|
I find the VSWR reading on the Diagnostics reading is incorrect for low VSWR. It reads high and is frequency dependent.
On my VSWR=1.0 dummy load, the 20-meter VSWR reading is 1.18, on 30 it is 1.11, on 40 it is 1.07, on 60 it is 1.04 and on 80 it is 1.03.
Similar reading show up on my antenna. Is there some adjustment that can be done to the VSWR circuit to null the reading?
Gary
W9TD
|
Hello Gary
I agree, SWR reads higher and higher as the frequency goes up. I saw this too. I wonder if any theoretical clues exist? Just applying a fudge factor per band seems a little uncomfortable.
On Sun, Oct 1, 2023 at 6:58?PM Gary W9TD < w9td@...> wrote: I find the VSWR reading on the Diagnostics reading is incorrect for low VSWR. It reads high and is frequency dependent.
On my VSWR=1.0 dummy load, the 20-meter VSWR reading is 1.18, on 30 it is 1.11, on 40 it is 1.07, on 60 it is 1.04 and on 80 it is 1.03.
Similar reading show up on my antenna. Is there some adjustment that can be done to the VSWR circuit to null the reading?
Gary
W9TD
|
Hans,
If I were to guess, it would be due to the diode threshold and power levels at higher frequencies.? The VSWR is a ratio, so the value will be off if the power levels get close to the diode threshold or the ADC measurement point.? Can you display the raw values of the detector inputs?? That might give a clue.
73
Evan
AC9TU
|
The SWR gets worse as the band goes up, 17m, 15m and 10m.
Tandem bridge tends to have the coupling coefficient increase and the directivity decrease with frequency, and it might have something to do with this. It is one thing I noted for myself to revisit later but as long as the SWR is below 2 it is not my priority concern at this point...
|
I might add that my SWR reading is 1.17 on 20m, and 1.48 on 10m (17/15m in between, monotonically) on diagnostics screen when my dummy load is flat all the way up to UHF.
|
I should also have added that the increased coupling and decreased directivity with increasing frequency is primarily due to capacitive coupling between the windings. You see versions using a short section of RG316-type thin coax cable (only one side is grounded, leaving the other open... this is to make a shield not a transmission line) with a large ferrite core to mitigate this problem. It may not be too practical to do so in QMX.
|
Hi,
as it is meant to be a VSWR indication rather than measurement maybe it would be enough to use a simple linear equation to compensate for the frequency error.
Something like that:
y = -0.2 - k * x
y - VSWR correction
x - freq
73 Bojan S53DZ
|
Hans,
From what I have read about the history of the bridge, adjusting the nominal 50-Ohm resistors can balance the bridge. For real accuracy at low power a compensated diode is needed, i.e. using a diode in the feedback path of an op-amp to linearize the detector diode response at low power levels. You probably can do this in software with a look-up table or an equation.
Gary
W9TD
|
Another suggestion would be to adjust the reverse power input voltage divider to be twice the value of the forward power and divide the value by two in the software.? You might need to clamp the max voltage on the reverse input with a Zener to protect it.
The above assumes that the current values are at the last few bits of the ADC.
73
Evan
AC9TU
|
I don't see a voltage divider in QMX schematic. Maybe it is in a different rig. But the SWR reading error problem is in the RF couplers and not the A/D quantization error.
It is also not a bridge balancing issue. Tandem bridge works by adding and subtracting the RF voltage and RF current. Forward and reflected waves have the same sign in voltage but opposite in current. The resistors are there to do the addition and subtraction.
|
I think it would be more indicative if the first vertical bar for PWR and SWR would be fixed and would consequently mean 0W and 1:1 respectively.
73 Bojan S53DZ
|
On Sun, Oct 1, 2023 at 06:49 PM, Bojan Naglic wrote:
Hi,
as it is meant to be a VSWR indication rather than measurement maybe it would be enough to use a simple linear equation to compensate for the frequency error.
Something like that:
y = -0.2 - k * x
y - VSWR correction
x - freq
73 Bojan S53DZ
This is a well known problem with SWR measurement at QRP power levels.? There is a very useful information in Mike, G8GYW's article on GitHub (URL https://g8gyw.github.io/). There's also a good deal of info in David Stockton, G4ZNQ's article in "Sprat" (URL https://www.sm7ucz.se/Meters/Stockton_pwr_meter.pdf) and detailed description at K5JCA (URL https://k6jca.blogspot.com/2015/01/notes-on-directional-couplers-for-hf.html).
Mike, G8GYW explains in detail about how he solved the problem of the the low output of the reverse detector for well matched loads.? Basically, he found that the output followed a quadratic equation of the form y = a(x.x) + bx, with the parameters of a = 1.0 and b = 0.75.? (that is x.x is x squared)
For the unit that I built, I found that the quadratic formula had different parameters to Mikes design, with a value of a = 1.02 and b = 0.42. (In this case, y gives the power in watts.)? The difference may be because I used different Schottky diodes.
I calibrated the unit using a string of silicon diodes from the 12V? supply with my QDX so I could vary the power output over 7 stages, measuring the power in a dummy load.
I found the sensitivity of the reflectometer was essentially constant between 3.5 to 14MHz.? I did use screened primary "windings" for the tandem bridge using short lengths of thin coax (screen grounded at one end only) which may have helped to achieve this.
? -- Peter Lee G3SPL
|
Below is a picture of the tandem bridge reflectometer that I built.? (OK, I admit it is not pretty!)? It's built in a tin box (obtained from Amazon in a twenty odd pack quite reasonably priced).? The advantage of using a tin box is that the tinplate acts as a good ground plane that you can solder to.? My reflectometer is used in my remote QDX setup: the forward and reverse readings go into an Arduino Uno that telemeters the data back to a PC in the shack.? The Arduino also provides remote control, like antenna selection and power supply voltage switching.
--
Peter Lee
G3SPL
|
You could use AD8310 or 8319 or others in that series to save you some trouble... though those chips are a bit expensive.
|
I read G8GYW's github page now but it talks about linearity problem, not the frequency response. The latter is the problem here.
Your T1 uses a coax as the single turn, with the braid grounded on one side, so your SWR bridge is largely free of this problem in HF. In QMX, it's not built that way (and there is no space to do that).
|
Ryuji,
On close inspection of my dead QMX, I could get an RG317 through the center.? I was careful to wind the secondary close and tight.
Is there a better small coax that could be used?
73
Evan
AC9TU
|
On further thought, the small error is not worth the effort.? It is to validate, tune, and protect.? Not to be used for exacting measurement accuracy.??
73
Evan
AC9TU
|
Evan, FWIW if you have a salvage WiFi antenna from an old router/AP, it probably has some 50 ohm micro coax inside.
JZ
On Mon, Oct 2, 2023, 9:19 AM Evan Hand < elhandjr@...> wrote: On further thought, the small error is not worth the effort.? It is to validate, tune, and protect.? Not to be used for exacting measurement accuracy.??
73
Evan
AC9TU
|
There is a issue also with single core vs separate cores for that bridge.
Generally while the binocular core results in a useful sensor its precision
is lower and its ultimate directional resolution is limited.? Part of the error is
a result of magnetic leakage? between the elements along with general
construction.? In this case the difference between 1.18:1 and 1:1 is
insignificant enough that the error is not a limitation.??
Making the 1 turn coax is possible, make sure the wire is heavy enough
for the power (current).? Using two cores might be better.? Some -43 tubes
are available that would be close in size to the each half of the binocular
core.?
Also using diodes near their conduction threshold is ripe with error.
This is more significant on the reflected leg due to lower RF voltages.
That and I'll bet most antennas and tuners cannot match that closely.
We are talking pointer width fudge factor.
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
I agree Evan, I think it's not a very significant issue and it's still useful to adjust the antenna or the tuner or detect an issue. I guess we just need to keep seeing the same reports coming up over and over.
A very thin coax could go through there but we still have to trim the coax in a very short length accurately without shorting one hair of the braid, and then have a ground pad on the board for the braid. There are probably 1:10 turn ratio coupler as components we can buy, though that may be expensive for a 95USD kit.
If I were dealing with 1.5kW I would be obsessed about SWR because of risk of overheating something, but at 5W level, anything below 2 is fine, and if the meter shows 2.5 or something, I wouldn't worry about it. In most portable operation situations, radiation efficiency and ground loss are a much bigger factor than the mismatch loss anyway.
What I would rather want instead of an instrumentation level accuracy of SWR reading is a tiny character based band scope of SWR at a very reduced power output so that I know the antenna is too long or short, which way to turn the manual tuner dial, etc.
|
Gary W9TD
Ryuji Suzuki AB1WX