Ahhh, such a trap it is woodturner-fran. The CCS using the LM317 for the DIYparadise amp is designed for that amp, which does not use a phase splitter. The phase-splitting occurs within the EL84 output section itself, which works exactly the same as a LTP phase-splitter with a CCS, of which there are many examples on this site.
So when you have a more conventional amplifier, like the tubecad design, which includes a phase splitter, one LM317 CCS for each output tube will not work, unless you have a bypass capacitor. This is why: as the grid voltage rises, the tube will start conducting more current, but the CCS will oppose any increase in tube current by increasing the voltage from cathode to ground. Since the cathode voltage has now risen, and the grid voltage has risen, there is no net input signal to the tube and the tube will not conduct any more current, therefore no output. With a cathode bypass capacitor in parallel with the CCS, say 100uF to 1000uF, it will charge and discharge as tube current changes, holding the cathode at a relatively stable voltage, so then you get an output.
So your choices are: (1) a CCS with a bypass cap for each output tube, (2) a CCS without bypass cap for each pair of output tubes which will limit you to Class A only, or (3) a CCS with a bypass cap for each pair of output tubes.
Shoog knows a lot about this stuff, search some of his posts for more info.
With simple EL84 designs, you will see that some designers prefer a resistor and cap for each individual output tube and others prefer a resistor and cap for each pair of output tubes. And still others prefer LED bias modules (Red Light District). My opinion only, for the best results for the least extra effort, is convert to fixed bias (if the B+ isn't too high that is - fixed bias will add another 11V or so to the voltage across the tubes.) I have found fixed bias (or LED modules) to give deeper bass and more detailed highs, but if you never heard the difference in the first place, you would not miss it.
So when you have a more conventional amplifier, like the tubecad design, which includes a phase splitter, one LM317 CCS for each output tube will not work, unless you have a bypass capacitor. This is why: as the grid voltage rises, the tube will start conducting more current, but the CCS will oppose any increase in tube current by increasing the voltage from cathode to ground. Since the cathode voltage has now risen, and the grid voltage has risen, there is no net input signal to the tube and the tube will not conduct any more current, therefore no output. With a cathode bypass capacitor in parallel with the CCS, say 100uF to 1000uF, it will charge and discharge as tube current changes, holding the cathode at a relatively stable voltage, so then you get an output.
So your choices are: (1) a CCS with a bypass cap for each output tube, (2) a CCS without bypass cap for each pair of output tubes which will limit you to Class A only, or (3) a CCS with a bypass cap for each pair of output tubes.
Shoog knows a lot about this stuff, search some of his posts for more info.
With simple EL84 designs, you will see that some designers prefer a resistor and cap for each individual output tube and others prefer a resistor and cap for each pair of output tubes. And still others prefer LED bias modules (Red Light District). My opinion only, for the best results for the least extra effort, is convert to fixed bias (if the B+ isn't too high that is - fixed bias will add another 11V or so to the voltage across the tubes.) I have found fixed bias (or LED modules) to give deeper bass and more detailed highs, but if you never heard the difference in the first place, you would not miss it.
Last edited:
Though it would be good to add a reference to skgpkg's other thread:
http://www.diyaudio.com/forums/tubes-valves/171386-anyone-ever-made-el84-p-p-circuit-referenced-tubecad-journal-site.html
Fran
http://www.diyaudio.com/forums/tubes-valves/171386-anyone-ever-made-el84-p-p-circuit-referenced-tubecad-journal-site.html
Fran
EL84 PP with global NFB
Has anyone built one of these with an UL output transformer and global NFB?
I want to build one of these like the MJ Bevois Valley amplifier with a 6N1P in place of the E88CC and apply NFB from the 16ohm output tap of an 8K to 16ohm UL output transformer.
I need to know how to calculate the feedback resistor value back to the cathode of the input stage with a 340 ohm Rk.
Thanks
Martin
Has anyone built one of these with an UL output transformer and global NFB?
I want to build one of these like the MJ Bevois Valley amplifier with a 6N1P in place of the E88CC and apply NFB from the 16ohm output tap of an 8K to 16ohm UL output transformer.
I need to know how to calculate the feedback resistor value back to the cathode of the input stage with a 340 ohm Rk.
Thanks
Martin
I want to use this driver stage (see post #1) and drive a pair of 6L6 or 5881's with it. I think it's needs a bit more gain to drive it though. I have ~360-370 volts to work with so I can adjust the 4.3K for the working voltage. Thinking I can change the 30K resistors to 15K so I can get 200V on the plates instead of 150V then the voltage divider will take it down to 100V for the split load splitter. I think the cathode resistor for the 6N1p should be around 500 Ohms.
Will this give me enough gain or am I completely off.
Will this give me enough gain or am I completely off.
The gain of the 6N1P is about 35. Putting a LED in place of the cathode resistor should ensure that you get all of that gain. The .1uF cap that bypasses the 1M resistor ensures that all of the AC signal gets to the concertina splitter. So with a standard 2v peak to peak signal you should get +/-35 volts of signal which is enough to drive a 6L6. But you may need higher current than that concertina can supply, especially if you want to go into class AB. Replacing the concertina tube with a suitable MOSFET and lowering the value of the concertina resistors would probably help with that.
The 340R resistor in the cathode gives degenerative feedback insuring low distortion of the input stage (leave out the LED int he cathode unless you really need the additional gain).
The paraphase phase splitter is also a low distortion stage due to degenerative feedback.
The output stage in class A triode mode with cancellation of second harmonic distortion results in a low distortion output.
What is not to like?
The paraphase phase splitter is also a low distortion stage due to degenerative feedback.
The output stage in class A triode mode with cancellation of second harmonic distortion results in a low distortion output.
What is not to like?
I question whether you need 3 stages, since the EL84 is a relatively high mu tube. My Leak Stereo 20 uses 6S4 triodes instead, in PPP. They have a mu of 16.5. I reworked the chassis slightly. Currently it has an ECC40 front end as a diff pair. Mu 32 and sounds nice, but different socket. I'll find an alternative, and currently thinking of a directly heated 3a5 for that.
As regards bias I've been getting very interesting results with SIC diodes, specifically C3D02060F. Something I got off Ale Moglia's 01A preamp in a recent shootout. This is my bias of choice right now. Each diode is giving me .92v in my 26 preamp. In that preamp I add a 120R resistor to the chain for a little feedback. Sound is very clear and clean.
Andy, Mark was talking about driving 6L6 tubes rather than EL84. That's about twice the voltage drive and probably more current if going into AB. And once you go into AB, those 6L6 grids can draw some current. The one weakness of the concertina using a tube is that due to the Rp, the output signals have uneven current capabilities. A MOSFET has a far lower Rds which gives more balanced current capabilities.
I've run this circuit with a 6N1P, EL84s and a MOSFET in place of the second tube, but without the bypass cap in order to lower the gain. It works brilliantly. You could do two stages instead of 3, but it's the same number of tubes and the first diff pair would need a CCS to get good results - which may require another power rail.
One of the subtle brilliances of the original circuit is that the resistors in the first two stages are matched. Any current increase in the first stage is matched by a current decrease in the second stage, and vice-versa. The same goes for the balanced output stage. This results in a constant current circuit, which keeps the power supply from being modulated by changing currents.
This circuit and variations thereof are probably my most commonly built schematic. I'v done many builds based on the same basic idea, just with values changed to suit.
Other than my flea amp (a real giant killer!!!!) my current daily use amplifier is this circuit, with 6N2P input/phase inverter, 6CW5/EL86 outputs, with garter bias cathodes. All toroidal power and ouput iron. It's powerful, affordable, easy to drive, and sounds fantastic. LED biased 6N1P up front would be a great way to go if you don't want/need the extra gain for feedback, or if you have hot sources feeding it.
Other than my flea amp (a real giant killer!!!!) my current daily use amplifier is this circuit, with 6N2P input/phase inverter, 6CW5/EL86 outputs, with garter bias cathodes. All toroidal power and ouput iron. It's powerful, affordable, easy to drive, and sounds fantastic. LED biased 6N1P up front would be a great way to go if you don't want/need the extra gain for feedback, or if you have hot sources feeding it.
I'm building this amp and I've got one channel assembled so I was doing my initial testing and measurements on it. My voltages are pretty close, but I'd like to tweak them a bit to get closer to the values in the schematic. Both halves of the 6N1P have a 30k resistor at the anode, but I'd get closer to nominal voltages by replacing one of those resistors with a slightly lower value resistor. My question is would this do more harm than good? I'm only about 10V off and I was wondering if there's a particular reason why the anode resistors are the same value.
I don’t want to highjack the thread, but I’m really interested in more details of your EL86 amp. Do you have a link, or please send me a PM. Thanks
Last edited:
I don’t want to highkack the thread, but I really interested in more details of your EL86 amp. Do ypu have a link, or please send me a PM. Thanks
Yeah, I don't think I'll bother. All the voltages are close enough.
But still the question remains, is there a reason why the anode resistors are both 30k? I have a faint recollection of someone mentioning something about this, but I can't remember where I saw it. I don't understand what difference it would make even if they were different values as long as the voltages are right.