Hello to all 
I would have to choose between some valves for a phase splitter
(For directly drive a couple of EL84 or EL34)
In Long Tail Pair it is better a hi mu or a low mu triode ?
12AT7
(Plate 250 V, 10mA)
u = 60
Ri = 10k Ω
gm = 5.5 mMho
12AU7
(Plate 250 V, 10.5mA)
u = 17
Ri = 7.7K Ω
gm = 2.2 µMho
12AX7
(Plate 250 V, 1.2mA)
u = 100
Ri = 62.5K Ω
gm = 1.6 µMho
Perhaps it would be better a low mu and low plate resistance as 12AU7 ?
Thanks.
I would have to choose between some valves for a phase splitter
(For directly drive a couple of EL84 or EL34)
In Long Tail Pair it is better a hi mu or a low mu triode ?
12AT7
(Plate 250 V, 10mA)
u = 60
Ri = 10k Ω
gm = 5.5 mMho
12AU7
(Plate 250 V, 10.5mA)
u = 17
Ri = 7.7K Ω
gm = 2.2 µMho
12AX7
(Plate 250 V, 1.2mA)
u = 100
Ri = 62.5K Ω
gm = 1.6 µMho
Perhaps it would be better a low mu and low plate resistance as 12AU7 ?
Thanks.
Simple answer, none of the above.
The challenge of phase splitting is the variation of output impedance between anode and cathode produces all sorts of drive/slew related harmonics, pushing distortion up.
Much better (Imo) to have a unity gain splitter (any of the above), feeding a pair of either anode followers, so that the impedances match up.
This is but one reason why a Transformer phase splitter is very handy
Have fun
Owen
The challenge of phase splitting is the variation of output impedance between anode and cathode produces all sorts of drive/slew related harmonics, pushing distortion up.
Much better (Imo) to have a unity gain splitter (any of the above), feeding a pair of either anode followers, so that the impedances match up.
This is but one reason why a Transformer phase splitter is very handy
Have fun
Owen
owen said:Simple answer, none of the above.
The challenge of phase splitting is the variation of output impedance between anode and cathode produces all sorts of drive/slew related harmonics, pushing distortion up.
Much better (Imo) to have a unity gain splitter (any of the above), feeding a pair of either anode followers, so that the impedances match up.
This is but one reason why a Transformer phase splitter is very handy
Have fun
Owen
He was actually talking about a long tail pair which with a CCS on the bottom is very symmetrical in terms of source impedance and ac parameters. I would tend to choose a type with a relatively low rp for this sort of service, of the types listed either the 12AT7 or the 12AU7A could serve. The 12AT7 will give you about 3X the overall voltage gain of the 12AU7A, neither type IMO is the best in terms of linearity but in PP service all of the even harmonics are largely cancelled - I think the 12AT7 might have a slightly more desirable spectra in this application. (Eli and SY both think highly of this tube for this sort of service, good enough for me IMHO.)
Another very interesting type to consider for this application, particularly with EL34 where the drive requirements are greater is the 5687, rp would be much lower as well - useful in particular if you are contemplating PPP triode connected EL34..
For a very interesting discussion of the problems you can encounter using IT in PP amplifiers see this thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=13480 Note also that even with a single IT if you intend to use negative feedback in this amplifier design getting good stability with meaningful levels of feedback and a "bogey" load is going to be difficult. ("bogey" in this case = any typically random speaker load)
12AU7 has terrible linearity, so I wouldn't use it outside of guitar amp service. If you want a medium mu LTP, 6SN7 or 6CG7 are much better choices.
12AX7 has issues with any loading heavier than a CCS, and the high mu means high Miller capacitance. But if you can use very high plate resistors and a high supply voltage (say, 300k and 400V), buffer the output and drive the input from a low source impedance, it is surprisingly linear.
12AT7 tends to be less linear than a 12AX7 run optimally, but... running the 12AX7 optimally is a challenge. For most practical circuits, the 12AT7 will show better linearity.
Using a good CCS in the tail is a must. That will take care of any balance issues.
12AX7 has issues with any loading heavier than a CCS, and the high mu means high Miller capacitance. But if you can use very high plate resistors and a high supply voltage (say, 300k and 400V), buffer the output and drive the input from a low source impedance, it is surprisingly linear.
12AT7 tends to be less linear than a 12AX7 run optimally, but... running the 12AX7 optimally is a challenge. For most practical circuits, the 12AT7 will show better linearity.
Using a good CCS in the tail is a must. That will take care of any balance issues.
hey-Hey!!!,
Type 12A4 is another single triode with good gain/gm combination. Also has good anode dissipation and voltage tolerance...extremely linear too. Also consider 6H30, perhaps also 6S45( a single triode so you'll need a pair just like 12A4 ).
And then there's pentode LTP too...
cheers,
Douglas
Type 12A4 is another single triode with good gain/gm combination. Also has good anode dissipation and voltage tolerance...extremely linear too. Also consider 6H30, perhaps also 6S45( a single triode so you'll need a pair just like 12A4 ).
And then there's pentode LTP too...
cheers,
Douglas
Johan Potgieter said:
Try this one: http://www.diyaudio.com/forums/showthread.php?s=&threadid=134804
I feel sure the 12AT7 is the right choice here; second would be 6SN7/6CG7/6FQ7. I agree the 12AU7 is unfit for audio purposes (yet people still use it!), and the 12AX7 is too lightweight for driving EL34s, especially if you use fixed bias because that will mean a low-ish value for the grid resistors.
I am conscious of Morgan Jones's warning in his Valve Amplifiers against using an LTP splitter to drive OP tubes in Class AB, but I'm unsure what problems it is likely to cause.
I am conscious of Morgan Jones's warning in his Valve Amplifiers against using an LTP splitter to drive OP tubes in Class AB, but I'm unsure what problems it is likely to cause.
the lemmings syndrome
In retrospect, it's a true shame that many of you weren't around decades ago to guide and mentor all of the successful designers who unleashed hundreds of inferior audio products that used this horrible tube to the general public, who in turn, had no idea what horrible sounding equipment they were getting stuck with. The list of these shameless manufacturers is long and includes names like Altec, Bogen, Eico, Fisher, Harmon-Kardon, Heathkit, Jadis, Leslie, McIntosh, Pederson, Pilot, RCA, Scott, Sherwood, Stromberg-Carlson and untold others. What they hell were they thinking?? Oh, the "audio crimes" you could have prevented! For those of you who were unfortunate to have fallen victim and purchased any of these products based on sound quality or marketing hype, do yourself a favor... dump it quick!
Regards, KM
In retrospect, it's a true shame that many of you weren't around decades ago to guide and mentor all of the successful designers who unleashed hundreds of inferior audio products that used this horrible tube to the general public, who in turn, had no idea what horrible sounding equipment they were getting stuck with. The list of these shameless manufacturers is long and includes names like Altec, Bogen, Eico, Fisher, Harmon-Kardon, Heathkit, Jadis, Leslie, McIntosh, Pederson, Pilot, RCA, Scott, Sherwood, Stromberg-Carlson and untold others. What they hell were they thinking?? Oh, the "audio crimes" you could have prevented! For those of you who were unfortunate to have fallen victim and purchased any of these products based on sound quality or marketing hype, do yourself a favor... dump it quick!
Regards, KM
kmaier: Simmer down. The 12AU7 IS actually very non-linear; just take a deep breath and Google search some plate curves for yourself. A non-linear tube = distortion. There are ways of working around that, as many of those hi-fi amps probably do, but they also had the motivation to use super-common tubes that people could replace easily. We in the DIY community have the luxury of using whatever weird tubes we want, so why not start a design with a linear one?
The rest of you, carry on.
The rest of you, carry on.
ray_moth said:
I don't think he tried a LTP with common-mode feedback for AC self balancing though:
http://dogstar.dantimax.dk/tubestuf/driver02.htm
SS CCS version:
http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=2592.msg32023#msg32023
Cheers!
Just to throw another concept into the pot: Have a look at Arthur Bailey’s article on the improved phase-splitter [Wireless World September 1962 and should be on the internet somewhere] that uses a triode pentode [ECF82]; Arthur Radford used this idea to drive a pair of EL34’s in his very successful STA15 and STA25 amplifiers. I’m about to experiment with this myself using a CCS in the tail - could be interesting.
one last shot
Mike,
No rocket science.... I've used them as LTP phase inverters and they work fine. Nothing fancy either, simple resistor biasing and reasonable current. The Philips data sheet for the ECC82 shows distortion levels of around 1% for input/phase inverter circuits... shows the values too. Like any tube, they can be used and can perform well depending on the circuit.
I would discount an earlier comment that they were used because they were popular. Back in the heyday of tubes, (equipment) manufacturers bought in bulk directly from the tube manufacturer... even had their logo printed on the tube as well... it made little to no difference if they bought 12AX7, 12AU7 or anything else. Hence, the companies that designed and used specific tube types ARE what made them popular.
I did a clean-sheet design for a Class-A SET a few years ago. I settled on a 2-stage direct-coupled cascade input/driver topology with low mu triodes. My first choice was a 6FQ7 as it's an old favorite... and it's a great choice in many ways. I was not able to get the performance I wanted and filament hum coupling was also on the high side (tried over 2 dozen NOS tubes... RCA, GE and Sylvania). In the end I was just not happy.
I eventually grabbed a 5814A (premium version of the 12AU7) and re-biased the stages (using standard published plate curves). After some testing, measuring and refining, it worked out better (yes, I was a bit surprised). I also went thru a couple dozen (again, mixed brands... RCA, ECG Philips, Mullard and others) to ensure the circuit was consistent. Overall gain is about 43dB and virtually zero filament hum coupling. The design can output a clean 150 volts peak-to-peak and drives a 45 or 2A3 with very low overall distortion and excellent gain linearity. At 1-watt output, THD measurements are below 0.4% and less than 1% at 2-watts output, zero feedback. The amplifier is flat within 1dB from 25Hz to 45KHz at 1-watt. Output noise with good tubes is below 250 micro-volts with all AC filaments.
So, is the 12AU7 perfect, of course not... but to make such blanket statements against it are misleading and unrealistic.... I do prefer the 5814A version however, but just my $0.02 based on personal findings. I've no desire to engage in any more bashing for or against it however... everyone has their own views and opinions which is fine... but hopefully they are based on personal experience, not someone else's.
Regards, KM
mashaffer said:
Mike,
No rocket science.... I've used them as LTP phase inverters and they work fine. Nothing fancy either, simple resistor biasing and reasonable current. The Philips data sheet for the ECC82 shows distortion levels of around 1% for input/phase inverter circuits... shows the values too. Like any tube, they can be used and can perform well depending on the circuit.
I would discount an earlier comment that they were used because they were popular. Back in the heyday of tubes, (equipment) manufacturers bought in bulk directly from the tube manufacturer... even had their logo printed on the tube as well... it made little to no difference if they bought 12AX7, 12AU7 or anything else. Hence, the companies that designed and used specific tube types ARE what made them popular.
I did a clean-sheet design for a Class-A SET a few years ago. I settled on a 2-stage direct-coupled cascade input/driver topology with low mu triodes. My first choice was a 6FQ7 as it's an old favorite... and it's a great choice in many ways. I was not able to get the performance I wanted and filament hum coupling was also on the high side (tried over 2 dozen NOS tubes... RCA, GE and Sylvania). In the end I was just not happy.
I eventually grabbed a 5814A (premium version of the 12AU7) and re-biased the stages (using standard published plate curves). After some testing, measuring and refining, it worked out better (yes, I was a bit surprised). I also went thru a couple dozen (again, mixed brands... RCA, ECG Philips, Mullard and others) to ensure the circuit was consistent. Overall gain is about 43dB and virtually zero filament hum coupling. The design can output a clean 150 volts peak-to-peak and drives a 45 or 2A3 with very low overall distortion and excellent gain linearity. At 1-watt output, THD measurements are below 0.4% and less than 1% at 2-watts output, zero feedback. The amplifier is flat within 1dB from 25Hz to 45KHz at 1-watt. Output noise with good tubes is below 250 micro-volts with all AC filaments.
So, is the 12AU7 perfect, of course not... but to make such blanket statements against it are misleading and unrealistic.... I do prefer the 5814A version however, but just my $0.02 based on personal findings. I've no desire to engage in any more bashing for or against it however... everyone has their own views and opinions which is fine... but hopefully they are based on personal experience, not someone else's.
Regards, KM
The best that I used in the past is 7044. Second is 6463 and 5687.
Now I am trying with 6H30.
The problem with them is the gain that is non high.
It is always better to use a low impedance tube, in this case the output impedance can give you the possibility to have a good swing for a large BW indipendently from the load (grid of the output tube and its associate Cag).
Ciao
Walter
Now I am trying with 6H30.
The problem with them is the gain that is non high.
It is always better to use a low impedance tube, in this case the output impedance can give you the possibility to have a good swing for a large BW indipendently from the load (grid of the output tube and its associate Cag).
Ciao
Walter
Thanks KM, I appreciate the explanation. I like to look at approaches that are off the beaten path as they are often very interesting. I am not sure what topology you are describing as input/driver cascade. Is that a common cathode direct coupled to a cathode follower, some sort of totem pole or something else?
Thanks again.
Thanks again.