Re my previous message, I have downloaded a GR bridge manual from BAMA. It
reports that:

- the bridge measures the series capacitance
- if D is low, the series capacitance almost coincides with the parallel
capacitance
- but if D is high, they differ significantly. A chart is provided to convert
series capacitance into parallel capacitance.

This confirms that, if D is high and if the loss is caused by a parallel
resistance (as it actually is), one must convert the measured capacitance value
using the chart.

73

Tony I0JX

"Antonio Vernucci" wrote in message
. ..
Re my previous message, I have downloaded a GR bridge
manual from BAMA. It reports that:

- the bridge measures the series capacitance
- if D is low, the series capacitance almost coincides
with the parallel capacitance
- but if D is high, they differ significantly. A chart is
provided to convert series capacitance into parallel
capacitance.

This confirms that, if D is high and if the loss is caused
by a parallel resistance (as it actually is), one must
convert the measured capacitance value using the chart.

73

Tony I0JX

I think you are looking at a manual for a later model
bridge. My 650A manual has the formulas but not charts.
I remeasured a bad cap and calculated the parallel
capacitance, series resistance, and parallel resistance.
This is a paper cap rated at 0.02 uf. The values I got a
Cs = 4.8 uf
D = 0.3
Cp = 4.3 uf
Rs = 994 ohms
Rp = 12 kohms

Not a very good cap.

New plastic film caps measure very close to the marked value
and have a D which is below the residual of the bridge
(essentially zero)

While there is an error from the rather high D it is not
significant in terms of this measurement, that is, the value
of the cap measures nearly three times its marked value. I
have not dissected one of these but suspect the winding is
distorted. That would also affect the voltage rating. What I
mean is that the plates of the capacitor are closer together
than originally, probably because of loss of the wax
impregnant. I found other caps in this RX which had high
values so this one is not unique.

I have not measured the caps at RF but I seems like an
interesting project and a practical use for my Boonton
Q-Meter:-)

BTW, I think my math is OK but maybe not.


--

--
Richard Knoppow
Los Angeles
WB6KBL

I think you are looking at a manual for a later model bridge. My 650A
manual has the formulas but not charts.

Yes. However, in the 650A manual there is a similar statement. At page 3, item
9, they say that the bridge measures the series capacitance, and they also give
the formula for calculating the parallel capacitance (that is what we need, as
the leaky capacitors have a resistence in parallel).

I remeasured a bad cap and calculated the parallel capacitance, series
resistance, and parallel resistance. This is a paper cap rated at 0.02 uf. The
values I got a
Cs = 4.8 uf
D = 0.3
Cp = 4.3 uf
Rs = 994 ohms
Rp = 12 kohms

Not a very good cap.

Your Cp/Cs ratio corresponds to that calculated using the formula at page 3.
However the other figures do not tie up with what my spreadsheet gives at 1kHz,
that is:

- for measured Cs= 4.8uF and D=0.3 (that is a reactance / resistance ratio =
3.33), then Rs should be about 10 ohm, rather than 994 ohm

Moreover:
- the series of 4.8uF and 994 ohm would corresponds to Cp= 5,335 pF and Rp= 995
ohm
- the parallel of 4.3uF of 12 kohm would corresponds to Cs= 4.3uF and Rs= 0.1
ohm
I get values close enough to yours only if I set a frequency close to 10 Hz, not
1 kHz (unless I did something wrong).

Anyway, you may measure the parallel resistance of your capacitor with an
ohmeter, and check that you really read a value as low a 12 kohm.

New plastic film caps measure very close to the marked value and have a D
which is below the residual of the bridge (essentially zero)

While there is an error from the rather high D it is not significant in
terms of this measurement, that is, the value of the cap measures nearly three
times its marked value.

why just three times? I would say that the ratio between 4.3uF and 0.02uF is
more than 200

I have not dissected one of these but suspect the winding is
distorted. That would also affect the voltage rating. What I mean is that the
plates of the capacitor are closer together than originally, probably because
of loss of the wax impregnant. I found other caps in this RX which had high
values so this one is not unique.

I have not measured the caps at RF but I seems like an interesting project and
a practical use for my Boonton Q-Meter:-)

BTW, I think my math is OK but maybe not.

73

Tony I0JX - Rome, Italy

"Antonio Vernucci" wrote in message
. ..
I think you are looking at a manual for a later model
bridge. My 650A manual has the formulas but not charts.

Yes. However, in the 650A manual there is a similar
statement. At page 3, item 9, they say that the bridge
measures the series capacitance, and they also give the
formula for calculating the parallel capacitance (that is
what we need, as the leaky capacitors have a resistence in
parallel).

I remeasured a bad cap and calculated the parallel
capacitance, series resistance, and parallel resistance.
This is a paper cap rated at 0.02 uf. The values I got
a
Cs = 4.8 uf
D = 0.3
Cp = 4.3 uf
Rs = 994 ohms
Rp = 12 kohms

Not a very good cap.

Your Cp/Cs ratio corresponds to that calculated using the
formula at page 3. However the other figures do not tie up
with what my spreadsheet gives at 1kHz, that is:

- for measured Cs= 4.8uF and D=0.3 (that is a reactance /
resistance ratio = 3.33), then Rs should be about 10 ohm,
rather than 994 ohm

Moreover:
- the series of 4.8uF and 994 ohm would corresponds to
Cp= 5,335 pF and Rp= 995 ohm
- the parallel of 4.3uF of 12 kohm would corresponds to
Cs= 4.3uF and Rs= 0.1 ohm
I get values close enough to yours only if I set a
frequency close to 10 Hz, not 1 kHz (unless I did
something wrong).

Anyway, you may measure the parallel resistance of your
capacitor with an ohmeter, and check that you really read
a value as low a 12 kohm.

New plastic film caps measure very close to the marked
value and have a D which is below the residual of the
bridge (essentially zero)

While there is an error from the rather high D it is
not significant in terms of this measurement, that is,
the value of the cap measures nearly three times its
marked value.

why just three times? I would say that the ratio between
4.3uF and 0.02uF is more than 200

I have not dissected one of these but suspect the
winding is
distorted. That would also affect the voltage rating.
What I mean is that the plates of the capacitor are
closer together than originally, probably because of loss
of the wax impregnant. I found other caps in this RX
which had high values so this one is not unique.

I have not measured the caps at RF but I seems like an
interesting project and a practical use for my Boonton
Q-Meter:-)

BTW, I think my math is OK but maybe not.

73

Tony I0JX - Rome, Italy
I will recalculate, I may have misplaced a decimal
point.




--
Richard Knoppow
Los Angeles
WB6KBL

"Antonio Vernucci" wrote in message
. ..
I think you are looking at a manual for a later model
bridge. My 650A manual has the formulas but not charts.

Yes. However, in the 650A manual there is a similar
statement. At page 3, item 9, they say that the bridge
measures the series capacitance, and they also give the
formula for calculating the parallel capacitance (that is
what we need, as the leaky capacitors have a resistence in
parallel).

I remeasured a bad cap and calculated the parallel
capacitance, series resistance, and parallel resistance.
This is a paper cap rated at 0.02 uf. The values I got
a
Cs = 4.8 uf
D = 0.3
Cp = 4.3 uf
Rs = 994 ohms
Rp = 12 kohms

Not a very good cap.

Your Cp/Cs ratio corresponds to that calculated using the
formula at page 3. However the other figures do not tie up
with what my spreadsheet gives at 1kHz, that is:

- for measured Cs= 4.8uF and D=0.3 (that is a reactance /
resistance ratio = 3.33), then Rs should be about 10 ohm,
rather than 994 ohm

Moreover:
- the series of 4.8uF and 994 ohm would corresponds to
Cp= 5,335 pF and Rp= 995 ohm
- the parallel of 4.3uF of 12 kohm would corresponds to
Cs= 4.3uF and Rs= 0.1 ohm
I get values close enough to yours only if I set a
frequency close to 10 Hz, not 1 kHz (unless I did
something wrong).

Anyway, you may measure the parallel resistance of your
capacitor with an ohmeter, and check that you really read
a value as low a 12 kohm.

New plastic film caps measure very close to the marked
value and have a D which is below the residual of the
bridge (essentially zero)

While there is an error from the rather high D it is
not significant in terms of this measurement, that is,
the value of the cap measures nearly three times its
marked value.

why just three times? I would say that the ratio between
4.3uF and 0.02uF is more than 200

I have not dissected one of these but suspect the
winding is
distorted. That would also affect the voltage rating.
What I mean is that the plates of the capacitor are
closer together than originally, probably because of loss
of the wax impregnant. I found other caps in this RX
which had high values so this one is not unique.

I have not measured the caps at RF but I seems like an
interesting project and a practical use for my Boonton
Q-Meter:-)

BTW, I think my math is OK but maybe not.

73

Tony I0JX - Rome, Italy
Turns out to be a couple of misplaced decimal points.
First of all I mis-typed, the measured value is 0.048uf, not
4.8uf. Recalculating I get:

C parallel = 0.044 uf
R series = 99.5 ohms
R parallel = 1205 ohms


--
Richard Knoppow
Los Angeles
WB6KBL

"Richard Knoppow" wrote in message
m...

"Antonio Vernucci" wrote in message
. ..
I think you are looking at a manual for a later
model bridge. My 650A manual has the formulas but not
charts.

Yes. However, in the 650A manual there is a similar
statement. At page 3, item 9, they say that the bridge
measures the series capacitance, and they also give the
formula for calculating the parallel capacitance (that is
what we need, as the leaky capacitors have a resistence
in parallel).

I remeasured a bad cap and calculated the parallel
capacitance, series resistance, and parallel resistance.
This is a paper cap rated at 0.02 uf. The values I got
a
Cs = 4.8 uf
D = 0.3
Cp = 4.3 uf
Rs = 994 ohms
Rp = 12 kohms

Not a very good cap.

Your Cp/Cs ratio corresponds to that calculated using the
formula at page 3. However the other figures do not tie
up with what my spreadsheet gives at 1kHz, that is:

- for measured Cs= 4.8uF and D=0.3 (that is a reactance /
resistance ratio = 3.33), then Rs should be about 10 ohm,
rather than 994 ohm

Moreover:
- the series of 4.8uF and 994 ohm would corresponds to
Cp= 5,335 pF and Rp= 995 ohm
- the parallel of 4.3uF of 12 kohm would corresponds to
Cs= 4.3uF and Rs= 0.1 ohm
I get values close enough to yours only if I set a
frequency close to 10 Hz, not 1 kHz (unless I did
something wrong).

Anyway, you may measure the parallel resistance of your
capacitor with an ohmeter, and check that you really read
a value as low a 12 kohm.

New plastic film caps measure very close to the marked
value and have a D which is below the residual of the
bridge (essentially zero)

While there is an error from the rather high D it is
not significant in terms of this measurement, that is,
the value of the cap measures nearly three times its
marked value.

why just three times? I would say that the ratio between
4.3uF and 0.02uF is more than 200

I have not dissected one of these but suspect the
winding is
distorted. That would also affect the voltage rating.
What I mean is that the plates of the capacitor are
closer together than originally, probably because of
loss of the wax impregnant. I found other caps in this
RX which had high values so this one is not unique.

I have not measured the caps at RF but I seems like an
interesting project and a practical use for my Boonton
Q-Meter:-)

BTW, I think my math is OK but maybe not.

73

Tony I0JX - Rome, Italy
Turns out to be a couple of misplaced decimal points.
First of all I mis-typed, the measured value is 0.048uf,
not 4.8uf. Recalculating I get:

C parallel = 0.044 uf
R series = 99.5 ohms
R parallel = 1205 ohms


--
Richard Knoppow
Los Angeles
WB6KBL

Oh, Yikes! I did it again. The correct measured value
of the capacitor is 0.048 uf, D = 0.3

I calculate:

C parallel = 0.044 uf
R (AC) series = 995 ohms
R (AC) parallel = 12050 ohms
Xc, at 1000 hz = 3315 ohms

Someone please check this.

Formulas a

Cp = Cs / 1+D^2

Rs = D/wC where w = 2*pi*f

Rp = (1+D^2)/D^2)*Rs

All measurements and calculations for f = 1000 hz

--
Richard Knoppow
Los Angeles
WB6KBL

Turns out to be a couple of misplaced decimal points. First of all I
mis-typed, the measured value is 0.048uf, not 4.8uf. Recalculating I get:

As a matter of fact a value of 4.8uF seemed real odd to me.

Oh, Yikes! I did it again. The correct measured value of the capacitor is
0.048 uf, D = 0.3

I calculate:

C parallel = 0.044 uf
R (AC) series = 995 ohms
R (AC) parallel = 12050 ohms
Xc, at 1000 hz = 3315 ohms

Someone please check this.

Your calculations seem correct to me (assuming that by Xc you mean the reactance
of Cs and not that of Cp, which is 3,617 ohm).

At this point, one would still have to explain how a capacitor marked 0,02 uF
can grow up to 0,044 uF, that is more than twice its value.

Before formulating hypotheses (e.g. that the plates of the capacitor are closer
together than originally because of loss of the wax impregnan) I would rather
try to reconfirm the measurement results.

Measuring the resistance of the capacitor by means of a plain digital ammeter,
do you obtain a value close enough to 12 kohm? Repeating the measurement on a
different scale, do you obtain similar results? My experience with lossy
capacitors is that the apparent Rp varies quite a lot with the scale.

Also it would be useful to repeat the test with the GR set at a diffierent
frequency (should this be possible).

73

Tony I0JX

"Antonio Vernucci" wrote in message
. ..
Turns out to be a couple of misplaced decimal
points. First of all I mis-typed, the measured value is
0.048uf, not 4.8uf. Recalculating I get:

As a matter of fact a value of 4.8uF seemed real odd to
me.

Oh, Yikes! I did it again. The correct measured value
of the capacitor is 0.048 uf, D = 0.3

I calculate:

C parallel = 0.044 uf
R (AC) series = 995 ohms
R (AC) parallel = 12050 ohms
Xc, at 1000 hz = 3315 ohms

Someone please check this.

Your calculations seem correct to me (assuming that by Xc
you mean the reactance of Cs and not that of Cp, which is
3,617 ohm).

At this point, one would still have to explain how a
capacitor marked 0,02 uF can grow up to 0,044 uF, that is
more than twice its value.

Before formulating hypotheses (e.g. that the plates of
the capacitor are closer together than originally because
of loss of the wax impregnan) I would rather try to
reconfirm the measurement results.

Measuring the resistance of the capacitor by means of a
plain digital ammeter, do you obtain a value close enough
to 12 kohm? Repeating the measurement on a different
scale, do you obtain similar results? My experience with
lossy capacitors is that the apparent Rp varies quite a
lot with the scale.

Also it would be useful to repeat the test with the GR set
at a diffierent frequency (should this be possible).

73

Tony I0JX

There is no DC resistance, that is, its open circuit
for DC but I think there is an AC resistance component in
parallel with the capacitance (have to look this up). The
capacitance definitely measures high as do a couple of other
paper caps from the same RX. Measured on the TEK multimeter
the capacitance measures even higher. I checked the TEK
meter on a General Radio decade capacitor which is known to
be accurate and it measures correctly. There is definitely
something strange here. The hummer in the GR bridge is
definitely on frequency and it shows correct values on both
the GR decade box and on single precision caps. I really
think something has happened to the cap internally. Also, I
am not at all sure of the tolerance of these caps
originally, probably quite a lot on the high side. My Xc
calculation was made for the series cap value.
FWIW, I measured several other old paper caps including
a non-leaking Black Beauty. All were within reason of marked
value, all somewhat higher but nothing like the one in the
original thread. The dissipation factors were high compared
to new film capacitors but a couple of them were probably
still good caps. D ran from a minimum of around 0.03 to
around 0.5. C values were within about 20% of marked value,
all on the high side. Measured value of leaky (I mean the
oil has leaked out) BB caps shows them to usually be on the
low side.
Probably most of the paper caps in the S-40A did not
need to be replaced but I had the caps and it was not a
difficult job. Its difficult to know how much, if any, this
improves the performance. As mentioned two caps were
thoroughly gone, one a dead short and the other completely
open but most of the others were probably still servicable.
It would be interesting to know what the D of these caps was
when they were new.
It would be interesting to know what the new film caps
will be like in fifty years but I probably won't be around
then (but you never know what developments there will be in
medicine).
And, last but not least, thanks for checking my math.



--
Richard Knoppow
Los Angeles
WB6KBL

Richard Knoppow wrote:

It would be interesting to know what the new film caps
will be like in fifty years but I probably won't be around
then (but you never know what developments there will be in
medicine).

I suspect they'll be hanging in there. Some, actually many, of the film
caps from the 1950s are still reliable and I can only guess that modern
ones will be better still...with the exceptions of some that may turn
out to have come from crummy manufacturers.

There is no DC resistance, that is, its open circuit
for DC but I think there is an AC resistance component in parallel with the
capacitance (have to look this up).

Well, in that case I believe that the calculations we did cannot be strictly
valid. Anyway, it is an interesting issue.

By the way I also have an HT-40 and found no need for changing capacitors, as
the leaky capacitors are placed where they make no harm (e.g. screen grid
bypass).

73

Tony I0JX
Rome, Italy