new design v1

[DCameronMauch]

Been looking for a single chip single-ended to differential conversion solution. Ran across the fairly fresh out of the oven OPA1632 over at TI. This chip does exactly what I want. Even designed specifically for audio. Here is a little design I came up with. Still working on the PCB. But attached is the schematic for one channel. The design has a 4x gain. Designed with the HD580/600/650 in mind. Since even the stock cable is 4 conductor all the way to the plug, it is simple enough to put on a 4 conductor XLR plug. I did this with my Cardas cable a while back. The schematic does not show any of the power stuff. Also two of the pins on the OPA1632 are not accounted for. The enable pin will be left unconnected, thus defaulting to enabled. The Vocm pin needs a low impedance ground reference. That will also be connected to the AD797 ground buffer. Half of the Meier crossfeed circuit is shown in the schematic. Connect the two Meier pins between the channels to xfeed. To me, the effect is pretty minor, but it helps out a lot with a few songs that are a little too super stereo. Note with this design, there should never be more than 1uA of current flowing on the ground plane.

In version 2, I will replace the HA5002s with diamond buffers. Also look into how a Jung multi loop topology would work with the OPA1632. Might put in some PPA style current sources. My experience is with the AD8610/20 it does not make much if any difference for low currents. And the currents are pretty low. But I will build version 1 first. Already got the chip samples from TI.

 

[Budgie]

Very interesting. I have wondering lately if any of the single-end to diff. chips would have enough juice to drive a set of phones directly(kinda of like a CMoy amp for high impedance phones). I had not thought about buffering it (mainly cause I am lazy

).

I am curious as to why a buffered ground would be better then a direct connection to ground? (Because you using a single supply voltage or am I missing something?)

 

[DCameronMauch]

You might be able to drive some HD580/600/650s in single-ended mode. The chip would see roughly 300 ohms. But no way with balanced mode where the chip would see 150 ohms. Plus you have thermal variations impacting the performance of the chip. A buffer addresses those issues nicely. It occured to me the OPA1632 could be used nicely with the TPA6120A2 too. I might try my hand at a layout for that. I could throw together a schematic of what it should look like if you want.

I am a big believer in keeping all currents off the ground plane. That way you insure the ground actually has the same potential everywhere on the plane. That made sense in my head. Did it make sense to you? Any ground plane does not have zero impedance. Not exactly sure what it is. Depends on a great many variables I imagine. But by buffering it, I get a consistent sub milli-ohm impedance. Even with several mA of current.

Update: Attached is the same schematic but using TPAs instead of HAs. I could not find a clean way to connect all the ground buffer points in the schematic, so I used tags instead. So all the GB points would be the connected on the PCB.

 

[Budgie]

I have not had much luck finding any good solid info on the idea of buffered ground, but I see were your coming from. (I guess maybe it's time to order some opamps and play around with it.) I have usually grounded conventionally (star ground), and had no complaints with the results, but maybe this is a better way.