From Brian Willoughby Sent Mon, Jun 18th 2018, 01:13
From Brian Willoughby Sent Mon, Jun 18th 2018, 01:13
I believe that a modular system should be capable of 20 Vpp. That would = be with the op-amps supplied from filtered +/-12V supplies, and then as = much as 2 V headroom on either side, depending upon the specific op-amp = model. I think that=E2=80=99s equivalent to +19.2 dBu (*), so there are = mixing consoles that could handle full modular audio signal levels. I = suppose you might need to engage the Pad on a mixer for +19.2 dBu = inputs. My computer audio interface has a digital pot for the pad, and = can handle +24.5 dBu, although I have yet to test it against full = modular outputs. Other modulars use regulators that bring down the 12V, so they=E2=80=99d = have less than 20 Vpp. Maybe only 16 Vpp or 14 Vpp. The Befacto uses the = modular rails directly, although they=E2=80=99re heavily filtered to = avoid noise coming in from other modules over the power rails. The 1N5817 reverse polarity protection diodes on the Befacto drop the = rails by 0.45 V each, so the op-amps see +/-11.55V. I=E2=80=99m sure = they can still put out close to 20 Vpp given they have 23.1 Vpp on their = power. I did a poor job of describing the headphone circuit because I was going = from memory. The two op-amp stages are not literally in parallel. The = first stage has gain of 1, although that gain is modified by the pot = setting and certainly varies according to resistor tolerances. The = second stage is a voltage follower which should have an exact gain of 1 = without any variation since there=E2=80=99re no resistors to set the = gain. This allows the second stage to more closely match the first = stage, I=E2=80=99m sure, although they still use summing resistors in = case they don=E2=80=99t match exactly. The only purpose of the second = stage is to double the current capacity for driving the headphone = speaker. Brian * I sometimes do those dB conversions wrong, so don=E2=80=99t quote me = on that exact +19.2 dB figure. I still want to find a standard that = tells how to calculate balanced signal levels from peak-to-peak = voltages. The important thing to remember is that fully balanced audio = would have 20 Vpp on the (+) *and* 20 Vpp on the (-). When combined, = that=E2=80=99s the equivalent of 40 Vpp at the output of a unity gain = differential input stage. It=E2=80=99s fairly easy to convert unbalanced = signals to dBV or dBu, whether using RMS Voltages or peak-to-peak = voltages. It=E2=80=99s the balanced dBu measurements that I=E2=80=99m = frequently getting wrong. On Jun 17, 2018, at 5:39 PM, Quincas Moreira <xxxxxxx@xxxxx.xxx> wrote: > Brian, I'm super happy with the sound of the Befaco, both the main = outputs and the headphone outs. Loud enough, clean enough and = transparent enough. I think they take advantage of the huge modular = audio levels (aprox 10vpp). In fact the first amp has a gain of 0.4, to = bring that modular madness down to line level. The one thing I don't = quite get is why they have the 5532 as a voltage follower after the = inverting amp stage. Wouldn't that first stage be enough thus saving = them a whole 5532? > Thanks for the detailed response BTW! Nice to know there's noise = cancellation, even if it's not true balanced. > Cheers > Q >=20 > On Sun, Jun 17, 2018 at 6:44 PM Brian Willoughby = <xxxxxx@xxxxxxxxxxxx.xxx> wrote: >> Hi Quincas, >>=20 >> The Befacto employs the first and most effective aspect of a balanced = line, even though it is not fully balanced. Many people focus on the = identical but opposite polarity signals in a fully balanced line without = realizing that it=E2=80=99s also very important that the impedance to = ground for each of the (+) and (-) lines be equal. If the impedance is = different, then the noise picked up by the cable and other parts of the = interface circuit will have a different amplitude in each of the (+) and = (-) lines, and then it won=E2=80=99t cancel out so well at the balanced = receiver. Analyzing a circuit to make sure that the DC and AC impedance = is completely matched on each line can be quite difficult - many = products from well-known brands have failed at this, as surprising as = that may seem. >>=20 >> The Befacto circuit saves money by not actually bothering to put the = inverted signal on the (-) line. Instead, they focus on making the = impedance of both (+) and (-) match, at least at the first order, by = using identical resistor values. I think that the Rane link that someone = else shared might actually mention this circuit setup as a sort of poor = man=E2=80=99s balanced. It=E2=80=99s not the best. >>=20 >> So, the Befacto outputs are set up so that any noise you might pick = up on the cable will get canceled out by the balanced input that you = connect it to. The signal will be 6 dB lower in amplitude, because the = (-) is 0 V, and the S/N will therefore be 6 dB worse than it could be, = but you get noise cancelation for the cost of a resistor and whatever = extra cost for the balanced connector. An unbalanced setup would not = have any noise cancellation. >>=20 >> I haven=E2=80=99t had time to completely review the circuit to see = whether there are some other impedances contributing to the total in an = unbalanced way. A quick guess is that the PCB layout might ruin its = potential if the ground paths are shared between the (+) and (-), but = probably only when significant current flows. >>=20 >> The best descriptions of true balanced audio come from Bill Whitlock, = of Jensen Transformer fame. Somewhere, he has a paper that diagrams the = various popular audio circuits for unbalanced, fake balanced, poor = balanced, and fully balanced topologies. Each diagram has a depiction of = how the current flows for each of the signal and noise currents, and = shows how the noise gets canceled (or not, as the case may be). I tried = looking for his paper, but haven=E2=80=99t found it yet. Thanks to the = AES (Audio Engineering Society) for supporting people who research all = of this stuff, find ways to make audio better, and then teach the = industry how to avoid common mistakes. I feel like I=E2=80=99ve received = a free continuing education thanks to the PNW Chapter of the AES. >>=20 >> Good eye, Quincas! >>=20 >> Brian >>=20 >> p.s. The Befacto uses two op-amps in parallel, but with resistors to = mix the outputs properly, all so that the headphone output has enough = current to drive headphones directly. I haven=E2=80=99t heard the = Befacto in person, but it probably sounds better than a lot of systems = I=E2=80=99ve suffered through which use specialized =E2=80=9Cheadphone = amplifier=E2=80=9D chips with way too much noise for my tastes. If this = dual op-amp design is loud enough, it probably has better S/N = performance. I=E2=80=99m just guessing, though. It can be tough to get = enough gain for headphones to be audible on stage, what with all the = competing sounds. >>=20 >>=20 >> On Jun 17, 2018, at 3:25 PM, Quincas Moreira wrote: >> > And actually, I was just looking at the schematics, and the output = is TRS and reportedly balanced, but I don't see an inverted signal at = the ring, only a 1k resistor to ground... Is this legit balanced? Or = just a way to send unbalanced out through TRS? Brian? >> >=20 >> > https://www.befaco.org/docs/Output/Output_V3_schematic.pdf >> >=20 >> > On Sun, Jun 17, 2018 at 4:43 PM Quincas Moreira wrote: >> >> Hey, kind of on topic here, I just did a review for a module that = is basically a dedicated DI box for modulars, with the added bonus of a = Headphone output with a switchable cue input for auditioning sounds you = don't want your audience to hear yet... >> >>=20 >> >> https://youtu.be/SN1BNx3amM4 >=20