50Ohm 11W Dummy Load with an attenuator.

[OnurCan]

Hey guys,

I am doing a dummy load project for my Radio Frequency lecture and I am going to need help for this project.

I first started with 11 parallel resistors design, each resistor is 560 ohms and 1 watt. There is a must in this project and that is the signal strenght should not exceed -10 dBm so I need to place an attenuator to my design here.

If I can just do a normal attenuator with 50 ohm it will be fixed

To make it consistent over frequency, I need to make sure the PCB has a characteristic impedance of 50 ohm from the connector to the load, and that the resistors themselves are capable of being resistive up to my top frequency which is 1kHz. The resistors I used are surface mounted resistors and they can only handle 1W of power so there comes the problem, so I need to make a parallel or series connection with multiple resistors, this is a design challenge for me and I am open for your professional ideas to solve this dummy load and attenuator problem.

Thank you four time.

 

[JimB]

OK,

First a few questions:

1 What is your signal source?
A transmitter, what power output.
A simple oscillator, some kind of test equipment?

2 What is the frequency?
You say the top frequency is 1kHz, that is hardly RF. (OK, it could be but you will have a very difficult job to make it radiate).

We need more information before a sensible answer can be given.

JimB

 

[OnurCan]

I am sorry for being unclear, this is my first thread here and Radio Frequency technologies are my weak spot.

My signal source is going to be a signal generator, i will generate the desired signal from there and I will analyze it with an antenna analyzer.
But you are definitely right 1kHz is not rf I just wrote it wrong my top frequency is 1GHz.

This is the given spesification for my project.
'Design and build a dummy load with 50 ohm impedance for the interval of 100-1000 MHz. The load shall have a power rating of at least 5 W in continuous mode, and have a measurement connector to connect to instruments. The signal strength in this connector should not exceed -10 dBm'


And due to Covid-19 it is not possible for me to go to University and test my device so unfortunately it is has to be just a theoretical project.

 

[JimB]

OK, all understood.

Give me a few hours to have a think and I will get back to you.

I will analyze it with an antenna analyzer.

Are you sure about that?
Do you mean a spectrum analyser?

JimB

 

[OnurCan]

Thank you a lot for your time.

And yes i mean spectrum analyser.

 

[JimB]

You have designed a 50 Ohm dummy load.
That is OK as far as it goes, and would be OK to provide a load in order to test a low power transmitter.
But, it would be a bad thing to connect some kind of test equipment or attenuator to the "Output connector"
Most RF test equipment is designed to have a 50 Ohm input, so if you connected a 50 Ohm attenuator to your 50 Ohm dummy load, electrically the two would be in parallel, and the load presented to the transmitter would be 25 Ohm.

You could make an attenuator which had a high impedance input and a standard 50 Ohm output. Connect the high impedance side to the dummy load and the 50 Ohm side to the test equipment.
The problem with that is that in practice, things with high impedance can have very poor frequency response due to small amounts of stray capacitance.

Re-thinking the problem.
Your load must be able to dissipate 5 watts, that is +37dBm.
The output to the test equipment must be no more than -10dBm.
So you need a total attenuation of 47dB.

To make the numbers easy, call it 50dB of attenuation.

50dB attenuators that work up to 1GHz don't exist, because you need high value resistors and the stray capacitance around those high value resistors makes a mess of the frequency response.
It is better to make two 20dB and one 10dB attenuator.

Design one of the 20dB attenuators so that it can dissipate at least 5 watts.
The output of that attenuator will be only 50mW, so the other 20dB and the 10 dB attenuators only need to be built with single resistors, rather than the multiple resistors needed to dissipate the 5W.

OK, that is enough for now, does this help?

JimB

 

[RadioRon]

Further to Jim's advice, most attenuator circuits use a Pi or Tee configuration of resistors to achieve transmission line matching at input and output, as well as a specific attenuation value. There are many online calculators for determining the resistor values needed. Here is one that I came across:

Matching Pi Attenuator Calculator

This will give you the resistor values needed. In order to achieve correct results, the physical construction will have to be high quality and carefully done to maintain transmission line impedance and to insure isolation from input to output. Jim suggests two 20dB and one 10 dB attenuators cascaded to form a 50 dB attenuator. This is good advice because output to input isolation can be reasonably achieved for 20 dB but is nearly impossible for 50 dB per stage. The first attenuator will have to handle the majority of the power, so you will probably find it necessary to use three 5W resistors. The power to the second attenuator will be down to 0.05 Watts so the resistors for the follow-on attenuators can be small ones. Each stage, that is, the 20, 20 and 10 dB attenuators should be separately sheilded. In practice, we often buy or build these attenuators with coax connectors into and out of each stage so that we can use the individual stages or cascade them as we see fit.

The heat from each 5W resistor has to be dissipated somehow to the outside air, so consider this to be a significant problem in this project.

 

[JimB]

Well said Ron.
I was anticipating saying most of that in my next post on the subject.

This is good advice because output to input isolation can be reasonably achieved for 20 dB but is nearly impossible for 50 dB per stage.

I had a quick look at a cook book table of resistor values for attenuators,
for a 50dB attenuator,
when built in a pi configuration the top resistor has to be 7905 Ohms,
and when built as a Tee, the resistor in the stalk of the Tee has to be 0.32 Ohms
Both cases present their own set of difficulties.

For a 20dB attenuator, those resistors become 247 and 10 Ohms respectively.
Much easier.

JimB

 

[OnurCan]

I appreciate all the answers. I learn a lot about the subject.
But I have to ask;
Designing just one attenuator and putting it together with the dummy load is fine but how can I add the other two attenuators with single resistors to my circuit?

And since I can not use 5W resistors I am thinking about making them with parallel or series connection with my 1W resistors, would that be okay?

 

[JimB]

how can I add the other two attenuators with single resistors to my circuit?

Like this:

Just build three attenuators, the first one with multiple resistors to be able to dissipate the 5W, the second two attenuators can be built with single surface mount resistors.
Connect them all in series.
Connect the transmitter at one end, the test equipment at the other end.
Job done!

JimB

 

[RadioRon]

I want to comment on jargon, for a moment. We are mixing the terms Dummy Load and Attenuator. Normally, a dummy load is a one-port device. That is, it usually has only an input, and its only function is to turn all input power into heat and not radiate anything. An attenuator's job, on the other hand, is to reduce power by an accurate fixed ratio, while maintaining correct impedance matching at the input and output. Now, in this particular project, we seem to be talking about a combination of the two concepts, where we have a dummy load that happens to have an instrument connection port. Theoretically, this is exactly the same as an attenuator, but practically, we could simplify the internal circuit if we allowed some leeway in the specifications. For example, there is no requirement to deliver exactly -10 dBm output with a 5W input. I would think that any attenuation value 50 dB or more up to a practical limit of, say, 80 dB would be acceptable. I also did not see any SWR specification, so perhaps it would be acceptable to compromise somewhat to simplify the internals.

This leads us to consider perhaps making a dummy load with a simplified output tap in the design. In other words, let's say, that we build the dummy load part using the eleven parallel 1W resistors as originally described. Then we add an instrument tap consisting of a high value resistor linking from the dummy load to a follow-on L (L shaped, not an inductor) circuit. The mismatch of addding this tap would add a very small mismatch to the input if the high value resistor were, say, 1000 ohms or more. And you could arrange the metering port to have a 50 ohm shunt resistor to establish a decent impedance match at the output. Then between the 1000 ohm series resistor and the 50 ohm shunt resistor, add one or two more resistors to arrange a suitable amount of attenuation. One could even be sloppy with these and still achieve the goal.

 

[JimB]

Ron,
I had considered that, but the specification is that the thing works from 100 to 1000MHz.
The high impedance tapping could have a problem with its frequency response.

JimB

 

[RadioRon]

Yes, I see your point, Jim. I envision implementing the high impedance tap using a cascade of individually shielded series-shunt resistor pairs made with 0603 surface mount resistors. At 1 GHz, I think that this size of resistor and a good layout would not corrupt the frequency response very much. And it may be worth noting that there is no specification for frequency response other than that the output not deliver more than -10 dBm. So if the response was lumpy, it would still be functional, as long as the lumpiness were calibrated out.

I would start by leaving out one of the input shunt 560 ohm 1 Watt resistors, giving me an input shunt R of 56 ohms, then follow with a series R of about 560 ohms (could also be 1 Watt if convenient) followed by a shunt 56 ohms (0.1 watt 0603). This would deliver the first -20 dB and deal with the power dissipation.

I realize that this is simply a Pi attenuator, even though it sounded a bit different in my head when I typed my previous post.

What do you think will happen to the frequency response with this configuration?

 

[unclejed613]

if you use Tee or Pi attenuators, you can find a handy calculator here: https://www.rfcafe.com/references/electrical/attenuators.htm

when i was in the Army, many calibration procedures require measuring power into a dummy load and monitoring the signal with a spectrum analyzer... we often used not a plain Tee connection for a tap, but a directional coupler with a known attenuation and 50 ohm impedance at the measurement port. some directional couplers DO have a very wide frequency range, and as long as they are properly terminated, have a very flat response...

this might also be helpful https://www.minicircuits.com/app/AN30-001.pdf

 

[JimB]

I would start by leaving out one of the input shunt 560 ohm 1 Watt resistors, giving me an input shunt R of 56 ohms, then follow with a series R of about 560 ohms (could also be 1 Watt if convenient) followed by a shunt 56 ohms (0.1 watt 0603). This would deliver the first -20 dB and deal with the power dissipation.

I realize that this is simply a Pi attenuator, even though it sounded a bit different in my head when I typed my previous post.

Ron, I think that you just re-invented to 20dB high power attenuator!

What do you think will happen to the frequency response with this configuration?

It will probably be OK.

JimB

 

[RadioRon]

For best frequency response, the high power resistors should be surface mount too, like this:
**broken link removed**

 

[OnurCan]

Thank you all for your time and answers I really appeciate it and thanks to you I finished my project today.

Onur.

 

[JimB]

That's good to hear, thank you for the feedback.

JimB

 

[johnbeck7799]

I am sorry for being unclear, this is my first thread here and Radio Frequency technologies are my weak spot.

My signal source is going to be a signal generator, i will generate the desired signal from there and I will analyze it with an antenna analyzer.
But you are definitely right 1kHz is not rf I just wrote it wrong my top frequency is 1GHz.

This is the given spesification for my project.
'Design and build a dummy load with 50 ohm impedance for the interval of 100-1000 MHz. The load shall here have a power rating for of at least 5 W in continuous mode, and have a measurement connector for fortnite merch to connect to instruments. The signal strength in this connector should not exceed -10 dBm'


And due to Covid-19 it is not possible for me to go to University and test my device so unfortunately it is has to be just a theoretical project.

thanks for the solution!