My own research into various topologies shows that it is quite possible to obtain figures like 0.03% THD and IM with no more than 20 dB of GNFB. It's not easy, not a cakewalk, but it can be done using standard, off the shelf parts. I assumed a nominal power output of 28.3 Vrms into both 8 and 4 Ohms.
Another thing which came up, and which did surprise me, is that the new breed of power devices from just about everybody, hitting well over 30 MHz, and sometimes as high up as 60 MHz, do NOT sound as good as some of the older devices, which can do 6 MHz at best. Logically, higher Ft devices should be considerably faster for Ton, Tstore and Toff, and lamost always are faster, although some older devices you would never expect it from (e.g. BD 249/250 C, Ft 3 MHz, Ton=350 uS, Tstore=250 uS, Toff=450 uS) are suprisingly fast. Two differences. One, the older devices are less to much less prone to HF oscillation, which is logical as they are slower and with lower Ft. And two, not logical, they can provide lower THD and IM figures in measurement, and actually sound with much more autrhority and presence, leaving little or nothing out. Unfortunately, most older devices are Vce open limited to around 100V or so, with the most notable exception of Motorola/ON Semi MJ 21193/21194 and 21195/21196. These are metal can TO-3 250W packages, but there are TOP plastic package versions, which are good for 200W dissipation.
In my view and hearing, these devices provide more authority and music physical presence than any of the new breed I have ever come across. The added power handling more than offsets the hassle of dealing with T0-3 package.
As you can see, I am something of a blasphemer by modern standards.
Another thing which came up, and which did surprise me, is that the new breed of power devices from just about everybody, hitting well over 30 MHz, and sometimes as high up as 60 MHz, do NOT sound as good as some of the older devices, which can do 6 MHz at best. Logically, higher Ft devices should be considerably faster for Ton, Tstore and Toff, and lamost always are faster, although some older devices you would never expect it from (e.g. BD 249/250 C, Ft 3 MHz, Ton=350 uS, Tstore=250 uS, Toff=450 uS) are suprisingly fast. Two differences. One, the older devices are less to much less prone to HF oscillation, which is logical as they are slower and with lower Ft. And two, not logical, they can provide lower THD and IM figures in measurement, and actually sound with much more autrhority and presence, leaving little or nothing out. Unfortunately, most older devices are Vce open limited to around 100V or so, with the most notable exception of Motorola/ON Semi MJ 21193/21194 and 21195/21196. These are metal can TO-3 250W packages, but there are TOP plastic package versions, which are good for 200W dissipation.
In my view and hearing, these devices provide more authority and music physical presence than any of the new breed I have ever come across. The added power handling more than offsets the hassle of dealing with T0-3 package.
As you can see, I am something of a blasphemer by modern standards.
Last edited:
Again, it is not just the topology, BUT the design philosophy.
Look back at the Otala amp from 1973. What you will find is a number of changes, I suspect primarily from Otala that depart from the audio design philosophy of the day. In truth, the actual topology used by Lohstroh and Otala is not the best. Complementary differential would have worked as good or better, but it is the HIGH Iq, compared to its competition at the time to reduce crossover distortion, high slew rate (about 5 times faster than the competition in most cases, and it just so happens, a high open loop bandwidth was built in with an exotic compensation used (almost totally impractical in most designs) and several other tweaks and mods.
Now what is IMPORTANT? We know (at least those of us who hear differences that this was an exceptional design, even though it was made relatively poorly. What made it as good as it came out to be is the real question.
Look back at the Otala amp from 1973. What you will find is a number of changes, I suspect primarily from Otala that depart from the audio design philosophy of the day. In truth, the actual topology used by Lohstroh and Otala is not the best. Complementary differential would have worked as good or better, but it is the HIGH Iq, compared to its competition at the time to reduce crossover distortion, high slew rate (about 5 times faster than the competition in most cases, and it just so happens, a high open loop bandwidth was built in with an exotic compensation used (almost totally impractical in most designs) and several other tweaks and mods.
Now what is IMPORTANT? We know (at least those of us who hear differences that this was an exceptional design, even though it was made relatively poorly. What made it as good as it came out to be is the real question.
OT (just a bit): Harry Pearson passed away yesterday. Another legend bites the dust...
Met him a few years back, at an RMAF showing off my very first L|A volume, managed to get him to sign it.
I think he felt sorry for me in some way, which I slowly start to understand.
jan
RIP
Why I Quit The Absolute Sound | The High Fidelity Report
George
Sorry Jan, I keep getting Greiner and Griesinger mixed up. In any case this is not worth the trouble. The misconception keeps propagating and if you read Greiner's paper he uses a Laplace transform of an idealized amplifier to make his conclusions. There is no mention of topology or any large signal amplifier properties, but he states his point in a way that is unfortunately too easy to misinterpret. It is easy to see from his argument that the important point was the magnitude of the input error signal of the amplifier with feedback applied and having sufficient open-loop gain 20-20kHz is what matters not that it is flat.
No need for you, the link is to "manchester" again, you have already tried it. I wanted to show measured difference linked with the files. You know there are many posts here.
Anyway, I have been very curious if someone could tell which file is with 60 ohm load and which is with not loaded output. You know these debates about 600 ohm loads etc.
The graphs show near identical result, although curiously the unloaded seems to show a higher 24kHz harmonic than the loaded. I'm guessing well within the confines of measurement error and perhaps physically moving things to make the load change.
Logic says they should be indistinguishable. I will give them a fresh listen in the morning, its to unsettling right now to concentrate (bonfire night, fireworks).
Logic says they should be indistinguishable. I will give them a fresh listen in the morning, its to unsettling right now to concentrate (bonfire night, fireworks).
Hello Pavel.Karl, I have used your Direct WAV file to create 2 files, one with unloaded preamp output, second with heavily loaded preamp output. If anyone would like to try and give his opinion which is which, here is the link:
https://www.dropbox.com/s/sndn8jpl5hlfigo/manchester.zip?dl=0
This test is completely unscientific, only two files and almost everything revealed
The differences are quite distinct (Toshiba 18'' laptop, in built harman/kardon 1'' speakers)
Blomley sounds wrong...there is an unnatural attack sounds brightness/suddenness, with harmonically wrong sustains and decays.
This characteristic conspires to obscure lower level detail.
This characteristic also causes this clip to sound slightly sped up/rushed.
Is this what PRAT is referring to ?.
Bolton is the unloaded preamp output.
Dan.
BTW, can you give schematic/photos of the preamp output stage please.
Last edited:
The distortion is mostly from ADC, then from DAC and almost nothing from the preamp (about -120 dB). At 24 kHz, it is most likely the distortion component phase. "Physically moving" is of no importance here, everything remains the same only the terminating resistor was added to the T-adaptor. Preamp output impedance is 10 ohm, 50 ohm terminator was added, and volume set to original value to keep the files matched.
Last edited:
Last edited:
This is the same amplifier, but the one wih lower LG uses local negative feedback around VAS, resistors R16 R34 connected or not to the feedback input. Distortion of the amp with higher LG is lower of course, but harmonic content is different. Does it help to choose the better sounding one?
Attachments
I know that paper and have been looking at the input error ever since I read it. Quite enlightening. No longer hung up on raw THD+N numbers
jan
IMO the version with R16, R34 connected might sound better. What is the real world result?
Today, I got my Marantz 10 vacuum tube tuner running! It has taken months of replacing caps, diodes, cleanup, but I was truly stumped for awhile. It appears that a critical connection to ground was forgotten, and not it works! Does it sound good. I first tested it with STAX headphones and it is wonderful. IF ONLY we could make audio that good with digital. '-)
Speaking for myself, John, you are knocking on a door wide open.
As a sideline, I find your sentence related to the topology they used rather self defeating. Yes, it can be argued that there were more ellegant topologies available even then, but the fact remains that in your own words it stood a head above others at the time. I could be wrong, but I doubt that going all complementary would have improved it to any meaningful level.
They did what you named and got what they got, an amp which like it or not made history and has influenced amp design to this day. Otala was the one who started analyzing loads in a later paper and poinitng out just how complex they can be and in fact actually be like nothing ever mentioned in the ads. He used AR 3A Improved and Yamaha NS 1000 as examples of very difficult loads, and later on still, even more difficult, I'd say even deadly loads appeared, like the Infinity Reference.
Their work was eventually rounded off by bodies such as IEC introducing totally new specs, like power available into 2 Ohms (albeit in a ridiculously short period of 20 mS).
While there were no doubt wide bandwidth amps before them, their publication very immediately started the Slew Rate War among the Japanese mass manufacturers, headed by Snasui and Kenwood. Over time, Japanese amps became less and less load sensitive, at least from the manufacturers of good repute for amps. In short, they got the ball rolling, they made people pause and think, that's what I percieve as their greatest achievement.
And, no doubt, they got some things either not quite right, or downright wrong. Nobody can do that much without taking a wrong turn somewhere. And it seems what we thought we knew was not necessarily so; for example, I have read several times in different magazines that the late James Bongiorno invented the DC Servo circuit, and learnt here that in fact it was not James, but our very own Richard Marsh. I like getting the facts straight, so I gladly stand corrected.
It seems the principal beef some have with Otala and Lohtroh's amp is the aspect of a wide open loop bandwidth, and consequently the amouint og GNFB which needs to be used. Now, the word "needs" is my beef with those people. Is it the actual circuit which needs a lot of GNFB, or is it a desired market position, goven that for deacdes the audio public was brainwashed that lower THD means better sound? I say it is not so, and in fact, it is often quite the opposite, because designers are pressured into delivering ever smaller THD and IM figures? Who you gonna ask to pay $5k for a SS amp delivering "only" 0.05% THD and IM?
I must have read at least 10 papers from various sources I do not wish to remember, claiming that the threshold of audibility was anything from 0.3 to 0.003 % THD. I don't know what it is, but I do know what my ears like better, and far too often, they like what is measurement wise inferior to something else better to much better. Eventually, personally I settled on the test limit of 0.1% peak distortion, 20-20k Hz, that's the point at which I say my amp will go, although it may well be able to go further before clipping.
Thus, you could say this discussion whether something has 0.008% or 0.0000008% of THD is simply ridiculous. I would like to be shown how anyone can tell the difference between the two, all else being equal. This is art for art's sake. But that's just my opinion.
I would never reject an amp made using 80+ dB of GNFB just because it does so, if it sounds good to me. Using much or little GNFB is not an end unto itself, it's just a tool to get where you want to be. It's neither good nor bad by itself, it's that depending on how it's used. In my view, using it sparingly (but not too sparingly) is good, using none is in my view not good, it leaves the sound wanting, at least the ones I had the opportunnity to hear.
Max, I heard an earlier (?) version of this circuit using BJTs - Pavel, please correct me if I'm wrong, it has been a long time since.
And let me say that in my view, that circuit will produce DAMN good sound (the BJT one). That was my first and unfortunately only meeting with Pavel's work, and after that, Pavel was a "made man" in my eyes. Because of that circuit, I am always interested in his comments, I know for a fact that he knows what good sound is.
No different than with John, I had his 1205 power amp with me at home for about 10 days, and since then, I am willing to hear his comments because again, to me, he has demonstrated that he knows what's it all about. Unfortunately, I hesitated a day too long, so instead of keeping that amp, I had to return it.
- Status
- Not open for further replies.