jim said:
[snip]
If you want to interpret all losses as if they were a result of an
actual series resistance, and the total losses are low, the formula is
fine. If you want to know what the parallel losses, the series losses
and the hysterisis losses are, it is not much use. I think the best
way to measure the actual series resistance is to subject the cap to
its series resonant frequency and and measure its impedance.

Are you saying that the best model would be a series R-L-C evaluated
at resonance?

Yes. I think ESR shows up most clearly under those conditions.
 
M

Mike Engelhardt

Jan 1, 1970
0
Jim,
Are you saying that the best model would be a series
R-L-C evaluated at resonance?

Maybe I should interject here that the equivalent circuit
of an AL electrolytic is really a ladder network:

---/\/\/\---o---/\/\/\---o---/\/\/\---o...
| | |
--+-- --+-- --+--
--+-- --+-- --+--
| | |
------------o------------o------------o...

LTspice's capacitor database just uses 1 R and 1 C
as an quick and effective approximation. But if you
use 2 R's and C's, you can model the phase angle of
the impedance correct within a few degrees over many
decades of freq. Three 3 R's and 3 C's should let
you model more accurately than you can measure with
any component analyzer I know of.

--Mike
 
W

Winfield Hill

Jan 1, 1970
0
Mike Engelhardt wrote...
Maybe I should interject here that the equivalent circuit
of an AL electrolytic is really a ladder network:

---/\/\/\---o---/\/\/\---o---/\/\/\---o...
| | |
--+-- --+-- --+--
--+-- --+-- --+--
| | |
------------o------------o------------o...

LTspice's capacitor database just uses 1 R and 1 C
as an quick and effective approximation. But if you
use 2 R's and C's, you can model the phase angle of
the impedance correct within a few degrees over many
decades of freq. Three 3 R's and 3 C's should let
you model more accurately than you can measure with
any component analyzer I know of.

Do you know of a good way to obtain the values of the
3 R's and 3 C's, given the component-analyzer data?

Thanks,
- Win

whill_at_picovolt-dot-com
 
M

Mike Engelhardt

Jan 1, 1970
0
Win,
Do you know of a good way to obtain the values of the
3 R's and 3 C's, given the component-analyzer data?

For 2 R's and 2 C's, a good initial guess was each C
was half the cap's "DC" capacitance and each R was
equal to the cap's ESR. So my initial guess for 3 R's
and 3 C's is each C is 1/3 the total cap an each R is
equal to the nominal ESR.

--Mike
 
J

John Woodgate

Jan 1, 1970
0
I read in sci.electronics.design that John Popelish <[email protected]>
If you want to know what the parallel losses, the series losses
and the hysterisis losses are, it is not much use.


You can separate parallel and series losses by measuring at more than
one frequency; two if inductance is negligible, three if it isn't. I'd
need to think about hysteresis loss: it isn't something that is normally
significant in capacitors.
I think the best way
to measure the actual series resistance is to subject the cap to its
series resonant frequency and and measure its impedance.

That still gives you a lumped 'equivalent series resistance' figure, and
often not a useful one, because you won't be subjecting the capacitor to
frequencies anywhere near the resonance.
 
J

John Woodgate

Jan 1, 1970
0
I read in sci.electronics.design that Mike Engelhardt
Maybe I should interject here that the equivalent circuit of an AL
electrolytic is really a ladder network:

---/\/\/\---o---/\/\/\---o---/\/\/\---o...
| | |
--+-- --+-- --+--
--+-- --+-- --+--
| | |
------------o------------o------------o...

LTspice's capacitor database just uses 1 R and 1 C as an quick and
effective approximation. But if you use 2 R's and C's, you can model
the phase angle of the impedance correct within a few degrees over many
decades of freq. Three 3 R's and 3 C's should let you model more
accurately than you can measure with any component analyzer I know of.

If you use more than one R and C, you need to add inductors as well.
 
Top