Giaime said:
There are two fixes for that problem: strong final grid drivers and gNFB. Some of the complaints regarding the 300B seem to be related to inadequate grid drive. Yeah, yeah, yeah: feedback is bad and "heresey" in a SET design. Well, if more designers used it, there'd be far fewer complaints about SETs and poor performance.
An externally hosted image should be here but it was not working when we last tested it.
Be careful there. A lot of those so-called "great amps" were designed to minimize the cost of production, and sonic quality was the first thing to suffer for it. Also, there was the "numbers game" that the manufacturers played: trying to get the lowest possible THD measurements. That led to excessive use of high gain stages (high gain small signal triodes and small signal pentodes tend to produce a lot more of the objectionable high order harmonics than do lower gain triodes) to support excessive amounts of gNFB. As with everything else, if a little gNFB is a good thing, then lots of gNFB must be even better. Ain't so! One of those "great amps", the Williamson, really sucked. Williamson, himself, agreed.
Really ain't no "tricks" to it. Select good, linear operating points, lots of headroom in the front end, adequate drive to get the slew rate up, judicious use of NFB, and, most important: good OPTs. Especially really good OPTs, that's not the place to be cheap. A $300 OPT and a $50 final works a helluvalot better than a $50 OPT and a $300 final.
Triodes are good if you can tolerate the rather low output power. Pentode mode with decent screen voltage regulation, preferrably from an active regulator, will give good distortion performance, and local + global feedback will give you decent woofer damping for bass with "authority", clarity and detail. I don't completely trust the claims being made for Ultralinear. It seems very much to be an "all or nothing" proposition. Get the feedback ratio exactly right, and it seems to work quite well. Get it wrong, and it just makes things worse. Problem is: once the OPT is wound, the die is cast, and you can't tweak it for optimum performance.
Pentodes: local, parallel off the plate(s) to get the r(p) equivalent down so that the inductive reactance of the OPT primary becomes a larger percentage of the effective r(p). Improves linearity, and damping. Goes a long way to cleaning up that pentode nastiness. Add gNFB off the OPT secondary to compensate for xfmr nonlinearity, and get the Zo of the amp down. Used right, it does wonders to improve the sound. Too much will give a dull, solid statey sound.
The metal in xfmrs is subject to work hardening, being kicked around is bad for its magnetic properties. Magnetostriction may contribute, and would hit power xfmrs and chokes more than OPTs. The laminations (and the copper for that matter) are annealed during production for optimal magnetic or electrical properties. Some exotic materials such as mu-metal are particularly susceptible to work hardening related damage.
I run the risk of ridicule but if you want a very good power amp at more than thirty watts, why not try a ST-70? The tubes are relatively cheap, the iron is great, and EVERYONE can give you tips on upgrading/altering them. Just my $.02
So, if you are going for big buck, an extra 50 euros for, well, you know the book I am talking about, isn't much and well, really worth it.
Erik
Re: Re: A little research: old tube amps from the past
Hello Miles,
sorry, probably it's a language problem, but I can't understand what kind of circuit you are proposing. Are you thinking about feedback from the output tube's plates to the drivers cathodes?
And for screen regulation, what do you suggest? shunt or pass regulator?
Miles Prower said:
Hello Miles,
sorry, probably it's a language problem, but I can't understand what kind of circuit you are proposing. Are you thinking about feedback from the output tube's plates to the drivers cathodes?
And for screen regulation, what do you suggest? shunt or pass regulator?
To Giaime:
I've included a couple of schemos. For the basic amp topology, you can see how the inner feedback loop is connected: from the 807 plates back to the driver grids. This helps avoid the 807's reverse transfer capacitance which would otherwise appear in parallel with the feedback resistor. I used 10% (6.95db(v) ) od feedback as recommended in a "white paper" that was ostensibly a discussion of these "new fangled" beam tetrodes, but read more like an advert. Since this appeared about five years after the introduction of this type, I figured they must have known what they were talking about, so far as recommended local feedback. THEY WUZ RIGHT! That worked out quite well in the finished product. 🙂
The screen regulator is a basic series pass type largely adapted to glass from a very common SS design. That, too, performs very well: the output voltage decreasing by 0.8Vdc when the positive rail is taken down by some 45Vdc.
807 Amp Basic Topology
[img=http://img107.imageshack.us/img107/4249/807ampivsx0.th.gif]
Active Screen Regulator
[img=http://img59.imageshack.us/img59/2525/screenreglx9.th.gif]
I've included a couple of schemos. For the basic amp topology, you can see how the inner feedback loop is connected: from the 807 plates back to the driver grids. This helps avoid the 807's reverse transfer capacitance which would otherwise appear in parallel with the feedback resistor. I used 10% (6.95db(v) ) od feedback as recommended in a "white paper" that was ostensibly a discussion of these "new fangled" beam tetrodes, but read more like an advert. Since this appeared about five years after the introduction of this type, I figured they must have known what they were talking about, so far as recommended local feedback. THEY WUZ RIGHT! That worked out quite well in the finished product. 🙂
The screen regulator is a basic series pass type largely adapted to glass from a very common SS design. That, too, performs very well: the output voltage decreasing by 0.8Vdc when the positive rail is taken down by some 45Vdc.
807 Amp Basic Topology
[img=http://img107.imageshack.us/img107/4249/807ampivsx0.th.gif]
Active Screen Regulator
[img=http://img59.imageshack.us/img59/2525/screenreglx9.th.gif]
Hello all 🙂
after much scribbling and looking at various designs, I finished up a preliminary schematic. What do you think?
Spice works real bad with partial feedback, from the output tube plates to 6CG7 grids, it appears that distortion INCREASES instead decreasing, and I can't understand why 😕
So I made the amp with the usual kind of negative feedback, but I'm willing to implement partial feedback since now nfb is too high, 16dB, and may cause trouble.
after much scribbling and looking at various designs, I finished up a preliminary schematic. What do you think?
Spice works real bad with partial feedback, from the output tube plates to 6CG7 grids, it appears that distortion INCREASES instead decreasing, and I can't understand why 😕
So I made the amp with the usual kind of negative feedback, but I'm willing to implement partial feedback since now nfb is too high, 16dB, and may cause trouble.
Attachments
Giaime said:
If you mean this literally, then it's no wonder the distortion increases. Going from the finals to the driver grids gives positive feedback. That does the exact opposite of what you want to do here: narrows bandwidth, increases output impedance. Using pentode finals, the output impedance is already too high, and you certainly don't want to make that worse. Positive feedback there simply undoes what the Ultralinear connection is trying to do. Since you are going with an Ultralinear topology, you don't need any parallel feedback off the finals' plates anyway.
You don't need another inner feedback loop for the finals. That's what Ultralinear accomplishes. It's not my choice in my designs because it's not tweekable, so I opt for parallel feedback off the plates which is quite tweekable. However, both topologies work the same way: local NFB, and accomplish the same things: lower the effective r(p) of the pentode finals, and convert their dynamic characteristic to be less pentode-like and more triode-like.
I suspect that 16db(v) of gNFB may be too much. The one and only way to know for sure is to actually build the amp, and listen. When I did this, I quickly discovered that the 12db(v) of gNFB I had originally selected was way too much. If you get the "solid statey" sound: bass over emphasis, loss of clarity to the mids, vocals that seem to receed into the background, almost nonexistant highs, then you'll know it's too much.
Sorry Miles, I got it wrong: I wanted to say feedback from teh 6550 anode to the 6CG7 anode.
I know the other way is positive feedback, in fact SPICE give no output signal that way.
I know the other way is positive feedback, in fact SPICE give no output signal that way.
Giaime said:
If you're going with an Ultralinear connection, which the schemo seems to indicate, then you won't be needing any extra feedback from the 6550s to the 6CG7s. You already have the local feedback.
Miles Prower said:
Not necessarily. A couple of years ago I made an amplifier for which I tested about twenty different output stages using different pentodes ranging from triode-strapped with various forms of feedback to pure pentode with various forms of feedback. Finally, I found a configuration that worked for the valves I'd chosen. UL N78 plus cathode feedback. It was far better than any of the other options I'd tested. But only for N78 - didn't work for EL84 or any others.
One of the design goals is to have a DF = 10.
I don't know if I have made all calculations right, but looking at the ultralinear plate characteristics I measured a rp (dVa/dVg) about 1.6k for each tube. Thus the reflected output impedance, plus a hundred ohms primary DC resistance, and 0.2ohm secondary, gave me about 5.5ohm, so to reach 0.8ohm, 16dB are necessary.
Is there any other way? I'll investigate on cathode feedback, since I will have the transformers custom made.
I don't know if I have made all calculations right, but looking at the ultralinear plate characteristics I measured a rp (dVa/dVg) about 1.6k for each tube. Thus the reflected output impedance, plus a hundred ohms primary DC resistance, and 0.2ohm secondary, gave me about 5.5ohm, so to reach 0.8ohm, 16dB are necessary.
Is there any other way? I'll investigate on cathode feedback, since I will have the transformers custom made.
Giaime said:
I wouldn't worry so much about that. You aren't going to get the same levels of DF that you can get from SS, and that design goal may very well conflict with others. There is a good chance that 16db(v) of gNFB will sound really bad. I found that 12db(v) in one of my designs was very bad.
An externally hosted image should be here but it was not working when we last tested it.
4.0db(v) sounded much better, even though the DF was obviously better with the higher feedback. Having tried both Technics SB-K25 (OK) and Bose 201 (Vast improvement) speakers with the lower level feedback with the "worse" DF, I don't have any problems related to this. Bass has plenty of "authority" and clarity, with no trace of any of the underdamped "sloppiness" that was all too obvious when running with no feedback at all.
DF isn't the end-all and be-all of how that's going to sound. I wouldn't sacrifice other sonic qualities for that.
You know Miles, I would agree with you, but that amp has to drive 38cm Tannoy Dual Concentric, and they need hefty DF.
But he's now using a trioded EL34 PSE, and he likes that (aside from power output), and in that case DF should be in the order of 4.
I'd like however to keep 16dB of *total*feedback, but dividing between 6dB of global and 10dB of local, I have to find out where and how.
But he's now using a trioded EL34 PSE, and he likes that (aside from power output), and in that case DF should be in the order of 4.
I'd like however to keep 16dB of *total*feedback, but dividing between 6dB of global and 10dB of local, I have to find out where and how.
You didn't say whether the Tannoy 15s were reds or golds. The reds really didn't like sealed enclosures (though quite a few went into them) Since I don't recognise the cabinet name, I'll assume they're after my time, ie golds.
Still, their efficiency was quite decent, considerably higher than most modern designs (that's not a criticism of modern speaker designs; indeed, it's most of the reason why that extra damping factor is nescessary)
And here we have an example; when I was working for Tannoy, a lot of studios used the Radfords, and they sounded great. (and very rarely blew out the speakers, which I appreciated) Yet the amplifier was basically an optimised Mullard 5-20 circuit, well done and with an excellent output transformer, but, unlike the Quad (for example) no esoteric circuitry, just conventional design done well, with good components. Modern technology could have improved them considerably (indeed, I picked up a batch of Radfords dirt cheap, because they were rack mounting and the output transformer had a 100volt tap, so no-one wanted them, and got considerably better impulse response out of them: but I was using them as high frequency drivers in a biamped system, so was uninterested in their bass response)
Still, their efficiency was quite decent, considerably higher than most modern designs (that's not a criticism of modern speaker designs; indeed, it's most of the reason why that extra damping factor is nescessary)
phn said:
And here we have an example; when I was working for Tannoy, a lot of studios used the Radfords, and they sounded great. (and very rarely blew out the speakers, which I appreciated) Yet the amplifier was basically an optimised Mullard 5-20 circuit, well done and with an excellent output transformer, but, unlike the Quad (for example) no esoteric circuitry, just conventional design done well, with good components. Modern technology could have improved them considerably (indeed, I picked up a batch of Radfords dirt cheap, because they were rack mounting and the output transformer had a 100volt tap, so no-one wanted them, and got considerably better impulse response out of them: but I was using them as high frequency drivers in a biamped system, so was uninterested in their bass response)
Hello Chris,
I'll ask him precisely what model of dual concentric it is.
As far as I remember, they are Tannoy Mainsfeld, about 1974.
I'll ask him precisely what model of dual concentric it is.
As far as I remember, they are Tannoy Mainsfeld, about 1974.
Vintage amps
Per original post, question was directed at vintage amps.
I have owned a few vintage amps & for some reason the Dynaco SCA-35 is a very good sounding amp with a few simple mods. Good bass response, warm yet detailed midrange & accurate hiigh frequencies. No bite or edge, just a sweet amp.
A few mods are necessary to achieve this good sound from the SCA-35. Replace the 12AX7 preamp output coupling caps & the four EL84 tube coupling caps on the grids with AuriCap. The original caps were green in color on the circuit boards.
Next, install a Solen 82uF capacitor on the EL84 cathode resistor & remove that nasty 100uF section from the circuit at the large electrolytic can. The Solen is large, this locate on top of chassis next to the electrolytic cans with two simple standoff terminals plus relocate that 95 ohm cathode resistor to the same location. Bring back the ground wire to the original location on the electrolytic can to prevent ground loops.
If you can, replace those old can type electrolytics with JJ 40-20-20-20uF @ 525 volt electrolytics. I super glue the caps in place for a super strong mount, or just use metal clamps. By connecting various caps in the JJ, one will end up with 60-40-20uF just like original & 40-40uF for the other can that had 50-50uF. Seeing the JJ measures a little over advertised microfarid specs, the little difference here makes no difference.
Finally, must use GE 6BQ5 grey oval plate, Sylvania 7199 grey plate & believe it or not Sovtek 12AX7LPS tubes.
I been thru about every EL84 ever made & even Mullard 1950s square getter fall far short of the GE. RCA 7199 black plate sound lousy compared to the Sylvania 7199. The Sovtek 12AX7LPS outperformed Telefunken 12AX7 smooth plate & Mullard 12AX7 short plate among other tubes- go figure.
I also owned a Dynaco Stereo 35 with same EL84s, power & audio transformers, but different input design, After mods & tube rolling, that Stereo 35 could not compete against the Dynaco SCA-35. The Dynaco SCA-35 should not perform as well as it does breaking rules such as tone contols, a pentode tube section in the audio chain, etc.
Per original post, question was directed at vintage amps.
I have owned a few vintage amps & for some reason the Dynaco SCA-35 is a very good sounding amp with a few simple mods. Good bass response, warm yet detailed midrange & accurate hiigh frequencies. No bite or edge, just a sweet amp.
A few mods are necessary to achieve this good sound from the SCA-35. Replace the 12AX7 preamp output coupling caps & the four EL84 tube coupling caps on the grids with AuriCap. The original caps were green in color on the circuit boards.
Next, install a Solen 82uF capacitor on the EL84 cathode resistor & remove that nasty 100uF section from the circuit at the large electrolytic can. The Solen is large, this locate on top of chassis next to the electrolytic cans with two simple standoff terminals plus relocate that 95 ohm cathode resistor to the same location. Bring back the ground wire to the original location on the electrolytic can to prevent ground loops.
If you can, replace those old can type electrolytics with JJ 40-20-20-20uF @ 525 volt electrolytics. I super glue the caps in place for a super strong mount, or just use metal clamps. By connecting various caps in the JJ, one will end up with 60-40-20uF just like original & 40-40uF for the other can that had 50-50uF. Seeing the JJ measures a little over advertised microfarid specs, the little difference here makes no difference.
Finally, must use GE 6BQ5 grey oval plate, Sylvania 7199 grey plate & believe it or not Sovtek 12AX7LPS tubes.
I been thru about every EL84 ever made & even Mullard 1950s square getter fall far short of the GE. RCA 7199 black plate sound lousy compared to the Sylvania 7199. The Sovtek 12AX7LPS outperformed Telefunken 12AX7 smooth plate & Mullard 12AX7 short plate among other tubes- go figure.
I also owned a Dynaco Stereo 35 with same EL84s, power & audio transformers, but different input design, After mods & tube rolling, that Stereo 35 could not compete against the Dynaco SCA-35. The Dynaco SCA-35 should not perform as well as it does breaking rules such as tone contols, a pentode tube section in the audio chain, etc.
- Status
- Not open for further replies.