Re: ESR,ESL for common capacitors

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>

Hi Jim,

Cornell Dubilier usually gives this data straight away (one of the few that 
do!):

http://www.cornell-dubilier-dot-com/film/9422000.htm

You can struggle and deduce the numbers from the other manufacturers too, 
but it is not easy to explain all the tricks needed...

In general, the series inductance is 20 to 30nH which is far below what we 
care about unless one is working with high frequency RFI problems.  The 
series resistance is much more of a concern since that is what burns the 
caps up.


It is interesting to find from these numbers what the peak current would be 
if we suddenly and directly shorted a Geek Group cap of 0.15uF, 0.005 
ohms  32nH...

Assuming our sudden short is perfect.  Xl = SQRT(L/C) = 0.461 ohms 
reactive.  The series inductance is by far the dominate current limiter in 
a short.  The frequency is 2.3MHz.  2000/0.461 = 4338 amps peak!!  About 
ten times what the peak current of the cap is rated at.  If one shorts 
these caps over and over, you will eventually damage them.  Shorting 
commercial poly caps is a definite killer!!  A spark gap directly across a 
commercial cap will kill them fast, as many people have found :-(

BTW - Hard shorting the caps at low voltage is a nice way to find 
Lseries.  Just watch the frequency on a scope.  L = 1 / (4 x pi^2 x f^2 x 
C).  Note that you have to keep the lead lengths very short if you are 
measuring small nH.  If you have a great way to short them with very low 
resistance, you may be able to find Rseries too from the ringdown.  But 
that is darn tough with poly caps.

Scott made a chart you may be interested in too of his testing of many caps:

http://hot-streamer-dot-com/TeslaCoils/MMCInfo/dissipation%20factor%20test%20data%205-19-00.xls

more follows...

At 11:29 AM 4/9/2003 -0700, you wrote:
>Does anyone have suitable equivalent series R and L for common TC 
>capacitors (as one might use in Spice, for instance)?
>
>I ran a quick calculation using the dissipation factor for the Ceramite 
>715C type (2% DF -at- 1kHz)..
>
>If you turn C into X and calculate it as DF = R^2/(R^2+X^2) a 2000 pF cap 
>has an ESR of about 11.4K (-at- 1 kHz)..

Rcap = DF / (2 x pi x C x f)

0.02 / (2 x pi x 2000e-12 x 1000) = 1591 ohms.  The real problem here is 
that 2000pF is a very small cap value.  Our poly caps usually have a DF in 
the range of 0.1% at 100kHz:

0.001 / (2 x pi x 150nF x 100000) = 0.0106 ohm


>If DF stays constant with frequency, then R would be proportional to X, 
>and hence 1/F (i.e. at 100 kHz, ESR would be 114 ohms).  I don't think 
>this is necessarily realistic, though...

See the paper at:

http://hot-streamer-dot-com/TeslaCoils/MMCInfo/MMCPower4.html

DF is fairly dependant on frequency.


>This just seems kind of high..

Small value caps will have very high series resistance.  Thus the use of 
NPO and other very low ESR dielectrics in higher power ceramics like ATC makes.

http://www.atceramics-dot-com/pdf/700e.pdf

These puppies will get you back to 0.01 Ohms in 1000pF at a few MHz.  But 
get out your check book$$$$  We are talking a 1000V 1000pF cap that costs 
like $15!!


>Maybe a series resistor isn't a good model, although, a constant series 
>resistor would have increasing DF as frequency goes up, which IS somewhat 
>realistic.

Series R is fine.  Once you know the Fo frequency of a coil, it is a 
constant.  See the ESR I use in my models for primary caps:

http://hot-streamer-dot-com/TeslaCoils/MyCoils/BigCoil/BigCoilSCH.gif

http://hot-streamer-dot-com/TeslaCoils/MyCoils/SmallCoil/SmallCoilSCH.gif


>And, of course, what sort of model can we use for the CD "geekgroup" type 
>caps?

http://hot-streamer-dot-com/TeslaCoils/MMCInfo/GeekCaps/GeekCaps.htm

"ESR" is not the only important factor.  What we really want to know is if 
they will burn up or not.  For that we need to know four things:

1.  How hot can the caps get?  About +10C is fine, +15 is getting "close".

2.  What is the RMS current the caps see?

3.  What is the ESR given the operating frequency?

4.  What is the thermal dissipation of the caps?  For Geek Group caps, 
every Watt of power dissipated raises their temperature 22C.

Also, we have to insure that the peak dV/dT rating is not exceeded (end 
caps burnout).  But with Geek Caps that should never be an issue.

I should mention that Cornell Dubilier typically "lies" about the values 2X 
(power).  So if they say it can take 10 amps, it can probably do 14 just 
fine.  Also, they are constantly "playing" with improving caps.  Recent 
testing of the 0.1uF version shows a very low ESR!  (I still have not 
retested the 0.15s yet Mark ;-))  They can modify dielectric materials, 
film thickness, construction...  in small ways that really increases 
current handling.

Note that many poly caps now go to 105C (decreased voltage) where they used 
to stop at 85C.  They recently figured out how to arrange the poly 
molecules in the film to withstand higher temps without changing anything 
else.  A cool trick which shows there are still many improvements awaiting 
us.  They still have not figured out how to wave solder surface mount 
polypropylene caps :o)))

Cheers,

         Terry