Review and Measurements of Massdrop THX AAA 789 Amp

[amirm]

Sorry for barging in like that, but I found the distinction important to make.

No that was fine. My reference was to feature size, not the actual makeup of the process itself. Clearly there are optimized processes for analog which may be superior to ones used for digital.

 

[amirm]

And oh, welcome aboard Tom. I have updated your title.

 

[amirm]

@tomchr It would be nice if you could send some amp of yours to Amir for test. It would be good marketing for you, I guess.

It would be good and it has been done. I will review his headphone amp soon (on the road right now).

 

[badboygolf16v]

One advantage of Tom's amp over the THX AAA is that you can build it yourself, if that takes your fancy...

 

[Grave]

I looked at https://www.neurochrome.com/hp-1-ultra-high-end-headphone-amp/ and the THD+N is 0.0006% or -104 dB, which is excellent, but nothing special, correct? The Topping A30 has a lower THD+N and it is $100.

 

[Sythrix]

I looked at https://www.neurochrome.com/hp-1-ultra-high-end-headphone-amp/ and the THD+N is 0.0006% or -104 dB, which is excellent, but nothing special, correct? The Topping A30 has a lower THD+N and it is $100.

I'll be interested in how it measures, but even from a DIY standpoint ($225 bare minimum for just the board, then add in ~$250 worth of components + $250 for the case, if you don't find and drill your own). You could get an entire setup (albeit, probably not balanced) for the price of the PCB alone, not to mention a completed unit. So maybe the A30 + D30 if you want them separate, or another combo.

It's really cool looking and I would love to build one... but probably not going to even consider it unless Amir measures it as the best amp ever built... Especially considering the THX 789 did so well for $350.

 

[tomchr]

Hi Tom, nice to read you here!

Thank you. It's good to see some familiar faces here too.

Here are Amir measurements on the balanced output:

Thank you for the summary. I did look at those graphs when I skimmed through the thread. Here's my conundrum: Massdrop claims of output power are consistently inflated. For example, they claim "200 mW into 300 ohm" (single-ended). Amir measured 150 mW. Massdrop also claims "6000 mW into 32 ohm" (balanced), but based on Amir's measurements (13.2 V max. voltage swing, 0.5 A current limit), I don't see how they can possibly get more than 4 W into 32 Ω (balanced; P = 0.5^2 * 32/2, current limited). It would be nice with that measurement to confirm. Their claim of "1800 mW into 32 ohm" (unbalanced) is also inflated. Amir measured 1.3 W.

Now, some of this can be explained by Massdrop's use of the 1% THD number rather than the highest output power obtainable without clipping that Amir and I use, but not all of it. It looks to me like Massdrop rounded up and then added precision, which is not a honest spec in my opinion. I do understand that 5.6 rounds up to 6, but 5.6 does not round up to 6.000.

Tom

 

[tomchr]

I looked at https://www.neurochrome.com/hp-1-ultra-high-end-headphone-amp/ and the THD+N is 0.0006% or -104 dB, which is excellent, but nothing special, correct? The Topping A30 has a lower THD+N and it is $100.

Depends... In case of the HP-1, the THD+N is dominated by the +N (i.e. the noise floor of the amp). There are many ways to get a good THD+N number. One such way is to limit the noise bandwidth of the measurement. I use 60 kHz bandwidth as that includes three harmonics of 20 kHz. Unfortunately, it also leads to the integration of more noise, so the THD+N number will be worse than had I used a lower measurement bandwidth. I've noticed that Amir uses 45 kHz bandwidth, which will lead to lower noise, hence better THD+N, so I expect his measurements of the HP-1 to be better than mine in this regard.

One must be careful when comparing specs. This example also shows the dangers of focusing on just one spec. For a headphone amp, I'd look at the integrated noise (20 Hz - 20 kHz) first. Just make sure you're comparing unweighted vs unweighted or A-weighted vs A-weighted. I'd also look at THD vs frequency and make sure the THD at 20 kHz is low. Finally, I'd look at the IMD - in particular the multi-tone IMD as I find that to be more indicative of a positive listening experience (at least for those who like a clean amp).

Tom

 

[tomchr]

@tomchr It would be nice if you could send some amp of yours to Amir for test. It would be good marketing for you, I guess.

That's already in the works, actually. I have plans to ship one of my Modulus-286 kits to Amir once it becomes available, which should be in early December (crossing fingers!) He already has my HP-1 in his review stack.

Tom

 

[amirm]

Depends... In case of the HP-1, the THD+N is dominated by the +N (i.e. the noise floor of the amp). There are many ways to get a good THD+N number. One such way is to limit the noise bandwidth of the measurement. I use 60 kHz bandwidth as that includes three harmonics of 20 kHz. Unfortunately, it also leads to the integration of more noise, so the THD+N number will be worse than had I used a lower measurement bandwidth. I've noticed that Amir uses 45 kHz bandwidth, which will lead to lower noise, hence better THD+N, so I expect his measurements of the HP-1 to be better than mine in this regard.

Hi Tom. My dashboard measurements of SINAD are actually at 22.5 kHz. Some of the other measurements use a wider bandwidth. So if yours are at 60 kHz, then my SINAD numbers will certainly be a lot better.

 

[Grave]

Why include the noise at >20 KHz?

 

[tomchr]

For SINAD, I use the 20 kHz AES17 filter, so our numbers should be comparable. It's the THD+N vs frequency where I use 60 kHz BW (I use the elliptical filter in the input setup).

Tom

 

[tomchr]

Why include the noise at >20 KHz?

I measure THD+N vs frequency from 20 Hz to 20 kHz. The predominant harmonics are the second and third, so I have standardized on measuring THD+N vs frequency at 60 kHz bandwidth as this will ensure that the second and third harmonics of 20 kHz are included.
I do not have the option to include the harmonics but not the noise with my current setup. I may in the future be able to provide a good FFT measurement that shows the harmonics with the fundamental subtracted, but that's a future project. It'll require some external hardware, which given the nature of low distortion measurements is not trivial to design.

You can argue that human hearing ends at about 20 kHz (at age 7 or so; and lower after that), so why should anything above 20 kHz be included. That's a fair argument. That just means you don't get to see THD or THD+N past 10 kHz (if you include only the second harmonic) or 6.67 kHz (if you include both the second and third harmonic). That's the approach taken by Class D manufacturers as their amps often have switching frequencies just above the audible spectrum. I find the THD+N in the 5-20 kHz range to be indicative of the quality of the PCB layout (low THD at 20 kHz requires that you pay attention to the PCB parasitics in the layout process), so I like to see it.

Tom

 

[Grave]

https://en.wikipedia.org/wiki/THX

So, it was started by George Lucas, then it was owned by Creative (not a good brand IMO) and now it is owned by Razer (also not a good brand IMO).

 

[wadec22]

https://en.wikipedia.org/wiki/THX

So, it was started by George Lucas, then it was owned by Creative (not a good brand IMO) and now it is owned by Razer (also not a good brand IMO).

Creative and E-MU have always had good tech, just poor software

 

[westyjeff]

Quick question, even if the specifications are overstated by massdrop does that take away from the overall performance this amplifier provides? I am thoroughly enjoying mine, headroom for days.

 

[AnalogSteph]

Quick question, even if the specifications are overstated by massdrop does that take away from the overall performance this amplifier provides?

Not at this kind of level, of course. Performance is still exemplary whatever way you look at it.

 

[tomchr]

Quick question, even if the specifications are overstated by massdrop does that take away from the overall performance this amplifier provides?

Not at all. It does say something about the vendor/manufacturer, though.

Tom

 

[Sythrix]

Thank you for the summary. I did look at those graphs when I skimmed through the thread. Here's my conundrum: Massdrop claims of output power are consistently inflated. For example, they claim "200 mW into 300 ohm" (single-ended). Amir measured 150 mW. Massdrop also claims "6000 mW into 32 ohm" (balanced), but based on Amir's measurements (13.2 V max. voltage swing, 0.5 A current limit), I don't see how they can possibly get more than 4 W into 32 Ω (balanced; P = 0.5^2 * 32/2, current limited). It would be nice with that measurement to confirm. Their claim of "1800 mW into 32 ohm" (unbalanced) is also inflated. Amir measured 1.3 W.

Now, some of this can be explained by Massdrop's use of the 1% THD number rather than the highest output power obtainable without clipping that Amir and I use, but not all of it. It looks to me like Massdrop rounded up and then added precision, which is not a honest spec in my opinion. I do understand that 5.6 rounds up to 6, but 5.6 does not round up to 6.000.

Quick question, even if the specifications are overstated by massdrop does that take away from the overall performance this amplifier provides?

We should probably be asking the actual THX engineer who was here earlier to respond to this. I don't know if he's continuing to monitor this thread, but there are valid questions here that should be addressed. While the overall performance as stated, is exemplary, I am also curious if there's something we haven't looked at yet. Everyone seems to have a different method for measurements and best practices, making it difficult to establish a baseline.

@AndrewMason

Do you have any comments that could help us clarify between Massdrop's final specifications and the measurements here, keeping in mind Tom's stated concerns?

Not at all. It does say something about the vendor/manufacturer, though.

I think we should hear both sides of the story before we start insinuating ill intent or negligence.

 

[tomchr]

I'm not insinuating anything. All I'm saying is that the product does not appear to meet the published specs and I would like to understand why it doesn't. I have provided a guess to what might have caused this, but it is a guess and that's all.

As for your question regarding what hasn't been looked at yet: I would like to see more measurements of IMD. Specifically, SMPTE (60 Hz + 7 kHz @ 4:1), 18+19 kHz @ 1:1, 1+5.5 kHz @ 1:1, and the multi-tone IMD residual. The 32-tone test signal that comes with the AP software makes the latter pretty easy to measure. I'd like to see the harmonic composition of the IMD measurements rather than an IMD vs output power measurement.

Tom