impedance bridge, which one is better?

[neazoi]

Hello, which one of the two Rf impedance bridges is better?
http://www.neazoi.com/bridge
http://www.qsl.net/va3iul/Homebrew_RF_Circuit_Design_Ideas/RF_Bridge_for_Impedance_Measurements.gif

 

[betwixt]

The SV3ORA one should give better performance if you can manage the construction accurately. The one on qsl.net is easier to build but looks to be a very old design and I doubt you would find exact parts these days. I would guess it's around 40 years old. For low frequencies the qsl.net one should be OK but for anything above say 30MHz use the SVORA design.

Brian.

 

[neazoi]

The SV3ORA one should give better performance if you can manage the construction accurately. The one on qsl.net is easier to build but looks to be a very old design and I doubt you would find exact parts these days. I would guess it's around 40 years old. For low frequencies the qsl.net one should be OK but for anything above say 30MHz use the SVORA design.

Brian.

Will the sv3ora one be fine for lower frequencies as well?
Also I do not know the purpose of the variable capacitors.

 

[betwixt]

It should work down to quite low frequencies, the only thing that happens is it becomes a little less sensitive as the reactance of the 300pF capacitors becomes significant.
The variable capacitors are I would guess made from twisted insulated wire and probably are no bigger than 5 - 6pF. With a high frequency input ( >100MHz) and a good load resistance they can be adjusted (twisted more or less) to give lowest output. They are there to compensate for differences input and output capacitance due to construction differences and component tolerances.

Brian.

 

[neazoi]

It should work down to quite low frequencies, the only thing that happens is it becomes a little less sensitive as the reactance of the 300pF capacitors becomes significant.
The variable capacitors are I would guess made from twisted insulated wire and probably are no bigger than 5 - 6pF. With a high frequency input ( >100MHz) and a good load resistance they can be adjusted (twisted more or less) to give lowest output. They are there to compensate for differences input and output capacitance due to construction differences and component tolerances.

Brian.

Ok so basically, you connect a good 50R known load at z1 and a good 50R identical load at the z2. then you adjust the capacitors for lowest reading (perfect match). Then the bridge is balanced.

If the frequency is significantly changed this setting has to be done again.

Am I getting this right?

 

[betwixt]

Almost.
At low frequencies, provided it's built properly, the match should be good anyway. It's only at higher frequencies it might need a 'tweak' to make it symetrical and once adjusted it should be good at all frequencies.

Brian.

 

[neazoi]

Almost.
At low frequencies, provided it's built properly, the match should be good anyway. It's only at higher frequencies it might need a 'tweak' to make it symetrical and once adjusted it should be good at all frequencies.

Brian.

All right that is clear now,

A thought that I had is that, this adjustment may not be needed if I record the meter reading using two 50R at Z1 and Z2. Then I will try to read the same reading when using an unknown load, to match it.
Even if the bridge is not initially balanced by adjusting the caps, reading the same initial reading means that the unknown load is matched.

Is my thought wrong?

 

[betwixt]

Your thinking is correct. The amount of error voltage should always be the same when the error is the same so you can compare that way.

Don't forget that if you want to make the fine adjustments, you can exchange the input and load ports and adjust for the same readings.

Brian.

 

[neazoi]

hello I have made this. It works ok but I was wondering something:

I am testing it on 10MHz now.
If I put a 50 ohm dummy load at z2, then it reads zero volts. Now, If I put a 50 ohm UHF antenna at z2 (not resonant at 10MHz), it read a value of 200mv, which is expected.

I was wondering why is this happening?
A possible explanation I am thinking is that the UHF antenna is 50R at UHF because it is designed to be resonant at these frequencies. But at 10MHz it is not resonant and its impedance is not 50R at this frequency.

Is my thought correct?

 

[betwixt]

That's correct. The antenna impedance will vary as the frequency feeding it changes. It should be 50 Ohms at it's resonant frequency but will go above or below that as you move away from it. Note that antennas are rarely as specified for that very reason, when the manufacturer quotes 50 Ohms it will only be at one frequency and in the case of wideband antennas (TV antennas for example) it is quite normal for it to be widely off impedance at band edges. In a receiver antenna that doesn't matter too much but in a transmitting antenna you can see why we usualy employ an impedance matching network to maximize the radiation at the desired frequency.

Brian.

 

[neazoi]

Hello I made the bridge and tested it up to VHF and it works as it should. http://neazoi.com/bridge/index.htm
In UHF there is a strange behavior. When nothing connected at Z2 I read some voltage on DC out port. When I connect some load, even a 50R one I read a greater voltage (more negative), but I should read a lower voltage. any load should give a lower reading than a completely open port (full SWR). It does not make sense.
Any thoughts what might be wrong?

 

[betwixt]

The change as the frequency gets higher is because of two factors.
1. small changes in symetry of the construction and minor differences in the components play a larger part n the measurements and
2. the feed and load impedances are more critical. For example, is your load resistor rated to the UHF frequency you are testing? Proper UHF terminating resistors look like plugs with no cable hole, the resistor is carbon composition and entirely inside the plug body.

Brian.

 

[neazoi]

The change as the frequency gets higher is because of two factors.
1. small changes in symetry of the construction and minor differences in the components play a larger part n the measurements and
2. the feed and load impedances are more critical. For example, is your load resistor rated to the UHF frequency you are testing? Proper UHF terminating resistors look like plugs with no cable hole, the resistor is carbon composition and entirely inside the plug body.

Brian.

About no 1, I tried to make the PCB as symmetrical as possible, it is really symmetrical in shape. http://neazoi.com/bridge/DSC08159.JPG
The only thing that worries me as far as concern the PCB are these small pcb tracks that connect to the left and right connectors in the above picture (reference load and test port). I fear if they are forming some kind of microstrip that affects the UHF and up.

As far as concern no 2, I used a narda microwave dedicated SMA load (dc-18GHz) for the reference resistor and the test port.

Should I try to reconstruct the circuit in air-style, with no PCB at all and conventional components instead of SMD?
I also affraid about the two gimmick caps, as I have used standard 4.7pF SMD (non variable).

Any thoughts or ideas would be greatly appreciated, to solve the "mystery"

 

[betwixt]

I suspect the problem is those capacitors. The imbalance produces a greater error as the frequency increases. I would remove them (or at least one of them) and try a twisted wire capcaitor as a replacement. If you repeat your tests and adjust the capacitor by twisting/untwisting the wires you should be able to get a better balance and in theory it will still be OK at lower frequencies too.

To give you some idea of how critical it can be, a few years back I worked in the US for a major telecoms company on the development of cell phone base stations and in the antenna matching units we even had trouble with the uneven thickness of copper in the PCB tracks.

Looked at your web site - very nice! Where in Greece are you based?

Brian.

 

[neazoi]

I suspect the problem is those capacitors. The imbalance produces a greater error as the frequency increases. I would remove them (or at least one of them) and try a twisted wire capcaitor as a replacement. If you repeat your tests and adjust the capacitor by twisting/untwisting the wires you should be able to get a better balance and in theory it will still be OK at lower frequencies too.

To give you some idea of how critical it can be, a few years back I worked in the US for a major telecoms company on the development of cell phone base stations and in the antenna matching units we even had trouble with the uneven thickness of copper in the PCB tracks.

Looked at your web site - very nice! Where in Greece are you based?

Brian.

I did not know the imbalance could cause such a behavior. I thought i would still be able to make measurements with comparison to the 50R known at Z2 and then the unknown device at Z2.
I will try to replace these capacitors with variable types. I have been informed that in the original circuit these capacitors were simple metalic wires that could be allowed to be variably spaced to the GND walls of the enclosure.

In such low capacitance values, even the PCB material could cause imballance.
Do you think I should try smd low capacitance variables (to make the construction more rigid)?

My current location is near Patra City, 200km west of Athens.
Thanks for looking at my website! Any comment is kindly appreciated.

 

[neazoi]

Hello,
I think I corrected the problem by replacing the smd fixed resistors with two smd variable capacitors of 2.5-6pf.

The tuning process I did was this:

I connected two identical coaxial loads (dc-18GHz) to the Z1 and Z2. Then I fed a UHF generator to the RF port. Finally I tuned the two caps for zero voltage (minimum) on the DC out port, which indicates no mismatch.

I have also changed frequencies to determine minimum voltage reading in different frequencies in UHF.
When tuned correctly for UHF lower bands deeded no further tuning.

Do you think my tuning process is correct?

 

[neazoi]

Do you know roughly what king of impedance to expect from the Monitor and DC outputs? short of 10K?
I would like to invert the nagative voltage out os these diodes, using an opamp and knowing the bridge output impedance would be usefull I think.
The diodes used are 1n5712

 

[betwixt]

Sorry - I've been away for a few days and trying to access the Forum using mobile phone with bad reception.

The monitor and output ports are voltage sources so I wouldn't try to load them with a low impedance. If you want to monitor them and invert the signals at the same time I suggest a dual op-amp with high input impedance such as the TL082. This will also allow you to amplify the voltages if it helps. If you connect the TL082 with a dual supply you can measure down to very low power levels at the monitor port.

Brian.

 

[neazoi]

Sorry - I've been away for a few days and trying to access the Forum using mobile phone with bad reception.

The monitor and output ports are voltage sources so I wouldn't try to load them with a low impedance. If you want to monitor them and invert the signals at the same time I suggest a dual op-amp with high input impedance such as the TL082. This will also allow you to amplify the voltages if it helps. If you connect the TL082 with a dual supply you can measure down to very low power levels at the monitor port.

Brian.

What sort of impedance value would you expect? 10K?

 

[betwixt]

The actual output impedance is complex and will vary considerably with frequency, particularly the monitor output. The impedance of the DC output will be more than 10K at low frequencies by virtue of the resistor but in both cases the 1nF capacitor will drop the impedance at high frequencies. What is important is that the outputs are not unduly loaded because they will not provide much current. Adding a load resistance will help the capacitors at the ports discharge but 100K should be low enough. My point was that if you use op-amps, pick a type with a FET input stage so it's input current requirement is very low.

Brian.