Eric, you might enjoy this,
http://www.diyaudio.com/forums/ever...rstand-what-we-think-we-hear.html#post4396260
http://www.diyaudio.com/forums/ever...rstand-what-we-think-we-hear.html#post4396260
The whole idea is that all the amps don't cross over at once. In this application, there's no way to achieve class A operation at full power. Your biased amps are going to cross over, guaranteed.
If you look at tutorials for employing this method, you would realize that for this trick to work - class A all the time - you need to set the bias current higher than the maximum load current. Do you see what I mean?
Agreed. I have several of both the BUF and LME to try this with. I believe a 5-10W amp using an octet of buffers should be possible.
Next on order is LT1210.... 1.2 amps continuous
I use both buffers and they both work great. They are so fast that no Miller compensation is needed to stabilize the nested feedback loop with most op amps.
I looked at LT1210. It's even more expen$ive than the other buffers. Please share your experience with this device when you use it.
Thanks for your input
I haven't had much chance to experiment yet, BUT I am pretty novice really and I'm sure ill find issues! I agree these buffers aren't cheap, and the LTI device even less so. My broad plan is to try and make a small amplifier of 5W or more. The 1210 seems to make this aim slightly cheaper simpler and more compact. Some creative heat sinking will be required.
I'm not likely to have a good prototype very soon, but ill probably document it, and I'm sure ill end up seeking advice from the real experts here (I.e. not me
)quite interesting. Paralleling audio IC amps seems to lower noise , and make the stay longer in class A.
What if...we parallel 10xTDA2003 each in series with a 5 ohm sharing resistor before the output coupling cap?
quiescent current should be near 500mA wich means for sure more than 1 watt of pure class A single ended sound into a 8 ohm load .
Interesting experiment isn'it guys ? this was inspired by the massive paralleleld opamp made of ne5532
What if...we parallel 10xTDA2003 each in series with a 5 ohm sharing resistor before the output coupling cap?
quiescent current should be near 500mA wich means for sure more than 1 watt of pure class A single ended sound into a 8 ohm load .
Interesting experiment isn'it guys ?
this was inspired by the massive paralleleld opamp made of ne5532 
I prefer to have quite high sharing resistors, because I prefer losing energy as heat rather than havin unstability issues.
Yes, only part of the quiescent current will be used by the output stage. That is why I said that 500mA of quiescent current will give more less 1 watt of class A into 8 ohm rather than the 4 watts it would have been if all the quiescent current went to the load. 1 watt seems to be a resonable guess.
In your experience, is the soud of paralleled TDA good? do they sound warmer than non paralleled TDA? How many did you paralleled?
I paralleled 4 and ran them bridged, so a total of 8 chips per channel. I don't recall them sounding any better than a single, but of course they played louder.
If you want warmer sound from a chipamp, pay attention to the amount of noise entering from the power supply. Using a transformer on the output is one way to improve the PSRR - running bridged without a balanced supply is another.
If you want warmer sound from a chipamp, pay attention to the amount of noise entering from the power supply. Using a transformer on the output is one way to improve the PSRR - running bridged without a balanced supply is another.
The more devices you have in parallel, the higher you can make the sharing resistors. This can be a big advantage.
If the buffers are AC coupled, then you can use individual output capacitors on each buffers. 10 x 220uf = 1 x 2200uF. This is another nice advantage of paralleled buffers. And in this case, the DC feedback can be totally decoupled between each buffer. I prototyped a circuit like this and it worked as expected. No global DC feedback allowed for less critical component selection.
If the buffers are AC coupled, then you can use individual output capacitors on each buffers. 10 x 220uf = 1 x 2200uF. This is another nice advantage of paralleled buffers. And in this case, the DC feedback can be totally decoupled between each buffer. I prototyped a circuit like this and it worked as expected. No global DC feedback allowed for less critical component selection.
Agreed. I have several of both the BUF and LME to try this with. I believe a 5-10W amp using an octet of buffers should be possible.
Next on order is LT1210.... 1.2 amps continuous
Try AD815 - dual opamp with 500mA per side max Iout....
I have a power amp with 8 chips per channel, works wonders for my tweeters.
Flat freq response and flat THD response up to 100kHz+
Jan
^ $14 a piece! When do you guys say, "the heck with it" and just make a discrete OPS using really high Ft driver transistors (or, heck, small signal transistors) in some sort of EF2? Park your favorite, fast, opamp in front of it (EF2 for a low power OPS should keep the opamp thoroughly unloaded) and have yourself a ball.
I built headphone amps with both 49600 and BUF634. Everyone that's auditioned them say they're the best they've ever heard. They're the best I've ever heard too.
Obviously they have not hear a class A single ended headphone amp in their lifes. Those are truly THE BEST.
Try ESP A class amplifiers, you will be extremely surprised. Indeed Rod Elliot is much smarter, creative and knows so much more than Doug Self or Nelson Pass or John Curl etc etc etc....
This is beyond any doubt, nobody can successfully prove the opposite
I have considered a discrete option. I have bucket loads of BD139/140 I could parallel....
Obviously they have not hear a class A single ended headphone amp in their lifes. Those are truly THE BEST.
Try ESP A class amplifiers, you will be extremely surprised. Indeed Rod Elliot is much smarter, creative and knows so much more than Doug Self or Nelson Pass or John Curl etc etc etc....
This is beyond any doubt, nobody can successfully prove the opposite
I prototyped the JLH class A amplifier. It was stellar and I almost built it.
These buffers are class A up to 15 mA when biased for maximum quiescent current. And you can build them on a much smaller board, and they are more practical for many applications.
Indeed Rod Elliot is much smarter, creative and knows so much more than Doug Self or Nelson Pass or John Curl etc etc etc....
This is beyond any doubt, nobody can successfully prove the opposite
All four bring a wealth of experience, both theoretical and practical to the table. All have their own specialist skills and areas of expertise.
If you study the published works of these guys then you will see just how extensive their in depth knowledge really is.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.