Transmitter Output Impedance
 

On May 2, 5:33*am, Wimpie wrote:
Are you familiar with the concept of S-parameters where you determine
impedance by measuring of reflection coefficient?

Exactly how do you determine the s-parameters for a single-port black
box? It is my understanding that an s-parameter analysis requires an
input port and an output port to be able to measure the parameters.
Where is the input port on an RF source?

What you seem to be measuring is the effect of one or more physical
impedance discontinuities existing in an environment of interference.
Is what you are measuring the actual dynamic source impedance? If I
understand correctly what Walter Maxwell is saying is that whatever
combinations of physical impedance discontinuities from which you guys
are reflecting your test signals, it/they are *not the source
impedance* which is a V/I ratio that originates in the source. A V/I
ratio and a physical impedance discontinuity do not yield the same
reflection coefficients for a 2-port device.
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Transmitter Output Impedance
 

Hello Cecil,

On 2 mayo, 14:37, Cecil Moore wrote:
On May 2, 5:33*am, Wimpie wrote:

Are you familiar with the concept of S-parameters where you determine
impedance by measuring of reflection coefficient?

Exactly how do you determine the s-parameters for a single-port black
box? It is my understanding that an s-parameter analysis requires an
input port and an output port to be able to measure the parameters.
Where is the input port on an RF source?

I used a 2 port VNA frequently for antenna measurements, with the
difference that a single port calibration takes less time than a full
two port calibration.


What you seem to be measuring is the effect of one or more physical
impedance discontinuities existing in an environment of interference.
Is what you are measuring the actual dynamic source impedance? If I
understand correctly what Walter Maxwell is saying is that whatever
combinations of physical impedance discontinuities from which you guys
are reflecting your test signals, it/they are *not the source
impedance* which is a V/I ratio that originates in the source. A V/I
ratio and a physical impedance discontinuity do not yield the same
reflection coefficients for a 2-port device.
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

I would recommend you to measure something yourselves, or put it into
a simulation. You will see that it doesn't matter whether you use a
deltaV/deltaI setup (complex values, not magnitudes) or reflection
(time varying phase of VSWR) measurement. Just try to explore other
paths and discover other insights.

It doesn't matter when you are measuring a single port device that
contains a generator in it (as long as your VNA setup is able to
distinguish between the output generated by the single port device and
the reflection towards the VNA).

There is similarity with measuring antenna impedance (single port
measurement) when close to (broadcast) stations. Your antenna is a
generator in that case. You can't use the non-coherent type of VSWR
analyzers as the detector detects the signal from the broadcast
station also. However when using a device with a coherent
(multiplying) detector you can, as the detector doesn't respond to the
output because of the (broadcast) station.

With Tom's HP89410 setup, the injected signal for S11 can be well
within the modulation bandwidth of SSB (that means well within 1 kHz
of the transmitter's carrier frequency).

When you have good understanding of diode detectors, you can even do
it without a VNA by using heterodyning and put your focus on the phase
and amplitude of the beat frequency.

With kind regards,


Wim
PA3DJS
www.tetech.nl
without abc, PM will reach me very likely

Transmitter Output Impedance
 

On May 2, 8:18*am, Wimpie wrote:
I would recommend you to measure something yourselves, ...

Please take a look at the numerous measurements performed by Walter
Maxwell.

Let's take a simple example of the single-port s-parameter results.
There exists a black box with one exposed port. The impedance is
measured to be 50+j0 and s11 is assumed to be zero when driven by a 50
ohm source. There could be a 50 ohm dummy load in the box but there is
not.

Actually, inside the black box is a 1/4WL Z0=100 ohm transmission line
routed to a 200 ohm resistor. s11 is certainly not zero and is
measured, using the 2-port procedure, to be 0.3333.

Which s11 is correct, 0.0000 or 0.3333?
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Transmitter Output Impedance
 

On 2 mayo, 16:17, Cecil Moore wrote:
On May 2, 8:18*am, Wimpie wrote:

I would recommend you to measure something yourselves, ...

Please take a look at the numerous measurements performed by Walter
Maxwell.

Let's take a simple example of the single-port s-parameter results.
There exists a black box with one exposed port. The impedance is
measured to be 50+j0 and s11 is assumed to be zero when driven by a 50
ohm source. There could be a 50 ohm dummy load in the box but there is
not.

Actually, inside the black box is a 1/4WL Z0=100 ohm transmission line
routed to a 200 ohm resistor. s11 is certainly not zero and is
measured, using the 2-port procedure, to be 0.3333.

Which s11 is correct, 0.0000 or 0.3333?
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Hello Cecil,

Assuming that your black box is a 100 Ohms quarter-wave line with 200
Ohms termination (all inside the black box), S11 (50 Ohms based) = 0,
as the input impedance of your black box is 50 ohms. If you like, you
may see your room temperature black box as a source with about –174dBm/
Hz output and an output impedance of 50 Ohms.

Wim

Transmitter Output Impedance
 

On May 2, 9:57*am, Wimpie wrote:
Assuming that your black box is a 100 Ohms quarter-wave line with 200
Ohms termination (all inside the black box), S11 (50 Ohms based) = 0,....

This is what happens when one changes math models in mid-stream. s11
is NOT zero at the mouth of a stub. The impedance looking into a stub
is IEEE definition (1)(B) and cannot have an s11 of zero unless the Z0
of the stub is infinite, which it is not.

Thanks for proving my point. The single-port s11 is completely
different from the dual-port s11 and that is most likely what is
happening with your attempts to measure the source impedance of an RF
amplifier. How can you possibly promote an experimental approach where
s11 changes by an infinite percentage depending on whether one
measures it as a single-port parameter vs a dual-port parameter???
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Transmitter Output Impedance
 

On 2 mayo, 19:38, Cecil Moore wrote:
On May 2, 9:57*am, Wimpie wrote:

Assuming that your black box is a 100 Ohms quarter-wave line with 200
Ohms termination (all inside the black box), S11 (50 Ohms based) = 0,....

This is what happens when one changes math models in mid-stream. s11
is NOT zero at the mouth of a stub. The impedance looking into a stub
is IEEE definition (1)(B) and cannot have an s11 of zero unless the Z0
of the stub is infinite, which it is not.

From your text I didn't conlude it is a stub (but just a series line
section that transforms 50 OHms into 200 Ohms).

Thanks for proving my point. The single-port s11 is completely
different from the dual-port s11 and that is most likely what is
happening with your attempts to measure the source impedance of an RF
amplifier. How can you possibly promote an experimental approach where
s11 changes by an infinite percentage depending on whether one
measures it as a single-port parameter vs a dual-port parameter???

Please explain (or someone else), as I don't understand anything of
the above with regards to a PA. You described a single-port device and
now starts talking about a two-port device.

If you want to prove that for a two port device the impedance
corresponding to S11 may not be equal to the input or output impedance
of port 1, you are right. When S12*S21 isn't zero, the input
impedance depends on the termination of port 2. This is just S-
parameter math, nothing magic.

For the PA case, the active device is the termination for the two-port
matching network, so in that case S11 measurement equals the input/
output impedance of the PA (just a single port measurement). See it as
the output impedance of the active device is connected to port 2 (of
the matching network), and the VNA (or load) is connected to port 1
(that is the output side of the matching network).

The PA reduces to just a single-port network with a source in it
(compare it with the antenna example where a transmitter is in the
vicinity). As long as the injected signal (or slight mismatch
connected to transmission line with increasing length) is small, you
can apply the small signal approach.

As suggested earlier, you may dive into active load pulling:
http://www.focus-microwaves.com/template.php?unique=232

It doesn't matter whether you show some mismatch to a PA, or a perfect
matched load with a small source in it representing the signal that is
reflected towards the amplifier.

--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Kind regards,


Wim
PA3DJS
www.tetech.nl

Transmitter Output Impedance
 

On May 1, 6:01*pm, walt wrote:
On Apr 27, 2:13*pm, Cecil Moore wrote:



On Apr 27, 10:30*am, Wimpie wrote:

Depending *on the frequency resolution of your VSA, the frequency of
the injected signal can be well within 1 kHz of the carrier, so LC
filters in the PA will not distort the measurement. *In case of a 100W
PA and injection of about 100 mW, the difference in wanted signal and
signal to be rejected is 30 dB (not that large).

Would any competent optical physicist suggest that it is valid to
study the conditions associated with interfering coherent light waves
inside an interferometer by introducing an incoherent light source
into the system? Why would any competent RF engineer suggest that the
system source conditions associated with interfering coherent RF waves
can be studied by introducing an incoherent test signal?
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Cecil suggested reading Chapter 19A in Reflections to view the results
of my extensive measurements of the output resistance (impedance) of
RF power amps, but except for Jim and Richard, it appears that the
others have not. Actually, *Chapter19A is an addition to Chapter 19,
which when taken completely will *provide some information that *will
hopefully change some minds concerning the maximum power delivered. It
should be understood that 'maximum' power delivered is that power
delivered with a specified level of drive.

For example, if the drive level is set to deliver a maximum of 100w,
and the pi-network is adjusted to deliver that maximum power into its
load, the source resistance (impedance) will be the (complex)
conjugate of the load impedance. We're not *talking here about the
very maximum power that the amp can deliver, with max drive, max plate
current, etc.

If you review the 19A portion of you will see beyond a doubt that the
conjugate match exists between the output of the pi-network and its
complex load impedance, and that the maximum power delivered at the
drive level that allows only 100w to be delivered as the maximum.

Further review of all the data presented there will also show that the
output resistance of the amp is non-dissipative, while the dissipative
resistance is that between the cathode and plate. The reason the
efficiency of the amps can exceed 50 percent is because the cathode to-
plate resistance is less than the non-dissipative output resistance,
where that R = E/I appearing at the output of the pi-network.

The earlier portion of Chapter 19, that appears in Reflections 2, can
be downloaded from my web page atwww.w2du.com, click on 'Read
Chapters from Reflections 2', and select Chapter 19.

I hope the review of my measured data will clear up some of the
confusion concerning the output resistance (impedance) of the RF power
amp.

Walt, W2DU

When a source is tuned (e.g. through a pi network or any other
matching network between a PA and its load) such that maximum power is
delivered to the load, it's axiomatic that the source impedance is the
complex conjugate of the load. Is there really a need for a whole
chapter for that?

You say, "We're not talking here about the very maximum power that the
amp can deliver, with max drive, max plate current, etc." I beg to
differ. That is EXACTLY what we are talking about. We're talking
about modern amplifiers that would be destroyed if not for protection
circuits, if they were loaded with a load that resulted in maximum
dissipation in the load. We're talking about even old amplifiers with
enough grid drive that they COULD be loaded to higher power output,
but for reasons of wanting the active devices to survive for a
reasonable length of time (or possibly other reasons), are not loaded
so heavily. If you want to exclude such amplifiers from
consideration, then I would hope none would disagree about the
relationship of the load impedance and the source impedance. You
needn't have made any measurements to convince me of that.

Cheers,
Tom

Transmitter Output Impedance
 

On May 2, 1:41*pm, Wimpie wrote:
Please explain (or someone else), as I don't understand anything of
the above with regards to a PA. You described a single-port device and
now starts talking about a two-port device.

If you recognize the example as a two-port device, you correctly
measure an s11 of 0.3333, (100-50)/(100+50). If you happen to overlook
the second port to which the 200 ohm resistor is attached and treat
the example as a single-port device, you measure an s11 of 0.0000,
(50-50)/(50+50). The example has not changed between those two
measurements so which s11 is correct? Doesn't that fact give you pause
to wonder if you are making essentially the same mistake with the PA
measurements?
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK

Transmitter Output Impedance
 

Hello Cecil,

On 2 mayo, 23:17, Cecil Moore wrote:
On May 2, 1:41*pm, Wimpie wrote:

Please explain (or someone else), as I don't understand anything of
the above with regards to a PA. You described a single-port device and
now starts talking about a two-port device.

If you recognize the example as a two-port device, you correctly
measure an s11 of 0.3333, (100-50)/(100+50). If you happen to overlook
the second port to which the 200 ohm resistor is attached and treat
the example as a single-port device, you measure an s11 of 0.0000,
(50-50)/(50+50). The example has not changed between those two
measurements so which s11 is correct?

You want to know the output/input impedance of your black box, this
means you should look into the 100 Ohms line with 200 Ohms
termination. We will see 50 Ohms, no matter the reference impedance of
the VNA. If it was a 75 Ohms VNA, it would read S11 = -0.2. What the
200 Ohms resistor sees is not important if you just want to know the
behavior of the single-port black box.

Doesn't that fact give you pause to wonder if you are making essentially the same mistake with the PA
measurements?

I am very sorry Cecil, but I still don't see the point where the
discussed method may go wrong.

The only thing I could think of is that you have in mind a setup where
the input of the PA is port number 1 (and the output is port number
2 ) and you carry out a full-port measurement. For a full class-A,
AB, non-saturated PA, this may give useful results. When S12*S21
1, S22 will equal the output impedance, otherwise you have to do the
match.

In real world many amplifiers do not behave as a linear system (food
for discussion) and then the two-port setup will fail, as during S22
measurement the input port is terminated with 50 Ohms (so there is no
source that provides output).

Therefore carrying out a single-port measurement with a slightly off-
carrier frequency (to create non-coherence) under required output
conditions, will result in a meaningful output impedance. As
mentioned before, doing a (slow) manual load pull measurement may give
different results because of bias and supply voltage variations.

73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK


With kind regards,


Wim
PA3DJS
www.tetech.nl

Transmitter Output Impedance
 

On May 2, 5:23*pm, Wimpie wrote:
I am very sorry Cecil, but I still don't see the point where the
discussed method may go wrong.

Everyone seems to be charging ahead, willy-nilly, without seeing the
point which is that there are other effects present besides
reflections.

Therefore carrying out a single-port measurement with a slightly off-
carrier frequency (to create non-coherence) under required output
conditions, will result in a meaningful output impedance.

Nope, it won't because virtual impedances don't cause reflections.
Only physical impedance discontinuities cause reflections. The rest of
the redistribution of RF energy is caused by the superposed
interaction between forward and reflected waves, i.e. interference
effects. Most hams do not understand the role of interference in the
redistribution of RF energy. Hope this helps.

http://micro.magnet.fsu.edu/primer/j...ons/index.html

Please pay close attention to the last paragraph. "... when two waves
of equal amplitude and wavelength that are 180-degrees ... out of
phase with each other meet, they are not actually annihilated, ... All
of the photon energy present in these waves must somehow be recovered
or redistributed in a new direction, according to the law of energy
conservation ... Instead, upon meeting, the photons are redistributed
to regions that permit constructive interference, so the effect should
be considered as a redistribution of light waves and photon energy
rather than the spontaneous construction or destruction of light."

You guys are presuming that reflections are the only thing you are
seeing and that is just not true. You are also seeing interference
effects without realizing it so your conclusions are doomed to failure
unless you can differentiate between constructive/destructive
interference and reflected waves. Since there has been no mention of
interference effects, I am forced to conclude that you guys are
ignorant of such effects.
--
73, Cecil, w5dxp.com
"Halitosis is better than no breath at all.", Don, KE6AJH/SK