• 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!

Hegel H95 Review (Streaming Amplifier)

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
1) All the DAC measurements are in our mind misleading. They were measured at a level where the amplifier is clipping. In our view, they should be dismissed completely.

1.1) There has been some debate on how we can allow the RCA pre out to be optimized at 550mV, which actually is a fair question. The reasoning behind this is that we mainly see this as a subwoofer connection. Obviously you can debate wether that is a good or a bad choice.

When the pre amp output is more than 550 mV, the speaker output will be clipping and that will influence the performance of the pre out.

The pre-out on the H95 is taken directly from the power amplifiers input stage. See figure below:

View attachment 168201
When the power amplifier starts clipping, this will reflect in the RCA pre out as well. As such, the RCA pre out can NOT be used as a 2V DAC output. When we measure the DAC performance we connect Audio Precision directly to the output stage of the DAC board. You can of course also measure the DAC on the RCA output, but with a lower setting than 98 and it will include factors from the preamp.

index.php


That explanation makes perfect sense.
We can see the 'clipping alike' behavior we see is also seen in the knee at -11dB which not surprisingly calculates to 0.56V.

Indeed, as @KSTR also mentioned, the DAC measurements are not correct and would advice @amirm to either re-measure at 550mV as 0dB or remove the line-out measurements or make a note stating that the line-out is 550mV FSD instead of the usual 2V.

Now... below 560mV the performance isn't great either though and about similar in performance to the cheap Razor USB dongle.

And as you say the line-out with 0,55V is a strange decision. Why not add a simple opamp with say 4x gain and line-out would have complied more to current line-levels instead of 5-pole DIN plug line level values.
Even when using the outs for subs (could have called it sub-out ?) the 2V would still make sense.

Another thing.. why is there NOTHING in the sparse data sheet about the line output and headphone out level/impedance ?

Perhaps in the future create a bit better specifications ? This wcould have prevented the DAC measurements made on false premises that line-out would be a rather 'standard' 2V.

Devices look nice.. maybe get your specifications in order as well ?
 
Last edited:

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,811
Location
Netherlands
We promised our own findings today. I will therefore present what we disagree on and why.
Finally some decent feedback from your side. That is really appreciated!
1) All the DAC measurements are in our mind misleading. They were measured at a level where the amplifier is clipping. In our view, they should be dismissed completely.
How can the amplifier be clipping even with the lower level measurements? Certainly, the IMD sweep is representative up to the clipping point?
When we measure the DAC performance we connect Audio Precision directly to the output stage of the DAC board. You can of course also measure the DAC on the RCA output, but with a lower setting than 98 and it will include factors from the preamp.
Once more the question: isn't that what the IMD sweep already did?
Actually we used to run the output from the preamp at a level where it never really started clipping (also earlier models), but on a 60w/pc amplifier consumers did not like to have the volume control at “98” to play loud. And what if you played an old Pink Floyd piece, with low gain… it would never be loud enough.
So the clipping is to make sure that lower gain digital sources can still reach proper levels? Sounds sensible. Most AVR's also clip before they search maximum volume.
2) The stability measurement THD + N from ASR claims that the amplifier never reaches stability after warm up.

This measurement is, in our view, wrong. This is how it should look (our own test on AP 2722 today)

View attachment 168202
We have only almost managed to recreate the ASR measurement if we wiggle the speaker cables in the H95 terminals or tap the output relays.

We disagree with the comment about the power amplifier not being stable. The Hegel output stage is a no-feedback design, and since the distortion characteristics of the power transistor change somewhat with temperature, the total distortion will have a small increase with temperature. In a high open loop gain feedback-design this effect will be much smaller.
How can the measurements be wrong? From what I see here, you just have different measurements with a different sample and different circumstances. We can only guess as to the cause of @amirm's findings. He has done these many times, and it seems straightforward to do.
 

BossBunos

Active Member
Joined
Oct 17, 2021
Messages
123
Likes
140
A shame the H95 performs so bad. I read alot of arguments tough that you are better off buying a chinese amp etc.
That's only valid when you buy the amp purely for performance (on paper; I doubt alot of people can actually tell the difference in sq when doing a/b).
I own a Hegel and besides it sounded good to me the most important reasons to go for the hegel was the smart home integration for example and the clean look. I can't buy gear and give instructions to my wife with what source she has to select etc when she wants to watch tv or listen to music.
Another thing is that it has a streamer and starts up in 5 seconds. I also own the minidsp shd and it bothers me that it takes 30+ seconds before i can stream music when i turn it on from stand by.

At least I couldn't find any other integrated amp with these features at the time. (yes I know the H95 does not have IP control but I own the rost)
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Another thing.. why is there NOTHING in the sparse data sheet about the line output and headphone out level/impedance ?

Perhaps in the future create a bit better specifications ? This wcould have prevented the DAC measurements made on false premises that line-out would be a rather 'standard' 2V.

Devices look nice.. maybe get your specifications in order as well ?
It is interesting to see that Hagel Music Systems liked this rather critical post.
 

JiiPee

Active Member
Joined
Oct 23, 2021
Messages
252
Likes
482
First, I like to thank Hegel for the response. Not all manufacturers would make this kind of effort.

Second, I think what we have here is an example of a more general topic: Is it preferable to use strong global feedback to get better SINAD figures, or not. Everyone has their own opinion about this, and my point of view is that one should go and listen the amplifier to make one's own judgement.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
It is interesting to see that Hagel Music Systems liked this rather critical post.

Not really given the comment Hegel gave:
1.1) There has been some debate on how we can allow the RCA pre out to be optimized at 550mV, which actually is a fair question. The reasoning behind this is that we mainly see this as a subwoofer connection. Obviously you can debate wether that is a good or a bad choice.

I understand the reasoning from Hegel as seen from their own systems. They just do it differently from other brands/conventions.
As said a more elaborate set of specifications is highly desired and can't be that hard to do and prevents incorrect usage.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
How can the amplifier be clipping even with the lower level measurements? Certainly, the IMD sweep is representative up to the clipping point?

The reason is the output of the DAC section at the point the line-out is splitted will also load the DAC and the input of the amp will not be linear anymore and 'clip' the output of the DAC section somewhat.

They could have prevented this with a buffer before the power amp or use a series resistor for instance.
It won't ever be a problem when using Hegel amps/SW as they will clip themselves at the 550mV levels.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,592
Likes
10,728
Location
Prague
It makes extremely high sense to keep 2V as a link level for SE signal transfer. The higher the voltage level, the higher is S/N ratio between useful signal and interference/ noise voltage drop across cable shield ground return caused by loop or capacitive currents. There is no reason to decrease this level with current opamps and there is no reason not to use a dedicated opamp for line output.
 

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,752
Likes
4,633
Location
Sweden - Слава Україні
Are you sitting down? Hold on to something now. This will sound incredible but I promise it's true. I have come across those who take it for granted that if they buy a DAC, it should be possible to plug the DAC into an amplifier without having to plow through meter-thick technical manuals, to read about signal strengths, ohms, etc. That without with a lot of settings. They called it plug and play. Wow, have you heard anything so strange?
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,811
Location
Netherlands
Second, I think what we have here is an example of a more general topic: Is it preferable to use strong global feedback to get better SINAD figures, or not. Everyone has their own opinion about this, and my point of view is that one should go and listen the amplifier to make one's own judgement.
Sure, as long as it's double-blind.

As for the feedback discussion, I think it's rather nonsense. Given properly low output impedance, linear response and enough bandwidth, the major remaining objective quality factor is distortion. How you solve this problem is rather irrelevant. Sure some methods have advantages over others, but often things get oversimplified. In the end, the final result counts. All the rest is marketing!
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,229
Likes
17,811
Location
Netherlands
The reason is the output of the DAC section at the point the line-out is splitted will also load the DAC and the input of the amp will not be linear anymore and 'clip' the output of the DAC section somewhat.

They could have prevented this with a buffer before the power amp or use a series resistor for instance.
It won't ever be a problem when using Hegel amps/SW as they will clip themselves at the 550mV levels.
That why I said: up to the clipping point!
 

caught gesture

Senior Member
Forum Donor
Joined
Feb 19, 2021
Messages
444
Likes
936
Location
Italia
One more thing that I forgot. Yesterday we linked to Paul Millers (HiFi News) lab report in our post #101. His results looked more like our own. We were obviously not the first as it was presented earlier in post #59, and then @amirm dismissed Paul in post #62 insinuating he was a “has been” and asked how Paul could be satisfied with 86dB signal to noise ratio. First of all, Paul wasn’t happy with 86dB. He was happy with 100,9dB, if you look closer. Second, Paul is an expert at measuring with 40 years of experience and one of the few in this industry who has actually designed and built high performance measurement equipment.
This is disingenuous. He did not say anything about Paul Miller being a “has been”. He wrote, “Paul Miller is not who he used to be. He used to have tons of graphs with pass/fail criteria. Now he does the minimum to let the whole "review" essentially by a subjective piece” which quite clearly points to a difference in how his reviews are presented now compared to before. I don’t know why you keep having to make snide comments. It really does you no favours. Just let your facts stand or fail on their own merit. You will be respected more.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
What's a bit strange here is that the DAC itself obviously did not reach 0dBFS at 550mV but rather was able to put out 2V (as can be seen from the measurements by Amir) but the output voltage of the DAC itself is 'loaded' a bit by the poweramp causing the signal to flatten a bit above 550mV and higher distortion above 550mV.

Its a strange way to use a DAC (having 11dB of unused headroom)... buffering would have helped.
 

tw99

Senior Member
Joined
Mar 13, 2019
Messages
467
Likes
1,069
Location
West Berkshire, UK
A shame the H95 performs so bad. I read alot of arguments tough that you are better off buying a chinese amp etc.
That's only valid when you buy the amp purely for performance (on paper; I doubt alot of people can actually tell the difference in sq when doing a/b).
I own a Hegel and besides it sounded good to me the most important reasons to go for the hegel was the smart home integration for example and the clean look. I can't buy gear and give instructions to my wife with what source she has to select etc when she wants to watch tv or listen to music.
Another thing is that it has a streamer and starts up in 5 seconds. I also own the minidsp shd and it bothers me that it takes 30+ seconds before i can stream music when i turn it on from stand by.

At least I couldn't find any other integrated amp with these features at the time. (yes I know the H95 does not have IP control but I own the rost)

Nobody really needs to justify what they like or don't like, but anyway these sound like perfectly good reasons to me. For more complex products like these, Amir's testing still only looks at some basic parameters of the DAC or Amp, and hardly considers useability, features or the total package at all. And still people love to do the pile-on to insult the ability or ethics of the designers/retailers/manufacturer because of an issue with some measurement that no-one, or almost no-one, can hear under normal circumstances...

So even if some measurements may be a bit below SOTA, the whole package can still make sense, but you won't find any of that detail in a review here.
 

aj625

Senior Member
Joined
Aug 31, 2021
Messages
325
Likes
226
It was double blind, level matched, amps operating well below their max power, ABX switch.

45 min to listen as long as I wanted to each one, switch back and forth as much as I wanted, use whatever songs I wanted.
What other amp was with ahb2 ?
 

Ismapics

Active Member
Joined
Nov 21, 2020
Messages
213
Likes
285
Darko is too old to be a hipster. He is pushing 50. Not sure what his full time job is, though. Maybe he makes his entire living reviewing gear. However, that would be a rarity in the audio world.
Yes its his business that is why what he says comes tainted that way. Take his love for the Dragonfly DACs, which have consistently performed no better than the $9.99 Apple dongle. To me he is just another one clinging to likes and views so he can sell it. There are plenty there, remember "specs dont matter".
 

Lambda

Major Contributor
Joined
Mar 22, 2020
Messages
1,785
Likes
1,519
It makes extremely high sense to keep 2V as a link level for SE signal transfer. The higher the voltage level, the higher is S/N ratio between useful signal and interference/ noise voltage drop across cable shield ground return caused by loop or capacitive currents. There is no reason to decrease this level with current opamps and there is no reason not to use a dedicated opamp for line output.
I would even say it could be increased for "high end HiFi" gear.
For mobile and dongles that operate from a single 5V supply rails 2V it maybe ok but i don't see any reasoning why we shuld not send 10Vrms over interconnects.

@Hegel Music System
Since your company seem to also have an AP analyzer why don't you do exactly the same measurements but with the "right" amplifier settings to show us all how mush better your product can perform if you use it right and how wrong @amirm measurements are?
 

Ismapics

Active Member
Joined
Nov 21, 2020
Messages
213
Likes
285
He knows. Look at his lifestyle. His is living the hipster dream by reviewing audio gear.
I don't blame him either.
Most people have a job that is very compromised and the way most folks make a living involves significant BS. Fault no one particular cat.
Don't put Darko's head on a pike unless you are earing a very cleaning living doing...?
I am quite ok with my life though. That does not discount that he is just a talking head, claiming to have golden ears. remember that the brands need a place to unload all the A&P that used to go to more shows and print media. He just wants to have the brands happy so they buy ad space. Still no credibility,
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Thank you for the reply even though I find it pretty lame and in no way a justification on why the measurements are bad. One of your excuses is the amplifier sounds good. Good luck with that but I am not going to comment on it on an objective forum like ASR.

1) All the DAC measurements are in our mind misleading. They were measured at a level where the amplifier is clipping. In our view, they should be dismissed completely.

When the pre amp output is more than 550 mV, the speaker output will be clipping and that will influence the performance of the pre out.
When the power amplifier starts clipping, this will reflect in the RCA pre out as well.
What I understand is the measurements are misleading because @amirm had increased the level above 98 even if any owner can and that you have not put a notice or limit it.

When we measure the DAC performance we connect Audio Precision directly to the output stage of the DAC board. You can of course also measure the DAC on the RCA output, but with a lower setting than 98 and it will include factors from the preamp.
In other words: we know the power amplifier clipping issue hence we do not measure the DAC at the line outs.

When the power amplifier starts clipping, this will reflect in the RCA pre out as well.
ASR readers, do you remember the same issue with the Denon AVRs?

Notes on why;
We used to have a lower output level on the RCA Pre out (earlier models), but since we have 32 dB gain in our power amplifiers we had to raise it to allow for easier subwoofer integration.

Actually we used to run the output from the preamp at a level where it never really started clipping (also earlier models), but on a 60w/pc amplifier consumers did not like to have the volume control at “98” to play loud. And what if you played an old Pink Floyd piece, with low gain… it would never be loud enough.
Here comes the crunch: What he is saying is that they have changed the design but haven't realised that it fooked up the line outputs. Now we all know that it is the consumer's fault, we shouldn't want to play "an old Pink Floyd piece!"
 
Last edited:
Top Bottom