[threepointone]

Just came home from the national semiconductor seminar at AES. Turns out that they're coming out with a new headphone buffer IC that's being announced in two weeks, the LME49600. It's virtually identical to the BUF634 in terms of specs; it's capable of +/-1900V/us, has 190Mhz bandwidth, 3nv/hz noise, and +/-250mA drive current (of course, these are rather preliminary, and I might have copied something down wrong, but it's definitely close to or better than the BUF634 in specs). Considering how bad TI's been with keeping their things in stock, I think it's a good time to start considering this new buffer.

National seems to be pretty dedicated to audio now--they've been designing these chips with audio in mind from the start (unlike most of the op-amps we've been using). This also means they'll be spec'ing most things at reasonable audio conditions, unlike most of the other IC manufacturers.

btw, there's also a new audio current-feedback IC and a new CMOS audio opamp coming out from national.

 

[ezkcdude]

Quote:

Originally Posted by threepointone /img/forum/go_quote.gif Just came home from the national semiconductor seminar at AES. Turns out that they're coming out with a new headphone buffer IC that's being announced in two weeks, the LME49600. It's virtually identical to the BUF634 in terms of specs; it's capable of +/-1900V/us, has 190Mhz bandwidth, 3nv/hz noise, and +/-250mA drive current (of course, these are rather preliminary, and I might have copied something down wrong, but it's definitely close to or better than the BUF634 in specs). Considering how bad TI's been with keeping their things in stock, I think it's a good time to start considering this new buffer. National seems to be pretty dedicated to audio now--they've been designing these chips with audio in mind from the start (unlike most of the op-amps we've been using). This also means they'll be spec'ing most things at reasonable audio conditions, unlike most of the other IC manufacturers. btw, there's also a new audio current-feedback IC and a new CMOS audio opamp coming out from national.

Good looking out!

 

[Sarchi]

Thanks a lot for the update, this is good news.

 

[jsiau]

I have tested the new National LME49600 and it is indeed a drop-in replacement for the BUF634 in most applications. The LME49600 actually outperforms the BUF634 in some respects. I have approved these new LME49600 for use in our HPA2(tm) headphone amplifier (the HPA2(TM) is in our DAC1 and many of our other products). We use the BUF634 inside the feedback loop of a unity-gain amplifier. In this configuration, the BUF634 and the LME49600 perform identicaly. BTW I do not recommend using either part outside of a feedback loop.

 

[NelsonVandal]

Quote:

Originally Posted by jsiau /img/forum/go_quote.gif I have tested the new TI part....

What TI part are you refering to, a new buffer?

 

[gtp]

So how is this looking for the Pimeta???

 

[jsiau]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif What TI part are you refering to, a new buffer?

Sorry, TI was a typo. I was refering to the new National LME49600. I have corrected my original post, and added part numbers for clarification.

 

[lostspyder]

Anyone know when this IC should be released?

 

[Linear]

Quote:

Originally Posted by jsiau /img/forum/go_quote.gif I have tested the new National LME49600 and it is indeed a drop-in replacement for the BUF634 in most applications..............BTW I do not recommend using either part outside of a feedback loop.

Could you comment as to why open loop is not recommended?

 

[Linear]

Quote:

Originally Posted by lostspyder /img/forum/go_quote.gif Anyone know when this IC should be released?

I second the above question. There is still no mention of the LME49600 on the National web page. Can anyone give us an update?

 

[tangent]

Quote:

Originally Posted by Linear /img/forum/go_quote.gif Could you comment as to why open loop is not recommended?

Because if it's as much like the BUF634 as everyone's speculating, it's also an open-loop buffer, meaning it has no feedback of its own. This means high DC offset, and THD of around 1%, minimum. Both go away when you wrap such a buffer in an op-amp's feedback loop.

 

[Linear]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif Because if it's as much like the BUF634 as everyone's speculating, it's also an open-loop buffer, meaning it has no feedback of its own. This means high DC offset, and THD of around 1%, minimum. Both go away when you wrap such a buffer in an op-amp's feedback loop.

Thanks for your comments. I like buffers precisely because they don't have any feedback. My experience is that steady state distortion (THD and IMD) for complementary emitter follower buffers (like the BUF634) ) is below 0.1%. I believe that this is adequate for even the most demanding audio applications. As we all know, tube circuits can have relatively high levels of THD and IMD, but they can sound wonderful. Offset can be nulled out with a DC servo loop.

The reason that I like to avoid global feedback is that it can often generate dynamic distortion (TIM and SID). As a result, circuits with high levels of feedback can measure extremely well on the test bench, but still sound harsh.

A good example of zero global feedback design is the Wadia line of CD players. They use a BUF634 as the output buffer, a current conveyor IC of my design to create a virtual ground for the DACs without using global feedback, and a DC servo loop to null out the offset. I have a Wadia 861, and I think it sounds great, but then again (pun intended) I'm biased!

Linear

 

[Linear]

Quote:

Originally Posted by Linear /img/forum/go_quote.gif ..........My experience is that steady state distortion (THD and IMD) for complementary emitter follower buffers (like the BUF634) ) is below 0.1%..........Linear

My comment about complementary emitter follower buffers being below 0.1% THD and IMD applies to circuits where they are not driving very much steady state current, such as line output stages (eg Wadia 861 CD player). When driving lower impedance loads, such as headphones, open loop, the distortion could be significantly higher.

However, IMHO, the design goal should be to use ONLY enough feedback to reduce the steady state distortion to about 0.1%, NOT 0.001%. This trade-off minimizes the risk of introducing dynamic distortion due to the use of excessive feedback.

Linear

 

[Linear]

Looks like the LME49600 actually exists, but has only been sampled to certain customers. (I think that jsiau said, earlier in this thread, that he had tested some.) I have an inquiry into National about the delivery status of production parts, but no response yet.

Linear

 

[jcx]

Quote:

Originally Posted by Linear /img/forum/go_quote.gif ... However, IMHO, the design goal should be to use ONLY enough feedback to reduce the steady state distortion to about 0.1%, NOT 0.001%. This trade-off minimizes the risk of introducing dynamic distortion due to the use of excessive feedback. Linear

Which is of course exactly the opposite conculsion from the math - a little feedback creates many close in distortion products, a lot of feedback can give a more regular harmonic distortion structure and with much reduced distortion levels - properly applied high gain feedback really reduces all errors - including "dynamic" ones

we are way past early 70's transistor amp flaws