DB25 Ribbon Cables in Mixer from Inputs to PCB?

funkmuffin

Well-known member
Joined
Apr 4, 2009
Messages
57
Reaction score
2
Location
Akron, Ohio
Hi -- I've been working on a 24 channel "Stereotype" mixer using Hakanai's boards and I'm to the point where it's time to wire everything up.  As I'm working through this, I've decided to use DB25s to connect the 24 inputs from my patchbays, but I'd like to keep things tidy and well managed inside the case.  There are readily available (and cheap, but that's not actually important) DB25-to-ribbon cables for use inside PCs.

Like this one:
http://www.microcenter.com/single_product_results.phtml?product_id=0235524

What sort of issues would I be likely to run into using these?  Noise? "Bad" wire?  I'd be splitting the wires out individually at the PCB for connection to screw terminals.

Tell me why I shouldn't take the easy, clean way out!  If this isn't the way to go, then what is?  24 channels for balanced audio inputs in a relatively small space - 2U rack.

Thanks in advance!  
 

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,706
Reaction score
222
Location
Helsinki, Finland
If all the audio running through them is balanced, then this solution will be fine.

If unbalanced, you will quickly run into crosstalk issues. I have issues with my mixer internal wiring for example where they have done this, It seems even extra 20cm of unbalanced audio running together side by side unshielded does matter. What it means is that the further away from summing unit these channels are, the more crosstalk there is.
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,163
Reaction score
1,510
Location
Norfolk - UK
Kingston said:
If all the audio running through them is balanced, then this solution will be fine.

I am not sure this is quite true. Even with grounds between pairs there will be capacitive coupling from the + of one channel to the - of the next for example. The capacitance will be quite small, a few pF, but at high frequencies it could represent significant coupling. How much crosstalk will depend on the source impedances. If they are 600 ohms then a mere 5pF of coupling could produce crosstalk of around -75dB at 10KHz. Typical ribbon cable has a capacitance of around 14pF/ft and since capacitance is proportional to separation then you can expect about 7pF/ft coupling between pairs separated by a ground. The balanced input will help but the further input of the balanced pair will receive a lower crosstalk signal due to its increased separation. The balanced source may well help by cancelling out some of the coupling but not all sources plugged into balanced inputs are balanced.

All that said, for short lengths it may not be a problem at all. The only way to find out for sure is to measure the crosstalk. Has anyone done this?

Cheers

Ian
 

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,706
Reaction score
222
Location
Helsinki, Finland
But the above kind of capacitive coupling applies to all cabling. You can only minimise it by selecting a really good cable, but can't avoid it. Of course with ribbons the choice is limited, but with the distances we're talking here it's probably a non-issue.

Cross talk in this kind of setup would be often seen as "common mode" if the pairing is done correctly (order of tracks in the ribbon cable). For unbalanced cables with audio signals running close to each others tracks the performance is pretty awful. Like I said, even 20cm matters significantly especially for high frequencies. I'm battling with this trying to improve a kind of a budget mixer. I will probably end up splitting the ribbon in two: one ribbon handles channels 1-8, other takes 9-16. They both have the same target, but at least the crosstalk won't be quite as bad between these two "groups".
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
11,919
Reaction score
645
Location
Marcelland
ruffrecords said:
Kingston said:
If all the audio running through them is balanced, then this solution will be fine.

I am not sure this is quite true. Even with grounds between pairs there will be capacitive coupling from the + of one channel to the - of the next for example. The capacitance will be quite small, a few pF, but at high frequencies it could represent significant coupling. How much crosstalk will depend on the source impedances. If they are 600 ohms then a mere 5pF of coupling could produce crosstalk of around -75dB at 10KHz. Typical ribbon cable has a capacitance of around 14pF/ft and since capacitance is proportional to separation then you can expect about 7pF/ft coupling between pairs separated by a ground. The balanced input will help but the further input of the balanced pair will receive a lower crosstalk signal due to its increased separation. The balanced source may well help by cancelling out some of the coupling but not all sources plugged into balanced inputs are balanced.

All that said, for short lengths it may not be a problem at all. The only way to find out for sure is to measure the crosstalk. Has anyone done this?

Cheers

Ian
A simulation trying to take into account all stray capacitances gives -88dB @ 10kHz. In real life, 600R impedance is not likely. Typical impedance from sound cards is about 100R, so crosstalk should be around -100dB @ 10k.
Unbalancing sources confirms -75dB @ 10k. Again, with 100R source, it drops down to -90.
But you're right, nothing beats actually measuring. Simulation just gives an incentive to pursue.
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,163
Reaction score
1,510
Location
Norfolk - UK
Kingston said:
But the above kind of capacitive coupling applies to all cabling. You can only minimise it by selecting a really good cable, but can't avoid it. Of course with ribbons the choice is limited, but with the distances we're talking here it's probably a non-issue.

I do not think that is a true statement. Even two unbalanced signals running alongside each other in separate screened coax cables will not suffer from cross talk.

Cross talk in this kind of setup would be often seen as "common mode" if the pairing is done correctly (order of tracks in the ribbon cable).

That's the point, this is unscreened cable so there is a coupling to both of the balanced inputs but it is not an equal coupling so crosstalk still occurs.

For unbalanced cables with audio signals running close to each others tracks the performance is pretty awful. Like I said, even 20cm matters significantly especially for high frequencies. I'm battling with this trying to improve a kind of a budget mixer. I will probably end up splitting the ribbon in two: one ribbon handles channels 1-8, other takes 9-16. They both have the same target, but at least the crosstalk won't be quite as bad between these two "groups".

As noted above, unbalanced is not in itself a cause of crosstalk but lack of screening is. Clearly with unbalanced tracks on a PCB there can be no real screening but unbalanced screened cables will have less crosstalk than a balanced unscreened ribbon. The important factor is whether a capacitive coupling can exist. In a ribbon it clearly can and between screened cables it cannot.

All that said, in your application with balanced signals, provided you keep the ribbons short you should have no problem.

Cheers

Ian
 

EZ81

Well-known member
Joined
Sep 7, 2004
Messages
222
Reaction score
0
Location
.de
funkmuffin said:
I'd be splitting the wires out individually at the PCB for connection to screw terminals.

I cannot comment on crosstalk etc., but the screw terminal idea is not great if the gear is ever moved around IME. Ribbon cables have really thin wires and will break easily if not supported properly.  Please consider using crimp connectors. If you need to split the cable by channel, you can crimp each channel's 3 wires into 4 contact connectors like these linked below (the pictures are not correct):
http://www.tycoelectronics.com/catalog/bin/TE.Connect?C=1&M=BYPN&BML=&LG=1&PN=215460-4&PID=46549
http://www.tycoelectronics.com/catalog/bin/TE.Connect?C=1&M=BYPN&BML=&LG=1&PN=215083-4&PID=47211

Matthias

 

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,706
Reaction score
222
Location
Helsinki, Finland
ruffrecords said:
As noted above, unbalanced is not in itself a cause of crosstalk but lack of screening is. Clearly with unbalanced tracks on a PCB there can be no real screening but unbalanced screened cables will have less crosstalk than a balanced unscreened ribbon. The important factor is whether a capacitive coupling can exist. In a ribbon it clearly can and between screened cables it cannot.

Ah yes. I see it now. I'll think about fudging in screened cable at least for the furthest away problem channels. Lack of space will become an issue, as does finding a good solution for connectors. It's going to be ugly in there. I can live with -75dB crosstalk, but -40dB is just too much.

I wonder how the big boys do this.
 

tmuikku

Well-known member
Joined
Dec 24, 2008
Messages
481
Reaction score
20
Location
Finland
Kingston said:
...
I wonder how the big boys do this.

Avoid ribbon cables?:) Screening or placing conductors further apart reduce capacitive coupling. What goes to bus motherboards in consoles: I think my old Tascam had slim ground traces between buses that are connected only to starground, at least it looked like it :D Other than that I don't have a clue how big boys do it though..
 

Attachments

  • MotherBoard 3.JPG
    MotherBoard 3.JPG
    182.4 KB · Views: 52

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
11,919
Reaction score
645
Location
Marcelland
Considering all the practical and technical aspects and the fact that it's a one-off, I'd use screened cables from the DB25 to the PCB's.
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,163
Reaction score
1,510
Location
Norfolk - UK
Kingston said:
I wonder how the big boys do this.

When I was at Neve in the 70s we fed unbalanced screened leads from the routing modules to the mix bus. The mix bus consisted of an aluminium extrusion with several 'U' shaped channels in it. Along each channel ran a strip of one hole wide Vero strip - that's right Vero strip in a Neve - which was supported on insulated stand-offs.. Holes were drilled through the top of the 'U' sections into which grommets were fitted. Bus resistors were fed through the grommets and soldered to the Vero strip. The screened unbalanced signals from the routing modules were soldered direct to the resistor lead poking out of the grommet on the top side of the 'U' shape. When this was all wired up a flat plate was screwed to the bottom of the 'U' channel so every bus was totally enclosed in a surrounding screen. Transformer isolated mix amps (essentially Neve mic pres) were connected to the end of each bus.

That's how that particular big boy did it.

Cheers

Ian
 

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,706
Reaction score
222
Location
Helsinki, Finland
Thanks Ian! Very kind of you to share this type of information. Looks like I have a major chore to plan. Space is a big issue, another is finding a good alternative solution/extension for the ribbon connectors that are already there. If Neve doesn't shy away from veroboarding these kind of connections, so won't I!
 

Kingston

Well-known member
Joined
Nov 1, 2005
Messages
3,706
Reaction score
222
Location
Helsinki, Finland
On a related note, would anyone here happen to know of a very thin single core foil-shielded cable for this kind space-constrained task?

[edit]

By the way, sorry to have hijacked the thread like this. I just thought my specific case happened to be very close to what the original poster is doing as well.
 

funkmuffin

Well-known member
Joined
Apr 4, 2009
Messages
57
Reaction score
2
Location
Akron, Ohio
Kingston said:
On a related note, would anyone here happen to know of a very thin single core foil-shielded cable for this kind space-constrained task?

[edit]

By the way, sorry to have hijacked the thread like this. I just thought my specific case happened to be very close to what the original poster is doing as well.

No worries Kingston.  I have the same question -- the (very helpful) replies to this thread have me reconsidering my approach too.  Like you, though, I need some guidance on appropriate screened wire to use - size-wise it's going to be very tight, and not a little bit messy with 24 individual 3-conductor wires running over and under PCBs. 

In my case, it might make good sense to stagger board-height in order to facilitate running the cables under them.  Either that or clip the wire-bundles to the side of the case to keep them out of the way.

I'm also half-tempted to just use a few lengths of 8-channel snake cable.  Anybody have a recommendation for smaller-diameter brand/type of raw snake cable?

Thanks for the feedback so far!
 

Latest posts

Top