[Test-Equipment] Question about series/parallel capacitors

[Richard Knoppow]

----- Original Message ----- 
From: "Barry" <n4buq at knology.net>
To: <test-equipment at mailman.qth.net>; "J. Forster" 
<jfor at quik.com>
Sent: Friday, May 13, 2011 7:15 PM
Subject: Re: [Test-Equipment] Question about series/parallel 
capacitors


>> If the capacitors were perfect (zero leakage), the two 
>> are equal.
>>
>> With real capacitors, option A with a bleeder string to 
>> equalize the
>> capacitor voltages would be preferable.
>>
>> Why don't you buy a single cap to do the job and save a 
>> bunch of
>> trouble.
>> -John
>>
>
> This is an attempt to replace a "standard capacitor" in a 
> General Radio Model 1617A capacitance bridge.  The 
> original capacitor shorted.  I found some 0.5uF, 0.5% caps 
> that are rated at 250V so I figured making an array out of 
> nine of them would get me 0.5uF at 750V at 0.5% (or 
> hopefully better).
>
> I think adding bleeder resistors would upset the balance 
> in this application, right?
>
> I have my sights set on a true GR standard capacitor (a 
> 1409-X); however, I think at best it's only rated at 500V 
> and this bridge can develop 600V of bias so I'm not sure I 
> can use that as a replacement.  I do plan to use it to 
> check the accuracy of my home-made standard.
>
> Thanks!
> Barry - N4BUQ
>
    I really think the use of balancing or equalizing 
resistors is a non-issue here. These are very low leakage 
film caps, mostly equalizing resistors are used with 
seriesed electrolytics to prevent excessive voltage across 
one _when they charge up_, after that the current is fairly 
low. Film caps have such low leakage that I don't think 
charging voltage or static voltage can cause any problems at 
all.
    When putting capacitors in either series of parallel the 
errors in their values are averaged, however, that assumes a 
normal distribution of error and my not be true of a small 
sample, especially if they are all from the same 
manufacturing lot. You must really measure them to see at 
least in what direction the error is. If you have a known 
precise capacitor of the right value, no matter what its 
voltage rating, you can measure to a very high degree of 
accuracy by the substitution method where the cap becomes 
the standard. The overall accuracy depends on the precision 
of the measuring instrument, its accuracy is of no 
importance since that is established at the desired value by 
the comparison standard.
     I think in practice there is no significant difference 
between the two series-parallel arrangements. As a side 
note, it would be interesting to know what is in the 
original GR standard cap. The catalogue shows caps up to 500 
volts and some to 700 for much lower values of capacitance 
than the one in the bridge. It may be that the bridge 
standard is a specially made single stack cap or its 
possible its made up of two seriesed catalogue standard caps 
in a single case. Since the cap is bad you could open the 
case to find out what is in it. The fact that its shorted 
suggests its a single cap since its unlikely that both caps 
of a series would short (but possible).
     The accuracy of the bridge with the replacement cap can 
be established by measuring a known accurate cap. If the 
standard cap is off value the error will be the same on all 
ranges using that cap. The _precision_ of the bridge will 
not be affected.


--
Richard Knoppow
Los Angeles
WB6KBL
dickburk at ix.netcom.com