Output stage thermal distortion
We've discussed some about the possibility of distortion caused by program-dependent output stage temperature variations and how to measure the consequences. This is important because it could bring out some causes of audible differences in amplifiers that are not uncovered in conventional lab tests.
A related issue is the different tendencies of amplifiers to be vulnerable to this phenomena. High-bias amplifiers with a large number of output transistors, like the JC-1, will naturally tend to experience less junction temperature variations with program material.
It is also likely that amplifiers built with ThermalTrak output transistors will suffer less from this problem as the thermal feedback to the biasing network is faster and more faithful.
Interestingly it seems likely that amplifiers built with the LT1166 bias controller will have fairly high immunity to this problem, since bias is set with these devices by a real-time monitoring of the voltage across the emitter ballast resistors.
Cheers,
Bob
We've discussed some about the possibility of distortion caused by program-dependent output stage temperature variations and how to measure the consequences. This is important because it could bring out some causes of audible differences in amplifiers that are not uncovered in conventional lab tests.
A related issue is the different tendencies of amplifiers to be vulnerable to this phenomena. High-bias amplifiers with a large number of output transistors, like the JC-1, will naturally tend to experience less junction temperature variations with program material.
It is also likely that amplifiers built with ThermalTrak output transistors will suffer less from this problem as the thermal feedback to the biasing network is faster and more faithful.
Interestingly it seems likely that amplifiers built with the LT1166 bias controller will have fairly high immunity to this problem, since bias is set with these devices by a real-time monitoring of the voltage across the emitter ballast resistors.
Cheers,
Bob
john curl said:
Hi John,
Yes, you're right about that. The usual implementations with the LT1166, such as those shown in the datasheet and app notes, can contribute distortion. Some of those arrangements actually pass the signal through the LT1166.
There are better arrangements where the LT1166 can be used only as a dynamic bias spreader where distortion introduced by the 1166 is quite low.
If we can get the distortion down with a better application of the 1166, then I think it may do a pretty good job of mitigating thermal distortion in the output stage.
I continue to believe that such output stage thermal distortion is not likely an issue in a design like the JC-1.
Cheers,
Bob
I tried test setup like attached...At TP are with "good" amp only "conventional" products of DUT nelinearity (and so a set of "static" IMD products) and products of non-zero output impedance, fluctuations caused by cycled thermal power are negligible. In my test are "static" IMD products about -110 dB, changes not more than 3-5dB. Square signal can be replaced by any other signal.
Attachments
stinius said:
Hi Stinius,
I used it in a MOSFET amplifier experiment about three years ago.
Its too bad that the Linear Tech guys did not put a little more thought into alternative ways to use the chip. I'm pretty sure I'm not the only one who had success with it, but I did have to give it a lot of thought and sweat some details. I was fortunate enough to get my hands on a SPICE file of it and somewhat reverse engineer a schematic out of it that gave me some insight.
The LT1166 can also be used with BJT output stages where there is enough bias spread to bias it. I think a Locanthi T circuit is probably compatibe with it. Additional artificial spread can always be added.
One key to using the LT1166 is to just use it as a shunt bias spreader, just like a Vbe multiplier, but one that acts dynamically.
Another key is to use special sauce when compensating the amplifier with Miller compensation around the VAS.
The third is to properly compensate the LT1166 itself.
These do not require many components at all.
Cheers,
Bob
Bob Cordell said:
Did it myself, about the same time, with bipolars:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1107005&stamp=1168801187
Wow, I thought I was the first one reverse engineering the LT1166, to the device level:
http://www.diyaudio.com/forums/showthread.php?postid=1485140#post1485140
Anyway, even if I don't know how the THS spec sounds, I'm pretty sure I could meet it using LT1166.
john curl said:
John,
Let be more clear. You should not diss the part if you did not do your homework. Just looking at the data sheet and app note and giving up on the part is not what "a real engineer" would do.
I'm pulling your chain here because you like to use the "real engineer" language on others.
Cheers,
Bob
Bob, you have me at a disadvantage. I can barely see, and the stress of using the computer for more than a few minutes at a time, threatens to tear my retina for a 3'rd time. It is impractical for me, therefore, to go into some 'dialogue' with you at this time on the 1166. However, I found a few notes, from 1996, that I will put up, IF anyone can locate SY, or put it up themselves. It might at least, show something useful.
john curl said:
Hi John,
I'm really sorry to hear about your problems with your eyes. I'm no youngster, either. Nor is my wife. I just spent over a week in NYC with her for some surgery she needed. It's Hell to get old, but the alternative is worse.
Anything that you feel comfortable contributing on the LT1166 is welcome. I have been trying to contribute on the LT1166 in a positive and educational way. Bear in mind that there are others on this thread who can contribute both their positive and negative experiences with the LT1166. All I ask of you is that you contribute in a technically positive way if you contribute, rather than in a way that is rude or demeaning of the contributions of others.
Making the best use of the LT1166 is not straightforward. It is a very clever device, but the datasheet and app notes do not do justice to it. This is understandable to some extent, because the focus of the LT engineers was probably more on the benefit of eliminating the bias pot that on highest sound quality. I had to work very hard to harness the 1166 in the way that I wanted to.
Here is an analogy. The ThermalTrak transistors are a wonderful device, nearly game-changing in some ways. Yet the Onsemi folks did not do a good enough job on the spec sheet and app notes to make them optimally useful to achieve their best potential. Charles feely admits that he had to workj very hard to make good use of them. Yet he did not criticize the devices as bad because it required a lot of effort to get them right.
For this reason, other designers should not feel bad that they did not get the greatest results from the 1166. Maybe they spent their many hours on getting good results in a different way.
Get better,
Bob
- Status
- Not open for further replies.