Monday 22 May 2023

Retro DIY Headphone Amp - The Audio Amateur, Gary A. Galo design, built by Mitch Barnett circa 1991.

Every once awhile, I think it's fun to try something a little different!

While the vast majority of us audiophiles buy our stuff, as per all tech hobbies at one point, this began with hobbyists in their workshops, basements, and "caves" putting together electronics parts and enthusiast kits with soldering irons making their own stuff. I remember the days in the early '70s as a child when the hi-fi system at home was put together by my dad based on designs he found in the magazines. A time before commercialization of consumer electronics was anywhere like it is today.

With miniaturization and economy of scale, these days, it's more expensive to build one's own components (both time and money) and more than likely the sound quality would not be as good. Of course, there is an element of lasting pride in building something oneself which we can never really capture when purchased with a credit card. Major respect to the DIY guys and gals out there still designing and making their own gear.

So instead of a commercially available product, let's have a look at the device shown above. It's a home-made "prototype" headphone amp lent to me by Mitch Barnett (of Accurate Sound) built on a wooden board (I guess there might be some vibration control with that ;-). He built this something like 30 years ago based on a design published in The Audio Amateur (4/91, predecessor to today's AudioXpress) by Gary A. Galo. For a bit of background on The Audio Amateur, check out this thread and post from Palustris.

Since this is a rather unique piece, let's contextualize what we're talking about. Here's a PDF scan of the relevant "Ask TAA" columns where we see the original circuits of the amplifier and power supply:

Original "Ask TAA" articles - headphone amp & power supply (1990 & 1991)

In the first part of the article, notice that as a reviewer of the Grado HP-1 headphone, Galo used this design as his reference amplifier. Back in those days, options were clearly much more limited than the plethora of high quality products we can access now!

Mitch informed me that he corresponded with the author and was given permission to modify and potentially commercialize the design working with a manufacturer; alas, it appears that other priorities took over. While the article is copyrighted, I trust publishing the relevant portions in the PDF in this post would constitute "fair use" for the purpose of this article's intent in evaluation, and testing.

Left and Right channel amplifier "breadboards". Locking 1/4" phone jack right. Input single-ended RCA left.

LT1010 with small "clip cooler" heat sink. With the board open to air, ambient temperature 21°C, after an hour it stabilizes to 46°C.

As you can see, this device is a dual-mono Class A design utilizing Linear Technology's LT1122CN8 (gain) and LT1010CT (buffer) op-amps. The author mentioned that it was based on the "typical application" design of the technical brief in that LT1010 link.

There is discussion in the article about using high tolerance resistors and capacitors, and the performance of volume control pots:

I love the smooth feel and precision of good quality Alps pots! While excellent, for the amplifier measurements in this article, I used digital volume control instead. Since I was interested mainly in the quality of the amplifier resolution and different DIY builds would likely change how volume is controlled, direct-to-amplifier from the DAC reduced noise pick-up from the extra wires and potentiometer.

The design should work well with relatively low-impedance headphones although as indicated in the article, the intention is for 40+Ω.

+/-15V, 3A each rail, power supply - plenty for headphone amp duties. Note that the schematic to this linear power supply was also published here in Stereophile in Nov 1991.

There was no discussion in the TAA article though as to how much power the headphone amp can provide nor distortion characteristics, so let's have a listen and do some measurements to see what a 30 year old DIY headphone amplifier can do!

I. Subjective

Since this amplifier had been sitting unused at Mitch's place for something like 30 years, he checked it out to make sure it worked and brought it over to Vancouver at one of our meetings for me to have a look and listen.

Taking turns with a few of my headphones - AKG Q701, Sennheiser HD800, Drop+HiFiMan HE-4XX, modded Dekoni Blue - I can vouch that this design provides plenty of power to drive these dynamic and planar-magnetic 'phones with the DIY amp fed by a SMSL DO100 DAC (dual ES9038Q2M).


For fun, I revisited the Erich Kunzel & Cincinnati Pops' Scary Music (2002, DR13 SACD rip). The "SFX" tracks like track 1, "The Legend Lives" features the sound of a decapitation from The Legend of Sleepy Hollow as the Headless Hessian Horseman rides up to the victim. Excellent dynamics, clean swish of the blade, palpable thud as the head drops to the ground. Scary indeed - even scarier if you're alone late at night in the dark listening multichannel. This version of the fun "Monster Mash" is always a favorite around the house at Halloween. On underpowered headphone amps, the AKG Q701 lacks bass; not an issue here.


If you're a fan of violin solos, Akiko Suwanai's recent Bach: The Sonatas and Partitas for Violin (2021, DR13) is a nice 2-disc hybrid SACD set. I agree with a review I read complementing the Partita No. 2 in D Minor, "Chaconne" on disc 2 track 5. Beautiful playing, and the nuances and timbre of her violin excellently rendered through the amplifier. As one would expect, differences between headphones are much more evident than any limitations I would point my finger at for the headphone amplifier. For classical music like this, an open, "fast" headphone like the Sennheiser HD800 is a real treat especially with all that treble extension - to a fault as some complain of the highs being too harsh without some EQ/correction.


Let's listen to some nasty compressed modern pop ;-) - Miley Cyrus' Endless Summer Vacation (2023, DR5) - in time for the upcoming hot summer days. Well, back in 1991 when the headphone amp circuitry was being conceptualized, the sound of the pop scene would have been different. Nonetheless, for accurate, high quality audio reproduction gear, these kinds of productions that intentionally include clipped samples and >0dBFS "true peaks" are not particularly difficult to negotiate and the sound is just as expected through the headphones. The track "Flowers" is getting a lot of local radio airplay. There are some other noteworthy tracks on the album for Top 40 pop fans like "Thousand Miles" or "Rose Colored Lenses".


Finally, let's listen to the "audiophile" female vocals of Clare Teal's Get Happy (2008, DR8). While still a bit dynamically compressed and I would have liked to hear less "shouty" instrumentation at times, the smooth vocals here represent a massive shift from the grainy and synthetic-sounding Miley (no doubt much of this from vocal trauma).

Overall there's a nice collection of "happier" mixture of standard tunes (Cole Porter, Irving Berlin), some popular tracks like "Moondance" or "The Very Thought of You", and some original compositions on the album.

So, subjectively this headphone amplifier sounds nice. Maybe not the quietest amplifier I've ever heard (very slight hiss at extreme output levels with the Sennheiser and AKG headphones in particular), but quieter than the noise floor for the recordings I listened to. No objectionable hum with any of the headphones listed above with volume turned up. I did not try very sensitive IEMs; an amp like this would be overkill plus this was not intended for very low impedance drivers.

II. Objective

For the tests here, I used my measurement computer (Beelink SER4) and the SMSL DO100 DAC (dual ESS ES9038Q2M) RCA out. Measurements were captured using the RME ADI-2 Pro FS ADC for convenience.

Let's have a peek at a high impedance 560Ω load, 1kHz sine just before visual clipping:


The maximum voltage output is 8.9Vrms. The dual-mono construction achieves excellent channel balance with only 0.06dB difference.

The headphone amplifier gain is +34dB. Keep input <0.2Vrms to avoid clipping. 

How about the output impedance?


Measured over 8 points across the audible frequencies, we get an average of around 1.2Ω which is really quite good and this should be able to drive >10Ω headphones with little frequency response variance. The design however is intended for 40+Ω devices and we could see higher distortion with lower impedance loads.

Speaking of frequency response, let's have a peek with different loads:

Levels intentionally spread apart for clarity.

As usual, I subjected this amp to my standard resistive loads at 20/75/560Ω representing low, medium, and high impedance values along with an actual headphone, the Polk Ultrafit 2000 with variable impedance ranging from 15-55Ω as previously measured. As expected, the low ~1.2Ω output impedance does a good job keeping the frequency response stable with <1dB variation up to 80kHz even with the more challenging Polk headphone.

So how much power can this DIY headphone provide?

With the low impedance 20Ω load, the amplifier is able to maintain <0.1% THD+N up to 600mW, not bad.

Raising the output impedance to 75Ω then:


It's good to 850mW at <0.1% (-60dB) THD+N. Notice the lower distortion compared to the 20Ω load above. And finally let's have a peek at 560Ω:


Nice and clean (<0.1% THD+N) up to 9.2Vrms into 560Ω or 150mW for those of you with a need to drive some high impedance headphones. Looking at the harmonic amounts, notice that they're well controlled with lower order 2nd and 3rd predominating once we hit the higher voltage levels.

Here's the 1kHz into 75Ω linearity (1% clipping at 0dBFS):


Less than 0.5dB deviation from clipping down to -110dB
 (note that since I'm using digital volume control, it clips around -8dBFS). Excellent output level linearity likely close to the limits of the DAC; more than enough for listening even with highly noise-isolating headphones.

I'm seeing 70dB channel separation into 560Ω load with signal at 5Vrms estimated using dual 300Hz and 4kHz tones, played to each channel and looking at the amount of seepage to the other side this is totally fine, better than vinyl and likely analogue tape as well:

Right and Left graphs offset to show the signal seepage to the other side.

Here's the detailed 1kHz THD+N FFT's, left and right channels (both driven simultaneously) into 560Ω:


As you can see, nothing's perfect, especially with 30-year-old DIY builds not placed inside a proper enclosure. ;-)

We see some low-level 60Hz hum with harmonics below -110dB, and the right channel achieved slightly lower total harmonic distortion. It's great to see that this DIY amp is hitting around -100dB THD+N and about 150mW into that load. I don't know how much clean power headphone amps were able to achieve back in the early 1990's, but this looks (and sounds) great.

Let's check the harmonic distortions across the audible range:


At 5.6V into 560Ω, we see a consistent and predictable pattern of 2nd > 3rd harmonic distortion across the audible frequencies.

As an older design, let's have a look at the Transient Intermodulation Distortion (TIM) test signal constructed in 24/192 (96kHz bandwidth):


On the left, we can see the oscilloscope waveform consisting of a 1kHz square + 12kHz sine (5:1). If the square wave transients are not clean, this will result in elevated distortion sidebands. On the right, even with a strong 8Vrms output into the 560Ω load, the sidebands are way down at below -110dB from the 1kHz peak. TIM as a form of "slew induced distortion" is not a problem and the old LT opamps released in the late '80s - early '90s in this design performed well.

Finally, let me do my "standard" 0.5Vrms Triple-Tone TD+N FFT into 20Ω, both channels driven:


We're looking at a TD+N score of around -67dB. As intended, we can see the complex array of harmonic and intermodulation products created by those 3 tones. We have about 70dB distortion-free range (cursor). Noise floor is a bit higher than I'd like to see, and this would probably improve with a properly shielded enclosure.



III. Summary

Thanks Mitch for letting me borrow this old DIY headphone amplifier from the early 1990's! And patience since I hung on to this for awhile to get the listening and measurements done. ;-)

While far from the "latest and greatest" head-fi gear of today, it's always nice to have a listen to devices across the decades; hence the series of "Retro" articles posted over the years on this blog.

Subjectively, I thought this headphone amplifier sounded really good. I like the authority it had with bass frequencies and it had no trouble driving the various headphones I plugged in. Looking at the data, I guess this is not a surprise given the low ~1.2Ω output impedance, flat frequency response well beyond audible, ability to hit 500-800mW with lower/mid impedance headphones and a generous 150mW into 560Ω, all <0.1% THD+N, and showing primarily low-order 2nd and 3rd harmonic distortions at higher output levels predictably consistent across the audible frequencies.

While a score of -67dB TD+N on the 0.5V Triple-Tone is far from amazing in this day and age of devices like THX AAA headphone designs with better than -90dB on the same test, let's just say that human ears are relatively forgiving "instruments" if we can even call them that. ;-)

There's simply a huge range of "good enough" fidelity to enjoy before we complain about distortions. Remember there's nothing wrong with preferring a little bit of distortion - our vinyl and tube-loving friends embrace this fact all the time because it's true! There are modern headphone DAC+amps like say the R2R Cayin RU6 I measured a little while back with performance significantly lower than this DIY amp even (below -60dB), yet many still report loving the sound; just call it "musical" and all is forgiven.

Placed inside a fancy, heavy box, stick on a "Class A" label, market the linear power supply with toroidal transformer and lack of capacitors in the signal path, advertise that it's "handcrafted", suggest that it sounds like "tubes", find a good-sounding European-like company name on the fascia, use some fancy wires (may I suggest some Cardas hook-up wire?), slap on a 4-figure price tag, bring it to an audio show paired with respected headphones (I think the Focal Utopia would be awesome!); I doubt many audiophiles or Golden Eared reviewers would question the excellent sonics. Maybe Mitch should have given this "High End" path to audio bliss a go back in the day. He could be rubbing shoulders with the likes of the father & son Wilsons. ;-)

Parts for constructing this DIY headphone amp should still be readily available if one wants to build it. I'm sure over the years, this design must have been fun for many DIY'ers, with many hours of listening pleasure following...

Hats off to all the DIY audiophiles out there who know what they're doing with the electronics and have kept the traditional hi-fi (not "High End") hobby alive.

Off to get some "real work" done. ;-) Hope you're all enjoying the music!

2 comments:

  1. Thanks Arch for giving the head amp a listen and putting it through your measurement sequence. The results turned out pretty good given no shielded wiring or enclosure.

    The "wooden board" was a last minute hack as the old Radio Shack Archer breadboards were just hanging there and I could not give it to you like that ;-)

    If I remember correctly, there are no capacitors in the signal path, which is one of the reasons the -3dB point in the low end is 2Hz.

    After bringing up the power on a Variac to confirm it would at least work, I gave it a listen again. Love the solid bass and uncompressed transients. I am amazed it still works!

    Thanks again Arch and keep up the great writings!

    ReplyDelete
    Replies
    1. Hey there Mitch,
      Thanks again for letting me give this a try and putting it on the board. Probably saved some wires from disconnecting in transport ;-).

      Yes, surprised it worked so well after so many years! While objective measurements clearly demonstrate improvements over the years, it's certainly good to recognize just how well hi-fi devices have sounded for so many decades already.

      With all the hype and Commercialization of Everything, it's a nice step back to realize that we don't need the "latest and greatest" for stuff like amplifiers IMO (and probably everything else) to be very satisfied already.

      Delete