For me , the only real way to have a "shootout" between topologies is to build my modular voltage boards and plug them in one at a time. maybe do it in a semi-objective fashion , 1 week of the blameless , 1 week of the next ... etc.
The only correlation I can really hear is in very different topologies , symasym Vs. a basic bootstrap amp (like original DX , no CCS). Much better stereo imaging ... this is a no brainer for 2 reasons ... better PSRR and much lower HF distortion ... the sym will win by design.
I do something different now and it reflects the real world. Measure the THD harmonics just short of clipping at 1k / 10k/ 20k. The blameless and the leach will "spit" out lots of 5/7/9th in this setup while the sym and my GX will retain the dominant 2nd or at least have equal H2/3 , so will the bootstrap topo's like the DX amps.
The only attempts to address this issue I have seen in the "wild" are the electrocompiant amps which drive a leach with a symasym(single ended driving a fully complimentary). That is a lot of devices to use hoping to cancel out both even and odd harmonics (30 ??? or more) Not a good DIY endeavor.
The hawksford cascode does just about the same job with far fewer devices. I would not bother with it if I was not not VERY impressed.
So some topologies will sound better with different material ,but I am looking for one that sounds best with them all.
I might add that the differences are highlighted as you increase wattage and the quality of the loudspeakers.
OS
The only correlation I can really hear is in very different topologies , symasym Vs. a basic bootstrap amp (like original DX , no CCS). Much better stereo imaging ... this is a no brainer for 2 reasons ... better PSRR and much lower HF distortion ... the sym will win by design.
I do something different now and it reflects the real world. Measure the THD harmonics just short of clipping at 1k / 10k/ 20k. The blameless and the leach will "spit" out lots of 5/7/9th in this setup while the sym and my GX will retain the dominant 2nd or at least have equal H2/3 , so will the bootstrap topo's like the DX amps.
The only attempts to address this issue I have seen in the "wild" are the electrocompiant amps which drive a leach with a symasym(single ended driving a fully complimentary). That is a lot of devices to use hoping to cancel out both even and odd harmonics (30 ??? or more) Not a good DIY endeavor.
The hawksford cascode does just about the same job with far fewer devices. I would not bother with it if I was not not VERY impressed.
So some topologies will sound better with different material ,but I am looking for one that sounds best with them all.
I might add that the differences are highlighted as you increase wattage and the quality of the loudspeakers.
OS
Last edited:
Where are those simulations (ES and blame)?? where they removed??
OS
Some misunderstanding, i was talking of single ended
and fully symetrical differentials amp , namely those simulated
in this thread , although only the leach is a symetrical..
Last edited:
HI, Pavel
Agree on the usefulness of making a monte carlo computation ,
but i just admit that it would have been too time consuming,
not counting that we would have ended with 3D graphs which
are somewhat not easy to interpret for not used DIYers ,
so i elected to limit the sim to a few parameters.
Anyway, i ll give it a try if ever i ve got some time left.
regards,
w
Hi Pavel,
Your concerns are well worth considering of course, especially your point about FFT figuring too strongly in an assessment of an amplifier. It is but one guideline among many. I had no intention to suggest that this is all the information someone may need.
I use FFT to find problems, although with practice and more information you can get a very rough idea what an amplifier might sound like. As always, a listening test or just living with the amplifier for a while is a basic step.
-Chris
Your concerns are well worth considering of course, especially your point about FFT figuring too strongly in an assessment of an amplifier. It is but one guideline among many. I had no intention to suggest that this is all the information someone may need.
I use FFT to find problems, although with practice and more information you can get a very rough idea what an amplifier might sound like. As always, a listening test or just living with the amplifier for a while is a basic step.
-Chris
The Leach amp is good project, but choices of transistors can impair performance (0,2%- 20KHz in site the Prof M.Leach). Advantage of this amplifier is speed (60 Vus) more suitable for powerful amplifiers.
What was the Bias of the output TR's in simulation?
Hi Pavel,
I like your suggestion, and this has been bounced around before in various threads. I think the proper place for this would be in it's own thread. That might give us a feel as to how important these things are. From experience, I know it makes a difference in the diff pair and especially in the complimentary differential pair setup (where it's even more important). So finding out where the sensitive spots are in each topology might be an extremely informative exercise. It might also explain why some people like one design and others not. Execution can mean everything - or not.
Seeing which designs are sensitive to mismatch would be interesting. Putting some rough numbers on that might also allow people to figure out whether to worry about it or not as well. Both Douglas Self and Morgan Jones (transistor and tube) touched on this in their books. I think Bob Cordell probably will also. This has very probably been explained in several other electronics text books dealing with signal amplification, and simply examining the basic properties of the differential, or long tailed pair, will show that this is at least of some concern.
I would love to see a thread on this topic. Now all we need is someone who knows how to run simulations and has the time to go "cook" some numbers for us.
-Chris
Now there is a sticky mess waiting for someone to step in! In this thread at any rate.
I like your suggestion, and this has been bounced around before in various threads. I think the proper place for this would be in it's own thread. That might give us a feel as to how important these things are. From experience, I know it makes a difference in the diff pair and especially in the complimentary differential pair setup (where it's even more important). So finding out where the sensitive spots are in each topology might be an extremely informative exercise. It might also explain why some people like one design and others not. Execution can mean everything - or not.
Seeing which designs are sensitive to mismatch would be interesting. Putting some rough numbers on that might also allow people to figure out whether to worry about it or not as well. Both Douglas Self and Morgan Jones (transistor and tube) touched on this in their books. I think Bob Cordell probably will also. This has very probably been explained in several other electronics text books dealing with signal amplification, and simply examining the basic properties of the differential, or long tailed pair, will show that this is at least of some concern.
I would love to see a thread on this topic. Now all we need is someone who knows how to run simulations and has the time to go "cook" some numbers for us.
-Chris
Last edited:
While Wahab's simulations and results are useful, there is 2 much happening dynamically with amplifiers for such a simple simulation to judge it's sound. Personally i would like to see more listening from those that design and build instead of just pure simulations and it does appear most here don't even listen to their designs, instead satisfying there onus from some hyphenated simulation or playing them back thru the most inadequate of transducers and or setup ...
I'm sure some real world testing of power and distortion measured across a proper load 8,4,2, .. sine and Square wave @ 20 ,1K,20K, 100 K et al. is not 2 difficult . You guys have some if not most of these amplifiers already built there is no reason for not bench testing as a follow up to simulations....
regards,
I'm sure some real world testing of power and distortion measured across a proper load 8,4,2, .. sine and Square wave @ 20 ,1K,20K, 100 K et al. is not 2 difficult . You guys have some if not most of these amplifiers already built there is no reason for not bench testing as a follow up to simulations....
regards,
About 70mA per pair, since i simmed with two pairs.
The leach amp is somewhat dated , prove is the triple
emitter follower ,typical of a time where you could
barely achieve a 2K Hfe using a double EF once the current
reached a pair of A...
True that his VAS has low Iq , rendering the triple EF mandatory,
but at the expense of stability...
Yes , it would .....................
Thanks Wahab , Most interesting if accurate ..............
Yes, I agree with AKSA.
Comparing amplifiers using some numbers is more productive
than some individuals discussing the sound they experience.
At least this is what I have found.
On the other hand we should not make 8 Ohm THD at 2 frequencies at one power level too important.
There are plenty of more numbers and other things that can make an amplifier we like.
----------------------------------------------
Some other things to weigh into:
- Bandwidth & Phase
- Can it cope with heavy loads
- How handle complex, capacitive, reactive loads/loudspeakers
- How behave at Clipping, Overdrive
- Safety, Current, power limiting
- DC Offset stability
- Noise and noise immunity
- What parts are used. Easy to find replacements.
- Power Supply sensitivity, PSRR, CMRR
-------------------------------
Generally speaking there IS OFTEN A CORRELATION
between low THD and a good amplifier.
With some few exceptions.
(Take for example some of Nelson Pass's amps.
We can not truthfully say they are very low in THD.
Still it would be totally unfair to say they are bad amplifiers.)
A designer that strive for low distortion is most often intrested in quality regarding most everything.
I applaude Wahab's topic & work.
Interesting and we have always something new to learn.
Regards Lineup
I agree with lineup - THD figures in isolation are meaningless.
I would rather see IM figures swept from idle to full-power and clipping behavior.
It would also be nice to see the THD spectra swept from idle to full-power to identify the designs that minimize the generation of 5th and higher-order harmonics.
I would rather see IM figures swept from idle to full-power and clipping behavior.
It would also be nice to see the THD spectra swept from idle to full-power to identify the designs that minimize the generation of 5th and higher-order harmonics.
Wahab,
be the first who described and connected measurings and sound quality
maybe a manual?
Hi, Smartx
There should be no great difference between the two
designs you mentioned.
For comparison with Symasym, you can use the existing
graphs , since the scale is strictly the same..
Ostripper s AX work very well, set apart some minors
design rework needed.
In my tries, i did find his Symasym version no better
globaly than the AX.
The leach amp is not the better version of a symetrical
differential , as it is devices dependant in respect to
stability.
regards,
w
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.