I looked up the data sheet for the DRV134 and see that it is a self-balancing paraphase driver - if one output is shorted the other output rises so the overall gain is always +6dB. The chip has internal output resistors, so R33-36 are not needed. neither are the sense caps since the DC levels of the outputs will be very matched by the chip itself, and any DC on the actual XLR is blocked by the coupling caps.
That's good, I'm not an expert on DRV134 and had recommended using an ordinary op-amp, so I didn't look at the DRV134 closely. What is the value of the built-in output resistor? Because the coupling cap might have to be adjusted to give the designed high pass threshold of 20Hz.
Whoever said above that the "real XLR is twice as loud as the fake XLR" is incorrect. having twice the voltage is only a few dB difference in loudness (6dB, which is noticeable but not a factor of 2) loudness-wise).
Yes, I agree, I was typing quite quickly so I didn't realize I'd referred to loudness instead of voltage. Oops.
The input buffers are going to add noise even though the inverting mode is lower THD than non-inverting. It is simpler just to have a 50k leak resistor and a 1k gate-stop into the chip to have high-z input without the noise penalty of 50k series resistance. This change reduces noise by about seven times.
In my view the noise added by any modern op-amp is negligible (orders of magnitude lower than what could be detected by the human ear), so the choice of how many stages to have is governed by convenience rather than noise considerations.
The reason he's got the inverting input buffer stage is to provide a low impedance to the volume control stage. Since the volume control is a pot to GNDREF (I recommended a 1k log pot, not sure what he has now), if this was fed directly from the external input then the resistance of the externally connected source would affect the volume level.
Another reason he's got the inverting buffer stage is to bandpass the external input to reasonable audio frequencies, he has something like 20Hz..30kHz at the moment. While this isn't strictly necessary, I think it is a good idea to be cautious. And the 20Hz lower limit also serves the purpose of shifting the signal up to half the supply, as it's a single-supply circuit.
An alternative way would be to use a high-value pot for the volume, fed directly from the external input, and assume the resistance of the external source is much less than the volume control pot. Doing it this way makes it significantly harder to bandpass the input, and also doesn't provide a low impedance output to drive the summing stage. If you buffer it after the volume control you haven't saved any op-amps/stages.
Anyway, I would be interested to know if there's a better way to do it, so I will check back later for any developments.
The summing stage needs a feedback cap from output to virtual-earth node to stabilise the stage, as does the EQ.
I definitely think it is a good idea to put space on the board to fit this should it be needed, but I don't think it will be usually, because the op-amps are compensated for unity-gain stability, and he's only using them at unity-gain.
And he's already filtered out undesirable high frequencies at the input stage, so there is no need to do this again. (Each time filtering with a 30kHz cutoff leaves you 3dB down at 30kHz and this is undesirable if you do it too many times, you'd need to increase the cutoff in such case).
The supply splitter should have a cap across one of its input divider Rs so the opamp input is at AC ground, and thus the output will be similar.
This is another interesting chip which like the DRV134 I'm not that familiar with, and I recommended to use an ordinary op-amp for this. So I didn't look at it closely. I originally suggested to omit one of the capacitors, but he is using the recommended circuit so I think that's OK.
You say the 24V comes from a SMPS in which case you will need some serious filtering at the DC-IN jack: a series 10uH choke follwed by 22uF ceramic to ground in parallel with 220uF electrolytic, feeding a ferrite bead to to a second 22uF C to ground into a linear reg with 22uF C to ground, another ferrite bead feeding another 22uF C and 470uF E to ground. The linear reg could be a low-drop-out type or simply a standard type, or a discrete circuit. It seems like a lot of circuitry but SMPSs are truly horrible noisy devices and issue forth a tsunami of high-frequency noise. They are not good for supporting linear circuits directly and we've found that an interface circuit is needed to keep the SMPS happy and to provide a low, constant impedance over the audio band for the audio circuits.
Yes, that's a good idea. I can't help wondering if the linear regulator might be overkill though, could you get away with just the 2-stage LC filtering? And could you use ordinary inductors here? Or is there a particular reason why the ferrite bead is the right thing to use in this case?
I would certainly use a low dropout regulator, and try to set the operating voltage of the circuit just under the 24VDC output of the SMPS, to avoid lots of voltage drop and heating in the regulator. But obviously there needs to be a bit of margin in case the SMPS is outputting a bit low.
By the way, Jbliss: I noticed that you've got the GND connection of the incoming audio going to GNDREF. I have the feeling that it is dangerous to put such a signal to the outside world, unprotected. Since static on it could damage your supply splitter, etc. Also, in the originally sketched design that I suggested, there was no supply splitter and the 20Hz high-pass had a dual function of level-shifting the input so that the GND connection of the incoming audio could go to GND and not the half-supply reference. Since you have a supply splitter, it might change things.
cheers, Nick
Home
Help
Search
Login
Register
