4S Universal Preamplifier for 12A*7 Tubes
I am told that;
"This circuit is more akin to a properly designed set of tires that work well with a set of cars exhibiting similar characteristics"
I am I missing something here? To me these tubes aren't similar at all and should have their own circuit and not a "universal" one.
Thoughts?
I am told that;
"This circuit is more akin to a properly designed set of tires that work well with a set of cars exhibiting similar characteristics"
I am I missing something here? To me these tubes aren't similar at all and should have their own circuit and not a "universal" one.
Thoughts?
> these tubes aren't similar at all
Over the entire range of tubes ever made, these tubes fall together fairly "similar". Consider an '01 or a 6080.
They sure should all "work" with the same bias resistors. Tube guitar amp techs often swap AT7 and AU7 into AX7 sockets in the rare case they want less gain.
"tires that work well with a set of cars" Well, if the small cars have skinny wheels and the large cars have wide wheels, the tire fit and bulge will tend to similar (or not so different) performance even if you use the same size tire.
They do work different. Note there is 9dB gain difference across the set. Also a 2:1 range of plate currents.
This group should show significant difference in output impedance. Well, not very different because he has not allowed the high-current tubes to run at high current. Anyway whatever happens in the tube is masked (except at full gain) by putting a 250K gain pot on the output. (The resulting >62K Zout does not encourage output leads even 3 feet long: 65K against 100pFd is -3dB at 24.5KHz, -1dB near 12KHz.)
If the 12AU7 were opened-up to a fat 8mA and output taken direct, Zout would be near 5K, "12X better". 12AX7 will never get there (without more tricks). Is that really better? Well if you want to drive 30 foot cables it sure is.
It hardly "proves" anything, because all might work better with an optimized set of values.
It's good clean fun for people with lots of tubes to try.
It sure would be a handy "tester" for odd-lot tubes. Do they work at all? Is their gain similar to others of the same type? Does this tube hiss, buzz, or rattle? You get a rough-check without futzing with bias.
Over the entire range of tubes ever made, these tubes fall together fairly "similar". Consider an '01 or a 6080.
They sure should all "work" with the same bias resistors. Tube guitar amp techs often swap AT7 and AU7 into AX7 sockets in the rare case they want less gain.
"tires that work well with a set of cars" Well, if the small cars have skinny wheels and the large cars have wide wheels, the tire fit and bulge will tend to similar (or not so different) performance even if you use the same size tire.
They do work different. Note there is 9dB gain difference across the set. Also a 2:1 range of plate currents.
This group should show significant difference in output impedance. Well, not very different because he has not allowed the high-current tubes to run at high current. Anyway whatever happens in the tube is masked (except at full gain) by putting a 250K gain pot on the output. (The resulting >62K Zout does not encourage output leads even 3 feet long: 65K against 100pFd is -3dB at 24.5KHz, -1dB near 12KHz.)
If the 12AU7 were opened-up to a fat 8mA and output taken direct, Zout would be near 5K, "12X better". 12AX7 will never get there (without more tricks). Is that really better? Well if you want to drive 30 foot cables it sure is.
It hardly "proves" anything, because all might work better with an optimized set of values.
It's good clean fun for people with lots of tubes to try.
It sure would be a handy "tester" for odd-lot tubes. Do they work at all? Is their gain similar to others of the same type? Does this tube hiss, buzz, or rattle? You get a rough-check without futzing with bias.
"As can be seen from the calculated performance, the peak gain can be varied considerably by choice of tube and the distortion (virtually all second harmonic) is very well controlled."
"The volume control (P1 on the schematic) is on the tube output to help ensure that the noise figure for this Hi-Fi preamp is as low as possible."
"This is very good performance."
The 12AT7 has 5% distortion, that's as much or more than a single ended power section driving a speaker!
Novelty item? Yes. Hi-Fi? Nope.
"The volume control (P1 on the schematic) is on the tube output to help ensure that the noise figure for this Hi-Fi preamp is as low as possible."
"This is very good performance."
The 12AT7 has 5% distortion, that's as much or more than a single ended power section driving a speaker!
Novelty item? Yes. Hi-Fi? Nope.
> The 12AT7 has 5% distortion
That stands out. I would not expect it to be so different from AX and AU. Aren't these SPICE numbers? SPICE model accuracy varies, and some may be very bad. Some may be extrapolated from a few measurements (often far from where we want to work). There are tools to extract numbers from TUBE-BOOK GRAPHS, which IMHO are only "close" up-right and very poor down-left.
It won't stink like an old transistor radio. Good ears will note more "color" than a real low THD stage (even my tin ears hear 5% in a guitar preamp vs a high-NFB lo-THD preamp). The high output impedance seems awkward to me.
That stands out. I would not expect it to be so different from AX and AU. Aren't these SPICE numbers? SPICE model accuracy varies, and some may be very bad. Some may be extrapolated from a few measurements (often far from where we want to work). There are tools to extract numbers from TUBE-BOOK GRAPHS, which IMHO are only "close" up-right and very poor down-left.
It won't stink like an old transistor radio. Good ears will note more "color" than a real low THD stage (even my tin ears hear 5% in a guitar preamp vs a high-NFB lo-THD preamp). The high output impedance seems awkward to me.
The load and bias point are in a very nonlinear section of the curves for the 12AT7.
I want to state that I am not opposed to experimenting and or if someone prefers high distortion etc.... It's just this design has the "hi-fi" label that doesn't sit well with me.
In reality this design is no more "universal" than any other design using say a 12AX7, what I mean is you can put a 12A*7 in any circuit designed for a 12AX7 and get the same poor result
The 12AT7 is indeed pretty dire. Unless you use JJ "ECC81"s which I gather are actually ECC88s inside!
Is this a rumor going around?
http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/ECC88.pdf
http://www.drtube.com/datasheets/ecc81-jj2003.pdf
The emoji tells me you are joking and alluding possibly to an internet rumor? The JJ E88CC is like twice the price of the ECC81.
I was sent an email telling me I wasn't being fair to the design and that I needed to read the whole article. I went to the authors site and read through the whole thing, I found this tidbit down at the bottom in the comments.
"As to the published specs of some of today’s amplifiers, it is important to realize how this is commonly done. What the manufacturers do is the publish a “referenced” distortion number. Let me explain. The normal modern audio design approach is to produce an amplifier with very high open loop gain and then feedback a significant portion of that signal gain to cancel noise and distortion. So lets say the designers need 50W from a line level signal (i.e a standard power amp design). At 8Ω, 50 watts is 20v rms. A line level signal is +4dBu which is 1.228v rms. So the actual end to end gain they need is 20*log(20/1.228) or ≈24dBv. So they design an amplifier with an open loop gain of something large; like 80dB. Then the approach is to use 56dB of negative feedback to get back to 24dBv end-to-end gain. Now in addition they want really low low distortion, like the 0.001% figure you mentioned above. Thats -100dBv in all one harmonic (i.e. the dominant harmonic). But achieving and measuring that 0.001% distortion is very difficult. So how do they accomplish this?
Well, the theory of negative feedback says that the reduction in distortion is the same as the feedback factor. And the feedback factor (in our example) is 56dBv. So what they do is target the open loop distortion to a number they can achieve. In this case -100dBv + 56dBv or -44dBv. This is only 0.63%. Now the other thing that negative feedback does is significantly increase the bandwidth. So what they do, is design a high gain ≈80dBv but narrow band amplifier with a distortion factor of <0.6%. This is actually an achievable goal and a rather straight forward process. Then they measure the "open loop" distortion, convert to dBv, then subtract the feedback factor (in this case 56dBv) and convert back to %.
So lets say that the amplifier as designed has an open loop distortion of 0.57% (i.e. below the target of 0.6%). This measured distortion equates to -44.9dBv. Then they subtract the feedback factor of 56dBv to get a "theoretical" distortion of -100.9dBv. This corresponds to 0.0009% THD. And THAT is the number they publish even though the real measured distortion was about 1/2%.
The problem with this, of course, is you NEVER even come close to this theoretical distortion number in actual usage. But that doesn't keep the manufacturers from publishing it. This type of mathematical masturbation has been accepted practice in the audio equipment industry for decades. But once you understand what's actually going on, all those super small distortion numbers become just so much advertising foolishness."
"As to the published specs of some of today’s amplifiers, it is important to realize how this is commonly done. What the manufacturers do is the publish a “referenced” distortion number. Let me explain. The normal modern audio design approach is to produce an amplifier with very high open loop gain and then feedback a significant portion of that signal gain to cancel noise and distortion. So lets say the designers need 50W from a line level signal (i.e a standard power amp design). At 8Ω, 50 watts is 20v rms. A line level signal is +4dBu which is 1.228v rms. So the actual end to end gain they need is 20*log(20/1.228) or ≈24dBv. So they design an amplifier with an open loop gain of something large; like 80dB. Then the approach is to use 56dB of negative feedback to get back to 24dBv end-to-end gain. Now in addition they want really low low distortion, like the 0.001% figure you mentioned above. Thats -100dBv in all one harmonic (i.e. the dominant harmonic). But achieving and measuring that 0.001% distortion is very difficult. So how do they accomplish this?
Well, the theory of negative feedback says that the reduction in distortion is the same as the feedback factor. And the feedback factor (in our example) is 56dBv. So what they do is target the open loop distortion to a number they can achieve. In this case -100dBv + 56dBv or -44dBv. This is only 0.63%. Now the other thing that negative feedback does is significantly increase the bandwidth. So what they do, is design a high gain ≈80dBv but narrow band amplifier with a distortion factor of <0.6%. This is actually an achievable goal and a rather straight forward process. Then they measure the "open loop" distortion, convert to dBv, then subtract the feedback factor (in this case 56dBv) and convert back to %.
So lets say that the amplifier as designed has an open loop distortion of 0.57% (i.e. below the target of 0.6%). This measured distortion equates to -44.9dBv. Then they subtract the feedback factor of 56dBv to get a "theoretical" distortion of -100.9dBv. This corresponds to 0.0009% THD. And THAT is the number they publish even though the real measured distortion was about 1/2%.
The problem with this, of course, is you NEVER even come close to this theoretical distortion number in actual usage. But that doesn't keep the manufacturers from publishing it. This type of mathematical masturbation has been accepted practice in the audio equipment industry for decades. But once you understand what's actually going on, all those super small distortion numbers become just so much advertising foolishness."
Thanks for sharing.
I wasn't aware of any well regarded amplifier companies using this sort of tactic either. All of the gear I have had my hands on measure close to published specs so I don't see any numbers fooling.
I would like to hear others experiences on this, especially if you are in the field.
I deduced it from SY's suspiciously low distortion measurements:
The Red Light District, p3
And from looking inside the bottle where you'll see what look rather like big ol' frame grid triodes! Whatever is inside there, it ain't a 12AT7...
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
But the preamp original page does EXACTLY that, they publish "calculated/simulated/emulated" values and nothing else, not a single real tube was plugged there, measured and results posted to, at least, allow "predicted vs. real" comparison.
Those tables are as much "mathematical masturbation" (sorry, not my words) than anything I have ever seen.
Which of course leads straight to the "advertising foolishness" issue
.IF at least they had provided for 3 switchable pairs of proper cathode and plate resistors, this box would have been more than a toy.
Notice one important parameter they chose not to show is where will plate voltage rest with such different tubes plugged in.
I don't often visit Diyaudioprojects.com but the few times I've been there , I've left wondering what on earth was going on . I know someone has obviously put some time and effort into this as well as documenting , but a few things I find very odd about this 4S preamp are the massive 10k grid stopper , the pot on the output (!!!) and excessive gain even with a 12AU7 fitted . The designer also uses this pre with a power amp that also uses a pot between driver and output stage . A very strange implementation IMHO
316a
316a
People who talk nonsense often object to having that fact being pointed out by others. They think you have misunderstood them, when in fact you have understood them only too well. The things you quote are pure nonsense, of course, but I guess it helps the circuit designer feel he has contributed to the sum of human knowledge on audio.famousmockingbird said:
Once I've built a guitar effect pedal for a cousin. It involved a ECC81 in starved conditions - 9V on the anodes. A JJ valve was used but it never worked. No sound at all! The circuit was simplisity itself, double and triple checked and the valve was "OK" on high voltage circuits. The project was abandoned with enough disappointment... Now I see it would be interesting to substitute for another brand!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.