Hengki HeadAmp 1 (4/4/2008):

This is me bragging, blabbing, and remeniscing while providing information for people 
who want to build a similar headphone amplifier.... (mostly the former)

One day while I was looking at the top of my audio rack there it was a now 13 year old 
Audio Alchemy DDE A/D converter.  I haven't used it much since I haven't listened to 
a CD directly for a few years, mostly due to laziness since it's much easier to click 
a mouse of an MP3 or FLAC file.  That same week I was listening to my Grado SR325i 
headphone and thought hmm... it would be really nice if I can listen through the 
Macbook Pro's digital output using that ADC, but I would need amplification.  
Ghetto rigging it, I used an old pc speaker with headphone jack.  After a week, I 
thought, damn, this is hella ghetto listening to a Grado connected to this pc speaker 
like this.

So I searched to buy a headphone amplifier.  Years ago, I wanted to get a Musical 
Fidelity XCan for my Grado SR80.  I was at an audiophile convention and saw them at the 
company's booth selling it for $249.  Tempted, but I was broke and couldn't buy one.  
Now that same amplifier is $499 - what a rip off - oh, they changed the knob, whoopee.  
Other headphone amplifiers are around the same price.  And old ones like Creek is now 
$200 more than it was before.  I guess high end headphones are more popular now.

I've wanted to build my own high end headphone amplifier for years, but have been too 
lazy.  So now I have some motivation (from my thriftiness?).

Options:

A) I really want tube amplification.  At least a hybrid with tube. I always have 
dreams of owning a tube pre amp and amplifier.  While working at an audio store, 
during a slow day I connected $30k's worth of Audio Research tube pre amp and amplifiers 
(pair for bi amp) to a pair of Martin Logan Quests (now you know that it was YEARS ago), 
using $2k's worth of Audioquest cables, and I almost cried listening to it.   I've never 
owned tube equipment, off and on have been studying (not much) tube electronics theories, 
but never could afford one or built one.  Sounds like Class D amplifiers is promising to 
have close to tube like harmonics sound quality.  My prediction is 5 years from now, will 
be the day tube becomes obsolete due to Class D amplifiers to have the harmonics of tubes 
and the balls of discrete amplification.  And hopefully I'll have a couple of G's to buy 
a pair (always bi amp for me).  But I digress.

B) OpAmp based.  Easier to tweak.  Parts are abundant (except for some OpAmp which are 
always out of stock because everyone and their moms are building headphone amplifiers 
it seems like).  But based on experience on fixing and tweaking high end amplifiers, it 
would sound "cold".  bleh.

C) Discrete (transistor) based.  Not easy to tweak (desoldering them would be a pain in 
the ass).  And transistors now are like tubes in the 80's - they are getting more scarce 
to find.

D) Hybrid discrete and OpAmp based.  Still easy to tweak.  It would have closer to the 
sound of discrete amplification since it is usually on the second stage of amplification 
while the OpAmp circuitry is on the first stage.  (But still not tube, oh well).

I chose option D) and found a few circuit design.  The PPA design looked to be promising 
and there are lots of other people that have listened to them so it's not something too 
obscure.  Judging from the circuitry, it looks like an OpAmp based first stage with 
discrete based buffer (no amplificatino or gain) as the second stage.  Floating ground
and discrete ground channel are very nice icing on the cake.  And found tangentsoft.net
that sells PCB for it - circuit to bias to Class A, and board space for 9 capacitors for
DC regulation sealed the deal for me.  Look for PPA v2 amplifier.

After studying the circuit, I was off to carefully choose the components for it - this 
took 95% of my time.  I love the internet.  I ordered everything online (DigiKey, Newark, 
and Mouser) and the shipments arrived in 3 days.  I remember when DigiKey and Mouser were 
only catalog based, internet was not around for the public yet, and it took a week or two 
for the order to arrive and everything was done on paper and over the phone(there's 
another one of my "I walked in the snow barefoot" speech).  The OpAmps I chose were AD843 
and AD845 (OPA627's are out of stock everywhere, urgh), though the AD843's came dead on 
arrival and still awaiting replacements.  And, I messed up in getting the resistor size, 
more like mislead.  1/8 watt resistors that I ordered were actually 1/4 watt, and I was 
too lazy to search and returned them, so they're slightly larger in size so you'll see 
them mounted on typical 45% angles. oh well, they're still 1% film resistors.

I also built a regulated power supply with a toroidal transformer and choke circuitry for 
the amplifier, PCB also from tangentsoft.net.


These are the PCB's and all the components I chose and ordered.
All components and PCBs


It's been a while since I've ordered from digikey and mouser.  I had a box ready with 
separators to organize the components before building as I usually do, but I didn't 
need it because wow! they came in nice individual little anti static bags each with 
labels.  nice!!  I ordered online, from my laptop, in bed in my pj's, and now this!  
Have I told you that I love the internet?  I used silver solders all around, and the 
cables from the innards of an Audioquest Midnight speaker cables for connections to 
all the jacks.

The PCB for the power supply as I assembled them.  This took me about 2 hours.
Power Supply PCB Power Supply PCB
Power Supply PCB Power Supply PCB
Power Supply PCB


The PCB for the amplifier as I assembled them.  This took me about 6 hours.
Amplifier PCB Amplifier PCB
Amplifier PCB Amplifier PCB
Amplifier PCB Amplifier PCB



After checking AC and DC output voltages from the power supply.  I deemed the power supply 
to be healthy.  And adjusted its output to 20VDC.

Still frankensteined, I connected the power supply to power up the amplifier.  At first I 
was confused as to why the power supply voltage dropped to 1.9v now that the amplifier is 
its load??  I thought I reversed the polarity of some capacitors that is causing load but 
usually that would cause the voltage to drop close to 0v, and I would see some of it 
leaking (sometimes blow up like a small firecracker).  After a few minutes, I removed the 
OpAmps and was getting the right voltages to them (20v or 10v between it and the virtual 
ground).  I replaced them with the AD845's and the voltage remained the same, as it 
should, and the power supply voltage was at 20vdc, as it should.  I connected a $2 
headphone an old cd player to it, and I'm getting nice sound!  I forgot to strap the 
potentiometer to ground as tangentsoft.net hinted.  After that, I was not getting anymore 
hum.  Still frankensteined I started the break in process (and applied for an RMA to 
newark.com to get replacements of those darn AD843's that are DOA).  I connected my Grado 
headphones to it and my ADC to my Macbook Pro.   Aaaaahhhh..... so nice.  I couldn't wait 
until it is completely broken in.  I gave it another 5 days of 24 hours break in period.

Some nice Cardas phono jacks are coming in the mail.  I still don't know how to make the 
square opening for the IEC power jack to the power supply.  I hate metal work.  
But here it is.
Hengki HeadAmp


Macbook Pro digital out --> SPDIF optical cable --> 
  Audio Alchemy Digital Decoding Engine ADC --> Audioquest Ruby interconnects -->  
    Hengki HeadAmp aka PPA v2 --> Grado SR325i  = aaaaahhh....
aahhh...


My next headphone amplifier to build... a dicrete amplifier (if I can find all the 
transistors for it).  stay tuned.


Update (4/13/2008):  

The Cardas CTFA RCA jacks arrived and I finished off the case work for the amplifier including
installing the power jack.  I've omitted the power switch because the amplifier (like all
high end amplifiers) should be on 24/7.  The power supply case work is still not finished.  I
still don't know how to cut that rectangular opening for the IEC jack.

After about 5 days of burn in time with the AD845 OpAmp, the sound was even better as expected.
As usual with other audio components, the top end has rolled off and is no longer as harsh, and
the midrange is even sweeter.  The bottom end is deep and balsie.  The imaging is acurate and
the soundstage is decently wide.  Overall the sound reproduction is accurate, though I wouldn't
call it neutral (which is not always a good thing).  I really like the sound which is pretty
sweet sounding (i.e. not cold, not "sweeeet dude").   If the soundstage is wider while still
retaining the accurate imaging (or even more accurate) then this would be a super awesome
sounding amplifier.  I still have to change the gain since it's too much for the low
impedance Grado's (too much background hiss and my amplifier would not "go to 11" :) ).  Right
now, R3 is 1k and R4 is 10k which is a gain of 11.  I think I'll probably change R4 to 3.9k
which will be a gain of 4.9.

Some technical details:  With the AD845 OpAmps, I've settled with setting the power supply to
output at 26.00V.  I've biased the OpAmps at 1ma (math is V=IR or R10=1k 1V across) and biased
the discrete buffer stage to 20ma (R24 or R34=2.2 44mV across)


Update (4/17/2008):  

Great customer service by newark.com for replacing the dead AD843 OpAmps.  They immediately
shipped a new set no questions asked and I didn't even have to return the dead parts.  I
replaced all three OpAmps with the AD843 but got some nasty distortion (I used my $2
headphone to test of course).  And it seems like it would latch into a distorted state when
the audio output is high at a certain level - this sounds like an OpAmp oscillation problem
where the OpAmp would amplify the signal wildly (much like resonance in mechanics).  I
thought they needed more supply voltage, so I reconfigured my power supply to output at a
higher voltage, but at 30 to 35V, the output buffer transistors were burning up and the same
distortion results.  On a hunch, I replaced only the Ground channel OpAmp with the AD845.
This seemed to control the oscillation problem, but distortion is still slightly there.  So
I reconfigured my power supply again but to output a lower voltage.  At 14.00V, the
amplifier seems to be stable.  Output buffer transistors are not burning up, no more
distortion nor oscillation.  But the amplifier clips easier now, but still has to be
extremely loud volume level before clipping occurs.   

Initial sound impression is that the AD843's are not as sweet sounding, but do have a
slightly wider soundstage while the imaging accuracy remains the same.  I will perform burn in
for a few days and I hope the midrange sweetens and the top end tapers off (i.e. sweeter
sound).  I'm still not happy with the clipping at higher volume, but I guess that's better
for my ears since it forces me not to listen at a very very loud level :)

I've heard alot of good things about the OPA627 OpAmps but they are extremely hard to find.
The OPA637 is sort of the equivalent but people have reported that they have a slightly
different sound.  I'm planning to order some in the near future.

Update (4/23/2008):

Those darn AD843 OpAmps are not stable in the PPA v2 :(   With all three chips in L/R/Ground
channels, the amplifier oscillated like mad, no matter the power supply voltage.  I then
tried replacing the Ground channel OpAmp with AD845, and it stabilized it somewhat.  I
increased the power supply voltage to 26V and it was somewhat stabilized, though I can hear
very slight distortion on some music passages especially low and loud sustaining tones.  And
the noise level when there is no signal is increased dramatically as compared to the AD845.

I did a break in period of 5 days anyway, and from what I hear, the AD843's have wider
soundstage still from the AD845's and spatial rendition is slighly better than the AD845's.
The midrange sweetened just a tad, but not much.  Still not as lush or sweet sounding as the
AD845 so it was a bit more fatiguing to listen too but not too harsh (though the fatigue
could be caused by the slight distortion especially those that are not too audible or
apparent).  But the trade of the wider soundstage was nice!  

I got tired of the very slight distortion and the noise level after a while.  The noise level 
is also present with the AD845, though not as prominent as with the AD843.  So I changed R4 
from 10k to 3.32k which decreases the gain from 11 to about 4.3.  The amplifier is even more
unstable.  Clipping and oscillating wildly.  I really got tired of the AD843's and have
replaced back the AD845's.  But for experimentation, now I have AD845's on L/R channels and
AD843 for the Ground channel.  I can't hear any audible differences with the AD843 on the
Ground channel, but left it in there anyway.  I miss the wide soundstage of the AD845's
though :(  

On a side note, I can't find OPA627's anywhere :(  And only found OPA637 but only in surface
mount packaging (and I don't have the board adapters to convert to 8pin DIP pins).  I'll
keep searching.

Update (6/8/2020):

It has been years but I finally have the time to do more tweaks. I finally got the OPA627. 
I tried them on the left, right and ground channels. They sounded harsh suggesting that they 
were distorting. The STEPS power supply was configured to output 26.5 volts. I changed R4 in 
the STEPS power supply from 1.5k to 2.2k ohm (1/2 watt resistors) to get 34.9 volts output. 
The distortion went away but using the Grado SR325, they sound very bright, but soundstage 
was wider and retained the focus and imaging as the AD845. I replaced the ground channel with 
the AD845. The soundstage narrowed but not as much as AD845's in all channels and brightness 
reduced. It was a good compromise, but I could only listen to a couple of hours before my 
ears were in hurting because of the excessive brightness. I put back the AD845's but retained 
the 34.9 volts supply. I think the OPA627 would sound great with Sennheiser headphones since 
they are not as bright as Grado headphones.



copyright materials ©hengki.com