• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Benchmark HPA4 Headphone Amp/Pre

Ron Party

Senior Member
CPH (Chief Prog Head)
Joined
Feb 24, 2016
Messages
411
Likes
559
Location
Oakland
Let's please be a bit more respectful to Mr Siau, he's large part of the engineering excellence for one of the most respected high performance audio electronics manufacturers in the world. He's definitely not the marketing guy, though seems to of drawn the short straw in having to deal with the online audio forums lol

http://www.aes.org/aes/john-siau

It's a absolute privilege to have him and indeed other knowledgeable industry folks participate here so please treat them with the utmost respect and courtesy.

^ This. Thank you, Tom.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Tricked implies deceit. :rolleyes:
It's so much easier to be an expert critic than it is to fund, design, and deliver cutting edge products.

- Rich
Only if you know what I was talking about. It's sad that people like not to listen to the truth.
Let me point something out for you and everyone.
1, Matrix element x has less than 1uV noise. Which means it has less noise LA4/HPA4. Even without taking attenuation into account.
2, Analog volume couples noise and signal together meaning noise is attenuated along the signal. DIgital signal will decrease SNR when attenuate signal because noise doesn't get attenuated. This is very basic and if anyone doesn't know this will easily be thinking into the wrong way.
3, He wasn't deceiving. But it's very misleading. With condition that the benchmark system is very high performance, adding LA4 improves SNR even more implies a few things, LA4 is magical or super high performance piece of gear that does something anyone else doesn't, using non benchmark DAC or power amp will see even more improvements. Which both are not accurate. DAC3 doesn't have the lowest noise in the market. And the fact that DAC3 has high level output makes the noise floow goes up along the signal. That's the reason why he could use LA4 to improve SNR that greatly. And if a higher noise power amp is used, the noise will be always dominated by the power amp, then LA4 won't make any difference. He should have compared to a different preamp or a simple passive attenuator. And probably using 2V/4V output from DAC3, this way the LA4 will only improve a little bit. This is indeed very clever. But to me it's easily misleading.
4, This is not a matter of disrespect. It's about clarification. I don't care who I am. I don't like to see people got misleading messages and then get wrong conclusions. Or if no one listens to me, then great, I'm not the one that's spending money. Surely benchmark is great, but what if next time it isn't benchmark? Having the ability to think independently is important.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
I wish I understood this stuff. I heard it was common knowledge that using a 'normal' preamp circuit is better then using digital control. Something about digital control decreasing resolution for every step of volume you go down, I think. But I've also heard that properly implemented, a digital volume control can be indistinguishable and sound better, and actually improve the sound.

So correct me if I'm wrong, John does work for Benchmark?
That's a mix of different information from different perspective.
Firstly, it needs to be at least a few hundred of uV even 1mV at the power amp output to make audible noise from speakers. Which means 10uV or more input noise is acceptable assuming 26dB gain(which is used by most poweramps). In audibility it's not the ratio of anything that's important but the absolute level of the residue after transducer that's important.
Secondly, in measurements, the analog implementation or hybrid will generally be much better. Unless in some cases the dac has less noise than preamp. But it's still possible to make preamp with even less noise. A simple 10K attenuator will perform just as good as LA4, noise performance wise. But pot will have volume imbalance and wearing issues over time. Relays can also get worn but that's a different story, generally more reliable.

A analog active preamp, attenuates signal and noise at the same time, and then adds some noise. 1uV noise typical , output impedance <1 typical
A digital volume, attenuates signal but noise remain the same. 2uV noise typical, 100ohm typical
A simple Pot, attenuates signal and noise the same time, then adds some output impedance and some noise. 1uV typical , varied output impedance 1kohm typical
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
My comment was spot-on, and you contradict yourself.
No it was rude and obnoxious as were a bunch of others you wisely deleted previously.

Some people don't know things , we are here to help them discover knowledge not just call them names.
 

nemesisrobot

Member
Forum Donor
Joined
Jun 20, 2019
Messages
38
Likes
38
Location
Seattle, WA
I've had this amp for a little over a month now and I absolutely LOVE it. Apart from my favorite headphones, it's my favorite piece of audio gear.

My favorite thing about it is hands down the implementation of the stepped attenuator. My amp before this was the Violectric V281 with the stepped attenuator option and while I did like that amp as well, its implementation of the attenuator is not as good. It has less steps (128), and the pops are audible when changing volume. It's also harder to dial the volume in as it was still controlled using the potentiometer rather than digitally with a rotary encoder.

I wouldn't hesitate to buy another one if I ever had a need for another pre or headphone amp :).
 

ayane

Active Member
Forum Donor
Joined
Dec 15, 2018
Messages
182
Likes
684
Location
NorCal
DAC3 has 9.2uV noise at max volume. HPA4/LA4 has 1.9uV noise. HPA4/LA4 doesn't have the lowest noise possible. It's easily achievable to have less than 0.3uV noise.

I know this is is a little late in the conversation to ask, but how would you achieve lower output noise for an active pre-amplifier, let alone a headphone amplifying circuit?

And on that note, I'm not really convinced that the LA4 has superlative performance for a pre-amplifier. It's pretty impressive for its construction and design, and the stepped attenuator is fantastic, if not a little overkill, but it's possible to achieve an order of magnitude lower output noise in a pre-amplifier than what the LA4 can deliver.

I can't understand what is the source of all that noise in Benchmark's circuits. Coming back to headphone amplifiers, even though the HPA4 is a hair quieter than the Drop/THX 789 in gain 2 (zero gain) mode, it's still pretty noisy compared to to the IEM out of the RME ADI-2 DAC fs (the ADI-2 DAC fs has a significant 5-6 dB reduction in noise via the IEM output compared to the HPA4). Hell, even the JDS labs Atom is a little quieter than the HPA4! While the HPA4 may be the cleanest headphone amplifier on the market in the present in ters of absolute dynamic range, there's definitely room for improvement.

This is related to why I was wondering earlier in the thread if Benchmark is working on a lower-noise successor to the DAC3 (which has over 9 µV of output noise), perhaps using a newer D/A converter chip like the ES9038PRO that the Element X and Okto DAC8 are using (both of which outperform the DAC3 in SNR and SINAD). I'm also curious to see the upcoming Neurochrome HPA-1, the successor to the now discontinued HP-1, because that was another headphone amp that had tremendously low output noise. A headphone amp having <1 µV of output noise guarantees that even the most sensitive of headphones will have a noise floor so low that it is completely inaudible, enough to enjoy a fully audibly transparent headphone system in every use case.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
I know this is is a little late in the conversation to ask, but how would you achieve lower output noise for an active pre-amplifier, let alone a headphone amplifying circuit?

And on that note, I'm not really convinced that the LA4 has superlative performance for a pre-amplifier. It's pretty impressive for its construction and design, and the stepped attenuator is fantastic, if not a little overkill, but it's possible to achieve an order of magnitude lower output noise in a pre-amplifier than what the LA4 can deliver.

I can't understand what is the source of all that noise in Benchmark's circuits. Coming back to headphone amplifiers, even though the HPA4 is a hair quieter than the Drop/THX 789 in gain 2 (zero gain) mode, it's still pretty noisy compared to to the IEM out of the RME ADI-2 DAC fs (the ADI-2 DAC fs has a significant 5-6 dB reduction in noise via the IEM output compared to the HPA4). Hell, even the JDS labs Atom is a little quieter than the HPA4! While the HPA4 may be the cleanest headphone amplifier on the market in the present in ters of absolute dynamic range, there's definitely room for improvement.

This is related to why I was wondering earlier in the thread if Benchmark is working on a lower-noise successor to the DAC3 (which has over 9 µV of output noise), perhaps using a newer D/A converter chip like the ES9038PRO that the Element X and Okto DAC8 are using (both of which outperform the DAC3 in SNR and SINAD). I'm also curious to see the upcoming Neurochrome HPA-1, the successor to the now discontinued HP-1, because that was another headphone amp that had tremendously low output noise. A headphone amp having <1 µV of output noise guarantees that even the most sensitive of headphones will have a noise floor so low that it is completely inaudible, enough to enjoy a fully audibly transparent headphone system in every use case.
For the first question, low noise input buffer then 1kohm pot then low noise output buffer is pretty good. In a headphone amplifier circuit it's basically already existing in a headphone amplifier so basically 1kohm pot can guarantee ultra low noise on the attenuation part. 10kohm pot isn't far off either still under 1uV. Low noise buffer can be easily made from high performance opamps which are really not expensive. With competent design in power supply and pcb, it will perform really well.

Second paragraph, I basically agree with what you said. It's amazing that they made such stepped attenuator.

Then the noise, I believe it's published by thx that thx 888 will have higher noise than 789. You may find a chart of all those models and specs on google.

It may be unfair to speak about 9uV noise from dac3. It's because it has much higher output voltage. It would be 2uV ish or lower with 2V(unbal)/4V(bal) output level. And ahb2 has a very low gain setting which is 9db. In a speaker system this is a very clever design. Where it should actually be very weird that we always have 26db or 30db gain in power amp. In reality we only need 6db to 12db and more gain should be coming from the front end. That's what benchmark did. Higher output from dac and attenuate all the noise using preamp and then they achieved very very low noise at the output of the power amp. Because no one actually always listen at max volume that's impractical. So this method will almost always have benefit.

About other dacs. Element x is really hard to beat in any way. It's just an amazingly engineered product.The new okto pro should also have lower noise than the previous 9028pro model. There is seriously some competition here.

About headphone amplifiers, Tom's new headphone amp is looking really promising. But I don't know much about the noise performance. You may ask him for more detail. I can trust him that it's very low distortion design that's for sure. And it's still in prototyping phase, there can be more changes. Let hope that it beats 789.
 

ayane

Active Member
Forum Donor
Joined
Dec 15, 2018
Messages
182
Likes
684
Location
NorCal
I believe it's published by thx that thx 888 will have higher noise than 789. You may find a chart of all those models and specs on google.
Actually, seems like the THX-888 is slightly lower noise than the THX-788 (and therefore 789) - https://www.thx.com/assets/uploads/2018/02/AAA-Design-Specifications.jpg

It may be unfair to speak about 9uV noise from dac3. It's because it has much higher output voltage. It would be 2uV ish or lower with 2V(unbal)/4V(bal) output level.
Yes, thank you for correcting me. I still do stand by my statement that I think it's possible to push that 2 µV lower. as We've seen in the newer DACs. I'm keeping my eyes peeled for the Okto DAC8 Stereo, heh.

Tom's new headphone amp is looking really promising. But I don't know much about the noise performance. You may ask him for more detail. I can trust him that it's very low distortion design that's for sure. And it's still in prototyping phase, there can be more changes. Let hope that it beats 789.
I'm also hoping for some competition with the HPA4 :D
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Actually, seems like the THX-888 is slightly lower noise than the THX-788 (and therefore 789) - https://www.thx.com/assets/uploads/2018/02/AAA-Design-Specifications.jpg


Yes, thank you for correcting me. I still do stand by my statement that I think it's possible to push that 2 µV lower. as We've seen in the newer DACs. I'm keeping my eyes peeled for the Okto DAC8 Stereo, heh.


I'm also hoping for some competition with the HPA4 :D
I am surprised there isn't 789 here. Two things to note here:
1, 788 is completely different design from 789. Different chips are used. 789 and 888 on the other hand are very similar in design. 788 uses ad8397, 789 and 888 use opa564. (EIDT: confirmed it is opa564) https://cdn.shopify.com/s/files/1/0...ws_-_Nov_Benchmark_HPA4_Reprint_printable.pdf
2, As you can see, the noise is 1.6uV which is higher than thx789. So I think it's 789 that having lower noise than 888. I even remember seeing one table showing 789 has lower noise. But it's more than half a year ago. I could be remembering something wrong. But still 789 has lower noise.
 
Last edited:

ayane

Active Member
Forum Donor
Joined
Dec 15, 2018
Messages
182
Likes
684
Location
NorCal
788 is completely different design from 789. Different chips are used. 789 and 888 on the other hand are very similar in design. 788 uses ad8397, 789 and 888 use opa564. 888 could possibly using something different but according to the photo i looked at the chip has same package as 564 and has mark of opaXXX looks like 564. So possibly different compensation techniques or values choices that makes the difference.
Thank you for correcting me again. I mistakenly thought the 789 was using 788 technology. If you ever find the noise specifications for the 789, please do share them!
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Thank you for correcting me again. I mistakenly thought the 789 was using 788 technology. If you ever find the noise specifications for the 789, please do share them!
Thx 789 performance was published on the THX 789 massdrop/Drop page which is <1uV.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Well the initial question comes back. How did they achieve 445mW at 300ohm with opa564 which can only run on +-12V?
 

Sunship

Member
Joined
Oct 17, 2018
Messages
21
Likes
38
A lot happened since my last post. Back to my original question, the DAC2 and 3 have internal passive jumpers that allow good matching with the rest of the system. I use 96db speakers I enjoy listening to music at low levels. I find that the DAC3 and 2 allow me to listen at those levels while being in a good range of the volume control.

Those passive jumpers are great and separate the Benchmark from many other DACs with digital volume controls that may perform comparably but in reality are les flexible and adaptable to the rest of the system.

I'd be also curious to see a comparison of the analog volume control of the DAC2 and 3 vs the HPA4 as I can imagine the stepped attenuator is far better than the analog controls of the DACs.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
A lot happened since my last post. Back to my original question, the DAC2 and 3 have internal passive jumpers that allow good matching with the rest of the system. I use 96db speakers I enjoy listening to music at low levels. I find that the DAC3 and 2 allow me to listen at those levels while being in a good range of the volume control.

Those passive jumpers are great and separate the Benchmark from many other DACs with digital volume controls that may perform comparably but in reality are les flexible and adaptable to the rest of the system.

I'd be also curious to see a comparison of the analog volume control of the DAC2 and 3 vs the HPA4 as I can imagine the stepped attenuator is far better than the analog controls of the DACs.
If you set the output level low on dac3 using jumpers. You will likely get same performance as LA4. Also depending on what power amplifier you use. You may not reach the limit of DAC3 in the first place. If assuming AHB2, you won't see performance improvement at all from LA4.

In term of other DACs, surely not all paid attention to volume control. But some products like RME adi2dac, Matrix element x, they use combination of digital volume control and step analog control(basically automated jumpers), where you can preserve SNR across the volume without thinking about swapping jumpers.
 

Sunship

Member
Joined
Oct 17, 2018
Messages
21
Likes
38
Yes.
Hence my praise for the DAC2 and 3.
(I am not un-praising the LA4, it's a 'do I need it question?' not a 'is it good?' question)

RME and Matrix look equally great, they just lack an analog input which makes it necessary to add a preamplifier if one uses an analog source.
If you don't, great. If you do, bummer.

(I use a JMF-Audio HQS-6002 power amplifier with an SNR of -122db un-weighted 1khz)
 
Last edited:

laidick

Active Member
Forum Donor
Joined
Apr 22, 2018
Messages
161
Likes
127
Today I got a chance to listen to the HPA4 in a dealer shop located in Hong Kong, however there is no LA4 or AHB2. It just has DAC3 hooked with HPA4.

First thing I notice is, the HPA4 is quite hot, as a preamp it is hotter than I expected.
Another issue is the remote control, as reported by others in this forum, it doesn't not support direct select input. You can only use Left/Right to select input. And the problem is the left/right not able select input 1. As input 1 is specially bonded to the DAC inputs, you have to select "USB/analog/... " Etc to switch to the input 1, and the same time it set the DAC input (if you have one).
This is a bit odd , and the design is really targeting for DAC3 user ...

Otherwise, the preamp itself has a very good control of noise, basically I enjoy a lot when all inputs and volume are set properly.
 

John_Siau

Active Member
Technical Expert
Audio Company
Joined
Jul 17, 2018
Messages
184
Likes
1,407
Location
Syracuse, NY USA
Today I got a chance to listen to the HPA4 in a dealer shop located in Hong Kong, however there is no LA4 or AHB2. It just has DAC3 hooked with HPA4.

Another issue is the remote control, as reported by others in this forum, it doesn't not support direct select input. You can only use Left/Right to select input. And the problem is the left/right not able select input 1. As input 1 is specially bonded to the DAC inputs, you have to select "USB/analog/... " Etc to switch to the input 1, and the same time it set the DAC input (if you have one).
This is a bit odd , and the design is really targeting for DAC3 user ...

The unit you tried was just configured for use with Benchmark DACs having older software. When the DAC has current software, you can scroll through all four inputs on the HPA4.
 

laidick

Active Member
Forum Donor
Joined
Apr 22, 2018
Messages
161
Likes
127
The unit you tried was just configured for use with Benchmark DACs having older software. When the DAC has current software, you can scroll through all four inputs on the HPA4.

Hi John, thanks for pointing this out.

Did you mean the HPA4 has latest current software instead of DAC3?

It's weird if I need to have a DAC3 in order to scroll through the inputs of HPA4.

So instead of scrolling, can we have direct input selection on any remote codes? Such as Logitech Harmony. I expect this would be a very easy change on the HPA4/LA4 software ?

Thanks
 
Top Bottom