Hi everyone, this idea jumped in my mind yesterday when i'm surfing the web for "DIY High-End balanced signal cables", trying to find a valid alternative to unbalanced ones.
Check this out:
Do you think that could be a good idea ?
Constructive opinions/suggestions are always welcomed !
Check this out:
An externally hosted image should be here but it was not working when we last tested it.
Do you think that could be a good idea ?
Constructive opinions/suggestions are always welcomed !
forart.eu said:
Balanced audio cables in my thinking imply twisted pairs with a shield for low level and high dynamic range (ie XLR). So i suppose in a pinch you could use two of the RGB coax's leaving spare the one coax and all the wires since they arent twisted, thus only for a single channel, in my view not too economical compared to a proper XLR connection system.
The **IMPORTANT** part about a cable for balanced use is that the pairs are twisted! The screen is almost incidental most of the time, it is the twisting that resists the magnetic coupling.
In fact, you can stuff balanced line level audio over hundreds of meters of unscreened twisted pair and it will be fine if the receiver is right.
For a balanced line receiver to work well, the impedances need to be well matched (And typically the input impedance will be much higher then the output impedance to minimise the effect of tolerance of the build out resistors. It is NOT required that there be voltage balance, it is the impedance balance that does the magic.
If you are building balanced line IO into gear, please do it right, pin one goes to the case right at the connector, not to circuit 0V (That way lies hum)!
Regards, Dan.
In fact, you can stuff balanced line level audio over hundreds of meters of unscreened twisted pair and it will be fine if the receiver is right.
For a balanced line receiver to work well, the impedances need to be well matched (And typically the input impedance will be much higher then the output impedance to minimise the effect of tolerance of the build out resistors. It is NOT required that there be voltage balance, it is the impedance balance that does the magic.
If you are building balanced line IO into gear, please do it right, pin one goes to the case right at the connector, not to circuit 0V (That way lies hum)!
Regards, Dan.
unclejed613 said:
But that wouldn't be called "VGA"! Yours would be workable using custom cable/s and DB-15 connectors but then it may as well be UJA (Uncle Jed Audio) or something else.
dmills said:
Light gauge twisted pairs have a Zo of about 120 ohms. So the max BW is with a difference reciever terminated with the same. If you use coax it will be 2xZo or 150 ohms using video coax.
Dan right about without shields, but you then need to max out the signal levels to keep a decent SNR depending on the ambient interference levels.
Re minimize hum (ground loops and such)
With a shield connection to the source only and left floating at the difference receiver is best for audio in my experience. The shield connection is referred to as a drain in this case. No ground loops!
Why would I be even thinking about transmission line effects in an audio frequency application (unless I am the phone company)? Characteristic impedance is irrelevant for audio frequency applications over sensible cable lengths (say up to a KM or so).infinia said:
That has not been my experience at all, as long as the input stage was designed by somebody with some engineering chops, the screen is as near as dammit pointless most of the time (Note this means an input stage that does not rectify common mode RF and that is designed with a reasonable bandwidth limit).
I have almost 100M of unscreened cat carrying 4 differential pairs of audio to the amp room at work (it feeds >20KW of power amp), that signal actually travels via the data comms room patch bay and in the same cable bundle as Gb Ethernet, and it seems to work fine. It even works fine when you expose it to a few tens of watts at 433Mhz. I would guess those circuits average somewhere around -30dbu or so, and I dont think I have ever had an interference issue.
The screen does however make a dandy return path for phantom powered mics.
Measure of last resort IMHO, better to solve the pin one problems in the gear rather then band aiding it.
Regards, Dan.
audio cable doesn't begin acting like a transmission line until you get into lengths like 1875 meters or so (1/8 wavelength at 20khz), so the "Zo" of audio cable is not important, and neither is the velocity factor of the cable. back in the days of all passive studio mixers, having a standard set of impedances made sense, because everything was done with transformers, caps, and resistive pads, and passive mixers were designed for minimum losses. these days 99.99% of all audio is done with active devices and the old "impedance matched" methods have given way to "voltage matching" of systems.
also the shielded pairs are useful to keep common mode signals out of the system. while not absolutely necessary to have, they can be useful.
also the shielded pairs are useful to keep common mode signals out of the system. while not absolutely necessary to have, they can be useful.
unclejed613 said:
No actually there are two sets of Zo one for low frequencies and another for high frequencies and the length and physical construction of the twisted pair determines the frequencies of transition between the two. For example using Cat5 unterminated with a length of 100 meters the frequency of transition is from 50KHz to 1 MHz going from 600 - 100 ohms. If you terminate using 100-120 ohms then you effectively swamp out the impedance transition. Yes you are correct on using active circuits esp using newer op-amps (transimpedace amlifiers) lets you drive cables with lower loads. Using Zo termination techniques free up the BW dependance of cables esp at longer. than 100 meters. Connection of shields at a single point for elimination of ground loops is good system enginering design practice! Not a bandaid solution. Microphones should connect at both ends and are not succeptible to ground loops because the power source is either batteries or from the reciever end (phantom).
Designing the input to comply with AES48 is good system engineering, lifting screens may help at 50/60Hz and the first dozen harmonics, but leaves you with large RF currents potentially being pushed back into the transmitting device (A conductor connected at one end has another name, an Aerial!) .
I am aware of SCIN, but it is normally minor compared to the other issues that a floating screen can cause.
There is no reason for a screen bonded at both ends to induce excessive low frequency noise and any gear that suffers in that way needs some attention.
See papers by Muncy, Whitlock, Walderon & Brown for the gory details.
With the ever increasing levels of environmental RF, it behoves audio types to consider the RF implications of what they design if we are to get satisfactory behaviour in the presence of cell phones and the like.
Regards, Dan.
I am aware of SCIN, but it is normally minor compared to the other issues that a floating screen can cause.
There is no reason for a screen bonded at both ends to induce excessive low frequency noise and any gear that suffers in that way needs some attention.
See papers by Muncy, Whitlock, Walderon & Brown for the gory details.
With the ever increasing levels of environmental RF, it behoves audio types to consider the RF implications of what they design if we are to get satisfactory behaviour in the presence of cell phones and the like.
Regards, Dan.
dmills said:
I'm more concerned with low impedance ground currents traveling from box to box from self generated RF hash (ie diode recovery noise and other commutating circuits) as well as the proliferation of cheap SMPS than outside RF stuff getting into chassis from singly terminated drain shields. Proper unit design has RF filters on all in and outs. Now if you are in a nearby influence of a commercial RF transmitter then the band-aids are needed. The fact is if you are using balanced interconnects you are already ahead in the game, with or without shield connections.
Its funny, a lot of my background is in commercial broadcast sites and UK practice is to bond all the boxes (and the rack) together in as many places as possible, but to treat the screen as pretty much optional. A lot of the inter bay wiring is simple twisted pair with no overall screen (overall screens are difficult to terminate on the Krone IDC strips used for a lot of the wiring).
Now I will grant that common mode rejection tends to be much, much, better in stuff designed for the purpose then in many of the cheap commercial bits of kit (And almost all RF pickup is common mode), but the point still stands. Extensive parallel earth conductor networks also help a lot.
Regards, Dan.
dmills said:
Racks esp. shielded ones are great for dealing with many system problems mostly solved by the common chassis ground. but most balanced circuits are used for interface to non co-located gear and the audio/digital circuits have no RF common mode rejection to speak of anyways. Best practice is with good bypassing of all inputs/out ie ferrite beads and ceramic post caps to chassis. The whole reason for differential circuits is to eliminate the ground and potential of ground noise on the signal of interest not like single ended which needs a ground. When using signal cables w/grounded shields and chassis to mains earth tied together they form a loop which is harder to control where the bad noise currents go. When you have to start adding more connections to earth to lower the impedance you are playing a losers game IME. My background is radio/transmiter design mostly MIL avionic and Comm/MIL satcom. RF LO synthesizers esp.
That is why it is good practise to design the input stage with a common mode choke followed by a RF decoupling network that returns the RF current to the case ground via a short path, not to the internal single ended reference (seen that all too often)! I also like an active low pass filter right after the line receiver so that I know what bandwidth I have to deal with in the processing stages (No reason to have DC->Daylight in my processing just because the line receiver has a wide bandwidth), but that is optional.infinia said:
If you are not doing the single integrated line receiver thing, then FETs are normally better then bipolar parts in terms of not rectifying RF right at the inputs, but there are good bipolar parts as well.
Probably the single most critical point however is that pin one on the XLR should be bonded DIRECTLY (As in shortest possible path, with braid or copper tape) to the chassis. Do this and it pretty much does not matter where the screen currents go as the loops formed do not include the internal single ended reference wiring.
But I think we are pretty much in agreement about the input stage side of things.
As long as they don't flow in the internal single ended analogue reference traces, why should we care?
An amp or so flowing in the case metalwork is no threat to anything, it is only when the screen connects to the on card '0V' trace that things turn ugly (or of course when that trace appears in parallel with part of that case).
Granted, you don't need to connect the screen at audio frequencies, but being able to just grab an XLR cable and know it will work is worth something surely (As is knowing it will still work when the air search radar fires up)?
Lifting the screens in todays RF environment is IMHO the band aid (granted there is a lot of kit out there that is broken in this respect).
Audio is a bit weird, in that it suffers so much from 'not invented here', the instrumentation guys have been getting this stuff right for ever, and it is not clear to me that audio actually needs to differ from instrumentation practise in any way that matters.
Regards, Dan.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.