Back in 2012, Dana Brock, Nelson Brock, and Mark Cronander (aka Variac) organized an "Amp Camp" in Northern California at the Brock's ranch. Nelson Pass, well known among audiophile circles, contributed with one of his designs and I'm sure they all must have supervised the participants as they went about soldering and building those first kits. The result was a simple, solid-state, Class A amplifier to take home. This amp was of course the Amp Camp Amp (or ACA) which we'll be talking about and measuring here.
You can see Pass' article introducing the amplifier with schematics and measurements. What a great idea! There is no substitute for the value of experience and the memories of building something, especially while sharing with a loved one; regardless of the subjective or objective performance of the final product.
The ACA has developed a healthy following over the years - check out the long-lived discussion thread on diyAudio. Nothing official, but I heard over the grapevine that there have been >2000 kits sold, suggesting conservatively >2500 actual stereo amps out there with no accurate way to estimate non-kit PCBs, unofficial "clones", or just completely DIY builds with the published schematics.
The device I tested, pictured above, is one of the early kit designs assembled as 2 monoblocks by Mitch Barnett (aka Mitchco, mitchba) a number of years back (with his daughter). As you can see, it's in a nice Ferrari-red enclosure. The switching power supply he's using are high quality 24V/5A units (Mean Well GST120A24-P1M) with low noise, rated 180mVp-p, 90% efficient. These are an upgrade as the original ACA design used typical 19V/3A laptop computer power supplies. Remember that because this amp is a DIY project, it only makes sense especially for novices to avoid working on high voltage circuits connected directly to the wall - hence the external power supply.
Since I'm borrowing these units for testing, I did not open them up to poke around or take pictures of the innards :-). Mitch tells me that the only modification he made was this one, the "1.1" version that upped the power a little and supposedly lowered distortion. These particular boxes were built in 2013. Here's a "borrowed" picture of Mitch's four "stuffed boards" from his post on the diyAudio forum:
As you can see, this is a low power, single-ended MOSFET amplifier. Depending on how you want to rate this, using a 24V/5A power supply, it's said to be "8W" on the DIYAudio Store into 8Ω "with clipping at 3% distortion" using the recent 1.6 version. Since we're still at version 1, the core amplification circuitry hasn't changed significantly over the years. The other spec we're given is that distortion is "0.7% at 1 Watt" presumably at 8Ω as well. We'll see if this version 1.1 amp can keep up with a 4Ω load when we put it on the test bench. As you can see in the introduction link above by Pass, there are measurements in there already, and Audio Science Review has measurements / impressions up as well based on the 1.6 version.
Despite all that has been written over the years, since Mitch had the amps on offer for a loan, I figure it would still be fun to have a listen myself and see what my test bench shows. Here are a few more pictures of one of the boxes with heat sink and Mean Well power supply:
|Rear: Screw-on compression speaker terminals for bare wires. Newer versions can use banana plugs. RCA in. Power jack.|
Before I begin testing and making an even bigger mess with cables everywhere, I want to show you my little basement set-up :-).
I'm starting now to use my Intel NUC 6i5SYH for data acquisition and analysis which is faster than the Asus laptop and Microsoft Surface 3 Pro I had been using all these years. Noise level through the measurement system remains very low. You can see the Rigol DS1104Z oscilloscope on the left, the Raspberry Pi 3 B+ "Touch" running Volumio with signals on a USB stick - convenient if I want to use the Topping DX3 Pro as signal generator. There's a 4Ω/300W power resistor in the middle. Behind the 2 amps are of course my RME ADI-2 Pro FS DAC/ADC which is the heart of the ADC measurement system sitting on top of the Linear Audio Autoranger MK II which I updated to a 2V firmware a few months back (originally 1V nominal out) allowing even better resolution (I know there's now an EPROM selectable up to 3V - can't keep up :-).
Nope, it's not an Audio Precision set-up but with carefully selected components, tested to verify functioning, there's a lot of flexibility one can enjoy. I can use different software and find new ways to measure different characteristics, without spending a ton of money as a hobbyist. It does take some time getting the pieces working, creating my own test signals and standardizing the procedure. But that's part of the fun! Furthermore, the bits and pieces like the DAC and ADC can be used elsewhere as playback equipment and for vinyl rips when not testing. Refer to my post on the "Measurement Of Amplifiers Rig" (MOAR) for details on how I'm doing this.
As you can see in the test bench picture above, I've put stickers on the front of the ACA boxes to remind me which is "right" and which "left" for the measurements. Obviously it's up to each user how they want to orient the mirror-image amp/heat sink. I'm putting the heat sinks facing inward as if I were using these in a sound room since I prefer the amps to be located close to my speakers on each side and because they get warm, I'd rather have the heat sinks medially oriented away from speakers placed further out.
I. Basic Amplifier CharacteristicsAs a start, let's just get a few noticeable nuances out of the way. When I turn these amps on, they send some odd "gurgling" sounds to the speakers for about 2-3 seconds. Nothing too loud or scary. Likewise when turned off there is what sounds like a brief DC "thud". I was told by Mitch that this is normal for the amp and he uses blocking capacitors with his sensitive compression drivers for protection.
These amplifiers also need to warm-up. In fact, I do not recommend listening to them until at least 15 minutes of warm-up has been achieved. During the warm-up period, the sound is unstable, demonstrating a noticeable meandering soundstage - it's like the amplitude level and frequency response appear to be fluctuating over those first few minutes and the instability can be seen while playing a steady tone and watching the oscilloscope levels. Things generally look/sound OK from 15-30 minutes and I would consider the amps ready for listening (and measurements) after 30 minutes to really get an idea of their performance.
As expected, the heat sinks do get warm as an inefficient Class A device that's always "on". Uncomfortable to the touch but not going to burn flesh - the large heat sinks worked well. After 2 hours in ~20°C ambient room temperature on my test bench, the contact thermometer probe tells me the heat sinks measure 45-46°C (~114°F) for both units.
With the units adequately warmed up, let's start with a look at the Amplifier Voltage Gain:
- Right: +16.0dB
- Left: +15.9dB
That's essentially an identical reading between the 2 monoblocks. Note that this gain level is quite low compared to most amplifiers. For context, the unbalanced input standard for THX is +29dB. My Emotiva XPA-1L last year measured +35dB, the cheap Yeeco TI TPA3116 class D was +25dB. My Hypex NC252MP DIY box measured +26dB. This means the ACA will need to be driven with higher input voltage compared to the other amps. This is obvious when hooked up and listening on my system - I need to push the preamp volume up by at least +10dB to get this amp near the output level of any of the others.
Pass' ACA intro article lists a gain of 14dB and I wonder if the mod Mitch implemented might have bumped the gain a little on these he lent me. I see that the Audio Science Review measurements only found a gain value of +9.4dB on the version 1.6. No wonder Amir asked "Is this thing on?" if either inadequately driven or paired with lower-sensitive speakers. Since version 1.5 (2018), the gain has indeed dropped down to 10dB. This means that a typical 2V source (like say the RCA output from the Topping DX3 Pro) would not be adequate to drive the amp to full output power. Make sure to check that your preamp is up to the task depending on which ACA version you plan to use.
Amplifier Damping Factor:
The ACA has low damping factor consistently around 2.0 - 2.3 across the audible frequencies into a 4Ω load. This means a rather high output impedance which also means it's not going to have tight control over a reactive speaker load. We can see this in the frequency response (~1.5V signal into load):
|Note: Captured at 192kHz with RME ADC.|
Notice that the amps are able to achieve excellent flat frequency response to 20kHz with less than 1dB variation by 48kHz into a resistive load through my measurement system. When it comes to managing the Sony 8Ω speakers however, it's having some difficulty with around 2dB variation across the audio band (compare this to the Hypex's "iron grip" on those same Sony speakers!).
A side note to this is that by ACA version 1.5, as the voltage gain dropped down to +10dB, the damping factor increased to 10 from an initial "about 3" of version 1.0. Current ACAs therefore should be able to "control" the Sony speakers significantly better.
To mix it up a bit, here's the frequency response into an 8Ω load with phase data - nice and flat:
|Note: Captured at 96kHz hence the dip into 48kHz.|
Given that this is by no means a "powerful" amplifier, let's have a peek at the Harmonic Distortion vs. Frequency at 2V/1W into a 4Ω load using Room EQ Wizard's stepped sine function:
II. Single-Tone Harmonic Distortion and Noise
The noise floor (brown) is quite good down around -90 to -110dB from the 2V fundamental. While I cannot fully capture what is heard over the first 15 minutes of warm up time with the fluctuating soundstage, we can see a slight difference in these graphs between "cold" compared to after warming up for 1 hour. As the unit warms up, the most prominent harmonics do increase by about 1dB across the audible spectrum; not a massive change. The most important difference I've found is that the amplitude stabilized after 15 minutes as noted above.
Even at just 1W, we're seeing quite a bit of harmonic distortion present already! By design, this amp uses relatively low 9dB feedback. The 2nd harmonic is prominent up around -36dB at 1kHz after warm-up and basically floats there without much variation. In fact, due to the amount of 2nd order harmonic, THD and THD+N would be basically pinned at just a tiny bit over the red 2nd harmonic line.
Here are some THD(+N) graphs at a few output levels to examine (lots of detail in there - click on image to zoom):
As you can see, the graphs continue to demonstrate the high amounts of 2nd harmonic; not just in the frequency sweep above but also across power levels from 16mW all the way to 6.25W. I've highlighted the THD+N values but since noise is low, and harmonic distortions are high, there's essentially no difference between THD and THD+N.
Notice also that I'm using a 950Hz fundamental. Why you may ask? The answer is with those purple asterisks I've put over the 8kHz peaks. Initially, I thought these might be due to the USB packet noise we've talked about before. Nope. The 8kHz spikes and 16kHz harmonic are actually due to the Mean Well switching power supply. Evidence that at least with this amp, we can make out some noise contribution from the power supply selected. Note that while the 8kHz noise might affect measurement of a 1kHz fundamental thinking that there's a higher level of the 8th harmonic, that 8kHz tone isn't a big problem; it's only -90dB at 16mW so in reality nothing to worry about when listening. Notice by the time we hit 4W and 6.25W, the 8kHz noise is easily swamped by the higher order harmonics.
For completeness, here's a similar "matrix" for the same amp but with an 8Ω load attached:
In watt-to-watt comparisons, we can see that the amplifier prefers the 8Ω load over 4Ω with better THD+N. For example, at 4W into 4Ω, THD+N is a high -25.6dB/5.2%(!) compared to 8Ω THD+N of -34dB/2% (still high). This amp appears to be current limited with lower impedance loads. Definitely will be a challenge with low impedance speakers with difficult phase angles!
If we graph THD+N vs. voltage output to summarize, here's how it looks:
We can easily see the difference the higher 8Ω load makes to harmonic distortion given the same output voltage. The question of course is just how sensitive your speakers are and if the sound is loud enough in your room. Usually, when I measure amplifiers, I like to see 0.1% distortion or less as a rough threshold for being "clean" or "linear". Well, this ain't going to happen here! If we say that the "first watt" is most important, what we see is that at 1W into 4Ω, THD+N is already -35.5dB/1.7% distortion. And 1W into 8Ω has a THD+N of -42.1dB/0.78% (which is about the 0.7% advertised for current versions of the product). At 5W into 4Ω, we're looking at THD+N of -23.6dB/6.6%, and 5W into 8Ω is -31.4dB/2.7%; "impressive" numbers, but not in the usual objectively "good" way!
III. Multi-Tone Testing: Intermodulation Distortion and Triple-Tone TD+NAlready we can determine that this is a rather nonlinear amplifier with significant distortions even at low power. Let's try some intermodulation tones at my standard 2V/1W level into 4Ω:
As expected, we're looking at some rather significant amounts of IMD even at 1W into 4Ω. Across the different tests, we're seeing -23 to -33dB levels. For context again, a good modern, yet affordable, Class D like the Hypex NC252MP ranges from -78 to -97dB on these tests. My Class A/B Emotiva XPA-1L monoblocks sit around -80dB. Admittedly, these are more expensive amplifiers (the Hypex is actually only 2x the price of the ACA kit), but the difference is still rather stark if we just look at cost to objective performance.
Next, let's have a look at the synthetic TIM ("Transient InterModulation") signal consisting of a 1kHz square wave with 12kHz sine created at 192kHz samplerate (96kHz bandwidth), measured again at 2V into 4Ω:
We're seeing some anomaly here and I've put the cursor at the level of those distortions around -70dBFS. Remember, as I've said previously, this is a challenging synthetic test signal and while from an engineering perspective we ideally would not want to see those -70dBFS peaks, I'm not sure just how audible this is. If you want to see essentially perfect TIM signal results at 2V into 4Ω, check out the Emotiva XPA-1L or Hypex NC252MP measurements again; both also achieve good results at 10V.
[Of interest, check out this recent Audioholics interview with Bruno Putzeys (May 2020) where there is discussion of TIM in relation to amplifier gain-before-feedback and the interplay with feedback allowing for lower TIM around 35 minutes. Remember that feedback is not in general a "bad" thing despite the audiophile mythology and love the discussion around 43 minutes that DSD able to sound good is an example of massive feedback.]
IV. Square Wave and Wideband Noise
Consistent with the extended frequency response, the square waves are looking very good. That tiny overshoot at the leading edge is from the DAC I used as signal generator (Topping DX3 Pro V2 at 24/384), but otherwise the tops and bottoms are flat, nothing like the superimposed ultrasonic noise we see with Class D amps. Mitch and his daughter did a great job with the build along with good matching components to achieve excellent channel balance between the two monoblocks.
And for completeness, here's a peek at the wideband FFT up to 600kHz to make sure there are no nasty ultrasonic signals up there headed for your speakers:
Again, you can compare this with Class D and note the absence of things like 400kHz noise as found in the Hypex NC252MP. There's some stuff around 250kHz for example; nothing major and could be switching power supply related.
V. Impressions, Discussions and SummaryHere's the AMOAR Composite Score graphic for this amp:
Permit me to be blunt. This is the worst measuring amplifier I've done to date and these measurements will likely not be "outdone" by another device soon in this blog!
The damping factor is very low at just over 2 with a 4Ω load (I'd like to see 20+). Notice how the 8Ω Sony SS-H1600 speaker (impedance measurement here) frequency response fluctuates across a 2dB range with this amp. Distortion is high at -30dB/3.2% on the triple-tone signal at 2V output level into 4Ω. And when it comes to "clean" power, it appears to be incapable of <0.1% THD+N into 4Ω at any point from 16mW up, nor did I see <0.1% THD+N performance with an 8Ω load.
Noise level is however quite low and the frequency response is nicely extended well beyond the hearing range.
From an objective perspective, I trust that it would not be controversial at all to say that this is not a "high fidelity" amplifier. I assume Nelson Pass himself would agree that the gain provided by this amplifier is rather imperfect, hence the strong non-linear distortions. As he said - "when you run a signal through a device which is even slightly non-linear, you have changed the signal forever". Philosophically, as one who seeks out "transparency" in my audio pursuit, the performance of this amplifier would certainly not make the cut as a reference device in my system. The idea that even-order harmonics might be "tubey", or "sweet", or "benign" is of no consolation if uncompromised signal purity is the goal.
But as with many situations in life, this is not the end of the story...
Let's change gears a little and talk about the subjective side for a moment. What does something that measures so poorly sound like in my system?
It actually sounds good to me - so long as I don't ask the amplifier to do too much! :-)
As you can see, the system consists of my Oppo UDP-205 as DAC streaming from the Roon server computer in the other room into the Emotiva XSP-1 preamp (which can easily provide >2V to drive low voltage gain amps) into the dual Pass ACAs to my Paradigm Reference Signature S8 v.3 speakers with an in-room sensitivity of 92dB/W/m. I used basic 12AWG OFC wire between the ACA and speakers. Listening position is about 10' away. Because of the amps' low power, I cannot run a proper DSP room correction (which typically will attenuate the signal another 3-6dB), and I have the subwoofers turned off (I would need to readjust the subs for integration).
What I heard from this amplifier was clearly a device with much less power than I'm used to either with the Emotiva XPA-1Ls or the Hypex NC252MP. With the low gain, I have to push the preamp up to about "-10dB" just to approach normal listening levels when typically -20 to -30dB is all I need. These ACA amps will not "rock" like the others (remember, I "need" something like 70W into 8Ω in this room and with these speakers). Through my speakers, the ACAs sounded quite good with classical quartets, jazz ensembles, and intimate solo vocalists. Pushing the volume up higher with well-mastered, higher dynamic range rock (like say the DR11 version of Aerosmith), or pump up Eric Johnson's soaring guitar work on Ah Via Musicom's rock tracks (DR12, tracks like "Desert Rose" or "Cliffs of Dover") can sound strained on my system and I start getting worried about damaging the borrowed gear! Generally I listened with average of <75dB SPL in sound room with low ambient noise.
For electronic music, they're not as dynamic or evoke the same sense of "excitement" in me through these speakers, but transients are crisp (I was listening to some Depeche Mode Exciter and Ramin Djawadi's Westworld Season 3 soundtrack). Despite the low power, the bass certainly did not sound "lean" at all - certainly very adequate without the subwoofers, but won't shake the foundations of my home of course. I enjoyed the bass on some Funky Cold Medina. ;-)
As for the soundstage, it's nice with an extended sense of "air" around instruments. I noted to Mitch early on that compared to my other amps, for most vocal recordings, the image can appear more "distant" as if Eva Cassidy singing "Songbird" just took a step back by a foot. I can only make this observation when I quickly A/B between my Hypex and this ensuring they're volume matched; otherwise without context the difference is not exactly "night and day".
Male voices sounded particularly "present" to me as in the illusion of the person singing in my room - Johnny Cash's "The Man Comes Around" (American IV) and Serge Gainsbourg's "Melody" (Histoire de Melody Nelson) were particularly noteworthy. I can imagine that the added harmonics might be "euphonic" with lower voices making the complexities of the human vocalizations "fuller".
"Special effects" type sounds like the alarms and chimes at the start of "Time" (Dark Side) remain well-placed in the soundstage but presumably because of that "air", created a pleasant illusion of taking up more spatial "volume" compare to my more powerful and accurate amplifiers. A similar effect was also heard with demo tracks like "Moonlight on Spring River" (The Dali CD Vol. 3) which really highlights the top octaves. In my discussions with Mitch, we used the phrase "glassy quality" which speaks of the subjective clarity, although I would also use the word "glow" to suggest that there's a pleasant accentuation to the sound ("bloom" might be another non-visual description for this effect I've heard used for tube amps). To me these are examples of potential subjective qualities which each listener will need to judge for themselves. I suspect the "air" and even "glowing" qualities again are likely the result of the distortions (primarily 2nd harmonic accentuation of the "presence" and "brilliance" frequencies?) which can be judged as "euphonic" and thus of subjective value even if typically seen as objective liabilities. Although philosophically I lean towards building a reference system based on more linear/"accurate" amplifiers, nothing stops me from appreciating the effects of adding distortion or coloration in the service of achieving pleasure either.
While personally I would not be satisfied by the limits of these amplifiers for my main system (mostly the low power), I certainly would not fault those with very efficient, easy to drive higher impedance speakers choosing this amplifier. For example, a pair of Klipschorns with 105dB/W/m (recent Stereophile measurement suggest actual 101dB/W/m) would sip <150mW most of the time at an average listening level of 80dB SPL at 10 feet distance. Even into a challenging 4Ω load, distortion would be <1% THD although louder transients would push that.
Looking around the Internet, it seems that this is one of those devices sitting at the crossroad between the "objectivist" vs. "subjectivist" polarity among audiophiles. I'm glad this post is being published after the "THD Blind Test" series because I believe those results (and other blind tests over the years) remind us of the interplay between the objective and subjective sides of the human equation. Recall that listeners on average did not choose the sample with zero added distortion on the test but the one that contained a little bit of added anomaly sounded "better".
Not everything we can measure can be heard and in fact I would say much of what we find with high resolution test equipment (like jitter) are at levels beyond human perception. Finding anomalies can easily tell us the limits of the engineering, but it's more tricky to discuss whether these same results are audibly significant since human minds/ears are inherently non-linear. And even if high distortions can be heard, some will judge the distortions as "improving" the sound (as we found in the THD Test). Appealing to one's perceptual mind/ears is different from achieving objective accuracy or the cognitive satisfaction of arriving at what may be a theoretical ideal.
Exactly how much distortion is tolerable or even potentially enjoyable will vary for each of us, and perhaps best understood based on psychoacoustic principles applied to populations. The casual generalizations any single audiophile might subjectively describe will likely not apply to many others, likewise, many anomalies found using sensitive test equipment probably would be perceptually insignificant.
Balance is the only nuanced solution for so many simplistic polarities we face in life. It's not "either/or" objective vs. subjective; the answer for me has always been a bit of "both". Being comfortable with objective performance gives us the ability to understand why devices might sound different based on physical properties so we can determine the limitations, how it may match with other components, its utilitarian value, protect us from hype, and even "snake oil" ideas/products that make no difference. Embracing our own subjective experiences give us insight into ourselves, perhaps allows us to efficiently seek out what we desire, dare I say achieving psychologically minded maturity with enough humility to appreciate our biases and limits that modify, even determine perception.
The measurement results clearly tell me that these Pass Amp Camp Amp 1.1's are highly non-linear amplifiers that "color" the sound significantly to the point where they yield to the reactive properties of the speakers they're connected to and approach 1% THD+N at just 1W output into 8Ω, with even higher distortion into 4Ω. I do not consider the ACAs to be clean, "first watt" amps.
However, listening to these amplifiers is a good reminder that pristine test signals are not music, emotions need not be logical, and joy does not arise purely from objective results given all the unquantifiable variables that go into what is "good" for each of us. What is "good" often is also not just about something that "sounds good", but we should also consider the non-utilitarian functions a device may serve.
In many ways, these amplifiers have the "character" of low-power class A tube amps embodied in a hassle-free solid-state package. However it is missing the illuminative glow of actual vacuum tubes that some might desire. :-) In the days ahead, I'll get hold of a SET amplifier and run some comparisons.
Let's not forget why this amplifier was created in the first place - as a DIY build that can be done in an afternoon for the experience of making something that works. The build guides for this amp are first rate. The joy of being able to do that, especially if shared with loved ones, may already be worth the price of admission. Heck, I might even buy one for myself as a project with the kids!
A big thanks to Mitch for lending me the amplifier to listen to and measure. I appreciate his patience as he has parted with these for a few months now - he and his wife came over with the ACAs the weekend before the COVID-19 pandemic restrictions started in British Columbia!
PS: It looks like Nelson Pass has published some measurements from the recent version (1.8) of the ACA just the other day. Significantly lower harmonic distortion in parallel and balanced modes for that "first watt" although one needs to consult the table around differences in output power and damping factor. Remember though that since the unique sound of this amp is likely a result of those distortions, to drive the distortion level down might not be a good thing if you're after that 2nd harmonic sonic character!
"If it measures good and sounds bad — it's bad. If it measures bad and sounds good, you've measured the wrong thing."Considering the measurements of the ACA, I think it's worth thinking about these words to ask if they may be applicable. Clearly these amps do not "measure good", so we'd be considering the second sentence as potentially relevant. While the ACA clearly "measures bad" compared to an objective ideal, yet I believe it can sound good, I still don't believe that I'm "measuring the wrong thing". Rather, the measurements are truly the performance of this device and it only sounds "good" because within a certain level of tolerance, the mind/ears are adaptive without immediate context for comparison, forgiving of distortions, and likely have non-linear preferences as discussed above. Given the number of positive user reviews, this amplifier could in fact be sounding good because of those distortions, not in spite of them. There's really no mystery, I think...
Previously, I offered this suggestion:
"If it measures bad and sounds good, maybe your hearing isn't as good as you think."I think this is still applicable to the pseudo-Golden Ears who somehow think their hearing is more resolving than modern measurement equipment and there's some fundamental audio property still "missing" to be explained by physics (sure, psychoacoustic understanding may be incomplete, but that's not hard-core physics).
Let's modify von Recklinghausen's statement to be less acerbic, more nuanced then:
"If it measures good and sounds bad — check that you're using it properly, it may be inappropriate for your needs. If it measures bad and sounds good, double check the measurements first."
[Be sure to check out Mitch Barnett's OWNER'S COMMENTS on his Pass ACA as his response to these measurements and discussions.]