μPA... a super simple, all-tube, good sounding, very Stax-ious and very evil headphone amp for high-sensitive IEM's.
Apr 30, 2011 at 8:48 PM Thread Starter Post #1 of 22

AudioCats

Headphoneus Supremus
Joined
Feb 25, 2007
Posts
3,726
Likes
262
Location
CO, USA
 
 
 
-- 5/13/2011: revised the schematic (added transformer wire colors) and IEM sensitivity requirement (117dB/mW, impedance preferably >18 ohm).
 
** list of suitable IEM's for this amp:
 
*1964-D (22 ohm, 120db/mW)
*1964-Q (46 ohm, 118db/mW)

 
*Earsonic SM2 (16 ohm, 119db/mW, the impedance is a little low but should still be ok)
*Earsonic SM3 (34 ohm, 122db/mW)

 
*JH5 (21 ohm, 119db/mW)
*JH13 (28 ohm, 119db/mW)
*JH16 (28 ohm, 118db.mW) 

 
*Ortofon, E-Q5 (40 ohm, 117db/mW)
*Ortofon E-Q7 (31 ohm, 118db/mW)


*Shure SE535 (36 ohm, 119db/mW) 
 
*UE 11 pro (18 ohm, 119db/mW)
*UE 7 pro (18 ohm, 124db/mW)
*UE 5 pro (21 ohm, 119db/mw)
*UE triple-fi 10 (32 ohm, 117db/mW)

 
*Weston AC2 (27 ohm, 119db/mW)
*Westone ES2 (27 ohm, 119db/mW)
*Westone ES3 (25 ohm, 124db/mW)
*Westone ES3x (56 ohm, 124db/mW)
*Westone ES5 (20 ohm, 120db/mW)
*Westone UM2 (27 ohm, 119db/mW)
*Westone UM3x (56 ohm, 124db/mW)
*Westone W1 (27 ohm, 122db/mW)
*Westone W2 (33 ohm, 117db/mW)
*Westone W4 (31 ohm, 117db/mW)

 
 
 
Objectives of the design:
-- a good sounding yet super simple amp that works well with (but only with) high sensitivity IEM's such as UM3X, SM3. The phones sensitivity needs to be >=117dB/mW.
 
-- also works as an active LOD cap-box/pre-amp for an iMod/DiyMod. The circuit must have input high impedance so low value DC blocking input caps can be used, making it actually affordable to use super exotic boutique caps. (instead of the more common 2.2uf~4.7uf value in an iMod dock, this only need 0.22~0.47uf, so even the V-caps are within our reach).
 
-- Low voltage, safe for casual DIY and casual desktop use. Build cost to be within a couple hundred $$, a few hundred $ on the maaaaaaaxxxxxxx-ed version.  
 
-- Most importantly, the design shall utilize the mystical Stax transformers from a Stax SRD7 transformer box to (hopefully) gain that "Staxious" sound. So if this design become popular, large # of SRD7 transformer boxes on the used market will be hunted down and gutted to build these little evil creatures. Hopefully this will trigger the global extinction of SRD7, how about that! Wahahahahahahahahahaha........ Ok, I might need some more pratice on my evil laugh.
 
 
 
 
 
The amp circuit is simple, actually, super simple. A traditional circuit, nothing new.  
 

 
*6GM8 as the (only) tube, fix biased to about 0.8V (by a pair of power shockey diodes in series). idle current at about 1.5mA, the Stax transformers are p/p type but signal swing and dc current is so low they can handle that with no problem.
*
 
 
power supply:
-- filament supply uses a 7806 reg and FM caps;
-- B+ uses a ~5mA CCS to load a string of LED's forming a 20V shunt regulator; reservoir tank only needs to be 20uf or so.
 
picture showing the filament and CCS module ( I somehow decided to use three jFED in paralle, instead of just one, they should work the same).

 
Apr 30, 2011 at 9:27 PM Post #3 of 22
Build phase.... I use what I already have, which is a pretty nice looking wood poker box.
 
 

 
 
 
 
Finally....what a mess, don't know how I managed to use so many wires in such a simple amp. (the main circuitry is the same but I improvised the power supply section quite a bit becase.... I only have AC wallwarts in suitable voltages, had to add rectificatin stage). Notice the black-with-white-spot wires that are every where, those are ground wires, all connects into the same spot in the red heatshrink.
 

 
 
Due to space constrains, 0.1uf FT2 caps are used as C*, the input stages cut-off frequency is therefore at 16Hz, not as low as I want it to be. I guess I can live with that, though. I'd rather have small teflon than large something else.
 
May 1, 2011 at 9:27 AM Post #4 of 22
Look nice!!  I wish I was crafty as you.  I have no tools, maybe a drill, and thats it.
redface.gif

 
May 1, 2011 at 11:32 AM Post #5 of 22
Nice. How does it sound? Any idea what the actual gain might be?

Very clever use of the case. Is there a fair amount of heat build up? How does it dissipate the heat from the regulation?
 
May 1, 2011 at 4:31 PM Post #6 of 22
the gain is.... about 0.5x
ph34r.gif
 (the tube amplifies about 12x then the stax transformer give a 1:25 step down). Expected power output is about 1mW, truly a micro-power-amp.
 
dead silent back ground, no hiss, no hum (of course, all power supply cap values are pretty much way over-kill. In theory the B+ only need 50uF, and in this build there is a 2200uf Nichicon before the CCS, then another 15uf polyprop after the CCS).
 
Now the heat.... Originally I planned using a 9~12V to feed the 7806, but that was intended for a relativly large metal case, heat won't be a problem there. For this wood box I am using 8V to feed the 7806 so it is not doing that much voltage dropping. Now why wouldn't I use a 6V regulated wallwart (which I actually have) and feed it directly to the filament? I want the final regulator to be in the case, so if the wrong power is plugged into the filament power jack there is still the 7806 in the case to drop it to 6v, preventing the tube from being fried. 6GM8's are $30 a piece, 7806 is $1, I'd rather sacrifice the 7806.
 
I don't think I should comment too much about the sound, other than it is very good. This design really was intended for phones with >120dB/mW sensitivity, such as the SM3 that I have coming. I only have ER4P right now, the ER4P's powe need is 20x of a SM3, so it is not the right phone for this amp. Anyway, the sound with ER4P: good instrument seperation, good air, wide head stage, very good bass detail and the clarity is about the best I have heard from a iMod/diyMod (though I have not heard that many iMod combo's).  I can      da
 
May 2, 2011 at 10:39 AM Post #7 of 22
Interesting.
 
Is there a reason you wanted to use the stax output transformers other than "staxious" sound?
There is a whole world of output transformers out there..
 
 
 
May 2, 2011 at 6:02 PM Post #8 of 22
the SRD7 transformers are readily available (to me), I knew how they sound (excellent), and, to use hammonds or Lundahl you are looking at about twice the price.
 
Plus, for such a truly micro power/signal application, the larger hammonds or lundahl are not necessarily better.
 
component costs:
-- SRD7: about $80, (any condition, as long as the transformers are not physically damaged).
-- pot: Alps blue-velvet is about $20; (TKD about $80).
-- input caps: about $40 for a pair of RTX 0.22uf, (or $80 for a pair of PCU 0.22uf/200V, or $120 for a pair of Relcap TFT 0.22uf/100V).
-- 6GM8: $30
-- 7806, shockly diodes, LED's, resistors, filament caps, tube socket: $20
-- B+ power caps: about $10 using Muse for pre-CCS and polyprop for post-CCS. 
 
Then add the DC wallwarts, price ranges from $0 (say, you already have them?) or $25 (buy on-line). Metal enclosure will probably be $15.
 
Due to the simplicity, the difference made by components in the signal path is easily audiable. Change the pot from Alps to Bourns (both are conductive plastic), the sound definitly changes; so is changing input caps, (of course); The only thing that doesn't seem to make that much a difference is the tube, I have tried both "A-frame disc getter" type and "square frame ring getter" type, the sound is very similar.
 
May 4, 2011 at 1:43 AM Post #9 of 22
judging from the performence with the ER4P, I think the sensitivity requirement can be adjusted from >120dB/mW down to >115dB/mW.
 
no obvious sonic problem pushing the ER4P, other than the highs are pretty rolled off (probably more to do with the ER4), and bass quantity is not enough to my liking (again, more to dow ith ER4).
 
 
 
 
May 13, 2011 at 8:21 PM Post #10 of 22
 
 
SQ with EarSonic SM3 (this is the original version, not the new V2):
-- Clarity is great, much more transparent than the SM3 driven by a modded Sony D-11 PCDP (vit-Q in signal path, OS-CON as output). Sharper, lively, good details.    
-- sound stage is quite wide, about 1" extended from each ear. (in comparasion, the SM3/D11 combo produces cheekbone-to-cheekbone width).
-- somehow the voice is a little nasal, but from both amp/source, so I will say it is the SM3.
-- Very good bass, both detail and quantity. The SM3/uPA combo definitly can Rock (unlike with the ER4P.....).
-- Adaquate available volume, though I do have to crank it to 75% volume for live Rock recordings. Since the SM3's sensitivity is at 122db, I'd adjust the guestimated phone sensitivity requirement (for this amp to drive well) back up a little, to maybe 117dB/mW.
 
May 15, 2011 at 2:02 AM Post #12 of 22
Hey can you give me break down on how you tuned it specifically for a sensitivity value?  I like tube sound with IEMs, but with my UM3X my TU882 hisses even at low output impedance setting.  
 
May 15, 2011 at 10:55 AM Post #14 of 22
Have you done any distortion measurement comparison to headphone out?
 
Quote:
The signal at the iMod/Diymod  LOD is very clean., and this amp adds no hiss/hum, the end result is a dead silent background for the IEM's.



 
 
May 15, 2011 at 1:52 PM Post #15 of 22
do not have a distortion analyzer, I can't provide you a % number. plus, even if #'s are available, this is a non-feedback circuit, while the headphone-out amp uses negative feedbacks, how do you compare the #'s? A fixed load is used when running the analyzer, that is not the same as an active load like a real pair of phones. ( If distortion % can tell that much a story, then there is abosolutly no need for people to roll op-amps, pretty much all good op-amps "measured" <0.001% THD, they should sound exactly the same to our ears, no?)
 
 
 

Users who are viewing this thread

Back
Top