Balancing unbalanced connection - which transformer?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

jhaible

Well-known member
Joined
Jun 24, 2004
Messages
530
Location
Germany
My mixer has balanced inputs, but the group outputs and multitrack tape inputs are unbalanced. (-10dB, RCA).

This worked fine with an analogue 8-track tape machine.

Now I want to replace the tape machine with a 8in/8out sound card in the computer.

The problem is, how can I avoid ground loops. Many instruments and FX boxes are connected to Earth Ground - no problem so far, as th emixer has balanced line inputs. But the computer is also connected to Earth Ground, so with an unbalanced connection all kinds of noises will be in the audio just from that shared GND connection.

I made a little test with the on-board sound interface of my computer (befroe I buy the 8x8 soundcard): Recording some low level audio signal via mixer into the computer, and listening to the noise.

First test, connecting unbalanced RCA mixer group output to unbalanced Line input of computer: Very high level of "digital" noises in the recording.

Second test, same connections, but with an audio transformer inserted: Much lower level of digital noise (probably what the internal soundcard picks up from the processor etc.)

Conclusion: unbalanced conncetions on both sides are ok for me, I just have to break the ground loop. (This has been expected.)

My first idea was to buy a 8x8 soundcard with balanced inputs and outputs and use special cables to connect the mixer. If I had a soundcard with floating balanced inputs and (!) outputs(!), I'd connect it like this:

Mixer group out unbalanced RCA -> balanced TRS input of soundcard
RCA hot -> TRS tip
RCA GND -> TRS ring
no connection -> TRS sleeve.

and

Balanced TRS output of soundcard -> mixer unbalanced RCA tape in
TRS tip -> RCA hot
TRS ring -> RCA GND
TRS sleeve -> no connection (only to metal enclosure)

I'm pretty sure this would solve my problem. ONLY that I doubt
these f***ing soundcards have a balanced floating output!

I've tried to find out more about these so-called balanced outputs.
What I've seen so far has much discouraged me.

Echo Layla 3G: tip is actively driven, ring is just connected to GND
with a resistor. This will not help my problem! (Taken from the user
manual.)

And, btw, if they would drive the ring in the opposite direction as the tip, this would not help either, if both are referenced to GND. What I need is "floating balanced", i.e. where ever I connect the ring, the difference between tip an dring will be the driving signal. A transformer does this naturally. There are opamp circuits which will also do this (within some common mode range). But are these circuits actually *used* in a 500-Dollar-range soundcard??

Next I tried to find out about M-Audio 1010. They have the unbalanced 1010LT and the so-called "balanced" 1010 rackmount version. I war writing a long and polite email to their technical department, explaining my application in detail, and asking if the 1010 outputs are of the floating balanced type or otherwise. They replied the won't give away technical secrets!! I asked again, non-technical, how the signal level will be if I connect ring to GND. Not even an answer this time. So I must conclude that this will not solve my problem either.

Now what I plan to do is this: Don't spend useless money on a "balanced" interface that actually isn't. I'll buy an unbalanced soundcard with converters in a breakout box (Most likely will be the Terratec Phase 88 as I type this), and buy 16 audio transformer, or find some other solution.

As for other solutions, I think if one of the following (comments invited):

a) 16 audio transformers. Would 10k:10k be the right selection? I doubt
a 600R:600R could gracefully be driven by these outputs ...

b) A power isolation transformer for the computer?! Will this work?
I'm *not* talking of a GND lift (which will be dangerous), but about
running the whole computer over a 500Watt 230V:230V transformer.

c) Building an electronic version of (a). - what these soundcard
manufacturers should have done in the first place ...

Or are there any affordable 8x8 soundcards that have floating balanced outputs?

Or any other idea to wire this?

Thanks in advance,

JH.
 
First of all, I would try it with the "real" soundcard. Most PC built-in stuff is more than horrible, but with a mid-range soundcard, chances are better. At least my Mac works perfectly this way (altough it is a long time back I did not carry my Motu with me).

Bying 16 transformers seems not very cost effective to me - I'd rather spend the money on a better soundcard or mixer. You don't want cheap transformers here as all your audio will pass them. If you disagree, I would recommend input transformers with static shields (8 for the mixer inputs and 8 for the soundcard inputs). They have way higher CMRR than output transformers (which typically have no static shield). Check the Jensen papers on this.

As a very simple active solution I could think of an one-opamp-four-resistor diff amp at the inputs where one leg has high impedance (say 10k) and the other low impedance (say 100 ohm). This would correct for the impedance imbalance comming from the unblanced outputs and should give rather good CMRR.

Samuel
 
Hi Samuel, thanks for your reply.

[quote author="Samuel Groner"]First of all, I would try it with the "real" soundcard. Most PC built-in stuff is more than horrible, but with a mid-range soundcard, chances are better. At least my Mac works perfectly this way (altough it is a long time back I did not carry my Motu with me).[/quote]

Yes, I will try it as it is first. If the problem remains, I'll just connect one channel and use a transformer there.

Bying 16 transformers seems not very cost effective to me - I'd rather spend the money on a better soundcard or mixer. You don't want cheap transformers here as all your audio will pass them.

Good point. That's why I plan to go for an unbalanced soundcard from the start - the 200 Euros price difference to a so-called (but not really) balanced soundcard will not cover the cost for the transformers, but will make a good start.


As a very simple active solution I could think of an one-opamp-four-resistor diff amp at the inputs where one leg has high impedance (say 10k) and the other low impedance (say 100 ohm). This would correct for the impedance imbalance comming from the unblanced outputs and should give rather good CMRR.
Samuel

This is interesting! I didn't think about something like that. It is to take acount to the use of a single shielded wire instead of a shielded pair, isn't it? I can (vaguely) see that this will help reduce noise that is radiated into to cable. Will it also have benefits to suppress the Earth GND noise mentioned in my previous mail? I _think_ the main problem (without the transformers) is galvanic coupling from a shared GND connection (shared by the computer engine and the audio connection), rather than inductive or capacitive coupling.

Thanks,

JH.
 
Hi Roger,

[quote author="rf"]I have dealt with this a number of times. I put together a box with 16 SSM 2142 balanced line drivers and a power supply.
[/quote]

Ahh yes, I used these chips for balanced outputs on my VCS3 synthesizer clone! I wasn't aware they are still available. Just checked at analogue devices, and they are even cheaper than I remembered.

So, it's a shame the soundcard manufacturers don't include something like that when they boast their products as having "balanced" outputs!

The un-balanced connections go to the 2142 inputs and the outputs are the balanced connections. Wire them according to the spec sheet, and don't forget the .1uf bypass caps on each IC's power pins.

Yes, they work great. I thought about making a similar circuit from opamps when I was writing about opamp solution, but with these SSM parts still available, and for that price, I'd better just buy the SSMs.

Maybe I'll make 6 channels of SSM and 2 channels of small transformers for coloration in each direction.

The more I think about it, the better I like the diy solution!

This approach completely eliminated the "digital" hash in the connection, and also removed any ground loop hum, since there is no gnd connection between the soundcard and the mixer. Works great for MIDI racks too.

This is very encouraging!
BTW, my Phase 88 is on its way to me from Thomann. They told me I can always send it back if I don't like it.
Can't wait to try out PC recording! :cool:

JH.
 
BTW, my Phase 88 is on its way to me from Thomann. They told me I can always send it back if I don't like it.
I'm using the earlier one, EWS88MT and am happy with it. Once the 'difficult' multicore cable between PCI & breakoutbox got faulty and Terratec nicely replaced the whole package.

Bye,

Peter
 
You can also try the INA 134 (working from memory here, it might be INA 137) DRV 134 (or 137) combination. Works great, pretty much equivalant to the ssm chips, same small parts count. I bought a bunch of these but I can't recall if it was because of better specs or availability...
its an easy search though.
Kelly
 
[quote author="Samuel Groner"]First of all, I would try it with the "real" soundcard. Most PC built-in stuff is more than horrible, but with a mid-range soundcard, chances are better. [/quote]

Samuel, you were right.
What a huge difference, even without any balancing!
Maybe I won't need any extra stuff after all.

But I was thinking, if I were to balance it, I can still do this with 16 differential amplifiers, making the 8 nixer tape ins and the 8 soundcard inputs balaned, without using a floating balanced circuit for the outputs!
Still could put this in a single box with two power supplies.

But for the moment, I'll leave it as it is.

Thanks for all who responded - you gave me couarge to take a big step!
(A big step for me, after 24 years of making music without a computer.)

JH.
 
PS.: Is there a forum where I can ask stupid beginner's questions about computer ecording? ASIO, latency, plugins and that kind of stuff? I know it doesn't belong _here_, but maybe you can point me to a good forum somewhere else.

Thanks,

JH.
 
PS.: Is there a forum where I can ask stupid beginner's questions about computer ecording?
Hmm, don't know how uncool it'd be to suggest this, but there's the DAW-depts at www.recording.org (a.k.a. the old place...). Recently had some fruitful discussions/info there about HDDs.
 
PS.: Is there a forum where I can ask stupid beginner's questions about computer ecording? ASIO, latency, plugins and that kind of stuff? I know it doesn't belong _here_, but maybe you can point me to a good forum somewhere else.

Hey JH, you might want to check http://forum.keyboards.de/ which has a forum for audio computer topics, it is hosted by the german magazine "keyboards". You will probably get help there, I too visit that site on a regular basis.
 
Thanks for all the hints - I'll follow them.

I'm learning Tracktion now. Seems to do all I need at the moment.

JH.
 
>> try it with the "real" soundcard.

> What a huge difference, even without any balancing!

That's what I was thinking. Even if not "true floating/balanced", good soundcards "have" to give good clean sound.

I've had some bad ones. The on-board "24-bit" sound in my icecube PC not only has noise (digital hash) about 12 bits down, the top bit does not work (anything over -6dB fs clips). However, for a joke, I got a USB hub with sound-card feature. It cost $5 more than a USB hub so I didn't expect much. It is actually very good. It isn't really 24-bit as its drivers clain, but it is good 15&1/2-bit sound (about -92dB noise floor) and ruler-flat 40-17KHz. Better than my old reference card.

This interfaces a stack of old hi-fi gear, ungrounded, unbalanced, and just stereo on 1/8" mini-stereo-plugs. So I didn't know if Pro-grade multi-channel interfacing was harder. But I'm glad to hear that good sound cards DO work in such setups. There are just a lot of bad soundcards.

BTW: most older soundcards had a semi-balanced input for the CD-drive analog connection. Even though the signal is unbalanced, the ground at front-right of the PC is not the same as the ground at rear-left. So they differential-input the unbalanced signal to subtract "ground" noise.
 
[quote author="PRR"]BTW: most older soundcards had a semi-balanced input for the CD-drive analog connection. Even though the signal is unbalanced, the ground at front-right of the PC is not the same as the ground at rear-left. So they differential-input the unbalanced signal to subtract "ground" noise.[/quote]

Maybe the Terratec does a similar thing. I've noticed that the RCA connectors in the breakout box which contains the converters, are isolated from the breakout box' metal enclosure. Semi-balancing for high level signals is quite simple, too. If you have an inverting input opamp stage, it's just two extra resistors.

But that would only be one direction. The other direction, from Phase 88 to mixer, wouldn't work that way. (unless the RCA inputs of the mixer would be semi-balanced, too.)

But maybe thee's a way to "semi-float" the soundcard outputs? Maybe they can sense the ground of what's connected to the outputs, and somehow use this for the GND point of the DAC reference? Just guessing, and not having a specific circuit in mind. But whatever they do, apparently it works!

JH.
 
Back
Top