[Leny]

Hello dBs,

Yup, you have confused yourself a tad there. It will be either CCS on top of a amplifier tube, or mu-follower whereby the upper tube (or FET) approximates a CCS. Never a CCS on top of a mu-follower for the reason that you noted.

If you choose the CCS on top of an amplifier tube then the PSRR is very high. The current from the CCS is divided between load and tube, hence the current through the tube is not constant. The output impedance is dictated by Rp (the internal anode resistance of the tube).

If you choose mu-follower then the PSRR is less high, and the performance of the PSU is more important. The mu-follower diverts PSU current to the load while maintaining almost constant current through the lower tube. The output impedance is dictated by the follower section, which if done right is going to be significantly smaller than the internal anode resistance of the amplifying tube.

Both CCS loaded amp tube and mu-follower are fine for parallel-feed configuration. Surely the output cap value is determined by the inductance of the output transformer primary?

Cheers for now,
L.

 

[dBs]

I have been looking hard for a couple of days for what I could expect to be the average input voltage for my amp from your average DAC. I would like to think that something like this would be relatively standard, but more than likely the answer is "it depends". I have found no specifications from DAC models (Audio GD, Zero, etc.) on what their output voltages are. It's a little tough planning your input stage without knowing what input grid voltage you're going to be running.

At this point I might have to start over on the output stage as I suspect that my selected grid voltage there is too high for any DAC (if I made it only one stage), or too low for any tube input/gain stage. Basically, I have painted myself into a corner unless the output voltages of your average DAC are low enough.

 

[jcx]

desktop digital source consumer audio devices usually put out 2 Vrms full scale

older "standards" include 0 dBv and 0 dBu; 1 and ~.7 Vrms

 

[dBs]

Quote:

Originally Posted by jcx /img/forum/go_quote.gif desktop digital source consumer audio devices usually put out 2 Vrms full scale older "standards" include 0 dBv and 0 dBu; 1 and ~.7 Vrms

Ahh, thank you. I've gone crazy looking for that information. As I suspected, it is a bit too low for my grid to go one stage. I will have to do some redesign. No biggie

 

[nikongod]

What output voltage swing are you looking for?

 

[Leny]

...and you only need to set your input grid voltage to just above your expected max input peak. So if you use 2V rms as a working assumption for input then peak will be about 2.8V, so a bias point of 3V should be fine.

Obviously you can raise that if you want to get into a particular part of the anode characteristic curves for whatever reason, ... or if you happen to choose a tube that turns out to draw significant grid current when approaching 0V!

(3V bias will actually be -3V, but I am sure you know that already).

 

[regal]

Quote:

Originally Posted by Leny /img/forum/go_quote.gif ...and you only need to set your input grid voltage to just above your expected max input peak. So if you use 2V rms as a working assumption for input then peak will be about 2.8V, so a bias point of 3V should be fine. Obviously you can raise that if you want to get into a particular part of the anode characteristic curves for whatever reason, ... or if you happen to choose a tube that turns out to draw significant grid current when approaching 0V! (3V bias will actually be -3V, but I am sure you know that already).

Don't forget there will be a volume pot, many good headphone amps have enough gain that a 1V bias works fine. I always target 2V bias, makes using an LED on the cathode easy which to my ears always sounds better than an electrolytic capacitor.

 

[dBs]

Wow, a lot of insight in just a few short posts. You guys are great =D Confirming the conclusions I have been slowly coming to in some cases. I don't mind [too much] banging my head against a wall if it means I'm not going to forget.

 

[dBs]

Quote:

What output voltage swing are you looking for?

 Well I was looking at using the 2C22 tube at Ip=9mA with Vg=-5V so my Vp=176V and my Vp-p would have been ~100V. This was before the knowledge that I would be receiving roughly 3V in from the DAC. If I am willing to go single stage at Vg=-3V then I get a Vp-p of roughly 130V at the same quiescent. The mu of the 2C22 is 20 so with this topology I expect to get most of that gain.
 
Just finding it difficult to find a input/gain tube that can take ~-3 in and amplify it only like 10-15 to fit the usual low-mu tubes for the driver stage. If I go the multistage route, I am tentatively looking at the 6AS7 and equivalents.

 

[nikongod]

Quote:

Well I was looking at using the 2C22 tube at Ip=9mA with Vg=-5V so my Vp=176V and my Vp-p would have been ~275 to ~75. This was before the knowledge that I would be receiving roughly 3V in from the DAC. If I am willing to go single stage at Vg=-3V then I get a Vp-p of roughly 130V at the same quiescent. The mu of the 2C22 is 20 so with this topology I expect to get most of that gain.  
Just finding it difficult to find a input/gain tube that can take ~-3 in and amplify it only like 10-15 to fit the usual low-mu tubes for the driver stage. If I go the multistage route, I am tentatively looking at the 6AS7 and equivalents.

First off this new forum software is very confusing. I would like to break this quote up into smaller sections, but I cant. This makes me sad. Updated: firefox automatic spell check does not work in this f|_|c!<1|\|g text window. ****ity **** **** ****. now the world will know how I really spell. What a bandolier of carrots we have been sold into.
 
to put this into perspective without line by lining like I would have on the old site:
You are adding another whole tube for less than 6db of gain.
 
You wanted 275 to 75V (200vp-p) but only get 130vp-p. You lost less than 6db, not even a factor of 2!
 
Get creative, and come up with another way. 1:2 input transformers may yield better results than a smatering of another tube and the rest spent on wasteful attenuation.
 
This of couse feeds back to my question of how much output voltage swing do you need? YOU are the designer, you tell us.

 

[regal]

From past experience with a source of 2V P-P,   a gain of 7 wasn't quite enough for 600 ohm phones but a gain 10 works just about right.   These were Sextetts,  I believe one of the worst case for voltage swing needs (maybe AKG340 even worse).   Hope this helps

 

[dBs]

Quote:

to put this into perspective without line by lining like I would have on the old site: You are adding another whole tube for less than 6db of gain.
 
You wanted 275 to 75V (200vp-p) but only get 130vp-p. You lost less than 6db, not even a factor of 2!
 
Get creative, and come up with another way. 1:2 input transformers may yield better results than a smatering of another tube and the rest spent on wasteful attenuation.
 
This of couse feeds back to my question of how much output voltage swing do you need? YOU are the designer, you tell us.

Hahaha, yea, I noticed the spell check right away. That will be tough. As I've said before, math is my native language, not English XD
 
Unfortunately, I don't know what kind of voltage swing to aim for. College made the main attention the gain, not the swing (unless of course you wanted the max swing). I did note that most headphones tend around the 90-105dB/V but that somehow seems wrong to me as that would mean (105x50)dB for 100V swing. That seems way too large, so I am either missing some key piece of information or I am not understanding something simple (I tend to make the simple things harder than they are).
 
I am aiming less for quantity and more for quality. Not looking to push K1000s or anything. The output transformer should help with the low impedance headphones. I think I am starting to give up on the notion of a cheaper amp, I am probably just taking the easier road in that decision, hahaha.
 
At this point I am leaning toward a single stage. From the few instances of Vp-p for speakers that I have read about (haven't found any information on what is recommended usually for headphones), 130V would be sufficient and the current source portion of the mu follower should give enough current as well.
 
...uh oh. Don't resize the reply box, BTW. I just tried that and I can't scroll to the side to see/edit what I typed on the right side.
 
I think I gave enough of an impression here though. Sorry if it seems like you guys are holding my hands or something. I can be a bit slow in the beginning, but my grasping pace will pick up, I promise, hahaha.
 
Quote:
Originally Posted by regal View Post
From past experience with a source of 2V P-P, a gain of 7 wasn't quite enough for 600 ohm phones but a gain 10 works just about right. These were Sextetts, I believe one of the worst case for voltage swing needs (maybe AKG340 even worse). Hope this helps
 
That helps a LOT, thank you. Yea, you make me think that the single stage 2C22 will be sufficient. Think I will find a higher mu tube for the top tube in the mu follower though to up that performance a little. I will be avoiding pentodes though, quality>quantity!

 

[dsavitsk]

Firefox spell check may not work, but SeaMonkey spell check works fine.
 
You can design 100 amps on paper and never get it right, or you can build one, get it mostly wrong, but get a very good idea of how this all works.  I suggest you put down the calculator and get out the soldering iron.  And, don't worry about making a project for everyone else. Make one for yourself first.
 
Most phones are spec'd at ~98dB per mW which is the key bit of information you were missing. 
 
Anyhow, a 2c22 has an rp of ~6600. You want your transformer's primary to be at least 3x that.  Really, you want it a lot more, but a transformer with a primary much higher than 15K is really pushing things. Let's say you are happy with a 20K primary.  2VRMS, or 125mW, into grados will make your ears bleed, so let's make 2x that power the target to allow for sufficient headroom.  Ignoring copper losses, that is a 70VRMS swing into the 20K primary.  mu is 20, so you will need a 3.5VRMS signal to get that.  Practically speaking, due to copper and core losses, etc, figure to add at least 20% to that.  Basically, the rp is too high, and the mu is too low to use the 2c22 (or the 6SN7, 12AU7, etc.) in this arrangement.  It will drive headphones just fine, but it won't be the ultimate tube amp you are probably looking for as there are too many compromises.  Find something with a similar mu but a lower rp (5687, 6n6p, 6H30), or a higher mu (6DJ8, D3a, 5842, 6688, C3m, etc.)
 
Edit: And, don't dismiss pentodes because the audio boards told you they were bad.  One of my all time favorite amps I have built is a simple transconductance amp using a high Gm pentode transformer coupled to a set of Grados. Works awful with other phones, but great with Grados. I think 99% of the reason many people dismiss pentodes is because they are a little harder to use.

 

[cobaltmute]

Quote:

First off this new forum software is very confusing. I would like to break this quote up into smaller sections, but I cant.

 
Sort of like this?  You can do it, but you have to goto source, which is not great unless you can read HTML.
 

This makes me sad. Updated: firefox automatic spell check does not work in this f|_|c!<1|\|g text window. ****ity **** **** ****. now the world will know how I really spell. What a bandolier of carrots we have been sold into.  

Click on source and then do your correcting and then click on source again.

 

[dBs]

 
Quote:

Firefox spell check may not work, but SeaMonkey spell check works fine.
 
You can design 100 amps on paper and never get it right, or you can build one, get it mostly wrong, but get a very good idea of how this all works.  I suggest you put down the calculator and get out the soldering iron.  And, don't worry about making a project for everyone else. Make one for yourself first.
 
Most phones are spec'd at ~98dB per mW which is the key bit of information you were missing. 
 
Anyhow, a 2c22 has an rp of ~6600. You want your transformer's primary to be at least 3x that.  Really, you want it a lot more, but a transformer with a primary much higher than 15K is really pushing things. Let's say you are happy with a 20K primary.  2VRMS, or 125mW, into grados will make your ears bleed, so let's make 2x that power the target to allow for sufficient headroom.  Ignoring copper losses, that is a 70VRMS swing into the 20K primary.  mu is 20, so you will need a 3.5VRMS signal to get that.  Practically speaking, due to copper and core losses, etc, figure to add at least 20% to that.  Basically, the rp is too high, and the mu is too low to use the 2c22 (or the 6SN7, 12AU7, etc.) in this arrangement.  It will drive headphones just fine, but it won't be the ultimate tube amp you are probably looking for as there are too many compromises.  Find something with a similar mu but a lower rp (5687, 6n6p, 6H30), or a higher mu (6DJ8, D3a, 5842, 6688, C3m, etc.)
 
Edit: And, don't dismiss pentodes because the audio boards told you they were bad.  One of my all time favorite amps I have built is a simple transconductance amp using a high Gm pentode transformer coupled to a set of Grados. Works awful with other phones, but great with Grados. I think 99% of the reason many people dismiss pentodes is because they are a little harder to use.

Unfortunately, right now all I can do is calculator math gymnastics. I don't have the money yet for parts or testing equipment (I plan to get a cheap oscilloscope and signal generator even though I know that they TECHNICALLY aren't essential pieces of equipment) yet. It wont be until mid next month that I will have the money to start going in at this. So in the mean time it's calculate away while trying to grasp better the nuance of tube amp design. At this point, my weakest areas of knowledge are at the "ends", meaning input and output considerations and calculations, and tube selection itself. I found a GREAT resource on //feed but how that interrelates with the design of the tube stage and accompanying passive elements I have seen little. When it comes to tube selection, I just don't know that many yet. There are soooo many so having an idea of what tubes offer the gain I am looking for along with the Vg I am looking for, along with the Iq I am looking for, and the grid/V swing I am looking for (if I go pentode top) is more just google searches for "low-mu medium grid" and/or luck of the draw. Not the most expedient of means to find such things but also pretty interesting in its own right.
 
When it comes to using a pentode as the top tube, there are 2 things: finding a tube that has the same Iq as the lower triode (haven't looked too far into this so this may prove to be a non issue), and my 337 already has a pentode in it so I was hoping to go all triode and see how different it would sound =D
 
I probably sound like I am making a lot of excuses, and maybe I am. Just a very large subject that I am throwing myself into headlong, as is my nature. As such there are a lot of gaps of knowledge to fill. Besides, we all know that college really doesn't "prepare you for the real world", hahaha.