High-Gain Amplifier Circuit
- Thread starter NewYorkDave
- Start date
Help Support GroupDIY Audio Forum:
Just like you said, pins 4, 5 and 9 are for "power". Pin 9 goes to ground and 4 and 5 go to a 6V supply. Or float pin 9 and put 12V across 4 and 5.
There's some good articles online on how tubes work. Do some googling and you'll find loads of stuff. There's also some good books on Amazon and such.
Peace,
Al.
There's some good articles online on how tubes work. Do some googling and you'll find loads of stuff. There's also some good books on Amazon and such.
Peace,
Al.
NewYorkDave
Well-known member
Two good beginner's books--at least two good ones that are in print right now--are Beginner's Guide to Tube Audio by Bruce Rozenblitt, and Valve Amplifiers by Morgan Jones. Be warned that the Jones book gets into some really tweaky hi-fi stuff that will have you scratching your head, but it's a good read overall.
RCA receiving tube manuals always had good basic theory and operation info in their introductory chapters.
Of course, there's also the grand-daddy of tube books, the Radiotron 4, which is now available as a free download if you look around carefully...
RCA receiving tube manuals always had good basic theory and operation info in their introductory chapters.
Of course, there's also the grand-daddy of tube books, the Radiotron 4, which is now available as a free download if you look around carefully...
ye cheers dave - i need to have sort out n sit down of the mass of info I've been collecting n archiving - everytime I do tho, there comes along another summin else to make me want to get tools out and start building! I really need to print out the shit for it to sink in too - i just zone out after a while readin txt on screen otherwise...
ye i was on that mofo while the middle was still warm :green:
peesh!
ye i was on that mofo while the middle was still warm :green:
peesh!
I have both of Rozenblitz's books: Beginner's guide and Audio Reality. They're both now out of print, but Antique might still have a few copies. I kinda sorta read the Beginner's guide, but I need to re-read it with much more attention. Unfortunately I've been too busy as of late. But I do recall it was very informative and easy to follow for a beginner. The most difficult thing for me is knowing how to apply what I've learned (and of course knowing when they apply), and trying to find the answers to the gaps of why's and what if's... :? Just takes some time I guess...
PS: Dave, I always love your hand drawn schematics!!! Very clear and neat. Very AOE-like.
PS: Dave, I always love your hand drawn schematics!!! Very clear and neat. Very AOE-like.
dramadisease
Well-known member
dumb question:
seems like theres alot of feedback - theres 2 main gain stages each with its own feedback line.
is this normal?
im thinking i might throw a channel of this together maybe early in the year. got this awesome box i wanna do a single channel tube pre in.
where would a good DI input be on this thing? could you just go into the first grid and make that first grid resistor something between 2 and 10 meg? something to help set the input inpedance higher i guess, i dunno what the input impedance is to a 12av7.
really nice to see designs that get away from the 12ax7 - im getting a bit sick of em.
thanks dave!
-bryan sours
seems like theres alot of feedback - theres 2 main gain stages each with its own feedback line.
is this normal?
im thinking i might throw a channel of this together maybe early in the year. got this awesome box i wanna do a single channel tube pre in.
where would a good DI input be on this thing? could you just go into the first grid and make that first grid resistor something between 2 and 10 meg? something to help set the input inpedance higher i guess, i dunno what the input impedance is to a 12av7.
really nice to see designs that get away from the 12ax7 - im getting a bit sick of em.
thanks dave!
-bryan sours
NewYorkDave
Well-known member
NewYorkDave
Well-known member
[quote author="dramadisease"]is this normal?[/quote]
What is "normal?" Do we even care? :wink:
It's not that odd, no. It will seem unusual if you're used to guitar amps or other "minimum parts" designs, but this kind of topology is not unheard-of in recording equipment. In recording and broadcasting gear, the overarching concerns were minimum distortion and flattest frequency response, so it was common to use more tubes and more feedback as opposed to fewer tubes and little or no feedback.
As a first approximation at audio frequencies, a tube grid is assumed to have infinite input impedance. With high-mu triodes, you do have to account for Miller capacitance at the higher frequencies, but it's not much of a worry with medium- and lower-gain triodes like the 12AV7.
If you're using it as a DI for electric guitars or basses, you don't need to make the input grid resistor any higher than 1M. If you wanna use it with piezos, you can make it higher, but there's a limit imposed by grid current.
The 12AX7 is an excellent and versatile tube, which is why you see it in more circuits than probably any other small-signal tube. But I've been using the 12AV7 in most of my recent experiments because even though it's out of production, it's still cheap and plentiful, and it has some very useful characteristics. It's kind of like a higher-gain 12AU7.
What is "normal?" Do we even care? :wink:
It's not that odd, no. It will seem unusual if you're used to guitar amps or other "minimum parts" designs, but this kind of topology is not unheard-of in recording equipment. In recording and broadcasting gear, the overarching concerns were minimum distortion and flattest frequency response, so it was common to use more tubes and more feedback as opposed to fewer tubes and little or no feedback.
As a first approximation at audio frequencies, a tube grid is assumed to have infinite input impedance. With high-mu triodes, you do have to account for Miller capacitance at the higher frequencies, but it's not much of a worry with medium- and lower-gain triodes like the 12AV7.
If you're using it as a DI for electric guitars or basses, you don't need to make the input grid resistor any higher than 1M. If you wanna use it with piezos, you can make it higher, but there's a limit imposed by grid current.
The 12AX7 is an excellent and versatile tube, which is why you see it in more circuits than probably any other small-signal tube. But I've been using the 12AV7 in most of my recent experiments because even though it's out of production, it's still cheap and plentiful, and it has some very useful characteristics. It's kind of like a higher-gain 12AU7.
idylldon
Well-known member
What value did you have in mind for the cathode bypass cap on V2b? There's a 2.7K resistor with a cap drawn in, but I didn't see a value listed. I'm assuming it's the same value as that used for V1b, which is 220uf?
I'm almost done putting this together; unfortunately, I only have a 1:3 OT and I'm not even sure it's of the right variety. It's a Stancor model number A-83-C, and it came out of an old Vibrosonic unit that I had lying around. The seconday leads were attached to the grids of a pair of 6L6s, and the primary leads went to a Leslie style, 5-pin socket, which then connected to some other unit I suppose.
Just did some checking and it appears this is an interstage transformer according to the number.
Did you have any specific tranny in mind?
Cheers,
--
Don
I'm almost done putting this together; unfortunately, I only have a 1:3 OT and I'm not even sure it's of the right variety. It's a Stancor model number A-83-C, and it came out of an old Vibrosonic unit that I had lying around. The seconday leads were attached to the grids of a pair of 6L6s, and the primary leads went to a Leslie style, 5-pin socket, which then connected to some other unit I suppose.
Just did some checking and it appears this is an interstage transformer according to the number.
Did you have any specific tranny in mind?
Cheers,
--
Don
NewYorkDave
Well-known member
Don,
You're correct, that's supposed to be a 220uF cap.
Use any input tranny that suits your needs. With a 1:3 ratio on the input, you'll have almost 60dB of total gain. If it's an interstage transformer, it probably won't work very well for mic input, but it might be fine for line input. This amp has so much gain that a 1:1 input would probably be more appropriate for line-level use unless your input signal is on the weak side, or you use a pad.
Please let us know how it turns out! Like I said, this is a concept based on a breadboarded circuit, so it seems you'll be the first to actually build it up complete it real life.
You're correct, that's supposed to be a 220uF cap.
Use any input tranny that suits your needs. With a 1:3 ratio on the input, you'll have almost 60dB of total gain. If it's an interstage transformer, it probably won't work very well for mic input, but it might be fine for line input. This amp has so much gain that a 1:1 input would probably be more appropriate for line-level use unless your input signal is on the weak side, or you use a pad.
Please let us know how it turns out! Like I said, this is a concept based on a breadboarded circuit, so it seems you'll be the first to actually build it up complete it real life.
idylldon
Well-known member
It's the *output* tranny I don't have right now. I have a bunch of cool Webster input trannies, which are 1:14. I wired a socket in for it and also wired up a DI input.
It always seems that no matter how much my in-stock parts selection improves, I never have everything to complete a project without ordering something.
Cheers,
--
Don
It always seems that no matter how much my in-stock parts selection improves, I never have everything to complete a project without ordering something.
Cheers,
--
Don
NewYorkDave
Well-known member
Oh, sorry... You wrote "OT" and I thought "IT." :wink:
In that case: yes, I think your 3:1 interstage would probably make an OK output transformer. If it was intended for a plate driving push-pull 6L6s, the primary inductance will be nice and high. You'll have slightly less output with the 3:1 ratio, but probably a bit less distortion as well, since the WCF will be working into an easier load.
The circuit can drive a load directly (capacitor-coupled, of course) or can drive a load via a 1:1 transformer, but it's a lot happier with a stepdown transformer if it has to drive a 600-ohm load. Driving 600 directly, it can only do about +24 or +25dBM before clip.
In that case: yes, I think your 3:1 interstage would probably make an OK output transformer. If it was intended for a plate driving push-pull 6L6s, the primary inductance will be nice and high. You'll have slightly less output with the 3:1 ratio, but probably a bit less distortion as well, since the WCF will be working into an easier load.
The circuit can drive a load directly (capacitor-coupled, of course) or can drive a load via a 1:1 transformer, but it's a lot happier with a stepdown transformer if it has to drive a 600-ohm load. Driving 600 directly, it can only do about +24 or +25dBM before clip.
idylldon
Well-known member
Found some more info on the tranny:
Stancor A-63-C
Single Plate to Push/Pull Grids
10K/90K Z Impedance
3:1
So I guess I can just wire up the primary (the wires are marked B and P for B+ and Plate, respectively) connecting one of the leads to the 10uf (is this really a 10uf, non-polarzied or poly cap--not a 1.0uf?) and the other to ground, right?
Cheers,
--
Don
Stancor A-63-C
Single Plate to Push/Pull Grids
10K/90K Z Impedance
3:1
So I guess I can just wire up the primary (the wires are marked B and P for B+ and Plate, respectively) connecting one of the leads to the 10uf (is this really a 10uf, non-polarzied or poly cap--not a 1.0uf?) and the other to ground, right?
Cheers,
--
Don
> your 3:1 interstage would probably make an OK output transformer. You'll have slightly less output with the 3:1 ratio, but probably a bit less distortion as well, since the WCF will be working into an easier load.
Probably: no, no.
Lower voltage gain, yes, but not lower maximum output. If the WCF were a voltage source... but it isn't. For max-output calculations, it is a several K ohm source. So 600Ω is a heavy load for 6SN7-class tubes, even with 2:1 transformation, but less-heavy with 3:1 transformation.
And counterintuitively, the WCFs's distortion does not rise at low-Z load. It will drive a short cleanly. The maximum un-clipped voltage does drop, of course! But at any load, a few dB below clipping, THD is low and does not depend on impedance.
> 10K/90K Z Impedance
That's no good at all. The winding resistance is probably greater than line impedance. If you only ever drove 10K loads, you could maybe design a conventional push-pull output with like 6SN7 or even 12AT7 driving the 90K side. But at 600Ω, or even several 10K loads, this is no good.
Probably: no, no.
Lower voltage gain, yes, but not lower maximum output. If the WCF were a voltage source... but it isn't. For max-output calculations, it is a several K ohm source. So 600Ω is a heavy load for 6SN7-class tubes, even with 2:1 transformation, but less-heavy with 3:1 transformation.
And counterintuitively, the WCFs's distortion does not rise at low-Z load. It will drive a short cleanly. The maximum un-clipped voltage does drop, of course! But at any load, a few dB below clipping, THD is low and does not depend on impedance.
> 10K/90K Z Impedance
That's no good at all. The winding resistance is probably greater than line impedance. If you only ever drove 10K loads, you could maybe design a conventional push-pull output with like 6SN7 or even 12AT7 driving the 90K side. But at 600Ω, or even several 10K loads, this is no good.
> other tubes that will work good in a WCF besides the 12BH7?
For what purpose? What do you mean "good"?
Any tube will work. Triodes are much easier than pentodes. And opposite to what you think, usually you don't have enough plate-cathode voltage to get good current since each tube sees half the total supply voltage, and heater insulation often limits working voltage.
We use WCFs for two reasons: power and impedance.
For POWER, you must face the basic matching rule. Plate resistance slightly lower than load impedance. With a single triode, Rl=2Rp is a good point for power, higher is good for power with low distortion. The WCF reduces the distortion problem, and you have two tubes, so Rl can be taken as equal to one plate resistance. To drive 600Ω to high power, you want Rp around 600Ω.
You also want low losses in drive and bias, and this gets into the other reason we like the WCF. It is a feedback amplifier, but only if it has gain. So you need a decent Mu. But this conflicts with getting low Rp at reasonable heater power. In tubes, heater power determines cathode size which determines Gm and maximum current. Rp is Mu times higher than 1/Gm.
So for low output impedance, at a given cathode size (heater power), Mu is not very important. But for maximum power, Mu should be low. Probably 5-10 would be good, but the available tube types are very limited so we take what we can find.
Note that with the 250V-300V supplies needed for the voltage-amp stages, we actually DON'T need "good power output". A perfect WCF with 300V supply would deliver over 100V RMS, which is +42dBm or over 10 Watts or about 5 times higher than you need to blow-out most line inputs. We can often be happy with 1/5th or 1/10th of that much. This leads to a very low-efficiency design, but in many cases this is acceptable.
If you have 300V B+, 150V on each tube, and want 15V peak output, you could use a tube with diode-line (grid voltage = zero) plate impedance 9 times higher than load impedance. So Rp can be 5K for a 600Ω load. Which is about what we do when we use the 6SN7-class or TV-tuner tubes.
A few years ago I posted an essay at HeadWize comparing a number of tubes for driving low-Z loads. For several reasons I'll re-post it here:
------------------------
> i wanted to be able to drive low impedance headphones relitively easily, without resorting to output transformers or feedback.
Having worked on a similar problem for some years now, I do not believe this is possible with any sane plan.
The root problem is that a vacuum tube's conductance is related to its heater power and its Mu. High-Mu is worse, super-low-Mu gives little advantage and causes drive difficulty. Mu of 3 to 10 is best for direct drive of low-Z loads without transformers.
In a triode, the large signal conductance is roughly equal to the inverse of its plate resistance. We will probably pick a bias point at very high current and fairly low plate voltage.
Therefore the product of conductance and heater power is the factor of interest. And it has always been important: we always want less heater power to reduce cost and heat, we almost always want higher conductance. 70 years of research brought about some improvement, but there seems to be a basic limit we are fighting. Electrons really don't like to cross vacuum.
Here is a table for available tubes in various sizes and vintages, assuming no grid current. The last column of numbers is a figure of merit, where lower is better.
6AC7(half) ____ 300 ohms __ 8 watts === 2,400 (6080 same;hard to drive)
6h30(half) __ 1,100 ohms __ 3 watts === 3,300
6DJ8(half) __ 3,800 ohms __ 1 watts === 3,800 (ECC88 similar)
6c33c _________ 120 ohms __ 38 watts == 4,500
2A3 ___________ 800 ohms __ 6 watts === 4,800
6SN7(half) __ 7,000 ohms __ 1 watt ==== 7,000 (12AU7, ECC82 similar)
300B ________ 1,250 ohms __ 6 watts === 7,500
8417(triode)_ 1,000 ohms __ 10 watts = 10,000
6L6(triode) _ 2,000 ohms __ 6 watts == 12,000
ECC81(half)_ 14,000 ohms __ 1 watt === 14,000 (12AT7; easy to drive)
That's for large signals. A vacuum tube with 1 watt of heater power conducts like a 2,500 to 5,000+ ohm resistor, at best. To get a lower large-signal impedance, we need bigger/more tubes eating more heater power. To drive 32 ohms "well", we want conductance more like a 32 ohm resistor, which means about 100 watts of heater power!
This is clearly over-kill for a 0.1 watt signal. We could use less heater power, let the plate power be wasted, and manage 0.1 watts with say 30 watts heater and plate power. But the output resistance is high.
The real answer is a transformer. Solves "about" as many problems as it causes. But there are reasons to reject it.
Small-signal output impedance does not have to be the same as the large signal impedance. In a plate-loaded stage, it is (approximately). Plate-loaded also gives voltage gain, but not for very-small load impedances. Feedback IS the answer. (Heavy damping resistors can also lower effective output impedance, but waste huge amounts of precious signal power and usually increase distortion.) The simplest feedback, one that is tolerated by some feedback-phobic fanatics, is the Cathode Follower in various disguises. Generally, output impedance is Mu times lower, where Mu is Amplification factor.
Another chart, this time dividing by Mu to estimate the small-signal output impedance with feedback (local feedback as cathode follower or overall feedback; makes little difference):
6DJ8(half) __ 3,800 __ 32 ==== 120
6h30(half) __ 3,300 __ 15 ==== 220
ECC81(half)_ 14,000 __ 60 ==== 233
6SN7(half) __ 7,000 __ 20 ==== 350
8417(triode) 10,000 __ 17 ==== 600
6c33c _______ 4,500 ___ 6 ==== 750
6AC7(half) __ 2,400 ___ 3 ==== 800
2A3 _________ 4,800 ___ 5 ==== 960
6L6(triode) 12,000 __ 10 === 1200
300B ________ 7,500 ___ 4 === 1800
The higher-Mu tubes are crummy for power, but better for output impedance.
To "match" 32 ohms with 6DJ8 (if you can actually own a true 6DJ8 today) needs about 4 sections of 6DJ8 per ear, 4 watts of heater power. ECC81/12AT7 (or even the vaulted 6h30) needs twice as much, 8 watts heat per ear. Oddly the mighty 6AC7/6080 does pretty poorly: 24 watts per ear (3 sections per ear, 3 bottles for stereo!) to "match" 32 ohms.
There is a significant difference in "damping" between "matched" and "low-Z" drive. Many of the phones I have tested show a bump at 100Hz that can be quite large unless the source impedance is much lower than the phone impedance. That suggests about three times as much heater power (and bottles) to get the damping down. Say what you like, feedback is a useful and often necessary technique, because excess gain is much cheaper than raw impedance.
> I tossed up between a white cathode follower and a mu-follower for some time, but after reading some articles on the shortcomings of white cathode followers, decided to stick with an mu-follower.
The difference is:
* For large signals, very little. Either topology has the same limits, and for very-low-Z loads they have to be carefully tuned to come close to those limits.
* For small signals: the mu-follower has some gain, the cathode followers (in several forms) trade gain for low impedance. If you took overall feedback, they come out the same. With only local feedback, some version of the cathode follower is clearly best for impedance, and you can make-up gain in some earlier low-power stage. And of the cathode followers, the White is about the best of the bunch, if you can accept two tubes. For a topology that isn't really push-pull, it comes very close to push-pull performance: half the output impedance, nearly twice the output power.
> i had requiraments for the types of valves i would use in the amp- i did not want to resort to anything rare or overly expensive, hence i wanted a stage with a ECC81, 82 or 83.
12AT7/ECC81 is very bad for power, quite good for impedance, and dirt cheap.
6SN7/12AU7/ECC82 is medium on power and impedance, but very easy to get.
6DJ8, ECC88, and the other top-notch TV tuner tubes, are pretty good for power and best for impedance. As good as or better than some big and exotic tubes! But they are "small": you want several bottles to do well into 32 ohms. The thing to do is look for any of the better TV tuner tubes, like 6BQ7, 6BK7, from overstocked warehouses. And understand the games that were played in this market: there were maybe only 6 basic tuner tubes (best, good, and not-so-good; sharp-cutoff and remote-cutoff) but every tube maker and every TV set maker had their own number to try to monopolize the re-tube market. Also, there are bargains in the odd voltage versions like 4 volt heater models.
Since these tubes are usually non-linear (deliberately so for the remote-cutoff variants), a push-pull or quasi-push-pull is going to be lower distortion than a single-ended plan like Waarde's.
Three to six parallel ECC88 or 12AT7, per side, per channel (yes, 12 to 24 tubes!) will give 32 to 16 ohm source impedance at unity gain (White Follower, or Mu-stage with overall feedback). If you use fewer tubes, output impedance for 32-ohm load is so high that the driver is hardly damped at all. In that case, fewer tubes won't harm damping, only power. Then you want to go to my first chart and just get best conductance/heat factor, with 6080 or a couple ECC88s.
For what purpose? What do you mean "good"?
Any tube will work. Triodes are much easier than pentodes. And opposite to what you think, usually you don't have enough plate-cathode voltage to get good current since each tube sees half the total supply voltage, and heater insulation often limits working voltage.
We use WCFs for two reasons: power and impedance.
For POWER, you must face the basic matching rule. Plate resistance slightly lower than load impedance. With a single triode, Rl=2Rp is a good point for power, higher is good for power with low distortion. The WCF reduces the distortion problem, and you have two tubes, so Rl can be taken as equal to one plate resistance. To drive 600Ω to high power, you want Rp around 600Ω.
You also want low losses in drive and bias, and this gets into the other reason we like the WCF. It is a feedback amplifier, but only if it has gain. So you need a decent Mu. But this conflicts with getting low Rp at reasonable heater power. In tubes, heater power determines cathode size which determines Gm and maximum current. Rp is Mu times higher than 1/Gm.
So for low output impedance, at a given cathode size (heater power), Mu is not very important. But for maximum power, Mu should be low. Probably 5-10 would be good, but the available tube types are very limited so we take what we can find.
Note that with the 250V-300V supplies needed for the voltage-amp stages, we actually DON'T need "good power output". A perfect WCF with 300V supply would deliver over 100V RMS, which is +42dBm or over 10 Watts or about 5 times higher than you need to blow-out most line inputs. We can often be happy with 1/5th or 1/10th of that much. This leads to a very low-efficiency design, but in many cases this is acceptable.
If you have 300V B+, 150V on each tube, and want 15V peak output, you could use a tube with diode-line (grid voltage = zero) plate impedance 9 times higher than load impedance. So Rp can be 5K for a 600Ω load. Which is about what we do when we use the 6SN7-class or TV-tuner tubes.
A few years ago I posted an essay at HeadWize comparing a number of tubes for driving low-Z loads. For several reasons I'll re-post it here:
------------------------
> i wanted to be able to drive low impedance headphones relitively easily, without resorting to output transformers or feedback.
Having worked on a similar problem for some years now, I do not believe this is possible with any sane plan.
The root problem is that a vacuum tube's conductance is related to its heater power and its Mu. High-Mu is worse, super-low-Mu gives little advantage and causes drive difficulty. Mu of 3 to 10 is best for direct drive of low-Z loads without transformers.
In a triode, the large signal conductance is roughly equal to the inverse of its plate resistance. We will probably pick a bias point at very high current and fairly low plate voltage.
Therefore the product of conductance and heater power is the factor of interest. And it has always been important: we always want less heater power to reduce cost and heat, we almost always want higher conductance. 70 years of research brought about some improvement, but there seems to be a basic limit we are fighting. Electrons really don't like to cross vacuum.
Here is a table for available tubes in various sizes and vintages, assuming no grid current. The last column of numbers is a figure of merit, where lower is better.
6AC7(half) ____ 300 ohms __ 8 watts === 2,400 (6080 same;hard to drive)
6h30(half) __ 1,100 ohms __ 3 watts === 3,300
6DJ8(half) __ 3,800 ohms __ 1 watts === 3,800 (ECC88 similar)
6c33c _________ 120 ohms __ 38 watts == 4,500
2A3 ___________ 800 ohms __ 6 watts === 4,800
6SN7(half) __ 7,000 ohms __ 1 watt ==== 7,000 (12AU7, ECC82 similar)
300B ________ 1,250 ohms __ 6 watts === 7,500
8417(triode)_ 1,000 ohms __ 10 watts = 10,000
6L6(triode) _ 2,000 ohms __ 6 watts == 12,000
ECC81(half)_ 14,000 ohms __ 1 watt === 14,000 (12AT7; easy to drive)
That's for large signals. A vacuum tube with 1 watt of heater power conducts like a 2,500 to 5,000+ ohm resistor, at best. To get a lower large-signal impedance, we need bigger/more tubes eating more heater power. To drive 32 ohms "well", we want conductance more like a 32 ohm resistor, which means about 100 watts of heater power!
This is clearly over-kill for a 0.1 watt signal. We could use less heater power, let the plate power be wasted, and manage 0.1 watts with say 30 watts heater and plate power. But the output resistance is high.
The real answer is a transformer. Solves "about" as many problems as it causes. But there are reasons to reject it.
Small-signal output impedance does not have to be the same as the large signal impedance. In a plate-loaded stage, it is (approximately). Plate-loaded also gives voltage gain, but not for very-small load impedances. Feedback IS the answer. (Heavy damping resistors can also lower effective output impedance, but waste huge amounts of precious signal power and usually increase distortion.) The simplest feedback, one that is tolerated by some feedback-phobic fanatics, is the Cathode Follower in various disguises. Generally, output impedance is Mu times lower, where Mu is Amplification factor.
Another chart, this time dividing by Mu to estimate the small-signal output impedance with feedback (local feedback as cathode follower or overall feedback; makes little difference):
6DJ8(half) __ 3,800 __ 32 ==== 120
6h30(half) __ 3,300 __ 15 ==== 220
ECC81(half)_ 14,000 __ 60 ==== 233
6SN7(half) __ 7,000 __ 20 ==== 350
8417(triode) 10,000 __ 17 ==== 600
6c33c _______ 4,500 ___ 6 ==== 750
6AC7(half) __ 2,400 ___ 3 ==== 800
2A3 _________ 4,800 ___ 5 ==== 960
6L6(triode) 12,000 __ 10 === 1200
300B ________ 7,500 ___ 4 === 1800
The higher-Mu tubes are crummy for power, but better for output impedance.
To "match" 32 ohms with 6DJ8 (if you can actually own a true 6DJ8 today) needs about 4 sections of 6DJ8 per ear, 4 watts of heater power. ECC81/12AT7 (or even the vaulted 6h30) needs twice as much, 8 watts heat per ear. Oddly the mighty 6AC7/6080 does pretty poorly: 24 watts per ear (3 sections per ear, 3 bottles for stereo!) to "match" 32 ohms.
There is a significant difference in "damping" between "matched" and "low-Z" drive. Many of the phones I have tested show a bump at 100Hz that can be quite large unless the source impedance is much lower than the phone impedance. That suggests about three times as much heater power (and bottles) to get the damping down. Say what you like, feedback is a useful and often necessary technique, because excess gain is much cheaper than raw impedance.
> I tossed up between a white cathode follower and a mu-follower for some time, but after reading some articles on the shortcomings of white cathode followers, decided to stick with an mu-follower.
The difference is:
* For large signals, very little. Either topology has the same limits, and for very-low-Z loads they have to be carefully tuned to come close to those limits.
* For small signals: the mu-follower has some gain, the cathode followers (in several forms) trade gain for low impedance. If you took overall feedback, they come out the same. With only local feedback, some version of the cathode follower is clearly best for impedance, and you can make-up gain in some earlier low-power stage. And of the cathode followers, the White is about the best of the bunch, if you can accept two tubes. For a topology that isn't really push-pull, it comes very close to push-pull performance: half the output impedance, nearly twice the output power.
> i had requiraments for the types of valves i would use in the amp- i did not want to resort to anything rare or overly expensive, hence i wanted a stage with a ECC81, 82 or 83.
12AT7/ECC81 is very bad for power, quite good for impedance, and dirt cheap.
6SN7/12AU7/ECC82 is medium on power and impedance, but very easy to get.
6DJ8, ECC88, and the other top-notch TV tuner tubes, are pretty good for power and best for impedance. As good as or better than some big and exotic tubes! But they are "small": you want several bottles to do well into 32 ohms. The thing to do is look for any of the better TV tuner tubes, like 6BQ7, 6BK7, from overstocked warehouses. And understand the games that were played in this market: there were maybe only 6 basic tuner tubes (best, good, and not-so-good; sharp-cutoff and remote-cutoff) but every tube maker and every TV set maker had their own number to try to monopolize the re-tube market. Also, there are bargains in the odd voltage versions like 4 volt heater models.
Since these tubes are usually non-linear (deliberately so for the remote-cutoff variants), a push-pull or quasi-push-pull is going to be lower distortion than a single-ended plan like Waarde's.
Three to six parallel ECC88 or 12AT7, per side, per channel (yes, 12 to 24 tubes!) will give 32 to 16 ohm source impedance at unity gain (White Follower, or Mu-stage with overall feedback). If you use fewer tubes, output impedance for 32-ohm load is so high that the driver is hardly damped at all. In that case, fewer tubes won't harm damping, only power. Then you want to go to my first chart and just get best conductance/heat factor, with 6080 or a couple ECC88s.
This brings up a question I have been thinking about.
Why not use a power BJT (tip 50 say) as a EF but use a tip 50 as a CC in the emitter leg. The Standard 2 diodes from base to ground and emitter R picked as .6V divided bythe emitter R = the CC you want.
Is there a difference in a BJT EF and a Tube CF when not driven to clipping?
You will not have to worry about the cathode to ground voltage rating of the tube.
Why not use a power BJT (tip 50 say) as a EF but use a tip 50 as a CC in the emitter leg. The Standard 2 diodes from base to ground and emitter R picked as .6V divided bythe emitter R = the CC you want.
Is there a difference in a BJT EF and a Tube CF when not driven to clipping?
You will not have to worry about the cathode to ground voltage rating of the tube.
idylldon
Well-known member
Great discussion so far, though I'm still struggling to integrate it all into my gray matter!
So, what would be an ideal transformer to use in this circuit? Since ideals are seldom possible, what should I be looking for in a transformer?
I truly appreciate all your patience with these lower-level questions.
Thanks,
--
Don
So, what would be an ideal transformer to use in this circuit? Since ideals are seldom possible, what should I be looking for in a transformer?
I truly appreciate all your patience with these lower-level questions.
Thanks,
--
Don
> Why not use a power BJT
Because it is a transistor!!! How dare you put a transistor in a tube circuit???
> Why not use a power BJT (tip 50 say) as a EF but use a tip 50 as a CC in the emitter leg. ... Is there a difference in a BJT EF and a Tube CF when not driven to clipping?
For the same topology: slim to none. And the BJT can readily be designed to deliver MUCH more current than any practical vacuum device.
But you are proposing to replace a White Cathode Follower with a Constant-current emitter follower. The WCF pulls both ways, true push-pull (up to a point). The CC-CF only pulls up, and pull-down is a dumb current source.
For the same idle current, for loads lower than the tube's Rp, the WCF will deliver more load current. The peak load current for the CC-CF is the idle current, for the WCF it is almost twice the idle current.
To be fair: for the same TOTAL power (including heater) the BJT even with a dumb CC will meet or beat the VT WCF.
The output impedance of the WCF is half that of a plain CF. But the Gm of a BJT is about 100 times higher than the GM of a tube, both working around 10mA.
One "fault" with BJTs: they will happily try to flow infinite current. Hang a long piece of wire on the output (capacitance), feed a transient, POOF! goes the silicon. It is a lot harder to pop a tube.
But the real answer is: it is not fashionable to add BJTs.
Because it is a transistor!!! How dare you put a transistor in a tube circuit???
> Why not use a power BJT (tip 50 say) as a EF but use a tip 50 as a CC in the emitter leg. ... Is there a difference in a BJT EF and a Tube CF when not driven to clipping?
For the same topology: slim to none. And the BJT can readily be designed to deliver MUCH more current than any practical vacuum device.
But you are proposing to replace a White Cathode Follower with a Constant-current emitter follower. The WCF pulls both ways, true push-pull (up to a point). The CC-CF only pulls up, and pull-down is a dumb current source.
For the same idle current, for loads lower than the tube's Rp, the WCF will deliver more load current. The peak load current for the CC-CF is the idle current, for the WCF it is almost twice the idle current.
To be fair: for the same TOTAL power (including heater) the BJT even with a dumb CC will meet or beat the VT WCF.
The output impedance of the WCF is half that of a plain CF. But the Gm of a BJT is about 100 times higher than the GM of a tube, both working around 10mA.
One "fault" with BJTs: they will happily try to flow infinite current. Hang a long piece of wire on the output (capacitance), feed a transient, POOF! goes the silicon. It is a lot harder to pop a tube.
But the real answer is: it is not fashionable to add BJTs.
