# Impedance matching joint between cable and ground

1. Jan 25, 2013

### freddyfish

I would like to impedance match a few joints in a circuit and would appreciate some thoughts about this, since I would like the match to be as perfect as possible. The problem is complicated by the fact that the available equipment only consists of trimpot(s). The joints are mostly cables to conducting strips (and vice versa), cables to different kind of cables and cables to a copperlayer on a rectangular plate. The latter is also defined to be the ground/earth.

One method is to connect a resistor in parallel and a resistor in series (both at the joint), where the resistor in parallel is connected to earth/the copperlayer on the plate.

This approach seems to do the trick for the wire-wire and wire-strip connections, but it is not possible in the case wire-copperplate, since the copperplate is the ground itself.

The cables are both coax cables and ordinary pairwise twisted single conductors.

Thank you

2. Jan 25, 2013

### the_emi_guy

Are you talking about a broadband or narrowband match? i.e. is the match intended to manage signal integrity of a digital signal, or to provide good return loss for RF signal.

In either case, what frequency (or risetime) do you have?

3. Jan 25, 2013

### freddyfish

The frequency is 1GHz and the square-wave impulses' amplitude is 1V. The only condition there is, is to prevent the joint from reflecting the signal. The interest lies in building a circuit that reflects, most preferably, nothing at all.

4. Jan 25, 2013

### the_emi_guy

Square-wave at 1GHz?

Are you sure you don't have a sine-wave?

Do you care about insertion loss?

5. Jan 26, 2013

### freddyfish

Yes, the impulse source is supposed to be adjusted to give square impulses.

Insertion losses are acceptable, but there are not to many things to insert to prevent reflection though.

6. Jan 26, 2013

### sophiecentaur

An L matching pad will work well if the resistors are good quality (not spiral cut metal film). Alternatively, you can get good loss-less matching with quarter wave sections of the appropriate impedance microstrip for a coax / microstrip transition.
I am not sure what you mean by
Is this not a normal strip over a ground plane (unbalanced line)?

7. Jan 27, 2013

### the_emi_guy

freddy,

You are probably going to have to give is more info in order for us to help.

1GHz rectangular pulse implies a signal bandwidth north of 5GHz. Impedance matching at these frequencies is not trivial, and certainly cannot be achieved with the trimpots that you mentioned that you wanted to use.

Can you give us more detail on source of these pulses and your intended construction?

8. Feb 11, 2013

### freddyfish

This problem is supposed to be solved using elementary methods and all the relevant equipment at hand has been presented, unfortunately.

Basically, what I would like to do is connect a conductor to earth making joint free from reflection. I can't give any more information since this is all the information I have been given.

9. Feb 11, 2013

### sophiecentaur

Then the two lines need to have known impedances.?? Then you can match from the low impedance line into the high impedance line using the parallel resistor formula. The other direction requires a series resistor.
You did not make that clear, I'm afraid. I know it can be hard to formulate questions. This may have been better suited to Homework questions?

10. Feb 11, 2013

### sophiecentaur

This may be a language problem. What do you mean by "a joint free from reflection"? reflection of what? Are you trying to terminate a transmission line? Are you trying to join two transmission lines?
Try giving us the same information that you have been given, using the same words that you have been given? You are missing out something, I'm sure.

11. Feb 12, 2013

### freddyfish

Explicitly, the propagating signal should be completely transmitted through the joint. I would like to connect a copper plate (functioning as a conductor) to the negative pole of the impulse source, which is done by connecting the shield of the utilized RG-58-cable to the copper plate. It is the connection between this shield and the plate that is problematic.

More precisely, the task is to transmit an impulse through the axial conductor of a RG-58, let the signal pass through a circuit and then return the signal to the source through the shield of the RG-58. The circuit is completely impedance matched at 553Ω, so the 50Ω of the RG-58 should be matched to the circuit's 553Ω. What's left to solve is only the italic & underlined part above.

I'm sorry, but there is absolutely no more information. A direct translation would be: "All transmission lines must be terminated properly, since no reflections will be accepted."

12. Feb 12, 2013

### sophiecentaur

You appear to have a 50Ω co-ax line, feeding a 553Ω load (the input to a circuit) which is on an Earth plane(??) and the outer of the co-ax is connected to the Earth plane. This is a normal arrangement. You want a transition from a 50Ω 'unbalanced' line to a 553Ω unbalanced load - and would that be a microstrip circuit on PCB?.

If this is the arrangement then you can buy transitions with BNC / Type N (or even better quality) connectors to PCB or you can just solder the coax outer to a pin though the board to the earth plane and take the inner onto the circuit. The maximum frequency you are using will determine just how good the connection needs to be. You then need a matching network for the 50/553Ω transition on the board itself. A single shunt resistor would work but you would lose some significant signal. (Around 20dB of loss.) The figure "553" suggests, even more, to me that this is a theory exercise. Am I correct?
That statement means nothing, really. Every practical transition has reflections and it is necessary to specify a maximum level permitted. If this is just a paper exercise then everything becomes ideal and the Earth on the co-ax is just 'ideally' bonded to the Earth plane on the board.

13. Feb 12, 2013

### the_emi_guy

The nature of your question has me worried that you are not understanding certain basics that may cause you trouble.

However, if you just want to know how to bond coaxial shield to a plane in a manner that will support 10s of gigahertz here it is...

1 - Terminate the coax with a high quality SMA connector (good up to 17GHz).
2 - Mount high quality SMA connector onto your "plate" using footprint guidance provided by connector vendor (depends on whether it is edge mount, vertical mount etc.)
3 - Plug it in.

14. Feb 14, 2013

### freddyfish

Sophiecentaur: A microstrip/piece of copper tape was taped to the side opposite of the copper coated side of a board mostly consisting of some dielectric material, (the exact type of material was not specified), so it is a "simulation" of a PCB. Actually, it was/is a practical exercise accompanied by a presentation of the necessary theory. This is why this thread is so loose. The course focuses on experimental skills and we have not studied the involved concepts yet, and for this purpose we were given less than a week, while taking three more courses. Thank you for your help which, together with further studies, has helped me get some clarity.

t.e.g: Connecting the RG58 itself to the plate would be inappropriate and for this purpose we did use a SMA connector with a characteristic impedance of 50Ω to match the RG58. However, this would not solve the problem of reflection since the need to match the 50Ω to 553Ω persists.

15. Feb 14, 2013

### sophiecentaur

Of course.
You have a 50Ω line which is terminated by a 553Ω resistor. What value of extra shunt resistor is needed, to turn this 553Ω into 50Ω? (Because that will then provide you with a perfect match) Funnily enough - but not surprising, the answer comes out to be a nice round figure! (But, as I wrote earlier, that's how I spotted it was a h/w question!)
Note that, by doing it this way (resistively), the majority of the input signal power will be lost as it will be dissipated in the extra resistor but I think it is all that's required by the question you were asked on your course. If signal loss were a problem, then an alternative form of matching would be used (some form of transformer) - but I don't think that's your problem.

16. Feb 14, 2013

### freddyfish

I was looking for possabilities to use the method with transformers, but I soon rejected the idea since we were not be allowed to use the material needed, so I think I will accept the loss of signal. :)

17. Feb 14, 2013

### sophiecentaur

Important: Until you understand the content of your other thread and the real meaning of Characteristic Impedance, this assignment of yours is of little value. You must address this.
I am sure the resistive solution is what they are after. What answer did you get?