lme49600 buffering the output of a generic soundcard

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello, I have a generic Audigy soundcard which I use for my Sennheiser 558 headphones during the long hours I am working at the computer. I plan to finally finish my USB>DAC rig that is gathering dust but I don't have the time and I need a temporary solution to boost the low end of the headphones and give it more control.

The soundcard outputs plenty of voltage, I can't listen to them at full volume, so I was thinking to use a current buffer, such as LME49600. I would just use something like Fig.8 in the datasheet or the buffer module from Twisted Pear Ventus amp. No gain op-amp before the buffer, just the soundcard driving the LME49600 directly.

Of-course I will also throw in some high quality regulators. What do you think about this? Would this work? Probably the gain of the set-up will be a tad smaller, but that's just fine with me.
 
You'll always see the LME49600's THD stats quoted in a nested feedback config -- such as Figure 2 in the datasheet. It just means that you'll need a high quality opamp like the LME49710, or pick your favorite flavor from TI (National's new parent). If you're going for some regulators of the "off the shelf" variety, try the Linear Tech LT1963A/LT3015.
 
Thanks for pointing this out, much appreciated. I've noticed that too but after a quick look at Ventus it seemed to me they were operating the buffer outside a feedback loop. I took a look again now and no, the implementation is the standard one.

So why exactly needs to also have feedback? The reason I would like to stay away from an op-amp is because you can't operate them properly at unity gain and I simply don't need more gain.

So, any ideas?
 
I simply believe the soundcard output is weak and doesn't have the current to drive the 50ohm Sennheiser. It seems to me it should have more bottom end and sound more tight as I experienced with some other amps.

I could build a new amp, with some gain and use it with my U2I usb card and AK4396, but it then needs a case, planing, etc. I just want a quick fix.

Thanks!

edit: If you can suggest some other solution for this, then please do so. I'm not fixated on LME49600, I just noticed it became popular for headphone amps. Before it was BUF634 :)
 
Last edited:
If you look at LME49600's simplified circuit, minus the biasing and protection transistors, the main protagonists are the transistors Q5 and Q6 operating as emitter follower, which is essentially synonymous with buffer.

Now, output of emitter follower (buffer IC or discrete components) doesn't exactly follow the input because of *blah blah textbook content* in short you get distortion if you leave it open loop. If you have feedback then the op-amp knows the output is inaccurate and adjusts till it's correct.

Some people like the sound though... anyway though we say higher distortion if left open-loop it usually isn't *that* bad that it will make ears bleed, but measured performance definitely takes a hit.

Personally I'd prefer discrete output transistors and an op-amp designed for that, there is this LM part designed for this... what is it again...?
 
I simply believe the soundcard output is weak and doesn't have the current to drive the 50ohm Sennheiser. It seems to me it should have more bottom end and sound more tight as I experienced with some other amps.

I could build a new amp, with some gain and use it with my U2I usb card and AK4396, but it then needs a case, planing, etc. I just want a quick fix.

Thanks!

edit: If you can suggest some other solution for this, then please do so. I'm not fixated on LME49600, I just noticed it became popular for headphone amps. Before it was BUF634 :)

The specs for your soundcard should indicate the minimum headphone load. If it is higher than 50 ohms, then you do need something to drive your headphones.

The BUF634 comes in a TO-220-5 package, which is a through hole device, so it's easier for DIY to use. You will need some kind of opamp in front of it, as others have stated. There are plenty of headphone amp kits on this message board that you could use if you don't want to design one yourself.
 
Thanks dirkweight,

I don't have a problem with surface mount devices, it saves me from drilling to many holes in the pcb :). And I can design a better pcb for the amp (proper groundplane, etc.).

I just don't want to go after a build with high gain, meaning additional volume control, fancy case, added complexity, etc. Voltage gain is not trivial to get right and, well, my soundcard already has a gain stage, don't really need a new one. :) I thought a simple current buffer with quality regulators, tucked away in the pc case would be enough. And fast enough to prototype.

How about the standard LME49600 or BUF634 implementation, with feedback loop, but with a stable unity gain op-amp, such as OPA 820 ?

Maybe I could live with a gain of two.. Will the buffer remain stable with such a small gain?
 
Many op-amps are unity-gain stable, including the JRC 4556/4558 on Audigy, which are probably being used as unity/low gain.

3x4558 and 1x4556, I can only assume that the 4556 handles the front out (I've traced the connections before but long forgotten and threw away the paper), and it is a pretty good choice for headphone loads. But IIRC the output has a resistor in series (47 ohm?), which is pretty standard for sound cards, but will affect headphone's sound quality.

For a quick and easy mod I'd just short that resistor and see if the sound becomes satisfactory.
 
Thanks dirkweight,

I don't have a problem with surface mount devices, it saves me from drilling to many holes in the pcb :). And I can design a better pcb for the amp (proper groundplane, etc.).

I just don't want to go after a build with high gain, meaning additional volume control, fancy case, added complexity, etc. Voltage gain is not trivial to get right and, well, my soundcard already has a gain stage, don't really need a new one. :) I thought a simple current buffer with quality regulators, tucked away in the pc case would be enough. And fast enough to prototype.

How about the standard LME49600 or BUF634 implementation, with feedback loop, but with a stable unity gain op-amp, such as OPA 820 ?

Maybe I could live with a gain of two.. Will the buffer remain stable with such a small gain?

The BUF634 has a bandwidth of up to 180MHz, as does the LME49600. From what little I know, you want an opamp that has less bandwidth than the buffer for stability. So, the OPA820 has too much bandwidth. I'd go with a standard audio opamp like the LME49710 or NE5534 instead. Also, with such high bandwidth devices, you'll be wise to have a good RF proof box and related stuff to keep it from oscillating. Proper bypassing is a must also, from reading the datasheets.
 
Thanks for replying, much appreciated.

I'll bypass the output resistor first, to see if anything good happens and then I will probably go along and try a simple op-amp + buffer with a fain of two and check what happens. I'll use quality decoupling and regulators.

It's kind of a bummer to go that route because with two quality op-amps and two buffers I'm already almost to a complete headphone amp, kind of The Wire amp, with DC servo and all the rings and bells.

I would've like a more straightforward solution... If I'd had more info about running one of these current buffers open loop I might even try that. As long as the distortions don't add to a significant level..

So yeah, I'm still all ears for a simple way to boost the current output of this sound-card without increasing the gain.
 
Aside from reduced output resistance, you can also try adding buffer capacitance for the 4556's supplies. Simply adding a 470µ from V+ to V- has been reported to help on either a Live! 24-bit or an Audigy SE (don't remember, but they're very similar cards anyway).

On a Live! 24-Bit, the whole analog section appears to be powered from +/-5V generated by two LDO regulators (ST LD111750xx and TI MC79L05A or somesuch, good for 100 mA each), with four 22µF electrolytics floating around nearby, which I guess are in/out buffer caps. From regulator outputs to OP power pins I only measure a fraction of an ohm, which should be OK. Buffer caps and regulators strike me as a little wimpy though, hardly unexpected on a budget card.

Poking around the opamps reveals output series resistors of 220 ohms for the line-outs and 33 ohms for the headphone out (I think these are only 0402, really tiny stuff, presumably with questionable linearity at higher power). Some sort of 6-legged critter marked 02NG is in series with the headphone out and may be used for switching between 4558 and 4556 plus muting (I did not find any noteworthy direct connection to the jack with power off). Expect some extra series resistance there. Some more of these were planned on the line-outs but left out and jumped by 0-ohm resistors.

Man, I really need to get some finer probes. Mine are far too fat for this SMD stuff, and poking around on the card is really frustrating.
 
Last edited:
Two op-amps and buffers and you *are* at The Wire (there is no servo). :)

The easiest thing to try first would be to add an electrolytic across the supply pins as sgrossklass mentioned, I would try something like 220uF or 470uF. Bulk capacitance is very often less than it should be in mass produced items, and bass quality is what usually suffers.
 
I don't see why not just try out the stand alone lme49600 buffer? The output of the soundcard is a opamp that can drive the buffer without trouble. I really don't think that using the buffer outside of the opamp feedback loop is going to be such a big deal in this aplication. The only thing to watch out for is dc offset at the output of the chip
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.