Tom Christiansen Audio HPA-1 Headphone Amp Review

[digicidal]

Besides, he is someone with the highest level of integrity, and I know anyone who will buy this will both be enjoying this piece of equipment for decades (if not a century), and will be well taken care of.

Well done, Tom!

As someone right around the median age of ASR members (at a guess... maybe a few years younger than that) your statement makes it seem like this device may increase life expectancy by many years! I'd say almost any price would be inexpensive in that case... might even be enough to make me get some headphones too. LOL!

On a serious note... when you consider how many times the pot is the very thing that goes first on so many devices - just from the numerous complaints read on here and other fora (as well as my own experience)... that is far more important than the stellar performance shown in the measurements. Unfortunately, it's also the area where most volume-based operations first look to cut costs.

Now @tomchr needs to get working on a pre/pro+DAC with HDMI switching and 2.2 balanced outputs with DSP... at MSRP under $4K. Pretty please!

 

[Nango]

@tomchr .....which is your favorite headphone?

 

[lator]

@tomchr HPA-2 with XLR headphone plug and I have found my headphone amp holy grail.

 

[tomchr]

Do you know what volume pot it is in DROP's 789? I assume it must not be RK27 volume pots right?

Its a much cheaper and problematic pot they use there.
A financial decision and maybe even potmeter allignment.

I'm sure both play in. The RK271 is large (roughly 25 mm square), so it requires a larger chassis. If you want the volume pot centred vertically on the front panel for aesthetics, you're in a 50 mm tall chassis all of a sudden, which increases the chassis cost. It also increases the cost of shipping boxes, foam, and the shipping charge itself. My mind is blown every time I order boxes at U-Line. It seems I end up spending $300-400 easily on maybe 100-150 boxes and packing material. Think about that next time you get "free" shipping.

in Large quantity:
so RK-271 about $15
and RK97 about $3

but.. I mean it's only $12 difference?
does RK97 perform way worst than RK271?

Thankfully, HPA-1 uses RK271 and has an additional support bracket for the volume pot. I am not sure but I can imagine the support bracket for volume pot cost $10 a piece . (I am guessing here)

The RK097 has ±3 dB channel tracking. The RK271 has ±2 dB channel tracking. Those are the spec sheet numbers. As shown by Amir's measurements, they tend to perform better in actual use. A logarithmic pot is made from sections of linear resistance tracks. An expensive pot like the RK271 has 3-4 sections. An economy pot like the RK097 generally has two sections. It's at the transition from section to section that you see the tracking error. Basically one wiper makes it to the next section a smidge earlier than the other, so you get an error. I've tried to find these spots of larger tracking errors and succeeded with the RK097. It's only in one very specific spot that you can measure the error. If you nudge the volume knob even a little bit, you're off that discontinuity between tracks and the pot tracks well again. It's basically something like 8 o'clock to 8:00:05 that shows the error. You'll be extremely unlikely to notice in actual use - especially in a low-gain amp.
I have heard tracking errors on an RK271 pot. That was in a tube amp with high gain. The high gain made it necessary to turn the volume control almost all the way down to avoid blowing my ears. At attenuations below 60 dB, the tracking is worse. This is covered by the spec sheet. In my view, it is a design flaw to have such high gain that the volume control never makes it above 60 dB attenuation.

About your other point: Let's use your numbers for the sake of argument: $22 for the RK271 and mounting bracket. First off, that number does not include the assembly cost. You'll now have to pay someone to turn a screwdriver and a wrench, which is expensive (at least in the western world). Also, the RK271 does not like to be machine washed (the label blows off, which exposes the innards of the pot). So it can't go through the wave solder + flux cleanup like everybody else. It has to be hand-soldered using solder with no-clean flux. So you now have to pay a solder slinger to sling solder, which adds cost. I don't have hard numbers on exactly what that added assembly cost is, but let's for argument's sake say it's $5. Now that "just $12 extra" pot costs you $27 extra in manufacturing cost.
Then there's the concern about profit margin. Businesses, such as mine, exist for the purpose of making money for their owners and/or shareholders. I like to eat periodically. I would also like to receive a salary that's reflective of my skillset, education, and fair market value. I'm not looking to get rich and famous. If I could make, say, 80% of an engineering salary and have the freedom that comes with self-employment, I'd be happy. I have yet to make it there. I'm not even close, but I'm working to get closer. So... To make money, I have to charge more than cost. Sorry. That's how businesses work. If I charged at or below cost, I would not have a business. I would have a hobby.
How much to charge depends on your brand and your distribution channels. A well-known prestige brand might be able to charge 5-10x cost, or more. An unknown low-end brand with online-only distribution might only charge 2-3x cost. I run a one-man show, so having some distribution would be nice as it would allow my customers to buy and receive my products even if I'm on vacation or otherwise unavailable. Distributors of unknown brands, such as mine, generally charge around 40%. If I use a representative to engage with that distributor, there goes another 5-6%. So if I want to break even, I have to charge roughly twice the production cost. As covered earlier, breaking even is not a good business model, so I'd have to charge more. Suddenly that "only $12 more" pot ends up adding $80-130 to the final price of the product, not counting the added chassis cost and added costs related to shipping and shipping supplies.

That's why a $99 headphone amp cannot contain a $15 pot.

This concludes today's business lecture.

Tom

 

[tomchr]

@tomchr .....which is your favorite headphone?

I don't really work in terms of favourites. Here are the headphones and IEMs I own (and use or have used):

• Focal Elex
• Sennheiser HD-650
• Sennheiser HD-580
• Etymotic ER3
• Etymotic ER4

I use the Etymotics when I travel. The ER4 replaced a pair of ER6 where the cable wore out after 15 years of use. I thoroughly enjoy the Focal Elex. That's a bargain of a headphone! I will always have the HD-650 with me at shows as it is a well-established reference, and, honestly, a good middle-of-the-road headphone. In the past, I've used the HD-414 and one of the two-digit Sennheiser models as well.

If I could justify the cost, I'd love a pair of Focal Utopia.

That said, I have not measured or heard all headphones. If your favourite is not on the list above, it should not be taken as an indicator that I don't like it. Just that it's not in my current collection.

@tomchr HPA-2 with XLR headphone plug and I have found my headphone amp holy grail.

Good things come to those willing to wait. It would certainly be a logical extension of the product line.

Tom

 

[KSTR]

If everything is balanced from the XLR plugs to the potentiometer, what do you think about moving further and get another version with a 4-gang pot. and XLR plug as well?

A balanced input's first and foremost purpose is to make the signal unbalanced again, referenced to the local GND. We want all the common-mode signal (noise) to cancel *before* it even hits the amp internals. Only at the output one may want to make an inverted copy for signal-symmetry. But signal symmetry is not a requirement for balanced cabling, only impedance symmetry is.
Which boils down to: a fully balanced signal path doesn't make any sense because it misses the main idea of balanced technology which is that the noise has to be subtracted out early at every input and shall neither appear on the internal stages nor on the output (no matter if balanced or not).

@tomchr : Excellent product!

 

[pma]

Klaus, I agree, I would just add that sometimes a floating balanced output as DRV134 has is an excellent aid.

 

[Ratatoskr]

How is the headstage of the HPA-1 on the HD800S? Most solid state amps I have listened to present music on a flat plane, with little depth, layering or sense of space.

 

[dragonspit4]

I'm sure both play in. The RK271 is large (roughly 25 mm square), so it requires a larger chassis. If you want the volume pot centred vertically on the front panel for aesthetics, you're in a 50 mm tall chassis all of a sudden, which increases the chassis cost. It also increases the cost of shipping boxes, foam, and the shipping charge itself. My mind is blown every time I order boxes at U-Line. It seems I end up spending $300-400 easily on maybe 100-150 boxes and packing material. Think about that next time you get "free" shipping.


The RK097 has ±3 dB channel tracking. The RK271 has ±2 dB channel tracking. Those are the spec sheet numbers. As shown by Amir's measurements, they tend to perform better in actual use. A logarithmic pot is made from sections of linear resistance tracks. An expensive pot like the RK271 has 3-4 sections. An economy pot like the RK097 generally has two sections. It's at the transition from section to section that you see the tracking error. Basically one wiper makes it to the next section a smidge earlier than the other, so you get an error. I've tried to find these spots of larger tracking errors and succeeded with the RK097. It's only in one very specific spot that you can measure the error. If you nudge the volume knob even a little bit, you're off that discontinuity between tracks and the pot tracks well again. It's basically something like 8 o'clock to 8:00:05 that shows the error. You'll be extremely unlikely to notice in actual use - especially in a low-gain amp.
I have heard tracking errors on an RK271 pot. That was in a tube amp with high gain. The high gain made it necessary to turn the volume control almost all the way down to avoid blowing my ears. At attenuations below 60 dB, the tracking is worse. This is covered by the spec sheet. In my view, it is a design flaw to have such high gain that the volume control never makes it above 60 dB attenuation.

About your other point: Let's use your numbers for the sake of argument: $22 for the RK271 and mounting bracket. First off, that number does not include the assembly cost. You'll now have to pay someone to turn a screwdriver and a wrench, which is expensive (at least in the western world). Also, the RK271 does not like to be machine washed (the label blows off, which exposes the innards of the pot). So it can't go through the wave solder + flux cleanup like everybody else. It has to be hand-soldered using solder with no-clean flux. So you now have to pay a solder slinger to sling solder, which adds cost. I don't have hard numbers on exactly what that added assembly cost is, but let's for argument's sake say it's $5. Now that "just $12 extra" pot costs you $27 extra in manufacturing cost.
Then there's the concern about profit margin. Businesses, such as mine, exist for the purpose of making money for their owners and/or shareholders. I like to eat periodically. I would also like to receive a salary that's reflective of my skillset, education, and fair market value. I'm not looking to get rich and famous. If I could make, say, 80% of an engineering salary and have the freedom that comes with self-employment, I'd be happy. I have yet to make it there. I'm not even close, but I'm working to get closer. So... To make money, I have to charge more than cost. Sorry. That's how businesses work. If I charged at or below cost, I would not have a business. I would have a hobby.
How much to charge depends on your brand and your distribution channels. A well-known prestige brand might be able to charge 5-10x cost, or more. An unknown low-end brand with online-only distribution might only charge 2-3x cost. I run a one-man show, so having some distribution would be nice as it would allow my customers to buy and receive my products even if I'm on vacation or otherwise unavailable. Distributors of unknown brands, such as mine, generally charge around 40%. If I use a representative to engage with that distributor, there goes another 5-6%. So if I want to break even, I have to charge roughly twice the production cost. As covered earlier, breaking even is not a good business model, so I'd have to charge more. Suddenly that "only $12 more" pot ends up adding $80-130 to the final price of the product, not counting the added chassis cost and added costs related to shipping and shipping supplies.

That's why a $99 headphone amp cannot contain a $15 pot.

This concludes today's business lecture.

Tom

is there a pot that allows for ±1 dB channel tracking? If there is, how much would it cost?

 

[JJB70]

How is the headstage of the HPA-1 on the HD800S? Most solid state amps I have listened to present music on a flat plane, with little depth, layering or sense of space.

Eh?

 

[solderdude]

is there a pot that allows for ±1 dB channel tracking? If there is, how much would it cost?

Stepped attenuators can do that and perhaps some very expensive pots.
As Tom already mentioned and can be seen on many measurements by Amir the error is already typically smaller than 1dB over the biggest part of the volpot range anyway.
The tested HPA-1 had +/- 0.25dB imbalance over the entire volpot range (80dB) so would not worry about this at all.

 

[Ratatoskr]

Eh?

When you listen to well mastered recordings on a good speaker system you can appreciate the sense of space around the musicians, not only left to right placement but depth. Headphones aren't good at this, the most well known exceptions are the HD800(S) and the Raal SR1a.

On HD800 most solid state amps present little to no depth to the recording. Exceptions have been the Eddie Current Black Widow and the ECP Audio. Eddie Current tube amps are well reviewed and known for a holographic sound stage on headphones. I was merely asking @tomchr if his latest amp actually had depth to the sound stage on phones.

 

[graz_lag]

When you listen to well mastered recordings on a good speaker system you can appreciate the sense of space around the musicians, not only left to right placement but depth. Headphones aren't good at this, the most well known exceptions are the HD800(S) and the Raal SR1a.

On HD800 most solid state amps present little to no depth to the recording. Exceptions have been the Eddie Current Black Widow and the ECP Audio. Eddie Current tube amps are well reviewed and known for a holographic sound stage on headphones. I was merely asking @tomchr if his latest amp actually had depth to the sound stage on phones.

Eh?

 

[amirm]

On HD800 most solid state amps present little to no depth to the recording. Exceptions have been the Eddie Current Black Widow and the ECP Audio. Eddie Current tube amps are well reviewed and known for a holographic sound stage on headphones.

I have listened to some 150 headphone amps if not more. There is never a difference in their soundstage. Soundstage comes from the recording and to some extent from the headphone. Tube amps often obscure detail which actually lowers sense of space between instruments, not help it.

Now, if the tube amp exaggerates highs/distorts heavily, this may be mistaken for slightly larger sense of space. But it is not good space as it is distortion.

 

[tomchr]

Here's the block diagram of the TCA HPA-1.

Tom

 

[jtwrace]

So simple anyone can do it.

 

[tomchr]

Anything is simple if you know how and have the tools to verify the result. My 2.9 GHz spectrum analyzer comes out when I design headphone amps. I've had several high-speed output stages that would happily sing at hundreds of MHz during prototyping. That's the kind of stuff that eats headphones and is often not picked up on an oscilloscope.

First and foremost an amplifier must be stable. Otherwise, you just have a power oscillator (and probably one with crappy phase noise to boot). Once stability is ensured, one can start to worry about audio performance.

Tom

 

[trl]

A balanced input's first and foremost purpose is to make the signal unbalanced again, referenced to the local GND.

I disagree a lot here! Making the signal unbalanced again it's up to manufacturer and the customer will choose what design will prefer based on objective or subjective aspects, depending on his/her knowledge in the "audio field".

We all know that the main reasons for balanced audio systems is that the interconnects are auto-cancelling the noise injected from outside, so this is why all mic cables are balanced. After all, if a headamp accepts balanced inputs then the main reason of balanced operation has been achieved.

I do like a lot the idea of going balanced from the DAC-out till the potentiometer because this will minimise the inside-case noises too, but I don't see another reason than "price increase" to not move further. However, it's probably a very good price/performance ratio stopping in front of the potentiometer with the balanced signals, so probably price/perf. ratio was the reason why Massdrop and Monolith THX headamp manufacturers did the same.

I do have a fully balanced DAC and headamp too and there's nothing references to the GND on the XLR plugs. The signal simply "flows" as V+ & V- down to headphones drivers.

 

[tomchr]

I disagree a lot here! Making the signal unbalanced again it's up to manufacturer and the customer will choose what design will prefer based on objective or subjective aspects, depending on his/her knowledge in the "audio field".

And some of those choices and knowledge may be rooted in science and engineering. Others may be rooted in marketing or snake oil. I prefer to make my design choices based on engineering and science. I'm willing to make some choices to be able to add a bullet to the marketing blurb, but only if such choices don't negatively impact the performance of the equipment (as characterized by measurements). Others are free to disagree or to make different choices.

I do like a lot the idea of going balanced from the DAC-out till the potentiometer because this will minimise the inside-case noises too

This is precisely why the HPA-1 has differential signalling all the way to the volume pot. I placed the differential receiver all they way forward in the chassis and routed the differential traces far away from the power supply (which is the most likely source of noise/interference within the chassis). This reduces the total length of the single-ended part of the signal path to about 4-5 cm. I doubt this arrangement would be quantifiably different from an arrangement with the differential receiver further back in the chassis, but it gave me a nice marketing bullet. Most importantly, this marketing bullet came without impacting performance negatively – and at least a theoretical chance that the performance might be improved. Furthermore, it came at zero cost. The differential receiver had to be somewhere...

The noise reduction of a differential receiver only works if the noise is common-mode, by the way, so you may not always get as much noise reduction as you think.

I don't see another reason than "price increase" to not move further.

I suggest re-reading Post #64. It's never just a price increase. There's also the question of serviceable market. How many do you think care about a differential volume control or even know what it is? How many of those who care would be willing to pay extra to get that feature? Enough to provide a decent positive return on the ~$15k investment in a minimum order of custom volume pots from Alps that I'd have to make to provide that feature?

However, it's probably a very good price/performance ratio stopping in front of the potentiometer with the balanced signals, so probably price/perf. ratio was the reason why Massdrop and Monolith THX headamp manufacturers did the same.

It's not about price/performance ratios. Not the way I think you think anyway. If you ran differentially through the volume pot, you would get higher noise and higher price. So yeah, using a differential volume control does hurt the price/performance ratio. You would get to add a bullet to the marketing materials, but it would come at the price of degraded performance. That's not my idea of a good time.

In my view, what matters is the end-to-end performance. How you get there is secondary. As I've demonstrated in many of my designs, it is perfectly possible for a competent designer to achieve good performance in a single-ended system.

Speaking of competence... I just noticed that I have a bug in the block diagram. The output stage is actually a composite amp with a high-speed, high-current output. I'll fix it and update it in the post above (and on my website).

Tom

 

[Doodski]

@ tomchr

Are you aware that a web search of HPA-1 results in mostly links to the Pass Labs HPA-1? There is also a LD Systems HPA 1