15-01-2021, 10:30 PM
I would definitely go balanced for this. How well it'll work with your existing cable will depend on many variables, not least how well designed the input stage of the speakers is, but balanced operation will give you the best chance.
You don't need another op-amp section to generate an anti-phase signal.
People assume that "balanced audio" means "differential signalling". But that is entirely optional.
What balanced audio really means is "identical resistance and capacitance to ground". That means that any interference source will affect both "hot" and "cold" signals equally, meaning the subsequent input stage will be able to reject them. This has nothing to do with the wanted audio signals.
So from your NE5532, pick an output resistor. Perhaps something in the 22 to 47 ohms region. Send that down one leg, have it end up on pin 2 of the XLR.
For the other leg, just get another resistor of the same value and return it to ground. That'll go to pin 3 of the XLR.
From the point of view of interference rejection, this will perform just as well as a pair of op-amps would.
Some people disbelieve this, or argue that it's cheap and nasty or something else equally spurious. Really, the only downside is a 6dB loss in headroom, which is hardly relevant here. For the doubters, I gently point to the schematics of various microphones that use this scheme, and remind them that if there's an audio application where noise rejection is paramount, it's microphones! Also, plenty of sound desks use this scheme for outputs - if not the main outputs, then often aux outputs and similar. It's often referred to as "impedance balanced".
Personally, I would always include output caps - perhaps 47uF, and bi-polar if you have them. And some resistance to ground on the output side of them - 10 to 100k - so that they don't cause pops when plugging up.
The Alesis appears to be DC coupled at the input, via the volume control and onto the next op-amp. That's naughty. If the volume control starts to crackle, serves 'em right! Having seen that, I would definitely advise you to include those caps. Though the PDF of the schematic I'm looking at shows two versions, and one of those has incorporated caps either side of the volume control. No idea which you'll have.
As your cable is unshielded, I'd also consider putting 100-470pF in parallel with the output resistors grounding the output side of the series caps. Value not critical, as long as they're identical.
Of course, even if you're not building anything between the 34 and the cable, you could still go balanced if you know what Zout of the 34 is. Though IIRC, it's high-hundreds, which will leave it rather more prone to noise pickup than a modern pre-amp. Bear in mind the input stage of the speaker might only have a CMRR of 30-40dB, given that there's no trimming, so personally, I'd prefer to give it the best chance possible by minimising Zout. If you're not hung up on originality, there's no reason not to modify Zout of the Quad, but if it's using muting muting transistors at the output, there are limits about how far you can take that.
For a super-simple split PSU, put a pair of zeners in series, drive via an LM317 and 100R working as a current source, supply with whatever source of floating DC you have to hand, regulated or unregulated. Given the signal levels, you'd probably get away with a pair of 7V5 zeners. The LM317 dropout voltage is around 3V. You'd probably get away with an unregulated "12V" wall-wart, but higher is obviously better if you have it. Using a current source will cut noise and ripple really well - not that it matters as the 5532 has excellent PSRR (I've got an old Philips CD player which feeds the audio stages from unregulated DC straight from the smoother caps, and there's no ripple at the output). This arrangement was used by NAD for a phono pre-amp, and was criticised by a reviewer with a little knowledge (a dangerous thing). NAD were given a right to reply, and robustly (and correctly) defended their design choices. It's a nice bit of "simple is always best" engineering, and if it works for a phono pre-amp, it'll be fine for line-level.
You don't need another op-amp section to generate an anti-phase signal.
People assume that "balanced audio" means "differential signalling". But that is entirely optional.
What balanced audio really means is "identical resistance and capacitance to ground". That means that any interference source will affect both "hot" and "cold" signals equally, meaning the subsequent input stage will be able to reject them. This has nothing to do with the wanted audio signals.
So from your NE5532, pick an output resistor. Perhaps something in the 22 to 47 ohms region. Send that down one leg, have it end up on pin 2 of the XLR.
For the other leg, just get another resistor of the same value and return it to ground. That'll go to pin 3 of the XLR.
From the point of view of interference rejection, this will perform just as well as a pair of op-amps would.
Some people disbelieve this, or argue that it's cheap and nasty or something else equally spurious. Really, the only downside is a 6dB loss in headroom, which is hardly relevant here. For the doubters, I gently point to the schematics of various microphones that use this scheme, and remind them that if there's an audio application where noise rejection is paramount, it's microphones! Also, plenty of sound desks use this scheme for outputs - if not the main outputs, then often aux outputs and similar. It's often referred to as "impedance balanced".
Personally, I would always include output caps - perhaps 47uF, and bi-polar if you have them. And some resistance to ground on the output side of them - 10 to 100k - so that they don't cause pops when plugging up.
The Alesis appears to be DC coupled at the input, via the volume control and onto the next op-amp. That's naughty. If the volume control starts to crackle, serves 'em right! Having seen that, I would definitely advise you to include those caps. Though the PDF of the schematic I'm looking at shows two versions, and one of those has incorporated caps either side of the volume control. No idea which you'll have.
As your cable is unshielded, I'd also consider putting 100-470pF in parallel with the output resistors grounding the output side of the series caps. Value not critical, as long as they're identical.
Of course, even if you're not building anything between the 34 and the cable, you could still go balanced if you know what Zout of the 34 is. Though IIRC, it's high-hundreds, which will leave it rather more prone to noise pickup than a modern pre-amp. Bear in mind the input stage of the speaker might only have a CMRR of 30-40dB, given that there's no trimming, so personally, I'd prefer to give it the best chance possible by minimising Zout. If you're not hung up on originality, there's no reason not to modify Zout of the Quad, but if it's using muting muting transistors at the output, there are limits about how far you can take that.
For a super-simple split PSU, put a pair of zeners in series, drive via an LM317 and 100R working as a current source, supply with whatever source of floating DC you have to hand, regulated or unregulated. Given the signal levels, you'd probably get away with a pair of 7V5 zeners. The LM317 dropout voltage is around 3V. You'd probably get away with an unregulated "12V" wall-wart, but higher is obviously better if you have it. Using a current source will cut noise and ripple really well - not that it matters as the 5532 has excellent PSRR (I've got an old Philips CD player which feeds the audio stages from unregulated DC straight from the smoother caps, and there's no ripple at the output). This arrangement was used by NAD for a phono pre-amp, and was criticised by a reviewer with a little knowledge (a dangerous thing). NAD were given a right to reply, and robustly (and correctly) defended their design choices. It's a nice bit of "simple is always best" engineering, and if it works for a phono pre-amp, it'll be fine for line-level.
