Maker Pro
Maker Pro

Measuring IP3

J

J M Noeding

Jan 1, 1970
0
Hi
It seems to be some problem to build equipment to measure over
+35...40dBm IP3, and xtal filters seem to cause intermodulation
problem with this level

What I am after is constructions to build which works to measure
higher IP3. Would also be interested to see IP3 measurements for xtal
filters

73, Jan-Martin
 
J

Jim Thompson

Jan 1, 1970
0
Hi
It seems to be some problem to build equipment to measure over
+35...40dBm IP3, and xtal filters seem to cause intermodulation
problem with this level

What I am after is constructions to build which works to measure
higher IP3. Would also be interested to see IP3 measurements for xtal
filters

73, Jan-Martin

Padded-down spectrum analyser ??

36 to 63V P-P into 50 ohms ain't easy ;-)

...Jim Thompson
 
J

J M Noeding

Jan 1, 1970
0
Padded-down spectrum analyser ??

36 to 63V P-P into 50 ohms ain't easy ;-)

...Jim Thompson
it is sort of calculation to find the IP3, you could measure 1dB
compression or just the IM

jm
 
T

Tim Wescott

Jan 1, 1970
0
Jim said:
Padded-down spectrum analyser ??

36 to 63V P-P into 50 ohms ain't easy ;-)

...Jim Thompson

IM3 is generally projected up from measurements made at much lower power
levels -- on most RF circuits running at useful power levels the
3rd-order IM products vary as the cube of power, so you measure the
power-in/IM3-out levels at lower input levels and plot it all on a graph.

Not that I can answer the OP's question, and padding down the signal is
probably going to be a featured part of any setup.
 
J

Jim Thompson

Jan 1, 1970
0
it is sort of calculation to find the IP3, you could measure 1dB
compression or just the IM

jm

True. Most systems will extrapolate out to IP3 pretty well, so take a
series of data points at lower power levels.

...Jim Thompson
 
J

J M Noeding

Jan 1, 1970
0
True. Most systems will extrapolate out to IP3 pretty well, so take a
series of data points at lower power levels.

...Jim Thompson
Another poster has the same opinion, too

then, back to original question:

It seems to be some problem to build equipment to measure over
+35...40dBm IP3 (at receiver input), do you know how it could be made,
or is it some references which might work?

jm
 
J

J M Noeding

Jan 1, 1970
0
True. Most systems will extrapolate out to IP3 pretty well, so take a
series of data points at lower power levels.

...Jim Thompson
Another poster has the same opinion, too

then, back to original question:

It seems to be some problem to build equipment to measure over
+35...40dBm IP3 (at receiver input), do you know how it could be made,
or is it some references which might work?

7025 and 7050kHz tone frequencies would do

jm
 
T

Tim Wescott

Jan 1, 1970
0
J said:
Another poster has the same opinion, too

then, back to original question:

It seems to be some problem to build equipment to measure over
+35...40dBm IP3 (at receiver input), do you know how it could be made,
or is it some references which might work?

7025 and 7050kHz tone frequencies would do

jm

Well, let's see. You want the 3rd-order product to be above the noise
floor of the receiver. Assuming a 10dB noise figure (you _will_ do
better, yes?) and a 5kHz wide filter a -120dBm signal will be just
discernible -- say you want it to be at least -100dBm.

With a projected intercept of 40dBm that gives you a 140dB difference;
dividing that by 3 gives around 47dB down from 40dBm, which means that
you need to feed about 7dBm into the receiver from each source (this is
sounding pretty absurd, I should just stop now).

Oh well, persevere:

I think the issue is that you need to have some very stout sources, and
a damn good hybrid combiner.

Not only do you need to make sure that each source's output won't bleed
into the output stage of the other source through the radio connection
and intermodulate there, you also need to make sure that note of the
energy from source A gets into source B and visa versa.

The "amateur" approach would be to make two sources each capable of
generating well over 7dBm (1W?) so that you can protect their outputs
with attenuators. If I were doing this as a garage effort I'd make two
identical crystal controlled sources, well shielded and operated from
batteries (rechargables, if I'm going to be producing a frigging watt each).

I have no idea what to tell you about the hybrid combiner, other than
you should pay attention to 3IM in the core, and test it.

Assuming that you have two good sources and attenuators, and a really
stout hybrid combiner, you should be able to get clean, high-level RF
into your receiver. At this point I'd take your RF (at what, 10 or
20dBm?), attenuate the heck out of it, and check the 3IM products. If
you get a 3IM product that tracks the attenuation linearly then your
sources or your combiner are at fault. If they track 3x attenuation
then congratulations -- you're actually measuring your receiver!

Good luck. Maybe I'll stick to NE602's -- the measurements are easier.
 
J

J M Noeding

Jan 1, 1970
0
Well, let's see. You want the 3rd-order product to be above the noise
floor of the receiver. Assuming a 10dB noise figure (you _will_ do
better, yes?) and a 5kHz wide filter a -120dBm signal will be just
discernible -- say you want it to be at least -100dBm.

With a projected intercept of 40dBm that gives you a 140dB difference;
dividing that by 3 gives around 47dB down from 40dBm, which means that
you need to feed about 7dBm into the receiver from each source (this is
sounding pretty absurd, I should just stop now).

Oh well, persevere:

I think the issue is that you need to have some very stout sources, and
a damn good hybrid combiner.

Not only do you need to make sure that each source's output won't bleed
into the output stage of the other source through the radio connection
and intermodulate there, you also need to make sure that note of the
energy from source A gets into source B and visa versa.

The "amateur" approach would be to make two sources each capable of
generating well over 7dBm (1W?) so that you can protect their outputs
with attenuators. If I were doing this as a garage effort I'd make two
identical crystal controlled sources, well shielded and operated from
batteries (rechargables, if I'm going to be producing a frigging watt each).

I have no idea what to tell you about the hybrid combiner, other than
you should pay attention to 3IM in the core, and test it.

Assuming that you have two good sources and attenuators, and a really
stout hybrid combiner, you should be able to get clean, high-level RF
into your receiver. At this point I'd take your RF (at what, 10 or
20dBm?), attenuate the heck out of it, and check the 3IM products. If
you get a 3IM product that tracks the attenuation linearly then your
sources or your combiner are at fault. If they track 3x attenuation
then congratulations -- you're actually measuring your receiver!

Good luck. Maybe I'll stick to NE602's -- the measurements are easier.

It seems this is your field, while I've fallen off where math counts,
hi. Agree that it seems somewhat absurd, just wonder how it is done to
measure above 40dBm on the mixer input. Heard from some writers that
xtal filters has an IP3 of 40dBm, andof the two 18.5MHz xtal filters
tested, one homebrew and the other commercially made, the homebrew
appeared to be the best, but the commercial filter had some rather
small coils while the homebrew uses T37-2 iron dust cores with much
less field. See notes about the Redifon mixer and the homebrew crystal
filter on http://www.noding.com/la8ak/12345/n16.htm
The last mixer uses 4x HSMS-2802 and 3x 74HCU04

With an output of something near -6dBm from an U L Rohde type low
noise xtal oscillator to an ultra linear amplifier, with say 20dB
gain, it is somewhat low.
The constructions I have seen for measuring lower IP3 than in this
case uses 20dB attenuator after the oscillator/amplifier to feed the
combiner. So you need at least +33dBm from the linear amplifier.
He tried a E280F valve amplifier, but the screening was insufficient
using copper clad board on 3.7MHz, and had to stick to a solidstate
design, but now it is not satisfying.

Just wonder how they measured above +40dBm? Perhaps you need some
specially designed and proven HP ultra low-noise oscillators, which
nobody of us ever have heard about?

73,
Jan-Martin
 
Top