[Hammarlund] Paper caps (long)

[Gottfried Ira]

Bill wrote:
> Now with the web, I suggest pulling down the manufacturers data sheets.
> I strongly suggest this
> because there are many types of ceramics out there with various
> temperature and coefficient ratings.
> There are a host of different profiles out there, e.g. z5u, y5p, y5u,
> y5v, NPO. The better the cap, the more it
> will cost.   

Agreed. That implies that
   - the manufacturer is known and still in (the same) business
   - the datasheet of the specific type is still available and the
     relevant (!) data is provided therein.
Plessey and Roederstein ("Ero") are typical manufactures (in Europe) of good
"modern vintage" tubulars. Both are out of business.
> 
> Saying that you measured a ceramic is not sufficient, you need to report
> the type (material, type, tolerance temperature coeff).  

If I find an unkown ceramic capacitor which exhibits a capacitance change by
more than ~30% at elevated temperatures it is probably a type
Z5.. or Y5.. ceramic which I wouldn't use anywhere in my Hammarlund.
This applies to non-ceramics accordingly.
> 
> Also, loss and capacitance change are two different items.  A bypass
> capacitor
> providing -j0.5 ohms to ground, that changes capacitance by 50% changes
> the impedance either to -j1 or -j0.25.
> Likely not a big deal if it is bypassing a 100 or 1 Kohm resistor.  Now
> if it was a tuned circuit, that
> is a different story - and you would have chosen the wrong part for the
> job.
> 
> As to impedance to ground, it does matter if the it is inductive or
> reactive.

Not necessarily. See below.

> If you are just worried
> about one particular frequency, then you might get away with providing a
> low inductive or capacitive 
> reactance to ground. But, what about the other frequencies, esp DC.
> Providing a low inductive impedance
> to ground at RF, may also imply providing a DC short to ground. I doubt
> you meant that, but you were not clear.

I measured the impedance of capacitors from ~0.1 to 200MHz.

> On the other side, once the reactance of a capacitor turns inductive,
> the reactance increases with frequency.
> Generally this is undesirable unless the frequency that this occurs is
> way outside your frequency range of interest.

You are 100% right. My mention of "capacitive or inductive" bypass was much
too terse and misleading. But generally speaking, even an ideal bypass
capacitor will exhibit a frequency dependant impedance of -j/(omega*C) :-)

An HF radio is a narrow band system. The phase of the bypass
impedance can be seen as constant within a bandwidth of a few kHz.
My (maybe academic?) point is, that the impedance should be as
small as possible, disregarding its phase. It will necessarily
depend on frequency, but should not change with temperature,
time, supply voltage or the signal itself (nonlinearity).
When you look at impedance curves presented in
http://www.oe1ira.at/hc/hammarcapac.html you will find capacitors with
highly damped resonances (2..3 Ohms peak).

Tuned ciruits in boatanchors (HF) exhibit characteristic impedances
of several kOhm. A bypass capacitor (even with a with a flat resonance)
establishes only a small (but frequency varying) impedance in series
to the tuned circuit. The calibration/adjustment procedure will take
care of it, as with all the parasitic components (C,L,R) of the layout.
The engineers at Hammarlund knew their job I am sure!

But as Bill wrote, it is good engineering practice to choose a
capacitor with its first resonance well above your highest frequency
of interest!

> 
> You also need to be very careful with resonances. That cap that just
> started acting like an inductor may form
> a resonant circuit with another cap at some frequency. Even in modern
> systems, it is all to easy to get the 
> decoupling caps to resonate in the power plane, and low and behold the
> power rail oscillates.
> 
Yes! A common source of trouble.
BTW I also measured a tubular (made by Sprague) which appears to be one
of the highly regarded Orange Drops (although 100nF).
Go to http://www.oe1ira.at/hc/hammarcapac.html and scroll down to no.26.
It has multiple resonances between 20 and 80MHz. Could be the
reason why they sound so good... :-)

Happy New Year to All!
Gottfried Ira




-- 
http://www.oe1ira.at  <-- New Address!