The current is listed as 5mA per tube for ECC88/6DJ8 per tube for 100V, and 10mA per tube for 150V and 200V. Considering that current from Power Supply goes through 2 tubes for each channel, this means that you have 5 or 10 mA per channel. Which means either 10 or 20mA total. So an EZ80 is OK too.
So these are the values you have to model in PSUD2. Look at the Hanmmond website to make a "guesstimate" about chokes inductance vs. DC resistance.
This current is also low enough that you can possibly have a 20H choke, which is about the upper practical limit for this amount of current. The practical limit is the size of the choke relative to the current it can pass. Chokes that allow lower current have more turns of finer wire, but because of that, they have higher resistance too.
And you need 6.3V * 300mA for the heaters. I see a lot of people using PC88 (same as ECC88 but with 7V heaters) with 6.3 volts, I don't see a problem running them with a 6V 2A or 3A transformer.
So these are the values you have to model in PSUD2. Look at the Hanmmond website to make a "guesstimate" about chokes inductance vs. DC resistance.
This current is also low enough that you can possibly have a 20H choke, which is about the upper practical limit for this amount of current. The practical limit is the size of the choke relative to the current it can pass. Chokes that allow lower current have more turns of finer wire, but because of that, they have higher resistance too.
And you need 6.3V * 300mA for the heaters. I see a lot of people using PC88 (same as ECC88 but with 7V heaters) with 6.3 volts, I don't see a problem running them with a 6V 2A or 3A transformer.
Hmmm... For a first build, why would you go straight to an Aikido?
What about taking a single 5687 or 6N6P twin-triode and make a simple, one-tube-for-stereo, common cathode stage with a selector switch and a volume control, and call it a "stereo preamp"? It will have too much gain, sure, but not by so much that it will be unusable. It will have an output impedance of about 2k or so, which is about the same as a Dyna PAS.
I think that's a great way to learn about tube design, because you can design it yourself. It's the basic building block for everything, and makes a good first tube project. Those tube types are low distortion, decent stuff. If done right, it will sound darned good too. Or done wrong, it can give you that extra 2nd harmonic distortion, if that's what you're after.
Why not?
What about taking a single 5687 or 6N6P twin-triode and make a simple, one-tube-for-stereo, common cathode stage with a selector switch and a volume control, and call it a "stereo preamp"? It will have too much gain, sure, but not by so much that it will be unusable. It will have an output impedance of about 2k or so, which is about the same as a Dyna PAS.
I think that's a great way to learn about tube design, because you can design it yourself. It's the basic building block for everything, and makes a good first tube project. Those tube types are low distortion, decent stuff. If done right, it will sound darned good too. Or done wrong, it can give you that extra 2nd harmonic distortion, if that's what you're after.
Why not?
I very much agree with rongon. Aikido is (presumably) a very good topology, but difficult to "get" if you are not familiar.
Simple preamp build guide , and fun to read
How to build YOUR own tube preamp
Simple preamp build guide , and fun to read
How to build YOUR own tube preamp
Well, it's nice to know someone agrees with me! 
That diyparadise article is on the right track. A single 5687 is not expensive at all, maybe $10. I would change some things, however. The layout described in the article is not optimal.
1) Put the input jack and volume control close to the 5687 tube, and far away from the transformers and other power supply parts. Tube socket, input/output jacks and volume control at one end, power supply parts, transformers, power switch, fuse holder and AC cord at the opposite end. Transformers and power supply parts handle large AC currents which radiate hum fields. There is no reason to run the input and output signals directly through them.
2) Tightly twist those filament wires. Related -- It would be easy to make a +6.3V DC supply for the filaments. Even +6V would be close enough. Still, twist the filament wires to reduce hum injected from the heaters.
3) Consider changing the operating points of the tube to be able to hear the differences between them. (Can be very instructive.) I made this basic circuit with a +450V B+ supply so I could go crazy trying out stuff. I found that with a 10k plate load, getting 15mA through each triode sounded noticeably better than only 10mA. I found that I liked 15k in the plate even better. YMMV. While you're at it, plot the loadlines and listen to the differences as they relate to those loadlines.
4) All sorts of tweaks and/or fancy parts can be applied. It's a great way to learn which parts make a difference, and which just cost lots of money.
- One tweak would be to regulate the B+ voltage. You could use a Maida regulator (lots of info available on that, here and elsewhere).
- Try Mills non-inductive wirewound resistors in the 5687 plates. Or even garden-variety wirewounds, which are supposed to be low enough inductance that they shouldn't make any difference.
- Try LED's in the cathodes, for cathode bias. Three red LED's in series would make 5.1V. Two would make 3.4V. Or use a 5.6V zener. Any of those should sound better than the usual cathode resistor with electrolytic capacitor bypass. But you could try it and decide for yourself.
- Try using different rectifiers. Maybe high voltage Schottky diodes, or a larger rectifier tube like a 5Y3 (if your power transformer happens to come with a 5V winding). Maybe try a pair of damper diode tubes (supposed to be the best option). Try some method of bypassing switching frequencies to ground, and see if you hear a difference.
- Try different grounding schemes. You could do a star ground, where everything gets grounded to a single point on the chassis, or you could make a high current star ground for the power supply parts, then a low current ground point for the audio signal parts (the input and output jacks, the tube and its supporting parts).
You can have so much fun with this kind of super simple circuit, and learn a lot while you're at it. It'll also go together quickly -- Instant gratification.
Why not?
--
That diyparadise article is on the right track. A single 5687 is not expensive at all, maybe $10. I would change some things, however. The layout described in the article is not optimal.
1) Put the input jack and volume control close to the 5687 tube, and far away from the transformers and other power supply parts. Tube socket, input/output jacks and volume control at one end, power supply parts, transformers, power switch, fuse holder and AC cord at the opposite end. Transformers and power supply parts handle large AC currents which radiate hum fields. There is no reason to run the input and output signals directly through them.
2) Tightly twist those filament wires. Related -- It would be easy to make a +6.3V DC supply for the filaments. Even +6V would be close enough. Still, twist the filament wires to reduce hum injected from the heaters.
3) Consider changing the operating points of the tube to be able to hear the differences between them. (Can be very instructive.) I made this basic circuit with a +450V B+ supply so I could go crazy trying out stuff. I found that with a 10k plate load, getting 15mA through each triode sounded noticeably better than only 10mA. I found that I liked 15k in the plate even better. YMMV. While you're at it, plot the loadlines and listen to the differences as they relate to those loadlines.
4) All sorts of tweaks and/or fancy parts can be applied. It's a great way to learn which parts make a difference, and which just cost lots of money.
- One tweak would be to regulate the B+ voltage. You could use a Maida regulator (lots of info available on that, here and elsewhere).
- Try Mills non-inductive wirewound resistors in the 5687 plates. Or even garden-variety wirewounds, which are supposed to be low enough inductance that they shouldn't make any difference.
- Try LED's in the cathodes, for cathode bias. Three red LED's in series would make 5.1V. Two would make 3.4V. Or use a 5.6V zener. Any of those should sound better than the usual cathode resistor with electrolytic capacitor bypass. But you could try it and decide for yourself.
- Try using different rectifiers. Maybe high voltage Schottky diodes, or a larger rectifier tube like a 5Y3 (if your power transformer happens to come with a 5V winding). Maybe try a pair of damper diode tubes (supposed to be the best option). Try some method of bypassing switching frequencies to ground, and see if you hear a difference.
- Try different grounding schemes. You could do a star ground, where everything gets grounded to a single point on the chassis, or you could make a high current star ground for the power supply parts, then a low current ground point for the audio signal parts (the input and output jacks, the tube and its supporting parts).
You can have so much fun with this kind of super simple circuit, and learn a lot while you're at it. It'll also go together quickly -- Instant gratification.
Why not?
--
Last edited:
Thanx, so a psu able to handle about 40-50ma should do the trick. ez80 is ezily available
.will check it out.
I am not hung up on aikido. any good design will do. aikido can come later.
what about cathode followers with just a single twin triode tube. i dont need the gain. sources are cd players and soundcards.
i havent seen these two tubes locally. i will try to look for a suitable single tube design with something that i already have or i can easily get it here. first post in this thread has the list of some of the tubes that i have.
i would prefer a already tried and tested design, this being my first tube project, there's plenty to learn anyway. i would like to make a good power supply though, so that even if i change the design later, the psu should be reusable.
currently looking for a neutral one, effects box can be tried later.
I plan to start with generic resistors and electrolytics. i have a few wimas also lying here, so those also i will try to use. later one by one will upgrade to better resistors and russian PIOs. that way i will exactly know which component is doing what.
i am currently thinking of using a tube rectified with choke supply. later can add a vr tube.
trying to make a pure tube build, would try to avoid all silicon.
i will be getting a few rectifier tubes. will surely try the 6v ones. if i am custom ordering a trafo, will try to get both 5v and 6.3v supplies, and will try the 5v tubes also.
will try that also.
There is nothing more tried and true than a simple common cathode triode, or a cathode follower for that matter. These circuits are two of the most basic building blocks -- so simple that most people here will assume that anyone can design one with a little bit of thinking applied. They are really that simple.
Are you asking for a schematic to work from, someone else's design? Or do you want to learn how to design this?
For this type of single tube preamp, you can use any of the following common (and reasonably priced), medium gain, twin-triode tube types and get good results:
5687
6N6P
ECC99
6FQ7 or 6CG7
6SN7
6GU7
12BH7A
For cathode follower (but not common cathode), you could try
12AT7 or ECC81
6DJ8 (ECC88, 6922)
6N23P
6N1P
6AQ8 (ECC85)
6DT8
and so on...
For a PSU -- what are you looking for? Something very simple? Does it have to have a tube rectifier? Do you want to try a regulated supply? Does it have to use a choke?
Look up the Maida high voltage regulator. That's a very useful one. Someone here is selling printed circuit boards (pcb's) for a Maida. That could make things easier.
I highly recommend finding a copy of Morgan Jones "Valve Amplifiers". The older 3rd Edition is around the web in a PDF. The latest 4th Edition has some nice additions, but all the basics are covered in the 3rd Edition. Very worthwhile reading, especially if you have some electronics background but are new to toobz.
--
PS - In MJ Valve Amplifiers 4th Ed there is a power supply design called the Statistical Regulator. It's super simple, and should be very reliable, but tedious to make (uses a looooooong string of 5.6V zener diodes). Since you want a "keeper" power supply, that might be an option.
Last edited:
The most basic step in designing a common cathode triode stage is to choose the operating points for the tube you're using, using a "load line."
Here's a classic article on load lines:
Norman Crowhurst "Making Use of Load Lines"
Also, this looks helpful:
Vacuum Tube Theory
Part 4 covers load lines.
Also, if you use a cathode follower, it almost begs the question why you want this at all. If you make it right, it will be very transparent. A cathode follower should sound like 'nothing.' The only reason to make a cathode follower is if you need to make an impedance matching stage (buffer) using a tube. If you make a cathode follower wrong, to get that 'tube sound,' you'll just get a pretty bad sounding stage.
If you want to experience that 'tube sound,' you'll probably want to build some kind of common cathode stage. If you do it right, you can get a nice sounding preamp, with about 20 to 26dB of gain.
If you must have less gain, consider a grounded-grid stage (a 'long tailed pair' using only one of its outputs). That would require twice the number of tubes (2, instead of 1), and more parts, etc. You could get the gain down to about 10 to 12dB, without using negative feedback. Or you could use negative feedback, but that makes things more complex...
--
Do any of your power amps have balanced inputs? If so, why not make SY's ImPasse preamp? That's a complete design...
--
Here's a classic article on load lines:
Norman Crowhurst "Making Use of Load Lines"
Also, this looks helpful:
Vacuum Tube Theory
Part 4 covers load lines.
Also, if you use a cathode follower, it almost begs the question why you want this at all. If you make it right, it will be very transparent. A cathode follower should sound like 'nothing.' The only reason to make a cathode follower is if you need to make an impedance matching stage (buffer) using a tube. If you make a cathode follower wrong, to get that 'tube sound,' you'll just get a pretty bad sounding stage.
If you want to experience that 'tube sound,' you'll probably want to build some kind of common cathode stage. If you do it right, you can get a nice sounding preamp, with about 20 to 26dB of gain.
If you must have less gain, consider a grounded-grid stage (a 'long tailed pair' using only one of its outputs). That would require twice the number of tubes (2, instead of 1), and more parts, etc. You could get the gain down to about 10 to 12dB, without using negative feedback. Or you could use negative feedback, but that makes things more complex...
--
Do any of your power amps have balanced inputs? If so, why not make SY's ImPasse preamp? That's a complete design...
--
Last edited:
ok, i'm sold, i will give it a try
.i might be able to get my hands on a couple of used 6sn7. dont have any of these other tubes.
would like to start with a tube rectifier with a choke.
i have the 3rd ed. i am currently reading it, but done only 10%. did the neets manual first. have basic electronics knowledge.
thanx, will read these.
will try the common cathode. still would like it to be reasonably neutral, bit of warmth is not an issue, but not excessively sugar coated.
no, dont have balanced inputs.
Read in MJ about load lines, also those links to articles I posted. Use those to pick an operating point for your 6SN7 that is as linear as possible. Then you can listen to it and see what you think. You will read in MJ that the 6SN7 is one of the lowest-distortion tubes made.
The only problem may be that 6SN7 output impedance will be high, maybe 7k ohms or so. I don't think it will work well into a solid state amplifier. SS amps often have input impedance down around 10k to 50k. Rarely as high as 100k. With a common cathode 6SN7 as line amp, you will need the power amp to have input impedance of at least 30k, at the absolute minimum. At least 70k would be much better.
Tube amps typically have input impedance from low of about 50k to high of maybe 500k.
--
If you run the pre into a SS amplifier most of the times the gain isn´t needed.
So think twice if it´s a buffer or a gain stage you need.
The 6sn7 is a good choice and will work as both buffer/amplifier in the same box so you can experiment without having to hacksaw it apart.
Leave some space free on the topplate and you can experiment with output transformer later.
So think twice if it´s a buffer or a gain stage you need.
The 6sn7 is a good choice and will work as both buffer/amplifier in the same box so you can experiment without having to hacksaw it apart.
Leave some space free on the topplate and you can experiment with output transformer later.
You want low gain, so a mu (amplification factor) of 20 or lower is good.
You want a low output impedance, so a plate resistance of less than 3000 ohms is good. Lower is better, if you want to use the preamp with solid state amps.
You want low input ("Miller") capacitance, so it doesn't roll off the highs, especially if it will be used with a volume control potentiometer in its grid circuit (which is the norm).
You want it to be a linear tube, so 6BX7 or 6BL7 are probably not the best choices.
- Look at the mu of the tube as the plate current and plate voltage are varied. Does the mu remain constant? Or does it change quite a bit?
- Look at the characteristic curves (plate curves). The curves should be pretty evenly spaced, not all bunched up at the right side, nor should they get far apart along the left side of the graph.
You want the two triode sections in the twin-triode to be well-matched in mu, plate resistance and transconductance.
That's all I can think of at the moment.
--
If you can find some 6K6GT tubes, you could triode wire them and use them as common cathode amplifiers. That's sort of like a smaller version of 6V6GT. Should make a fine line stage, with not too much gain. I was once given a bag of them (back around '92), but eventually gave them away. Little did I know that they would turn out to be useful...
here's a preliminary design with the 6sn7. I will be using a sylvania 6sn7w tube as of now. I tried some tubes apart from the ones mentioned above, only ecc82 looked good but the distortion was very high at some 4.1%, while 6sn7 gave me some 0.58%.
Please let me know what you think.
Some calculations -
Common Cathode Preamp With 6SN7
Point A on 6sn7gt sheet, Loadline - 300V, 20mA, -4V
quiscent voltage – 167, current – 9 mA
Rp – 300/20ma = 15k ohms
Rp power = I^2R = 9 * 9 * 15000 = 1.2 watts
My sources give a 2vrms signal – 2.8Vp, lets say 3Vp or 6Vp-p
signal swing -1V to -7V, from point B to C
-1v = 11.6mA, 125V
-7V = 6.1mA, 210V
210-125=85Vp-p output signal
210-167=43
167-125=42
H2% = 100*(43-42)/(2*(43+42) = 0.58%
A = Vout/Vin = 85/6 = 14.16, 20*log(14.16) = 23db
Cathode resistor = -4V/9ma = 444 ohms approx 500 ohms – use 0.5w resistor
Actual bias point = 500*9 = -4.5 volts
gm at the operating point of -4v, 300v, 20ma
Vg from -1 to -7, change is 2.4 to 16.4mA (point E and F)
gm=(16.4-2.4)/6 = 2.33
from -1 to -7v, change is 231 to 108 (point H and J)
mu = (231-108)/6 = 20.5
for ra (use point G and D)
185v, 11.6ma and 145V and 6.1ma
ra=(185-145)/(11.6-6.1)=7.27K
calculated
ra=mu/gm = 20.5/2.33=8.798k
Why such a large difference in the two ra values?
A=mu*Rp/(Rp+ra) = (20.5*15000)/(15000+8798) = 12.92
Ck=1/(2*pi*Rk*f) = 1/(2*pi*500*20)=16 mfd, 25V or more
Zout with cap bypass
Zout = Rp || Ra = (15000*8798)/(15000+8798) = 5545
Grid leak resistor – 1M ohms – use 0.5w resistor
Co = 0.47 mfd, 400v
Rl = 1M, 0.25w
pot 100k
Please let me know what you think.
An externally hosted image should be here but it was not working when we last tested it.
Some calculations -
Common Cathode Preamp With 6SN7
Point A on 6sn7gt sheet, Loadline - 300V, 20mA, -4V
quiscent voltage – 167, current – 9 mA
Rp – 300/20ma = 15k ohms
Rp power = I^2R = 9 * 9 * 15000 = 1.2 watts
My sources give a 2vrms signal – 2.8Vp, lets say 3Vp or 6Vp-p
signal swing -1V to -7V, from point B to C
-1v = 11.6mA, 125V
-7V = 6.1mA, 210V
210-125=85Vp-p output signal
210-167=43
167-125=42
H2% = 100*(43-42)/(2*(43+42) = 0.58%
A = Vout/Vin = 85/6 = 14.16, 20*log(14.16) = 23db
Cathode resistor = -4V/9ma = 444 ohms approx 500 ohms – use 0.5w resistor
Actual bias point = 500*9 = -4.5 volts
gm at the operating point of -4v, 300v, 20ma
Vg from -1 to -7, change is 2.4 to 16.4mA (point E and F)
gm=(16.4-2.4)/6 = 2.33
from -1 to -7v, change is 231 to 108 (point H and J)
mu = (231-108)/6 = 20.5
for ra (use point G and D)
185v, 11.6ma and 145V and 6.1ma
ra=(185-145)/(11.6-6.1)=7.27K
calculated
ra=mu/gm = 20.5/2.33=8.798k
Why such a large difference in the two ra values?
A=mu*Rp/(Rp+ra) = (20.5*15000)/(15000+8798) = 12.92
Ck=1/(2*pi*Rk*f) = 1/(2*pi*500*20)=16 mfd, 25V or more
Zout with cap bypass
Zout = Rp || Ra = (15000*8798)/(15000+8798) = 5545
Grid leak resistor – 1M ohms – use 0.5w resistor
Co = 0.47 mfd, 400v
Rl = 1M, 0.25w
pot 100k
Woops!
Ck is in AC parallel with both the Rk and ra. So in this calculation R should be the value of these 2 impedances in parallel.
thanx. Just tried the larger eqn (the big sqrt one). gave me slightly larger cap at 17.44mfd. I would probably put in a slightly larger value anyway.
is the loadline good? I could squeeze out another couple of mA, though that would push it towards the max power dissipation line.
Last edited:
Just a general question --
I've read in many places that for greatest linearity, a higher value plate load resistance should be used (the larger the value, the better). So the ratio between Rload (Rl) and tube's internal plate resistance (rp) should be as large as possible. 4:1 minimum is usually recommended.
This proposed design uses a Rl/rp ratio of only about 2:1, with a B+ of 300V. This yields the most power out. Yet the predicted 2HD is 0.58%.
Would it make an order of magnitude difference to increase the B+ to 450V and use a larger value of Rl, maybe 33k? (That would be approx. 4x the 6SN7's rp.)
--
I've read in many places that for greatest linearity, a higher value plate load resistance should be used (the larger the value, the better). So the ratio between Rload (Rl) and tube's internal plate resistance (rp) should be as large as possible. 4:1 minimum is usually recommended.
This proposed design uses a Rl/rp ratio of only about 2:1, with a B+ of 300V. This yields the most power out. Yet the predicted 2HD is 0.58%.
Would it make an order of magnitude difference to increase the B+ to 450V and use a larger value of Rl, maybe 33k? (That would be approx. 4x the 6SN7's rp.)
--
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.