Message	Date	From	Subject
1097	2007-10-16 15:34:44	Kevin Purcell	Q, Q meters and polyvaricon Q
I've also posted this at a couple of QRP lists but this is a different group. I'd appreciate any ideas. It also related to W7ZOI's most recent Q related notes on his web site. A nice collection of Q measurements made on a range of -2 and -6 (and some -7 and -10) toroids measured by Zack Lau and published in QEX March 1995 p25 and later in "QRP Power" p2-20 is useful to check the range of values if you need an estimate of the Q. Zack's rules of thumb for type powdered iron toroid Q: if you think the Q is low then assume Q=100; if you think the Q is high then assume Q=300. Almost all the measurements fit in between these boundaries. I found this article by browsing in "QRP Power" over the weekend. And that was useful because I'm still thinking about the W8JI article on end-fed half-wave tuner. AA5TB was skeptical of his Q results. Initially I was skeptical of AA5TB (there are a couple of issues there) but now I'm not so sure. <http://www.w8ji.com/end-fed_1_2_wave_matching_system_end%20feed.htm> > I used the same wire gauge, core type and material, same turns > ratio, and a very similar brand new capacitor of the same > construction and style for measurements below. The measurements > below are on equipment certified to national standards. The primary > instrument is a HP-8753E network analyzer, and was verified by a > HP-4191A laboratory standard impedance test set. W8JI gets consistent results using a (calibrated) network analyzer on several caps but his polyvaricon Q is much lower than you would expect from the manufacturers spec (Q > 500 @ 50pF @ 10MHz so say Q ~ 180 @ 180pF assuming Xc decreases for constant loss resistance) and the loss tangent for polyethelene (the dielectric used). And from other peoples experience with PVCs in tuned circuits with toroids. But nothing solid to challenge that measurement with -- perhaps PVCs just suck? Until I read Zack's list and compared the W8JI measured Q for the inductor W8JI Q = 78 for the 4.12uH inductor (28t #24(?) on T-50-2) with Zack's measurements at 7MHz(ish) W1VT Q = 231 for 3.5uH (28t #22 on T-50-2) and W1VT Q = 224 for 5.9uH (35t #24 on T-50-2) There are more of Zack's measurements on a T-50-2 cores and they're in the same ballpark (Q = 200ish). The only caveat is "the Q meter isn't calibrated on a regular basis" but his measurements span 5 pages of measurements taken from his notebook over time. IMHO, it looks like W8JI Q for the inductor is off by factor of 3. That's a large amount. So it makes me wonder about the polyvaricon Q measurement too. Is that of by a similar fraction? I'm not posting this to rag on W8JI (I like his work, postings and his web site) but in this case I suspect a measurement problem or, more likley, an assumption that isn't true (unloaded rather than loaded? can't be that simple). I doubt the network analyzer hardware is the issue! If anyone has a polyvaricon and is willing to measure the Q over it's capacitance range I'd love to hear what results you get too (What is the Q? How does it vary with C? Is there a sliding contact component to the loss (though W8JI doesn't see Q change with power so probably not an issue)?). That measurement of Q versus C is important. Above I extrapolated from Q > 500 @ 50pF @ 10MHz to Q ~ 180 @ 180pF ASSUMING Xc decreases for constant loss resistance. But if the loss increased with the capacitance (i.e. a dielectric loss in parallel with the capacitor) then one might expect the Q to get worse as the square of C or to be 9 times worse at 180pF, a Q of about 50 which would be closer to W8JI's measured result. If anyone has any ideas on this I'd love to hear them. PVCs are "popular" in QRP ATUs so I think its significant issue. 73 Kevin

I've also posted this at a couple of QRP lists but this is a
different group. I'd appreciate any ideas. It also related to W7ZOI's
most recent Q related notes on his web site.

A nice collection of Q measurements made on a range of -2 and -6 (and
some -7 and -10) toroids measured by Zack Lau and published in QEX
March 1995 p25 and later in "QRP Power" p2-20 is useful to check the
range of values if you need an estimate of the Q.

Zack's rules of thumb for type powdered iron toroid Q: if you think
the Q is low then assume Q=100; if you think the Q is high then
assume Q=300. Almost all the measurements fit in between these
boundaries.

I found this article by browsing in "QRP Power" over the weekend. And
that was useful because I'm still thinking about the W8JI article on
end-fed half-wave tuner. AA5TB was skeptical of his Q results.
Initially I was skeptical of AA5TB (there are a couple of issues
there) but now I'm not so sure.

<http://www.w8ji.com/end-fed_1_2_wave_matching_system_end%20feed.htm>

> I used the same wire gauge, core type and material, same turns
> ratio, and a very similar brand new capacitor of the same
> construction and style for measurements below. The measurements
> below are on equipment certified to national standards. The primary
> instrument is a HP-8753E network analyzer, and was verified by a
> HP-4191A laboratory standard impedance test set.

W8JI gets consistent results using a (calibrated) network analyzer on
several caps but his polyvaricon Q is much lower than you would
expect from the manufacturers spec (Q > 500 @ 50pF @ 10MHz so say Q ~
180 @ 180pF assuming Xc decreases for constant loss resistance) and
the loss tangent for polyethelene (the dielectric used). And from
other peoples experience with PVCs in tuned circuits with toroids.
But nothing solid to challenge that measurement with -- perhaps PVCs
just suck?

Until I read Zack's list and compared the W8JI measured Q for the
inductor

W8JI Q = 78 for the 4.12uH inductor (28t #24(?) on T-50-2)

with Zack's measurements at 7MHz(ish)

W1VT Q = 231 for 3.5uH (28t #22 on T-50-2) and
W1VT Q = 224 for 5.9uH (35t #24 on T-50-2)

There are more of Zack's measurements on a T-50-2 cores and they're
in the same ballpark (Q = 200ish). The only caveat is "the Q meter
isn't calibrated on a regular basis" but his measurements span 5
pages of measurements taken from his notebook over time.

IMHO, it looks like W8JI Q for the inductor is off by factor of 3.
That's a large amount. So it makes me wonder about the polyvaricon Q
measurement too. Is that of by a similar fraction?

I'm not posting this to rag on W8JI (I like his work, postings and
his web site) but in this case I suspect a measurement problem or,
more likley, an assumption that isn't true (unloaded rather than
loaded? can't be that simple). I doubt the network analyzer hardware
is the issue!

If anyone has a polyvaricon and is willing to measure the Q over it's
capacitance range I'd love to hear what results you get too (What is
the Q? How does it vary with C? Is there a sliding contact component
to the loss (though W8JI doesn't see Q change with power so probably
not an issue)?).

That measurement of Q versus C is important. Above I extrapolated
from Q > 500 @ 50pF @ 10MHz to Q ~ 180 @ 180pF ASSUMING Xc decreases
for constant loss resistance. But if the loss increased with the
capacitance (i.e. a dielectric loss in parallel with the capacitor)
then one might expect the Q to get worse as the square of C or to be
9 times worse at 180pF, a Q of about 50 which would be closer to
W8JI's measured result.

If anyone has any ideas on this I'd love to hear them. PVCs are
"popular" in QRP ATUs so I think its significant issue.

73 Kevin