| Message | Date | From | Subject | ||||||||||||||||||||||||||||||||||||||||||||||||
| 10308 | 2014-09-27 10:11:39 | peter_dl8ov | Coax Connections Within Equipment | ||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
I need your help in settling an argument, well, an animated discussion that is taking place outside the group. Suppose you have two sections within a circuit that has a common chassis and these two sections are connected using a coax cable. Do you:
A) Ground the coax at one end using the chassis as the ground return. The problem here is that the chassis will have circulating currents from other sections and these could interfere with your signal? or B) Ground the coax at both ends risking a ground loop/hum loop? Gut feeling tells me that option (B) is the best but I cannot prove it. All assistance would be appreciated. Peter DL8OV | |||||||||||||||||||||||||||||||||||||||||||||||||||
| 10310 | 2014-09-27 10:36:42 | Andy | Re: Coax Connections Within Equipment | ||||||||||||||||||||||||||||||||||||||||||||||||
These are RF signals, right? Then I would definitely do (B). "B) Ground the coax at both ends risking a ground loop/hum loop?" Screw the ground loop. (Pardon my French.) The signal goes along the wire inside the coax, and its return path is in the shield of that coax. If you don't ground the coax shield at both ends, then you might as well just run a bare hookup wire between point A and point B. If your signals are such that you're OK with a length of bare hookup wire, then fine, choose (A). The current in the coax center conductor causes a mirror current on the inside of its coax shield, equal but in the opposite direction. When it reaches the ends of the coax, it needs a place to go. You must give it that path, normally to chassis or to the PC board where the next piece of circuitry is. If you don't give it a place to go, you open up the door to many problems. Always follow the currents. If you know where your currents go, and if they follow good paths, then you should be OK. Ignore currents at your own peril. Create big loops between currents and their return paths, and radiation results. Andy | |||||||||||||||||||||||||||||||||||||||||||||||||||
| 10311 | 2014-09-27 12:23:11 | peter_dl8ov | Re: Coax Connections Within Equipment | ||||||||||||||||||||||||||||||||||||||||||||||||
Thanks for your reply Andy and your arguments were similar to mine last night when we had the discussion. Those who argued for method (a) wanted all currents to return through the chassis and the cable braid with a ground at one end to act as an electrostatic shield. I can see their point which is why I made the original post.
Would it matter if the signals were RF or audio? Thinking about direct conversion receivers such as the R2 Pro Rick emphasizes careful layout and a need to avoid hum loops because of the high gain in the audio stage. This issue can get quite complex. 73 Peter DL8OV | |||||||||||||||||||||||||||||||||||||||||||||||||||
| 10312 | 2014-09-27 14:51:44 | Ephemeral | Re: Coax Connections Within Equipment | ||||||||||||||||||||||||||||||||||||||||||||||||
If we were to connect perfect boxes with perfect coax there would be no potential difference between the box cases and no magnetic field outside the coax? So no current flows in the chassis and the chassis is irrelevant and it doesn't matter whether we connect out boxes to the chassis or not? However if the boxes are imperfect or the coax is imperfect at RF the chassis is likely a second line of defence? Presumably we get best performance is all boxes are bonded to the chassis? Is there any mileage in slipping a ferrite core over the coax? Adrian
. | |||||||||||||||||||||||||||||||||||||||||||||||||||
| 10313 | 2014-09-27 19:04:10 | Russell Shaw | Re: Coax Connections Within Equipment | ||||||||||||||||||||||||||||||||||||||||||||||||
| 10314
| 2014-09-27 20:38:32
| duwayne@bellsouth...
| Re: Coax Connections Within Equipment
| If you are dealing with RF or Video signals, the coax is normally grounded at both ends. The input and out impedance of these type modules are usually low (50-75 ohms ) which helps eliminate some of the problems with noise pickup. If you are dealing with audio or or some type of low level sensor you usually find high impedance input circuits. In this case you will usually find single point grounding of the signal cable. In some very low level or high precision transducers you might find active driven shields instead of grounded shields. Analog devices has a good app note on grounding and shielding.
http://www.analog.com/static/imported-files/application_notes/41727248AN_347.pdf DuWayne 10315
| 2014-09-28 02:44:10
| T. Hills
| Re: Coax Connections Within Equipment
| Try both, see which works best. Hard to say anything else without such things as frequency(ies) involved, power levels, cable type, length of cables, Etc. Tim Hills Sioux Falls, SD 10316
| 2014-09-28 05:40:14
| peter_dl8ov
| Re: Coax Connections Within Equipment
| Now I am really confused. The post from Duwayne said that normally both ends of a length of coax should be connected and then he gave me a link to an Analog Designs paper on shielding and grounding. Figures 4&5 show the coax connected at only one end with the return via the chassis and above figure 7 it states:
Don't connect both ends of the shield to ground. The potential difference between the two "grounds" will cause a shield current to flow (Figure 7). The shield current will induce a noise voltage into the center conductor via magnetic coupling. The best idea I've heard so far is to run the shielded cable along the chassis or ground plane maintaining contact along its length (although sometimes this may not be practical). Does anybody else want to help solve this dilemma? 10317
| 2014-09-28 07:55:48
| David
| Re: Coax Connections Within Equipment
| There are other application notes from Analog Devices which shed more light on
the subject: http://www.analog.com/static/imported-files/application_notes/AN-202.pdf http://www.analog.com/static/imported-files/application_notes/6001142869552014948960492698455131755584673020828AN_345.pdf The usual solution when grounding the shield or one side of a balanced transmission line at both ends creates a ground loop issue is to either use a differential signal or use a balun at one or both ends. Any of these solutions will raise the common mode impedance cutting off the common mode current. To give a concrete example, old Tektronix oscilloscopes use both methods. When a single ended signal moves between modules using either balanced twisted pair (!) or coaxial transmission line, a balun (*) is used. Because the common mode impedance is raised by the balun, the return current *has* to go through the transmission line and any voltage differences between the grounds affect the transmission line ground and signal equally. When a differential signal moves between modules, the common mode rejection by itself rejects differences in the grounds but a balun might be used anyway to improve the situation at high frequencies where common mode rejection falls. The example you cite replaces the balun with a differential input amplifier on the receiving end which senses the different between the remote signal ground and local ground and subtracts the difference from the signal but the same thing can be done on the transmitting end using either an instrumentation amplifier to sense the remote ground (like a voltage balun) or high impedance current drive (like a current balun). Since instrumentation amplifiers and current output drivers have degraded common mode rejection at higher frequencies, RF baluns may be added to make up the difference. In AC coupled RF circuits, low frequencies are rejected anyway so only the baluns are needed. In tuned RF circuits, nothing is needed because anything except the frequency range of interest is rejected by filtering. (*) Tektronix used current mode baluns because they pass DC but a voltage mode balun will work also. On 28 Sep 2014 05:40:14 -0700, you wrote: 10318
| 2014-09-28 09:06:27
| Roelof Bakker
| Re: Coax Connections Within Equipment
| Hello Peter,
Your problem can not be solved by words as every situation can be different. Just try the various options and finds what works best. Never trust a random advice! I often use RG174 coax between modules and connect the coax shield at both ends. No problems encountered so far! 73, Roelof Bakker, PAoRDT 10319
| 2014-09-28 10:14:10
| duwayne@bellsouth...
| Re: Coax Connections Within Equipment
| Sorry, I should have referenced other app notes also. The one I chose applies mostly to low frequency low level signals.
When you go up in frequency the impedance of the return path becomes more important than the resistance. With RF, the impedance through chassis ground may go up to a high value as frequency increases. With a properly designed circuit using coax inter-connections you would normally have. 1.input and output of each circuit at the same impedance. 2 the coax shield connection very close to the input or output of the circuit. 3. the proper impedance cable connecting the two circuits. When these conditions are met the signal flow will mostly be through the coax inner conductor and returning through the shield. With very little through the chassis ground path to cause ground loop issues. Stray inductance and capacitance from mechanical layout and construction can significantly change the actual impedance of the circuit from what is expected. Relying on chassis ground between circuits could possibly cause problems with impedance mismatches. As frequencies go up these problems usually become much worse. At these frequencies the interconnection might change to balanced differential signals, or tuned transmission lines. Hope this does not cause more confusion. DuWayne | |||||||||||||||||||||||||||