[NelsonVandal]

I've always found this amp a bit on the dark side. It improved a great deal after the removal of the opamp, but still a bit dark and unclear. So now I've removed the small 47 nF capacitors going from output to ground. I can't see why they're there. At the same time I removed the 22 Ohm output resistors. I don't know if it's placebo, but I find it less dark and slow with a more extended treble.

Do you know why these small caps should be there? Isn't this a low pass filter?

 

[NelsonVandal]

I've done some more listening to it. I'm quite sure that the sound is brighter. SQ is not on par with Sijosae's buffers. Next thing I will do is swap those extremely cheap transistors for some real audio transistors like BD137/138.

I'm not sure what diodes those are. They are SMD and marked 47u. It's a double diode. It could be BAV199 or something similar. I wonder how much they affect the sound. Is there a better part?

I also wonder about the bias. It's about 1 mA now with the 3k resistor. Will it sound any better if I increase the bias? Maybe a CRD will be better? What do you think.

After those mods the final mod will be adding an active ground channel. I'll probably use LMH6654 for that. I'm quite happy with it in another amp, and it's lean on quiesent current. I don't think there's room for a bufferd opamp.

Do you have any suggestions for further improvement?

 

[FallenAngel]

No idea why that cap would be there, makes little sense to me, but I'm no engineer.

CRD or JFET CCS instead of 3K resistor should yield much better results. Increasing the bias might also be a nice idea.

 

[NelsonVandal]

It was a real bitch to roll those transistors but now BD137/138's are in place. A bit more DC-offset I'm afraid but acceptable at 12 and 18 mV. Gone are the grainy lack-of-detail sound. A bit brighter than I expected, nicer punchier bass. Synergizes well with Mylarthree, makes them sound less grainy. The treble is a bit synthetic compared to Sijosaes buffers when using HD650, but that is without any burn in time (I'm not sure I believe in burning in though).

Due to the size of BD137/138 there's no room left to tweak bias with common resistors. The SMD's used are very very small, and I don't have any at hand.

Still a bit "monophonic" soundstage due to crosstalk i assume. Hope I can cram in an active ground channel later.

 

[TzeYang]

maybe you should try raising the output resistors to 4.7Ohms. The offset could be due to the mismatch of the output transistors.

Nelson, why not just skip the whole Xtra X1 amp and go for discreet buffers that are waayyyyy cooler and sounds better?

 

[NelsonVandal]

It's not meant to be the amp. I'm just trying to get the Xtra to sound as good as possible. I already screwed it up while rolling opamps (PCB traces broken) so I can't sell it to anyone. I think it can become a good amp for a portable rig with CrossRoads mylarthree and a DAP. I wouldn't want anything bigger than this amp.

And yes, the transistors are mismatched. I bought 25 each, but the Hfe's doesn't match a bit

, even though all are ON-semiconductor BD137G and 138G. Do you really think changing output resistors to 4,7 Ohm would help.

 

[TzeYang]

would reduce the bias a little. I think it'll help stabilize the circuit a little bit.

Heck i use 4.7Ohms these days. They provide some amount of protection too.

C'mon you know you want to ditch this amp and start making tiny yet good sounding buffers on a perfboard instead!

 

[leeivan86]

Desolder the stock opamp and add a AD8620 into it. You will be surprised at its change. Larger soundstage and more airy sound with great clarity was the end-result I got.

 

[NelsonVandal]

Quote:

Originally Posted by leeivan86 /img/forum/go_quote.gif Desolder the stock opamp and add a AD8620 into it. You will be surprised at its change. Larger soundstage and more airy sound with great clarity was the end-result I got.

I did have AD8620 in it, and LM4562, and AD8058, and a ditched them all. I don't have any opamp in it, only the buffers. This is way better than AD8620/buffers. Better soundstage, more intimacy, more air, way better clarity. Even with HD650 I rarely need more than unity gain. The amp sounds very close to nothing, and nothing is my goal. I don't like coloration.

And TzeYang, I'm about to build a small perfboard amp. Buffers only of course. I'm eager to find a no-opamp solution, and I'm going to build a pseudo balanced amp. I hope this will cure the crosstalk problem when using an open loop buffer as active ground in a 3-channel configuration.

 

[NelsonVandal]

This amp really sounds good now. Probably better than Sijosae buffers, wich I think is thanks to the BD137/38. The output impedance is lower, and I believe it has less cross talk. But I'm not really sure, have to listen some more. Right now I'm not very eager to modify it further, it sounds so good right now.

 

[NelsonVandal]

I did the active ground channel mod as well. It's a LMH6654. It was hard work. I had to cut the ground plane to isolate the output jack, remove the gain setting pins and the tantalums to be able to fit the small board with the opamp, and drill a hole in the PCB for power wires. I added 4 SMD tantalums and a ceramic on the new board.

On the good side it sounds better, improved soundstage and less monaural sound. On the bad side there's about 15 mV DC offset. I'll have to find a way to reduce the offset, so it will be safe to use with in-ears. Sounds great with HD650.

 

[nikongod]

the 47nF cap is there as a LOW pass filter. if the amp oscillates or rings, the cap "sends" the HF oscillation through the cap into ground. better than HF oscillation hitting the headphones full force.

perhaps i read what you did incorrectly, but this is my bit:
the 3.3ohm resistors can likely be "shrunk" to a smaller value, but total removal is sketchy. as the value of these 2 goes down the distortions go down BUT the likelihood of oscillation goes up. matching them to a pretty fine tolerance should not be hard.

also, you can probably get away with swapping the 22ohm output resistor for a smaller one. like 11ohm. if you are NEVER going to short the output, remove this resistor, but its risky.

 

[NelsonVandal]

Those transistor aren't that big, but in this amp they do look big. Rolling transistors was the most major improvement except for removing (or rolling, best was AD8620) opamp.

Hard to see there's an opamp there. Tant's and switches had to go.

I think there's something wrong with this amp, there's still warranty, isn't it? I had to cut the ground plane on both sides, and get rid of some vias. The wires are voltage supply to the ground opamp.

 

[nikongod]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif I think there's something wrong with this amp, there's still warranty, isn't it?

if you send modified gear back to the MFR as defective, they will laugh themselves sore before they call you back and decline repair work.

it is also opposing a couple of the codes of "ethics" that has been posted for diy'ers and modders.

 

[NelsonVandal]

It was supposed to be a joke. There's nothing defective with it. It sounds very very much better than original. If Xtra had mounted some decent transistors in it and used a better opamp this would have been a killer amp that would be worthy of the name Pro.