Hi, I bought a bunch of cheap 6C45p on ebay and I figured with the high gm I can use them to build a parallel OTL headphone/IEM amp.
This is my 2nd amp that I designed from scratch. So still learning as I go. Here is the schematic and my derivation. Please let me know if this is going to work, or am I just out of my mind.
I have 2 target headphones that this will be used for: a LCD2cb with impedance of 60ohm and a Andromeda with 12ohm. I hooked up the LCD2cb to my current headphone amp and played some music while looking at the output. It started to get unbearably loud around 1-1.5Vrms. So I don't need a lot of voltage swing. The Andromeda is even more sensitive so anything that can drive the LCD2cb it will be more than enough.
I went backwards from the output. Each 6C45p has a gm of 45 so a single cathode follower should give output impedance of ~22ohm. If i put 5 of them in parallel, that will give 4.4ohm. Ideally the Andromeda need 2ohm but this will have to do for the first pass (I do have 20x 6C45p ).
I just arbitrarily picked 150V B+ and split the rail in half for the operating point. 75V and that give 12.5mA for each tube. 62.5mA total which means the shared cathode resistor is 1.2kohm. Is 150V too high? should I go lower? the curve on the datasheet doesn't give much detail as lower voltages and things starts to go non-linear below that.
On the grid side, I put 1k grid blocker on each 6C45p. I was also concerned about miller capacitance due to the parallel config. 1.1pF * 52 each, give 0.291uF. The datasheet doesn't say anything about grid to cathode. Can I just measure Cgk with a capacitance meter with no load on the tube?
It will be DC coupled from the driver tube. I haven't decided on the driver tube. I need something that can drive the miller capacitance, calculation based on 5Hz cutoff, anything stage with less than 120k output impedance will suffice. I don't think that's a problem. I don't need a huge amount of gain either. In fact, I think the less gain the better. my source is 2Vrms.
Any suggestions on the driver tube? I have various ones available, 6SN7, 12AT7, 6C45p, 12BH7, 6922, 5670 and others too.
The output decoupling capacitor. That is probably the biggest issue. With my calculation it shows I need 2700uF or so.
I will need a decently sized filament supply and B+.
I am most concerned with noise as the Andromeda is very sensitive. How do I estimate how noisy this will be?
Any thoughts / comments welcomed!
This is my 2nd amp that I designed from scratch. So still learning as I go. Here is the schematic and my derivation. Please let me know if this is going to work, or am I just out of my mind.
I have 2 target headphones that this will be used for: a LCD2cb with impedance of 60ohm and a Andromeda with 12ohm. I hooked up the LCD2cb to my current headphone amp and played some music while looking at the output. It started to get unbearably loud around 1-1.5Vrms. So I don't need a lot of voltage swing. The Andromeda is even more sensitive so anything that can drive the LCD2cb it will be more than enough.
I went backwards from the output. Each 6C45p has a gm of 45 so a single cathode follower should give output impedance of ~22ohm. If i put 5 of them in parallel, that will give 4.4ohm. Ideally the Andromeda need 2ohm but this will have to do for the first pass (I do have 20x 6C45p
).I just arbitrarily picked 150V B+ and split the rail in half for the operating point. 75V and that give 12.5mA for each tube. 62.5mA total which means the shared cathode resistor is 1.2kohm. Is 150V too high? should I go lower? the curve on the datasheet doesn't give much detail as lower voltages and things starts to go non-linear below that.
On the grid side, I put 1k grid blocker on each 6C45p. I was also concerned about miller capacitance due to the parallel config. 1.1pF * 52 each, give 0.291uF. The datasheet doesn't say anything about grid to cathode. Can I just measure Cgk with a capacitance meter with no load on the tube?
It will be DC coupled from the driver tube. I haven't decided on the driver tube. I need something that can drive the miller capacitance, calculation based on 5Hz cutoff, anything stage with less than 120k output impedance will suffice. I don't think that's a problem. I don't need a huge amount of gain either. In fact, I think the less gain the better. my source is 2Vrms.
Any suggestions on the driver tube? I have various ones available, 6SN7, 12AT7, 6C45p, 12BH7, 6922, 5670 and others too.
The output decoupling capacitor. That is probably the biggest issue. With my calculation it shows I need 2700uF or so.
I will need a decently sized filament supply and B+.
I am most concerned with noise as the Andromeda is very sensitive. How do I estimate how noisy this will be?
Any thoughts / comments welcomed!
Last edited:
1.1 pF × 52 × 5 = 286 pF or 0.286 nF.
Since you're already in the realm where, and I quote, “1.5 volts is painfully loud”, for both the least sensitive and obvious even more so for the more sensitive headphone, then you really don't need U1A to be doing any amplifying whatsoever. Since the input is labelled 2 VRMS or so.
The driver then becomes obvious: cathode follower for U1A, with a raised cathode operating point. 2 resistors in series, with the grid attached to the midpoint. Anode to B+. If VGK is what, 1.1 volts? or something like that nominally, then to get to about 70 volts requires ⁷⁰⁄₁.₁ = 63× the resistance for the lower (to ground) resistor. Place a nice 470 µF cap across it (suitable voltage tolerance), and you're ready to rock. Drive all the downwind grids off the kathode of the driver tube. It can be 'most anything you like capable of driving 6+ ma into ½ B+.
After that, I see that your 2,700 µF output final is a bit overkill, but without any real harm. Your poor head isn't likely to register much sound below 25 Hz or so, so setting the 3/4 V (–3 dB) point at 5 Hz is fine. You could use any value from 1,500 µF to 2,200 µF to 3,300 µF, and all be comfortably be in range. Remember: better to use a quality cap than an unusual value one. Err on side of quality.
Since the “driver tube” is in fact now a kathode follower, why not make it yet-another–6C45P? In KF mode, its quite a nice valve. At least in my experience.
I've gotta run, no telling when I might get back to this thread.
Good luck!
GoatGuy
Since you're already in the realm where, and I quote, “1.5 volts is painfully loud”, for both the least sensitive and obvious even more so for the more sensitive headphone, then you really don't need U1A to be doing any amplifying whatsoever. Since the input is labelled 2 VRMS or so.
The driver then becomes obvious: cathode follower for U1A, with a raised cathode operating point. 2 resistors in series, with the grid attached to the midpoint. Anode to B+. If VGK is what, 1.1 volts? or something like that nominally, then to get to about 70 volts requires ⁷⁰⁄₁.₁ = 63× the resistance for the lower (to ground) resistor. Place a nice 470 µF cap across it (suitable voltage tolerance), and you're ready to rock. Drive all the downwind grids off the kathode of the driver tube. It can be 'most anything you like capable of driving 6+ ma into ½ B+.
After that, I see that your 2,700 µF output final is a bit overkill, but without any real harm. Your poor head isn't likely to register much sound below 25 Hz or so, so setting the 3/4 V (–3 dB) point at 5 Hz is fine. You could use any value from 1,500 µF to 2,200 µF to 3,300 µF, and all be comfortably be in range. Remember: better to use a quality cap than an unusual value one. Err on side of quality.
Since the “driver tube” is in fact now a kathode follower, why not make it yet-another–6C45P? In KF mode, its quite a nice valve. At least in my experience.
I've gotta run, no telling when I might get back to this thread.
Good luck!
GoatGuy
using the 6C45p as driver is a great idea. the only reservation I have is if in the future I need to drive more voltage hungry headphone it won't work.
I guess I will just have to modify it then...
I am pricing out the parts. Considering lowering the B+ and the quiescent current, 62.5mA and 75V is drawing a lot of power and I only need a couple volts of swing. Let me see if I can get it down to 100V B+.
I guess I will just have to modify it then...
I am pricing out the parts. Considering lowering the B+ and the quiescent current, 62.5mA and 75V is drawing a lot of power and I only need a couple volts of swing. Let me see if I can get it down to 100V B+.
I just said elsewhere "any tool can be the right tool".
But if you are going to fret over the costs....
This is a high Mu tube (52).
POWER favors low Mu tubes. It will reduce the B+ you need.
Also: this is a 7 Watt tube and you are proposing to run it nearer 0.7 Watts.
It also has a heater as hungry as a 6V6; and you plan five of these.
This IS an exceptionally fat "high Mu" tube. You could do worse. You could do better.
THat's a good catch. Yes.... you definitely do not want to be shorting out the source.
Chris Hornbeck said:
Not a bad idea to try when setting it up. Just bypass (either) the KF or the degenerated amplification stage.
Don't forget to do the variable resistor "the right way". The center tap drives the grid of the first stage, not the source input.
GoatGuy
good catch!
yes. it is a bit of a waste/overkill. but I have 20 of these lying around.
oh. and I have to power 10 of them. 5 on each side.
priced it out. not cheap, but not bad. after reducing the B+ power and filament transformer from Hammond is <$40 each + ~$10 rectifiers and $20 of caps.
the idea is to have some fun and get my hands wet on different designs. not to be optimal in any shape or form. I am not the type who can learn and understand from pure reading. i need to read and then build it before I truly understand and appreciate something.
Last edited:
Are you talking about C2 and C3? If so, then yes, leakage would be where the 100 mv is coming from. No place else, actually. However, if you are relying on a digital voltmeter, it could well be interpreting some passed-through (inaudible) HF/RF too. Surprisingly common with breadboard layouts. No big all-encompassing metal case to shield ambient RF.
This is where our friend the oscilloscope comes in mighty handy. If you don't have one, by all means get either a very inexpensive (these days) used one off eBay, or either a new or slightly used one of the digital variety off Amazon (or eBay if you like it).
For instance: Tektronix 2225 50MHz Dual Channel Oscilloscope | eBay
Seems like a steal, at $100. If it doesn't come with probes, other sellers are flogging probe pairs for under $20, 100 MHz grade. I paid over $2,500 for my scope back in the 1970s, and IT was used. Just goes to show how egalitarian The InterWebz has become.
GoatGuy
Yes. I have have a Rigol 1054. Actually, on that side topic, I noticed the Rigol isn't very good with low voltages. even with the probe directly grounded, I get around 10mV of noise. while that's ok for most usage, 10mV will probably blow my ears when hooked up to a very sensitive pair of IEMs/earphones. And I cannot reliably judge how noise free my circuit is. I remember my old Tektronix was a bit better with noise.
Now I am planning the full build. Just ordered all the parts needed and laying out the PCB. I am hoping all 11 tubes + transformer + caps will fit into 20cm x 25cm case.
Now I am planning the full build. Just ordered all the parts needed and laying out the PCB. I am hoping all 11 tubes + transformer + caps will fit into 20cm x 25cm case.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.