[TowerTalk] Power ratings, vendor specs... Line Isolators (was: common mode chokes)

[Jim Lux]

>
> <<snip>>
> > I asked the proprietor of Radio Works about this by email.  After
> > initially replying to suggest that perhaps I'd been sent the wrong
> > model, he ceased responding to my email.
> >
> > I invite anyone owning one of these products to measure its impedance
> > and to report your results on this reflector.
> >
> > Caveat emptor!
> >
> > 73 de Chuck, W1HIS

>From Radio-Works' web page: (http://www.radioworks.com/nbalun.html)
Power Rating

All products made by the RADIO WORKS will handle the legal power limit,
unless it is specifically designed for low power or receiving applications.
Since The RADIO WORKS advocates adherence to the legal power limit, I do not
like to rate components above that level. However, since 2:1 and 3:1 safety
factors are often desirable, the RADIO WORKS does build heavy duty
components.

Rated power assumes an SWR of less than 2:1 unless otherwise noted. The
rated frequency is 3.5 MHz. Duty-cycle is CW or SSB with normal processing.
High duty cycle modes, like RTTY, may over stress a balun and require
improvement in load matching, lowering the power, or switching to a higher
rated balun.


---------------
Some comments:

I wouldn't think that showing component ratings  above the  legal limit (and
designing to accommodate same) would be a problem (except, perhaps for Max
output power on an amplifier, where there are regulations). In fact, as the
author points out, safety factors are desirable.  Certainly, if I were
designing a system to operate at 1.5 kW, I'd want to specify components
rated at somewhat higher levels, but that would be my conservatism for
wanting to accommodate "tolerance stack up", not a desire to run more power.
One hopes that the Radio Works folks specify and buy components that are
only rated to the nameplate design power (or, have some sort of statisical
process control or rigorous testing mechanism in place to allow them to
"design to the edge").

They do specify, after a fashion, the limitations on the power handling.
Unfortunately, they don't give enough data to properly figure out how one
should derate for your application (I don't want to pick on Radio-Works
here... most mfrs are the same).  They rate at 3.5 MHz, but give no
information on how one should adjust the rating for other frequencies... Is
it a linear derating? Is the loss flat DC to Light?  Duty-cycle is specified
as CW or SSB with normal processing, but what does that really mean?
Continuous CW sending with the usual key on/key off ratios, sending random
characters? Or, are they assuming send CQ 3 times, then wait 5 minutes?
Most duty cycle limits have a (thermal) time constant associated with them.
The SWR spec is 2:1, which seems fairly self explanatory, but, this is a two
port device, with a nominal 50 ohm impedance, so, from the specs, one
assumes that one can have a 2:1 mismatch on each port?  They also only
specify the match requirement "looking out" from the interface plane.  As a
system designer, one usually also wants to know what the impedance is
"looking in" to the device.

Some may argue that I'm asking for too much data, and that it's too hard for
an amateur to make the measurements necessary.  To that I reply: nonsense!

here's why:
1) They're selling a product, they're professionals, that's what separates
the "tinkerer in the garage" from the "business", and yes, it costs money to
do it, but again, that's part of being in business.

2) The measurements and analysis can be done with pencil and paper, a
calculator, a DMM, a transmitter that tunes all the relevant frequencies,
and a dozen discrete components (a diode, a few caps and resistors).  An
oscilloscope would make it easy. Sure, it might be a bit tedious, and take
the better part of a day to fully characterize the device, but again, that's
what comes with selling a product, as opposed to publishing an untested
design for all comers.  We're not talking about characterizing active
components over temperature at 94 GHz here.

A mfr able to support a website clearly has enough technical sophistication
to do simple RF engineering calculations, or to hire someone who can.  They
should be able to either measure the performance, or provide calculated
values, and say how they got either.

3) It costs nothing to clearly define the analysis conditions and
assumptions (i.e. duty cycles, time constants, etc. in quantitative,
numerical terms, not fuzzy "typical SSB with no speech processing" kinds of
terms)
---------------
So,  although Chuck points out that Caveat Emptor is the rule of the day, I
think that it's time that we, the customers, say "Caveat Vendor" and start
demanding real numbers backed up by real data. We demand it for tower
structural data (motivated by regulatory incentives, I admit), and the
manufacturers seem to have no trouble providing it. The entire professional
electronics world demands it for components, and the mfrs seem to have no
problem with it (except, perhaps, the paper-only, fabless,  RF/wireless
semiconductor companies selling chipsets).

Jim, W6RMK