Bottlehead Crack Headphone Amplifier Kit Review

[PeteL]

@A.West
possibly underperfoming a PC headphone jack at normal listening volume (a worthy face-off).

You bring some good points, but those are generalities i find. It's really not "a worthy face off" You won't drive a 650 (or similar) properly with a PC output, and you won't have great results with this on low impedance IEMs. "Normal listening level" by itself, don't mean much. Normal listening level with what? I can tell you one thing, the audio jack output of my older w series Lenovo is simply horrible. I'm quite certain measurments would confirm this but I wouldn't bother, it's just bad, audibly. That said I have absolutely no interest for something that measure that bad, I don't see how it could be a benefit, not into this, but I find that it's also for some (not you) an opportunity to go all out and bring down anything with tubes. I've experienced great hifi audio experience with SOME tube design, not really headphone amps but we never know. I have a Schiit Freya + at home, and I don't need the tube part, both options are included, truth be told I flip them on for most of my listening. But the thing is, It doesn't only sound good, It measure very good as well with the tubes. They are not all glorified FX boxes. Generally speaking it depend on how you look at this, but a very strong argument could be made that tubes are objectively a component that operate in a more "linear" way than transistors...

 

[cistercian]

No problem at all!

It just like the antithesis of the cheap ChiFi we like to buy, just a funny point.

Very true my friend!

 

[solderdude]

No it won't.

Yes it will. mad_economist already explained why and saves me the trouble of repeating what he said.

Indeed, not everyone will enjoy it nor do they have to, but those that do will be happy with it and in the end that's what counts.

 

[KaiserSoze]

... Sonically, when used with suitable headphones it may well be a more pleasant listening experience. ... You say it would only lead to a preference with one particular headphone/listener on one particular day with particular music at their preferred volume. This is complete nonsense. It has been proven a zillion times that the sound signature Amir described is reported and verified by measurements that this is not a 1x lucky shot that only works with HD650 using the 'Crack' happening to Amir on that day with his preferred music and loudness.

It has been proven a "zillion times" that, uh, what was it that was proven a "zillion times"? Oh yeah, that the sound signature that Amir measured is not a lucky shot that only works with HD650. This apparently has been proven a bazillion-gazillion times, and yet, it only works with "suitable headphones".

 

[KaiserSoze]

Ceasing to derail for a moment, if viewers at home would like to produce an equivalent to this plot looking at Zout behavior with a given headphone, note that you do not need a HATS or other ear simulator, you can do it with any ADC! All you need is a splitter that lets you have an amp's output connected to your (presumably very high Z) interface/ADC input and your headphone simultaneously. An example with a very cheap USB dongle (Creative Play! 3, which has high Zout):
View attachment 80864
The blue trace is the output of the Play 3 looped back into its input - as you can see, the FR isn't very flat, but that's fine for our purposes. The gold trace with an average level about 11dB lower is the same loopback with a 33 ohm resistor in parallel, the pink trace is a Koss KSC75.

Since the Play! 3 isn't very flat, we can subtract its loopback from the measurements to look directly at the losses from voltage drop on its Zout as a function of frequency here:
View attachment 80865
This gives us the grey trace (the resistor), which is essentially flat minus a little noise, and the orange trace (the KSC75), which corresponds with the impedance magnitude by frequency measured by an independent source.
View attachment 80866
This is, indeed, essentially how we measure impedance as a function of frequency for headphones in most cases - comparing the loss by frequency with a known resistance.

Edit: People who've been in the audio hobby for a while may remember NwAvGuy's article on Zout - all of his plots were acquired by this method.

Nice. I will elaborate on this effect, taking a more scholastic approach.

If the goal is to put a peak in the headphone's frequency response at the frequency where the headphone's resonance and impedance peak is found, this can be done by putting an impedance in series with the headphones. Moreover, if there is any appreciable amount of resistance in series with the headphones, this will happen whether you want it to or not.

Of course this is something that anyone with a basic understanding of electronics thoroughly understands, but since not everyone has a basic understanding of electronics, I will explain why it happens. The reason it happens is that when two impedances are connected in series, the voltage across the series pair is split between the two impedances in the same ratio as their individual impedances, at each given frequency. For example, at frequencies where the two impedances are the same, the voltage is evenly split between them. But if there is a frequency where the load of primary interest has impedance that is, say, twice the impedance of the other load, then at this frequency the voltage split will be in that same 2:1 ratio. The load of primary interest will receive 2/3 of the voltage, whereas at other frequencies where the voltage split is even, the load of primary interest will receive 1/2 of the voltage.

The easiest way to compute the difference in dB is by computing (for each of the two cases) the dB value that represents power relative to what it would be if there were no other impedance such that all the voltage appears across the load of primary interest. To do this calculation you simply square the voltage factor (because power is proportional to the square of voltage) and convert to dB. Thus, for the even-split case, .5^2 = .25 => -6 dB. For the uneven-split case where the impedance of the primary load is twice greater than the other load (and therefore twice greater than the primary load's nominal impedance): (2/3)^2 = 4/9 = .444 => -3.5 dB. Thus, in this purely hypothetical example, the power dissipated in the primary load, at the frequency where its impedance is twice the nominal value, and relative to its nominal power dissipation, will be -3.5 dB - (-6 dB) = -3.5 dB + 6 dB => +2.5 dB.

Is this desirable? Only for someone who wants a response peak at the frequency coinciding with the headphone's driver resonance and impedance peak. This is certainly not consistent with the Harman reference-preference curve, which calls for a de-emphasis at 200 Hz. It should be noted that what is good for a pair of speakers is not good for a pair of headphones. Speakers can benefit from an emphasis in the frequency range very roughly 75 Hz to 300 Hz because in this approximate range the prominent reflections from the walls near the speaker, especially from the wall behind the speaker, are mostly out-of-phase with the wavefront directly from the driver. Of course this depends greatly on the particulars, particularly the distance from the baffle to the wall behind the speaker. (It also assumes that this emphasis is not added onto "baffle step compensation".) But in general with speakers in a typical room an emphasis in this rough range can be desirable. Manifestly this effect does not apply to headphones, which begs the question of why anyone would want to have this kind of emphasis in a pair of headphones. In the past I have had headphones that had a natural emphasis in this rough frequency vicinity, and I did not like these headphones in the least. To my ear it only made everything sound as if there were an unnatural emphasis in the middle and upper bass. I suppose some people would describe it as "muddy". Headphones with this kind of emphasis do not sound good to me in the least. It seems a lot like homeopathy. As long as the emphasis is so mild that you can just barely hear it ...

 

[KaiserSoze]

so it seems the matching synergy thing in amp and headphones does work, I remember this amp was said to be the to go amp for HD650

Oh for Christ's sake. Said to be the to go (goto?) amp BY WHOM? By everyone who owned the HD650? Or by a handful of people who were vociferous in sharing their audiophile opinions on web forums? If I were to try and count all the silly faddish notions that groups of audiophiles have advocated on the Internet over the past thirty-three years ... hell I don't even think I can count that high.

 

[KaiserSoze]

It's not me that characterizes him as such. You need to see this in its context and understand the quotes it is in. It is not my opinion.

It you did not mean to characterize Amir in that way yourself and were intending an allusion to way he is characterized by other people, on other forums, this did not come through to me when I read what you wrote. I'm will to accept half the blame for the failure to communicate. Every once in a while I myself write in a way that makes it easy for other people to misunderstand what I intend to say.

 

[KaiserSoze]

Definitely agree with this. "Show me where the measurements explain my subjective impressions of X" is a fairly common question. Or, "how can I compare the sound of X and Y using measurements"

Unless we are talking about transducers (and even then room/ear dependant) I'm not aware that any of Amir's usual measurements on electronics would help either question one jot.

There isn't really any way for Amir to be certain what particular sort of sound the other person is talking about unless the other person figures out how it shows up in measurements and explains it to Amir in terms of measurements. So far as I know he does not have the magical power to get inside another person's head and figure out what other people mean when they use non-quantitative language to describe something they heard (or thought that they heard). Does the fact that Amir does not have such magical powers mean that his measurements of electronics are not useful? How would we ever hope to reach a stage where such communication difficulties do not any longer exist, except by continuing to measure and continuing to encourage everyone to study the measurements of the components they like and the components they do not like and figure out, by comparing the measurements, (1.) what measurements are most important and what measurements are less important, and (2.) what exactly it is about the physical behavior of a piece of equipment that determines whether they will like it?

 

[North_Sky]

Wow, it truly feels like AVS forum ... the perfect fit and form.
You ok kiddo...why so serious.

 

[Timbo2]

Wow, it truly feels like AVS forum ... the perfect fit and form.
You ok kiddo...why so serious.

For those of us in the US and of certain age it needs more...

SNL: Jane, you ignorant....

 

[SIY]

Here you go:
View attachment 80850

With a 32 ohm load, the bass rolloff frequency is 50 Hz. I'm always amused to see some of my design features pop up a few years later in their products.

Mods to improve this are easy. First, a safety resistor from the grid of the 12AU7 to ground. Second, bias up the heaters to 50V above ground and bypass each side to ground to eliminate common mode heater noise. Third, either improve the plate loading of the 12AU7 or (best yet) replace it with something more linear, making sure to re-bias it. 90V on the anode is far too low for it to run linearly. The output stage is a disaster, which would require some pretty extensive redesign to fix, but getting some feedback in there is bound to help- there's already excessive gain, so why not use it?.

 

[KaiserSoze]

It's nice to see a review of a tube amp coming from Amir; it's also good to see his subjective comments about an amplifiers.

Having reached advanced age, it's no longer a surprise to me to hear that amps with high(ish) levels of 2nd and/or 3rd order HD have "rich"sound. I would really like to hear Amir's subjective comments pertaining to panned amps like the Parasound JC2 preamp and Audio Research D300 power amp that have relatively high 2nd/3rd order HD -- especially the latter.

I don't agree. Amps, etc., with relatively high levels of 2nd and/or 3rd order harmonic distortion are going to sound more agreeable that amps with the same SINAD but relatively more higher order distortion.

(I struggle to remember that, "All amps sound the same". My bad.)

When people say this, they are only making a comment about the variability in the sound of amplifiers in comparison to the variability in the sound of speakers. In other words, compared to the difference in the sound of different speakers, all amplifiers sound the same. When put into this correct context it is difficult to disagree with it. Moreover, the inescapable implication is that for the great majority of people in the great majority of home settings, spending more than a few hundred dollars on an amplifier could only be justified on some other basis than the quality of sound. Or to say this in a way that most audiophiles would likely find more tenable, if you spent more on an amp than half of what you spent on your stereo speakers, you did not spend your money wisely. But of course there is more to it than just sound quality. Very many of us like to pretend that all we really care about is the sound quality when we all know perfectly well that this is not true. It is the aesthetics and intangibles that make us feel like we have been transported into some different realm, or so far into the future that we have lived beyond our mortal years. With the right aesthetics and intangibles, we are left salivating and more than willing to pay whatever price has to be paid for nirvana. Also, a pair of backlit dancing galvanometers helps a whole lot. (One thing that doesn't matter is the quality of the AC/DC conversion, i.e., the power supply. You don't have to be concerned with noise on the DC rails, because all you need to take care of that is one of those little LED doohickeys that Paul at PS Audio sells. )

 

[KaiserSoze]

I agree. If I had a degree of competency and it was Atom price I might be tempted to try, but at the cost of an A90 or similar, nuh huh.

People who live in the Great White North are known to be especially fond of tube amps during the winter months. They all have several under the bed and keep them running throughout the night.

 

[bobbooo]

It's less a matter of that particular headphone than that particular headphone operating principle. If you look at the electrical impedance of ex. the DT880, the AKG K612, or the Focal Clear- or the T1, HD800, K812, or Utopia - you'll see a similar rise in impedance magnitude around the bass, right where it tends to start to fall off. This is because the rise in impedance, in the fairly open moving coil case, is the driver's resonance frequency.

They might have broadly similarly shaped electrical impedance curves, but the rise is not always at the same frequency, nor by the same amount, nor of the same 'Q-value'. And of course the frequency responses of the headphones aren't the same either. This all means that the effect of this tube amp's high output impedance cannot be generalised for these headphones. Reference Audio Analyzer's excellent comparison tool illustrates this. Below are the frequency responses of some of the headphones you mentioned using the FiiO K5, which has minimal output impedance (0 ohm measured by RAA, 1 ohm as measured here on ASR):

And here are their frequency responses using the Laconic Lunch BOX HA-06 PRO tube amp (how do they think up these names?), which has an output impedance of 171 ohms as measured by RAA (close to the 188 ohms of the Bottlehead Crack):

As is pretty obvious looking at these two graphs, the effect of the tube amp's output impedance is quite different for each headphone, particularly the Focals. Here's just those two comparing both amps together:

Here we see a huge increases in bass, and shifting of the bass peak of the Utopia from just over 100Hz down to 70 Hz, completely changing the tonality of the headphones.

Here's the HD800 with the two amps, showing a much broader increase in bass than the other headphones, extending all the way out to the lower mids:

And the Beyers, showing less of a change between the amps than the other headphones:

Finally, as a cross-check, just the HD650 on its own with both amps:

This frequency response change in the bass matches very well with what Amir measured, corroborating both:

Yes it will. mad_economist already explained why and saves me the trouble of repeating what he said.

See above. Your claim that the Bottlehead Crack 'synergizes with all high impedance headphones' is not borne out by the above data, even for a small selection of headphones with similar operating principles. To claim that all high impedance headphones will sound good (even for a subset of people) using this tube amp is ludicrous, because as shown above the changed sound will be different for each headphone, depending on its impedance curve and frequency response.

 

[KaiserSoze]

I will see. Was out yesterday and this morning had to tend to the garden.

Was gonna use a link to "Tend My Garden / Garden Gate", but then when I came across this, of just the Garden Gate but with the picture, what was I supposed to do?

 

[KaiserSoze]

They might have broadly similarly shaped electrical impedance curves, but the rise is not always at the same frequency, nor by the same amount. And of course the frequency responses aren't the same either. This all means that the effect of this tube amp's high output impedance cannot be generalised for all these headphones. Reference Audio Analyzer's excellent comparison tool illustrates this. Below are the frequency responses of some of the headphones you mentioned using the FiiO K5, which has minimal output impedance (0 ohm measured by RAA, 1 ohm as measured here on ASR):

View attachment 80955

And here are their frequency responses using the Laconic Lunch BOX HA-06 PRO tube amp (how do they think up these names?), which has an output impedance of 171 ohms as measured by RAA (close to the 188 ohms of the Bottlehead Crack):

View attachment 80956
As is pretty obvious looking at these two graphs, the effect of the tube amp's output impedance is quite different for each headphone, particularly the Focals. Here's just those two comparing both amps together:

View attachment 80959
Here we see a huge increases in bass, and shifting of the bass peak of the Utopia from just over 100Hz down to 70 Hz, completely changing the tonality of the headphones.

Here's the HD800 with the two amps, showing a much broader increase in bass than the other headphones, extending all the way out to the lower mids:

View attachment 80965
And the Beyers, showing not much of a change between the amps really:

View attachment 80969
Finally, as a cross-check, just the HD650 on its own with both amps:

View attachment 80970
This frequency response change in the bass matches very well with what Amir measured, corroborating both:

See above. Your claim that the Bottlehead Crack 'synergizes with all high impedance headphones' is not borne out by the above data, even for a small selection of headphones with similar operating principles. To claim that all high impedance headphones will sound good (even for a subset of people) using this tube amp is ludicrous, because as shown above the changed sound will be different for each headphone, depending on its impedance curve and frequency response.

Most excellent proof of point there, bobbooo. I think you just won the award for the best post in this thread (following Amir's original post of course). I tried to think of some way to add to it by saying something clever and witty, but all I came up with was "normal eyes 2 won kay hurts".

 

[wwenze]

I see income opportunity in selling a headphone amp with adjustable output impedance.

Somebody should indiegogo this.

 

[majingotan]

I see income opportunity in selling a headphone amp with adjustable output impedance.

Somebody should indiegogo this.

It's been done for ages. There's an older amp called Project Solsctice by Garage 1217
http://garage1217.com/garage1217_diy_tube_headphone_amplifiers_005.htm

Specifications:

• Solid state output stage
• Non overall feedback - triode voltage gain stage
• Power supply: 24VDC (0.15A cont, 0.40A peak)
• Input Resistance: 61kO With default 50k gain module
• Input Sensitivity (6N23): 0.8V or 1.6V (dependent on gain setting and used tube)
• Gain: 12dB (selectable and dependent on tube)
• Max Output voltage: 7Vrms at 300R
• Output Resistance: Selectable 0.1R, 31R or 68R
• Frequency Response: 13Hz – 80 KHz (-0.5dB) with 32R load
• Frequency Response: 4.5Hz – 300 KHz (-3dB) with 32R load
• Signal to Noise ratio: 98dBA (dependent on tube)
• Crosstalk: -93dB (dependent on tube)
• THD: > 0.10% (dependent on tube)
• Suitable for: 16-600R Headphones, 32-300R recommended

 

[Mad_Economist]

They might have broadly similarly shaped electrical impedance curves, but the rise is not always at the same frequency, nor by the same amount. And of course the frequency responses aren't the same either. This all means that the effect of this tube amp's high output impedance cannot be generalised for all these headphones. Reference Audio Analyzer's excellent comparison tool illustrates this. Below are the frequency responses of some of the headphones you mentioned using the FiiO K5, which has minimal output impedance (0 ohm measured by RAA, 1 ohm as measured here on ASR):

View attachment 80955

And here are their frequency responses using the Laconic Lunch BOX HA-06 PRO tube amp (how do they think up these names?), which has an output impedance of 171 ohms as measured by RAA (close to the 188 ohms of the Bottlehead Crack):

View attachment 80956
As is pretty obvious looking at these two graphs, the effect of the tube amp's output impedance is quite different for each headphone, particularly the Focals. Here's just those two comparing both amps together:

View attachment 80959
Here we see a huge increases in bass, and shifting of the bass peak of the Utopia from just over 100Hz down to 70 Hz, completely changing the tonality of the headphones.

Here's the HD800 with the two amps, showing a much broader increase in bass than the other headphones, extending all the way out to the lower mids:

View attachment 80965
And the Beyers, showing not much of a change between the amps really:

View attachment 80969
Finally, as a cross-check, just the HD650 on its own with both amps:

View attachment 80970
This frequency response change in the bass matches very well with what Amir measured, corroborating both:

See above. Your claim that the Bottlehead Crack 'synergizes with all high impedance headphones' is not borne out by the above data, even for a small selection of headphones with similar operating principles. To claim that all high impedance headphones will sound good (even for a subset of people) using this tube amp is ludicrous, because as shown above the changed sound will be different for each headphone, depending on its impedance curve and frequency response.

Nobody is alleging that the Crack will produce a uniform frequency response with all headphones, though - obviously the Q of the driver resonance, and since it's not a perfect current source the ratio of the nominal load impedance to the output impedance are both significant. You will get a more significant voltage divider "equalizer effect" with a headphone whose peak impedance at resonance frequency is 12dB above its nominal impedance (e.g. Utopia) than one for which it's closer to 6dB (e.g. HD600), but a comparison at a static Zout across that range of nominal impedances isn't going to represent that terribly visibly through the relative noise of the vastly higher drop on the Utopia to begin with.

While I think it's fidelity failings are quite obvious - and, frankly, I've always considered tube designs in that style kind of a fire hazard - I've only heard Bottlehead advocate its usage with loads close to or exceeding its Zout, which will limit the effect significantly (the maximum possible difference, assuming infinite Z at fs, would be 6dB; within the practical range we see in normal cases, it'd be something closer to twice the effect you see on the HD800 at most.

A source with a Zout that's >.5x and <1.5x the nominal impedance of a typical open headphone will generally yield a similarly placed, relatively similar magnitude boost centered at low frequencies and specifically around the driver's resonance, if you'd prefer a more formalized statement equivalent to what I believe @solderdude intended. This isn't exhaustively accurate - some open headphones have very damped driver resonance, and conceivably one could implement any sort of passive internal filtering that would alter the impedance curve in unexpected ways, but the general expectation of "pair headphone with nominal impedance >150 ohm with Crack or similar, experience mild (1-4dB) boost to lower mids/bass" will hold in most cases.

Of course, if you wanted that, a 200 ohm resistor costs like $.10 on Mouser for one piece, and if you felt really adventurous you could get a whole box of different values and roll them from Amazon for $9.

Edit: Allow me to correct myself, I had forgotten that, technically, Bottlehead had recommended a low impedance headphone at one point...but they were of course kind enough to add a series resistor to make its load more suitable for the Crack, in their eternal creativity. At least the 8323's impedance wasn't that variable to begin with...

 

[AudioTodd]

Well, he doesn't know the plot. The measurements predicted the change in frequency response. And the high distortion killed the audible performance with a number of headphones. So the issues were real and audible.

It just happens that EQ effect it provided through high impedance was a positive one in one instance. This too was predicted in the measurements of the amp. And confirmed through headphone measurements.

Were it up to him, he would have falsely assumed that tube = warm sound and distortion = inaudible. Both are wrong. Measurements prove it. And my proper listening tests demonstrate it.

All of this aside, I provide listening tests for when it matters. So his assertion that we only go by measurements is absolutely false in these scenarios. When there is a potential for subjective testing adding value, that is what I provide.

Hey! What’s the big deal? This is just proper component matching and balancing of the system through careful assembly of the various parts!! An audiophile never ending quest that objectivists just simply do not appreciate!! LOL