Member
Joined 2009
Paid Member
I know some people don't like this kind of messing around with perfectly good SS amplifiers but it was too tempting to pass-up - and so I'd like to recommend you give it a try too 😀
There has been much chat over the benefits of tubes for their sound. An infamous paper by Russell Hamm was one of the early attempts to explain it in terms of the harmonic profile.
Here's an extract for a 2 stage triode amplifier. Key things to note:
Low Power: the output starts off mostly as H3 at low power with H2 and H4 close behind.
Medium Power: Overall distortion heavily dominated by H2 then a strong H3 followed by H4;H5 and above remain small.
High Power: dominated by H2, some H4 with H3 falling off. H5 and above remain small.
There has been much chat over the benefits of tubes for their sound. An infamous paper by Russell Hamm was one of the early attempts to explain it in terms of the harmonic profile.
Here's an extract for a 2 stage triode amplifier. Key things to note:
Low Power: the output starts off mostly as H3 at low power with H2 and H4 close behind.
Medium Power: Overall distortion heavily dominated by H2 then a strong H3 followed by H4;H5 and above remain small.
High Power: dominated by H2, some H4 with H3 falling off. H5 and above remain small.
Attachments
Member
Joined 2009
Paid Member
I wondered how to achieve this with my TGM amplifier. The first version TGM1 used a resistor to feed some of the output to the top of the LTP resulting in some self-modulation of the signal which creates plenty of H2 and some H3 [http://www.diyaudio.com/forums/solid-state/140461-tgm-amplifier-13.html#post1803703]. But the results became muddy at higher power. Too little injected distortion and it wasn't noticeable, too much and it was nasty.
TGM2 uses a CFP for the LTP and doesn't respond to self-modulation [http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier.html]. So I created some asymmetry in the feedback loop by taking a little feedback from the emitter of an output device instead of at the junction of the two output emitter resistors.
TGM2 was a little less musical than I had hoped so perhaps this change will help.
TGM2 uses a CFP for the LTP and doesn't respond to self-modulation [http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier.html]. So I created some asymmetry in the feedback loop by taking a little feedback from the emitter of an output device instead of at the junction of the two output emitter resistors.
TGM2 was a little less musical than I had hoped so perhaps this change will help.
Attachments
Last edited:
Member
Joined 2009
Paid Member
Simulated results that Hamm would appreciate:
I've built this version and it sounds very good indeed. The bass is strong, the highs are clear. The mids are very smooth, if a little darker and overall it's one of the first SS amps that I've heard that has zero traces of fatigue.
I've built this version and it sounds very good indeed. The bass is strong, the highs are clear. The mids are very smooth, if a little darker and overall it's one of the first SS amps that I've heard that has zero traces of fatigue.
Attachments
Cool. How did you produce that distortion graph? And I'd also like to see this paper if possible...
What about high order even harmonics though? It seems to me that high order harmonics would be produced during switching. Nevermind, this would surely show up in harmonics up to the seventh...
- keantoken
What about high order even harmonics though? It seems to me that high order harmonics would be produced during switching. Nevermind, this would surely show up in harmonics up to the seventh...
- keantoken
Last edited:
Member
Joined 2009
Paid Member
The distortion graph ? - just throw the answers from spice into a spreadsheet
The infamous Hamm paper is here (and it's contested by some people of course so don't take it as gospel):
http://www.amplificadores.com.br/Texto%20para%20publica%E7%E3o/Tubes%20versus%20transistors%20I.pdf
Yup there will be higher order harmonics, not just because it's Class AB but because it has gnf. It's a compromise of course, but sounds good and easy to try.
The infamous Hamm paper is here (and it's contested by some people of course so don't take it as gospel):
http://www.amplificadores.com.br/Texto%20para%20publica%E7%E3o/Tubes%20versus%20transistors%20I.pdf
Yup there will be higher order harmonics, not just because it's Class AB but because it has gnf. It's a compromise of course, but sounds good and easy to try.
Member
Joined 2009
Paid Member
Very nice, Bigun! You discovered a cheap modification while I was off in the corner messing the NTP. The NTP has its merits, but I'm swayed. This should serve as a nice modification to many class AB amplifiers existing already.
- keantoken
- keantoken
Member
Joined 2009
Paid Member
The TGM tubey approach can only add distortion so it's limited to amps that are low THD to begin with. It helps out with Class AB that rely on a reasonable amount of gnf to keep their output stages under control and it sounds very good.
Your NTP is probably a better approach because it generates a good profile with low distortion to start with.
Your NTP is probably a better approach because it generates a good profile with low distortion to start with.
Bigun, I've tried several times to reply to your post but for some reason I simply have not found an appropriate response.
Let's just say your blameless statement has thrown my brain into deep saturation negotiating and cataloging what things would change were we to switch between NTP and LTP. For me at least, it is not a trivial change.
What I can say is that I suspect net THD might decrease and I also think the spectrum would be more benign. I think the only way to know for sure is to simulate or build, since it depends on the implementation...
Could you post simulation files?
- keantoken
Let's just say your blameless statement has thrown my brain into deep saturation negotiating and cataloging what things would change were we to switch between NTP and LTP. For me at least, it is not a trivial change.
What I can say is that I suspect net THD might decrease and I also think the spectrum would be more benign. I think the only way to know for sure is to simulate or build, since it depends on the implementation...
Could you post simulation files?
- keantoken
NTP has more H2 and H3 distortion than an LTP. Overall, the thd of the NTP, I have found, is higher, though more benign.
It is better at high amplitude than the LTP, which because of the S shaped transfer function tends to compress yielding high order odd harmonics.
Anthony, does this resolve your cerebral dilemma?
Hugh
It is better at high amplitude than the LTP, which because of the S shaped transfer function tends to compress yielding high order odd harmonics.
Anthony, does this resolve your cerebral dilemma?
Hugh
Our findings agree, except that the NTP has higher 3rd. This depends heavily on the implementation, IE whether or not the NTP has the same gain and whatnot as the LTP it's replacing (if an NTP and LTP have the same gain, I've found the only difference is that the LTP cancels 2nd harmonics; and if we substitute the NTP without changing quiescent current, the NTP will actually have twice the gain as the LTP). As long as the substituted NTP has the same gain as the original LTP, THD will be higher (2nd harmonics constitute the vast majority of blanket THD, the LTP looks so good since it chops these out).
I think that in certain implementations, the NTP might actually have less distortion than the LTP. It depends on what variables you keep constant (constant OLG or constant something else? Convenience of substitution?).
Maybe I'm just filling the problem with hot air and fussing about it.
My implementation of the NTP with the level shifters on either side DOES have higher H2 and H3, and worse THD (perhaps you were talking about this, Hugh?). It has profoundly dominant 2nd harmonics. However using this to "sweeten" the sound comes at the expense of THD (however, when OLG is increased with this configuration, odd order harmonics plummet in relation to the 2nd and 3rd since the level shifters create the distortion, but leave the NTP to work on everything else).
In any case, it is VERY hard to compare the LTP and NTP without getting into apples and oranges. If you aim for benign profile AND low THD, which apparently was my angle, then weighing the two will be a long, hard struggle. Having low THD without offensive profile and without complex circuitry, is also a long, hard struggle of weighing options. Thankfully, it doesn't seem low THD is important for audio enjoyment.
- keantoken
I think that in certain implementations, the NTP might actually have less distortion than the LTP. It depends on what variables you keep constant (constant OLG or constant something else? Convenience of substitution?).
Maybe I'm just filling the problem with hot air and fussing about it.
My implementation of the NTP with the level shifters on either side DOES have higher H2 and H3, and worse THD (perhaps you were talking about this, Hugh?). It has profoundly dominant 2nd harmonics. However using this to "sweeten" the sound comes at the expense of THD (however, when OLG is increased with this configuration, odd order harmonics plummet in relation to the 2nd and 3rd since the level shifters create the distortion, but leave the NTP to work on everything else).
In any case, it is VERY hard to compare the LTP and NTP without getting into apples and oranges. If you aim for benign profile AND low THD, which apparently was my angle, then weighing the two will be a long, hard struggle. Having low THD without offensive profile and without complex circuitry, is also a long, hard struggle of weighing options. Thankfully, it doesn't seem low THD is important for audio enjoyment.
- keantoken
Ah, an intellectual tipping point exacerbated by feverish use of the simulator.....
You really need to build these circuits, have you gotten around to building that board I sent you? I don't want to heckle you, but this is the next logical step, you must have more empirical evidence!
Trust you and your family are well and options are appearing on the horizon....
Hugh
You really need to build these circuits, have you gotten around to building that board I sent you? I don't want to heckle you, but this is the next logical step, you must have more empirical evidence!
Trust you and your family are well and options are appearing on the horizon....
Hugh
That must be one of the most misinterpreted and misleading whitepapers in the field of audio. Just for kicks, see if you can find the D. O. Monteith & R. R. Flowers "rebuttal" article "Transistors Can Sound Better Than Tubes", in which the gentlemen present a preamp design that pretty much beats the tube in generating the harmonics of ideal order - at least according to principles defined by Hamm's paper.
Oh... Basically, any asymmetrically clipping / distorting circuit will do that. Too bad that in reality such circuits tend to sound like a broken radio.
Member
Joined 2009
Paid Member
that's why I said 'infamous' 😉
The proof is in the listening.
I'm not looking to reproduce a tube and especially not the clipping behaviour. I've only attempted to shift the balance of harmonics from odd to even and to emphasize the low order harmonics. It sounds good.
I have some money to burn, I think I could get those boards built within one mouser shipment, and I think I have a suitable heatsink, chassis and transformer, though it will not necessarily be pretty. I've got those 2SC3423's mounted on the minisinks already, waiting in the bin for the proper time, with the caps you sent me... I've started wiring up my projects, I've got an Allison class A, 1.3A bias running on my testbed, but my test speakers can't do it any justice. I was planning on getting a mouser order in to refresh my stock, but I can't remember what kind of SMD resistors I need (I don't feel like jerry rigging normal resistors in, and some 3W SMD ones might come in handy).
I think mine is a case of projectitis-induced asphyxiation of the motive cortex, coupled with the fact that the Texas sun makes you want to sit around all day and surrender to the heat. If I had it my way with electronics I'd build circuits and test out every topology I found, and not be sparing about it. Unfortunately, soldering takes more time than I'd like (though I'm getting better), and the heat builds up...
I'm thinking it would be a good idea to cover all the traces with solder, so they don't oxidize before I'm able to build them. Holler if this is a mistake!
- keantoken
I think mine is a case of projectitis-induced asphyxiation of the motive cortex, coupled with the fact that the Texas sun makes you want to sit around all day and surrender to the heat. If I had it my way with electronics I'd build circuits and test out every topology I found, and not be sparing about it. Unfortunately, soldering takes more time than I'd like (though I'm getting better), and the heat builds up...
I'm thinking it would be a good idea to cover all the traces with solder, so they don't oxidize before I'm able to build them. Holler if this is a mistake!
- keantoken
Member
Joined 2009
Paid Member
It's good to build, even better if you can finish what you start. I find the biggest problem is finishing the project but that maybe because I treat each project as something that should generate a box my wife can plug into the wall outlet. If you can get that Allison up and going as a full power amp - that would be great !
I use a chemical that you dip the pcb into and which puts a tin coating on the pcb traces. It's a nasty chemical. Covering the traces with solder works, but them you break in more fumes.
I use a chemical that you dip the pcb into and which puts a tin coating on the pcb traces. It's a nasty chemical. Covering the traces with solder works, but them you break in more fumes.
These traces don't have that pinkish tinge that copper does, I'm wondering if they're not gold-plated or something.
- keantoken
- keantoken
Hi, Gareth! Your posts are very instructive! Thank you!
By the way, would this thread be related by any means with this:http://www.diyaudio.com/forums/everything-else/167105-there-some-mystery-air.html#post2191795??
I would like to hear this misterious sound, for never had any contact with tube amps.
Thanks for any advice.
Cheers,
Max.
By the way, would this thread be related by any means with this:http://www.diyaudio.com/forums/everything-else/167105-there-some-mystery-air.html#post2191795??
I would like to hear this misterious sound, for never had any contact with tube amps.
Thanks for any advice.
Cheers,
Max.
hey Bigun,
I was just having a closer look at your schematic in post #2 and I reckon when you include the 120k tubemagic resistor you need to increase the value of R9 or decrease the value of R1 and R8 because 120k // 39k = 29k.
regards
I was just having a closer look at your schematic in post #2 and I reckon when you include the 120k tubemagic resistor you need to increase the value of R9 or decrease the value of R1 and R8 because 120k // 39k = 29k.
regards
Member
Joined 2009
Paid Member
Yes, Hugh's Aspen amplifiers were my initial inspiration for building my amplifiers and then tweaking the distortion profiles. But I'm afraid my brute force approach to meddling with the harmonics doesn't match the art that Hugh practices in his excellent designs.
Greg you are absolutely correct, if you want to add the magic resistor without affecting the gain/sensitivity you would have to adjust the feedback network resistors. I didn't do this because I only needed to ensure all the channels were matched rather than achieving a particular gain. Of course if you change the impedance of the feedback network you can change dc offset at the output so this has to be checked and adjusted for if necessary.
There is another way to do all of this of course and that is to string a potentiometer across the emitters of the output devices and then take the feedback from the wiper - continuous adjustment of the 'magic' resistor. I didn't try this though.
- Status
- Not open for further replies.