12AU7/12BH7 Headphone Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been thinking about using triodes with toroidal power transformers. I also have a shipment of a few 7AU7's coming. I'd originally biased for use with 12AU7's in the output stage but figured that a little more power (hopefully 3-ish watts per channel) can't hurt. My headphones are 32 ohms, which is VERY convenient for output impedance switching (4:1 turns ratio between series/parallel). Does this thing seem feasible? I designed an overkill PSU with 4 chokes, two LC's for the B+ and another for the B++, but I'm not even sure that critical inductance will be reached.

An externally hosted image should be here but it was not working when we last tested it.


Now the question is, what will 3 watts into 32-ohm headphones sound like :D
 
Hi Soren,

I also would apply maybe 6dB of gNFB. The grid of the lower 12AU7 section gives you an ideal injection point (use 10k resistor from grid to gnd).

Much more important than that, you must give the LM334Z constant current source a negative voltage of at least, say, -12V to breathe, or you certainly will encouter massive distortion due to stalling of the CCS, from which especially the LM334Z recovers very badly (MJ explicitely mentioned that LM334Z behaviour at ETF06 when discussing a quite similar circuit shown below).

Since it will be a headphone amp, you could use DC to power the heaters, which also will conveniently give you the negative voltage for the CCS. I have done so in a similar design, although using a normal PP OPT.

Link to amp schematic.

Link to PSU schematic, showing combined DC-heaters and splitter CCS negative voltage feed.

Regards,

Tom
 
Shoog said:
Meet you at the hospital when they try to stitch your ear drums back together.

Shoog
yeah I'll watch out.

Tubes4e4 said:
Hi Soren,

I also would apply maybe 6dB of gNFB. The grid of the lower 12AU7 section gives you an ideal injection point (use 10k resistor from grid to gnd).

Much more important than that, you must give the LM334Z constant current source a negative voltage of at least, say, -12V to breathe

you could use DC to power the heaters

Regards,

Tom
I assume you mean to the bottom grid of the output 7AU7, right?

As for supplying the negative voltage, the only way I can see is to run the ~6.8V filaments through a doubler, which is bulky. I'd rather not use DC power simply because I believe in AC filaments.

My PSU at the moment is two 115:6.3VCT at 6A filament transformers, run back to back with the second "115" (closer to 125) fed into a doubler, a 157M, etc etc. for B+. The 125-ish house AC means that the 6.3V runs closer to 6.8, perfect for the 7.2V of the 7AU7.
 
Shoog said:
The toroidal option sounds very workable. I have a single ended parafeed headphone amp which uses 12V toroidals _ they certainly aren't the limiting factor.

Meet you at the hospital when they try to stitch your ear drums back together.

Shoog
Cool, I have big hopes for them. I only wish they weren't so expensive ($25 each plus shipping). Anyone know a good source of dual-115V primary, dual 6V or 7V secondary, 15-25VA toroids?
 
My PSU at the moment is two 115:6.3VCT at 6A filament transformers, run back to back with the second "115" (closer to 125) fed into a doubler, a 157M, etc etc. for B+. The 125-ish house AC means that the 6.3V runs closer to 6.8, perfect for the 7.2V of the 7AU7.
Tubes4e4 Hi Soren,

I to use back to back transformers for my HT supply on my headphone amp. A bit wasteful and a proper HT transformer would be better though. Mine are 18V toroidals, and I use one of the 18V windings for the series connected heater supplies - with a little resistance padding to get to the 12.6V AC.
I agree with you about AC heaters, but a headphone amp might be one of those rare instances where DC is appropriate. My little headphone amp is very compact and unfortunately suffers from a slight bit of hum - which I think is inductively coupled off the power transformers. If I find a nice case I will definately be looking for a complete rebuild at some point.

Shoog
 
Shoog said:

A bit wasteful and a proper HT transformer would be better though.

I agree with you about AC heaters, but a headphone amp might be one of those rare instances where DC is appropriate.

Shoog

Yeah, but I wasn't going to use the transformers elsewhere, and I like the look of the two identical transformers next to each other, symmetry is big for me.

I'll try AC first. I'll be pretty anal about the wiring, and hope for the best.
 
2 problems. I'm guessing that your headphones are Grados given the 32 Ohm impedance. You have way too much gain, the volume pot's going to be stuck at the very bottom of the range unless you listen at some insane volume. Consider changing the tubes and/or redoing the circuit to give an overall voltage gain of around 2-5, that's all you'll ever need. A gain of 2-3 is plenty given the ouputs of most CD players.

Other potential issue is the output transformer. I've had the 15W Amvecos and there was enough capacitance & resistance in the windings to severly roll off the high frequency response. The 50W ones I'm using right now work just fine and have no issues. The 25W ones which you propose to use may have issues, but I don't know for sure.
 
aerius said:
You have way too much gain, the volume pot's going to be stuck at the very bottom of the range unless you listen at some insane volume. Consider changing the tubes and/or redoing the circuit to give an overall voltage gain of around 2-5, that's all you'll ever need. A gain of 2-3 is plenty given the ouputs of most CD players.
My biggest concern. Would changing the LTP stage to a split-load work? Gain of a bit less than one. I don't know what the gain of the output stage is, though. If I look at my load line, is there an easy enough way?


aerius said:
Other potential issue is the output transformer. I've had the 15W Amvecos and there was enough capacitance & resistance in the windings to severly roll off the high frequency response. The 50W ones I'm using right now work just fine and have no issues. The 25W ones which you propose to use may have issues, but I don't know for sure.
Well at 4W of plate dissipation for two 7AU7 triodes, I think 25W is the correct ratio. All else will just be left to change, I guess
:smash:

I might also just use different tubes. I only have a limited amount of power to work with. Is there a recommended way to use the 7AU7's without OPT's to power 32-ohm headphones? Mine are Sony MDR-V6's btw.
 
Here's another crack. The constant-current White CF, according to TubeCAD, has an output impedance of under 50 ohms, which should be pretty cool for my headphones and most other amplifiers etc. I tagged the grounded-cathode stage at the beginning to get a bit of gain, you never know when you might need it. Plus, using 3 tubes total instead of four gives me more headroom for the B+ with my back-to-back filament trafos :)

An externally hosted image should be here but it was not working when we last tested it.
 
Looks like a clever design. Can't really comment on technical issues. The only slight concern I have with direct feed from a tube headphone amp is the remote possibility that your headphones could go live if your output cap fails !! Unfortunately caps are more likely to fail in a dangerous way than transformers.

Have you been over to the Headwize website, they have some very good worked up tube headphone amps.

Shoog
 
Shoog said:
Looks like a clever design. Can't really comment on technical issues. The only slight concern I have with direct feed from a tube headphone amp is the remote possibility that your headphones could go live if your output cap fails !! Unfortunately caps are more likely to fail in a dangerous way than transformers.

Have you been over to the Headwize website, they have some very good worked up tube headphone amps.

Shoog

Thanks! Is there some kind of fuse or circuit breaker that I can put to prevent a spike in current over 10mA or something? I'd like to use a really really reliable film output capacitor, I'm thinking a Solen 630V as they're reasonably priced, are available in big uF-age, and I haven't heard any disaster stories about them...

I have checked out Headwize and this is kind of inspired by the ideas I saw there ( http://www.headwize.com/projects/showfile.php?file=strain1_prj.htm ) but I wanted a much lower output impedance, <50 instead of around 500.. Plus I have boatloads of 7AU7's so I wanted to use those. I'll build it pronto and report on how it sounds! I'll try AC wiring but I'll leave room for some hefty 2200-ish uF 16V caps in case hum is intolerable.

Until next time, I'm enjoying my newly modded PSX. You'll see a bit more about that in an upcoming thread.
 
I agree with the comments on gain,and eardrum-damage.
Seriously,be careful!

I once made a 1W/chan headphone amp with some 60FX5 tubes,and it was wwaayy overkill. 3W is insane! :hot:
Go easy on the volume control!

Another small piece of advice- NEVER 'debug' a circuit using headphones! If it were to break into oscillation,the volume control was accidentally set at MAX,or some other "unseen issue" -you could be deaf in an instant! :dead:
Use a small speaker on the workbench instead.
Not to mention you could damage the headphones with that much power.
 
DigitalJunkie said:
Use a small speaker on the workbench instead.
Not to mention you could damage the headphones with that much power.

Yeah, that's the plan. I don't know what impedance the iPod headphones are but I'll start with those. Leave them on the bench, I should still hear something coming from them, can put my head a bit closer for clarity. I have a scope coming in the mail though (Tek 2213) so a nice 50-ohm load resistor or something for the later, much-less-than-1-watt design should be OK to work with. :smash:
 
Oops. I realize that I had TubeCAD simulating paralleled 12AU7 triodes. This would explain the crappy frequency response and stuff when simmed in SPICE. Fixing this problem and recalculating gives the following:

Ra: 12.4K --> 14K 1W
Ra2: 21.5K --> 37.4K 2W
Rk: 143 --> 590, bypass with 31uF 16V nominally, but probably ~470uF 16V
Rk2: 8.87K --> 23.2K

Gain remains at .94. The cathode bypass cap enables a lower output impedance of around 38 ohms, despite only using single triodes.

B+ is still at 300V but total current draw for the CF section is now 10mA. As voltage increases to potentially 350V, cathode resistance increases to bias the tube more negatively to stay at 10mA dissipation.
 
Final schematic, both channels. I borrow Rod Eliott's balance pot idea from his website ("Better Balance Control " at http://sound.westhost.com/project01.htm ).

After a crash course in RC filters (yes, I didn't actually know how they worked :bigeyes: ) I learned that the nice big 4.7uF Solen lets pretty much no bass through. Replace the bypassed 4.7uF with a 220uF 450V electrolytic with negative connected to output, bypassed by a 2.2uF 630V Solen. It could have been 400V but whatever, it's a 20 cent difference at PCX.

I simulate around 10.4Hz -3dB into a 32 ohm load with TubeCAD. With my 63-ohm Sony MDR-V6's, the -3dB point drops to 7.2Hz. Both figures are acceptable, and as you can imagine, with higher impedance headphones, the frequency response just keeps getting better. There isn't a whole lot of room but maybe for headphones of 300 ohms or better I could switch that 220uF electrolytic out for 22uF for 21.4Hz -3dB. Otherwise, it's a 2.1Hz -3dB point which is all well and good but pretty impractical.

An externally hosted image should be here but it was not working when we last tested it.
 
I didn't even know that that tube existed. Looking at a datasheet, it seems a little mismatched for my purposes, though. Amplification factor of 100 vs 20 for the 12AU7. 1.5W plate dissipation vs 2.75 for the 12AU7. Its place resistance is around 32K as opposed to 7.7K. Finally, the plate curves looks INCREDIBLY non-linear, even compared to the 12AU7 ;)

Do you mean 12BH7? I might try them later but their higher heater current directly sucks power from my B+ (I'm using two filament transformers back to back)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.