[Wyloch]

I'd like to upgrade my CMOY.

Thinking of rebuilding using two OPA134s, instead of the dual OPA2134. Is the extra fidelity worth it?

I'd like to do something to maybe clean the output. What exactly is an "output stage?" And what's a "buffer" do? Can I just tack an output stage on to the existing amp from the two signal output lines?

 

[ericj]

As I'm itching to use my other OPA2132PA on a board with discrete diamond buffers and a simple crossfeed, I'm curious in hearing these answers too.

Right now my only excuses are "Because i can" and "Because it's more interesting than building the same cmoy Again, and will likely Sound Different."

 

[BlazerFRS]

The idea of a buffer is to seperate the op-amp from the load (your headphones). When the amp is driving headphones, it sees a fairly difficult load (the lower impedance your phones are the tougher it is on the op-amp), if the op-amp is forced to strain too hard to drive the load, distortion and general loss of SQ can be the result. By adding a buffer the op-amp is seperated from the difficult to drive phones, allowing it to sound better.

buffers also add a ton of output current, which also helps with sq in many phones.

 

[Wyloch]

Thanks Blazer - so, I know that buffers are 8 pin chips that look just like opamps - any suggested models? And how do I hook them up - do they require power or is there just a signal in/signal out?

 

[guzzler]

Quote:

Originally Posted by Wyloch I'd like to upgrade my CMOY. Thinking of rebuilding using two OPA134s, instead of the dual OPA2134. Is the extra fidelity worth it? I'd like to do something to maybe clean the output. What exactly is an "output stage?" And what's a "buffer" do? Can I just tack an output stage on to the existing amp from the two signal output lines?

1. No, it'll likely just cost you more. You'll measurably reduce the crosstalk, but not audibly so. The single chips will run cooler than a dual as they're driving half the load in the same package, but again, not anything major

2. Moving into this takes you away from the CMoy, per se. If you look at the datasheet for any opamp, you'll see they're happier driving reasonably high impedances, and low capacitances. A headphone, say 32R Grados, present quite a hard load for an opamp to drive, and also 32R headphones need a fair amount of current. Couple this with the intrinsic capacitance of any cable, and it presents an awkward load. The output stage in the opamp is that that provides the current to the output. A buffer is basically the same as the output stage, but with a high input impedance, and an output that provides more current than the opamp can by itself. It doesn't amplify the signal (probably actually slightly less than unity gain), but does provide the extra current.

So, ericj, to your point, you can't just put the buffer after the opamp. You need to wrap up the buffer inside the feedback loop of the opamp, and the opamp then compensates (within reason) for errors in the output of the buffer. The THD measurements of a BUF634 are terrible compared to an opamp, but placing it inside the loop reduces these errors. Also, putting it inside the loop lowers the output impedance.

If you want an amp that stays true to the original CMoy type simplicity, I suggest an A47. It's basically two opamps in parallel to provide more output current. Enjoy...

 

[Mod_Evil]

Can I use a transistorized buffers?

I will make this circuit to buffer my Cmoy... I'm going to assemble a circuit similar to a Pimenta. What the advantages to use transistors in output? ( in pimenta 2n5485 and 2n5486 ) ?

Thanks.

 

[guzzler]

Quote:

Originally Posted by Mod_Evil Can I use a transistorized buffers? I will make this circuit to buffer my Cmoy... I'm going to assemble a circuit similar to a Pimenta. What the advantages to use transistors in output? ( in pimenta 2n5485 and 2n5486 ) ? Thanks.

Yes you can use that as a buffer, not the best there is, but I do know ppl likes a similar type buffer (see LISA II at iPodStudio). The 2N5486 you refer to in the PIMETA schematic is the Class-A biasing network, nothing to do with the buffer, ie the BUF634. You can keep these in your design with this buffer. May I also suggest you go to HeadWize and search for the Walt Jung Diamond Buffer by sijosae and also digi01s implementation.

 

[Mod_Evil]

Guzzler, the Walt Jung Diamond Buffer is the buffer with two LM317? Can you post the schematic?

Is this more best than the frist buffer?

What the Class-A biasing network advantages?

Thanks in advance

 

[guzzler]

The buffers you've shown are nothing to do with the Jung Diamond buffer, just some alternative ideas.

Original JDB:

http://www.elecdesign.com/Articles/A...7209/7209.html

Most people replace the cascode current sources with LEDs, as per:

[ Link here to save bandwidth ]

As for the Class-A biasing, there are two "sides" in the opamp, positive and negative. In a Class-B design, each side turns on and off depending on the output. At the crossover point, this leads to a slight mismatch on the transfer, called Inter Modulation Distortion (IMD) [someone correct me if I'm wrong on the name]. Class-A biasing involves drawing enough current to keep the output permanently on, so removing the IMD. This is only really possible with a buffered design (in terms of headphone amps), as this current drawn reduces the effective output of the opamp further.