50 Ohm triax to dual coax pigtail

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
Hello all,

What I want to do is make a 50 ohm triax cable to dual coax pigtail. The reason is because the connector for the triax is too big to mount on my PCIe card, and I'd like instead to use two BNC type connectors to bring in the signal, while keeping the signal integrity throughout. I'd like a solution that is completely passive, so I thought about using an EMI shieded box. The triax connection would mount to the box on one end, with the shield connected directly to the box. On the other end, the two BNC connectors would mount, both of their shields are also connected to the box. Inside the box, each center conductor from the triax cable connects to one of the coax conductors. My only concern is the lack of impeadance control once the center conductors are broken out. Also, the triax will be terminated at the box entry.

I know this can be solved with active components, but I do want to keep it passive if possible. Can anyone comment on wether this will work or not?

Tannks!

PS: Details on the triax cable and connector are on the way!
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
The signals are differential ~2V p-p, 20mhz manchester encoded data. The cable is a multi-drop data backbone.
 

crutschow

Joined Mar 14, 2008
34,281
How long would the connection be between the triax connector and the coax connector load circuit? That determines whether impedance will be a problem

Triax is not commonly used for differential signals since it has an unsymmetric impedance to ground. Are you sure it's differential?

You might consider going from the triax to a twisted-shielded pair. That's more typical for carrying differential signals and would require only a single wire.

Edit: It might be easier just to use a miniature triax connector. Here are some that are no bigger than a BNC.
 
Last edited:

Ron H

Joined Apr 14, 2005
7,063
I never heard of transmitting differential signals on triax. It seems to me that the capacitance from the inner conductor to the outer shield (ground) would be much different from the inner shield to ground. I think this will cause a common mode signal to appear on the receiving end, which means the receiver needs better common mode rejection than would be required if you used twinax.
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
crutschow said:
How long would the connection be between the triax connector and the coax connector load circuit? That determines whether impedance will be a proble
About six feet.

Triax is not commonly used for differential signals since it has an unsymmetric impedance to ground. Are you sure it's differential?
I'm quite sure.
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
I never heard of transmitting differential signals on triax. It seems to me that the capacitance from the inner conductor to the outer shield (ground) would be much different from the inner shield to ground. I think this will cause a common mode signal to appear on the receiving end,
Don't know what to tell you. Here is the stuff http://milestek.com/p-16655-trc-50-2-trompeter-triaxial-cable-yellow-jacket-50-ohm-emerson.aspx Can't find much technical data, other than it's 50-ohm triax. It's what is used in the present system.
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
crutschow said:
Edit: It might be easier just to use a miniature triax connector. Here are some that are no bigger than a BNC.
Awesome! I'll need to see if they will work with our cable.

I don't think they are made for the thicker cable we're using, which means I'm back to square one, adapting from one type of connector to another.
 

crutschow

Joined Mar 14, 2008
34,281
Awesome! I'll need to see if they will work with our cable.

I don't think they are made for the thicker cable we're using, which means I'm back to square one, adapting from one type of connector to another.
How thick is your cable? The website says the series 70/340 can accommodate cable of to .475" outer diameter.

If not, it would still be a good choice to go from your adapter box to your circuit with smaller triax. At 20' you don't want to terminate the triax with resistors at the box. Terminate it at the board.
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
crutschow said:
How thick is your cable? The website says the series 70/340 can accommodate cable of to .475" outer diameter.
Yeah, missed that part.

If not, it would still be a good choice to go from your adapter box to your circuit with smaller triax.
Completely agree.

At 20' you don't want to terminate the triax with resistors at the box. Terminate it at the board.
I'll need to think about that. Problem is, the system is made only for "fly by" connections, and only terminated at the ends. As the system is now, each connection point uses an active circuit to transform from high input impedance to a low impedance twisted pair for the last 6' or so run. The connection point is unterminated if in the middle ( fly by ) and terminated at the ends. Taking coax the last 6' might not work, unless I have two pairs, one for return. But my predecessor was working on a passive impedance scheme that used diodes in series, which I assume were intedned to 'squash' any reflections. The whole idea here is to simplify connections to the bus, and eliminate the need for an active connection point. Might not be possible without using a return triax/coax pair. Anyway, the idea of using minature triax is good. With a pair of them, I can accomplish connection to my PC card, and still keep the integrity of the signal throughout.
 

crutschow

Joined Mar 14, 2008
34,281
I'll need to think about that. Problem is, the system is made only for "fly by" connections, and only terminated at the ends. As the system is now, each connection point uses an active circuit to transform from high input impedance to a low impedance twisted pair for the last 6' or so run. The connection point is unterminated if in the middle ( fly by ) and terminated at the ends. Taking coax the last 6' might not work, unless I have two pairs, one for return. But my predecessor was working on a passive impedance scheme that used diodes in series, which I assume were intedned to 'squash' any reflections. The whole idea here is to simplify connections to the bus, and eliminate the need for an active connection point. Might not be possible without using a return triax/coax pair. Anyway, the idea of using minature triax is good. With a pair of them, I can accomplish connection to my PC card, and still keep the integrity of the signal throughout.
I don't understand your need for a pair of triax cables. You only have one triax cable, right? Then you just need one triax from your box to the board.

I'm confused. :confused: Is you tap from the middle of the cable or the end? If it's the end, then it needs resistive termination at your board to avoid reflections. The idea is that a signal can be tapped along the cable with a short, high impedance tap and not affect the signal, but the end of the cable must be terminated with its characteristic impedance. You can not tap the middle with a low impedance tap using a long cable of any type (any long cable becomes a low impedance tap at high frequencies).
 

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
I'm sorry for the confusion. My idea is evoloving as we speak. I'll try to diagram it later for clarification.

PS: Let's forget about why I need a pair of triax connectors. That's a completely different issue ( I was thinking out loud ) I think we have at least a good possible solution to the issue of connecting the bus to my PC card.

Thanks!
 
Last edited:

Thread Starter

Brownout

Joined Jan 10, 2012
2,390
FYI: Here is why I need two triax cables. At the top of the drawing is what we have now. The triax, which runs under the floor, is connected to the interface electronics using a 'tee' connector. The resultant stub is very short. In the bottom drawing, the interface electronics have been absorbed into the new system. Now, an extension to the existing bus needs to come up through the floor to connect to the system. Since I cannot tap a long cable in the bus backbone, I'll need to extend the bus all the way to the new system's location.
 

Attachments

crutschow

Joined Mar 14, 2008
34,281
OK. Then you must have a very short, high-impedance, unterminated stub from the 'tee' to your circuit board. This means the connection from the tee should be free air wires, not coax or triax to minimize capacitance which would cause distortion of the signal.
 
Top