• 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 Hypex NC400 DIY Amp

landco

Member
Joined
May 7, 2019
Messages
69
Likes
7
Location
Russia
That is a good discussion to have, but totally unrelated to Hypex, class D or anything else in the thread topic.
A user with a nick ironhorse128 asked me about the alternative to chips from HYPEX, I answered.
From my point of view, this discussion makes sense because the people in this topic discuss things that have no practical (audible) meaning.
 

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
Does anybody know the exact power stage topology of the NC400? Is it 2 transistor half-bridge or a 4-transistor full-bridge? The datasheet indicates that it will accept a single-ended RCA input, which raises my suspicion that it is a simple half-bridge. The ICEPOWER 700AS1 mono module is a full-bridge that will only accept a balanced input signal. I'm currently considering both but aiming towards the 700AS1 due to its native balanced amplification, which can only be implemented by a full-bridge topology.

Thanks
 

JimB

Addicted to Fun and Learning
Forum Donor
Joined
Sep 19, 2019
Messages
731
Likes
493
Location
California
Does anybody know the exact power stage topology of the NC400? Is it 2 transistor half-bridge or a 4-transistor full-bridge? The datasheet indicates that it will accept a single-ended RCA input, which raises my suspicion that it is a simple half-bridge. The ICEPOWER 700AS1 mono module is a full-bridge that will only accept a balanced input signal. I'm currently considering both but aiming towards the 700AS1 due to its native balanced amplification, which can only be implemented by a full-bridge topology.

Thanks
Hypex recommends balanced inputs for the NCore amps. I believe they are true differential inputs. One can always feed a single ended ouput into a differential input. On pages 17-19 of the NC400 datasheet, Hypex provides advise on how best to do that when one must.
 

BobPM

Active Member
Joined
Dec 20, 2019
Messages
101
Likes
108
Location
Driftwood, Texas
NC400 DIY or NC500 (base model from Nord)? Lower distortion or more power? Trying to decide between these two amps and wonder which factor is better to emphasize for a pair of inefficient 86 db 4 ohm speakers. So what do you think, go for more power or the lowest distortion?
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands
Does anybody know the exact power stage topology of the NC400? Is it 2 transistor half-bridge or a 4-transistor full-bridge? The datasheet indicates that it will accept a single-ended RCA input, which raises my suspicion that it is a simple half-bridge. The ICEPOWER 700AS1 mono module is a full-bridge that will only accept a balanced input signal. I'm currently considering both but aiming towards the 700AS1 due to its native balanced amplification, which can only be implemented by a full-bridge topology.

It has a fully differential input stage, but the rest of the circuit topology is not fully symmetrical (not that it matters). I think you have the logic the wrong way - a fully balanced/differential input can be driven by an unbalanced signal, while one that is only "sort of" balanced, but not truly differential, can't.
 

Zoomer

Senior Member
Joined
Oct 5, 2019
Messages
323
Likes
468
NC400 DIY or NC500 (base model from Nord)? Lower distortion or more power? Trying to decide between these two amps and wonder which factor is better to emphasize for a pair of inefficient 86 db 4 ohm speakers. So what do you think, go for more power or the lowest distortion?

IMO the difference in power far outweighs the difference in distortion from these modules. So I'd focus on the power requirements. There's a recent thread about finding out how much power you really need. The Purify would give you a bit more power than the NC400 with even lower distortion.
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands
NC400 DIY or NC500 (base model from Nord)? Lower distortion or more power? Trying to decide between these two amps and wonder which factor is better to emphasize for a pair of inefficient 86 db 4 ohm speakers. So what do you think, go for more power or the lowest distortion?

The power difference is pretty minimal, but so is the distortion difference, so I think the best answer is "doesn't really matter". :)
 

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
Does anybody know the exact power stage topology of the NC400? Is it 2 transistor half-bridge or a 4-transistor full-bridge? The datasheet indicates that it will accept a single-ended RCA input, which raises my suspicion that it is a simple half-bridge. The ICEPOWER 700AS1 mono module is a full-bridge that will only accept a balanced input signal. I'm currently considering both but aiming towards the 700AS1 due to its native balanced amplification, which can only be implemented by a full-bridge topology.

Thanks



OK, got a reply from Hypex, the NC400 is based on a simple 2-transistor half-bridge topology, which means it is natively a single-ended amplifier. This is very disappointing. There is really no point for a balanced input in that case. There are many advantages to using a full-bridge topology. One important property is the fact that the effective PWM switching frequency doubles at the differential output, making the output filter more efficient.. Only a full-bridge class-d is a true balanced amplifier.


This really made it easier for me to choose the Icepower 700AS1 modules instead.
 

Attachments

  • NC400_reply.jpg
    NC400_reply.jpg
    77.9 KB · Views: 173

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands
OK, got a reply from Hypex, the NC400 is based on a simple 2-transistor half-bridge topology, which means it is natively a single-ended amplifier. This is very disappointing. There is really no point for a balanced input in that case. There are many advantages to using a full-bridge topology. One important property is the fact that the effective PWM switching frequency doubles at the differential output, making the output filter more efficient.. Only a full-bridge class-d is a true balanced amplifier.

I think you are mixing/confusing symmetric topology and balanced/differential inputs/outputs/connections. They are two very different things, and pretty much unrelated.

can you please explain how "the effective PWM switching frequency doubles at the differential output, making the output filter more efficient"?
 

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
I think you are mixing/confusing symmetric topology and balanced/differential inputs/outputs/connections. They are two very different things, and pretty much unrelated.

can you please explain how "the effective PWM switching frequency doubles at the differential output, making the output filter more efficient"?


First, I'm not just making assumptions, I happen to have a PhD degree in electrical engineering. The best known way to use a full-bridge topology is to use a differential balanced input signal. This means that any single-ended signal is first phase-split into two out of phase signals. Each signal gets modulated using one half-bridge leg of the full-bridge topology. The output filter and speaker is connected between the two half-bridge legs of the full-bridge power stage.

So if each half-bridge leg is modulated at a switching frequency of Fs, the full-bridge differential output will have an effective modulation frequency of exactly 2Fs

fulbidge.jpg
fulbidge2.jpg



As you can see in the attached images, not only that the effective switching frequency at the differential output doubles, the number of PWM levels increases from 3 levels to 3 effective levels.

Both doubling the switching frequency and increasing the number of output levels have positive effects on the sound quality, as the harmonic distortion decreases by more than 50%.


Instead of taking a single-ended audio signal and splitting it into differential, the best practice for full-bridge class-d is to feed it with a balanced audio signal, making the full-bridge class-d amp a native and true balanced amplifier


This is exactly how the Iceppwer AS series works. If Amirm could measure and review the Icepower 700AS1 modules, I'll bet that it will outperform even the purifi's 1ET400A , which BTW is also based on a single-ended topology.


Eli
 

Zoomer

Senior Member
Joined
Oct 5, 2019
Messages
323
Likes
468

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands
I happen to have a PhD degree in electrical engineering.

Excellent!

As you can see in the attached images

Out of curiosity, what is the source of those images?

This is exactly how the Iceppwer AS series works. If Amirm could measure and review the Icepower 700AS1 modules, I'll bet that it will outperform even the purifi's 1ET400A , which BTW is also based on a single-ended topology.

I hope you don't mind me asking, but do you have any affiliation, commercial or otherwise, with ICEpower?
 

rajapruk

Senior Member
Forum Donor
Joined
Sep 22, 2018
Messages
472
Likes
489
Does anybody know the exact power stage topology of the NC400? Is it 2 transistor half-bridge or a 4-transistor full-bridge? The datasheet indicates that it will accept a single-ended RCA input, which raises my suspicion that it is a simple half-bridge. The ICEPOWER 700AS1 mono module is a full-bridge that will only accept a balanced input signal. I'm currently considering both but aiming towards the 700AS1 due to its native balanced amplification, which can only be implemented by a full-bridge topology.

Thanks

You can bridge 2 nc400 togheter. Then you get ”full-bridge”, I think. And theoretically some more distortion cancellation due to fully balanced operation, maybe?
I did that once upon a time, so it is possible. There are guidelines from Hypex how to do it. An extra capacitator should be added between some speakerterminals.
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
Excellent!



Out of curiosity, what is the source of those images?


I hope you don't mind me asking, but do you have any affiliation, commercial or otherwise, with ICEpower?

No, not at all. I'm an academic staff member at Ben Gurion University in Israel. Our department used some of their modules for some projects. This is how I've learned about Icepower and bought two 125ASX2 modules that I currently use as dual mono amplifiers in BTL mode. Now I'm looking for an upgrade, and so far narrowed it down to the NC400 and the 700AS1, which is used in PS Audio Stellar 700M monoblocks. I'm also considering purifi's 1ET400A. Could a single-ended half-bridge sound better than a full-bridge class-d? Yes, it could. But there is no way to know without comparison.


Source for the images:

D. G. Holmes, Thomas A. Lipo, Pulse Width Modulation for Power Converters Principles and Practice, IEEE Press., 2003.
 

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
You can bridge 2 nc400 togheter. Then you get ”full-bridge”, I think. And theoretically some more distortion cancellation due to fully balanced operation, maybe?
I did that once upon a time, so it is possible. There are guidelines from Hypex how to do it. An extra capacitator should be added between some speakerterminals.


For best harmonic performance, two single-ended output stages should be bridged BEFORE the output low pass filter, not after. Nothing beats a true full-bridge topology.
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
3,004
Likes
3,998
Location
Amsterdam, The Netherlands
For best harmonic performance, two single-ended output stages should be bridged BEFORE the output low pass filter, not after. Nothing beats a true full-bridge topology.

Could we please keep the 3 separate things - bridge topology, symmetric topology, and balanced/differential inputs/outputs - apart?
 

boXem

Major Contributor
Audio Company
Joined
Jun 19, 2019
Messages
2,014
Likes
4,852
Location
Europe
First, I'm not just making assumptions, I happen to have a PhD degree in electrical engineering. The best known way to use a full-bridge topology is to use a differential balanced input signal. This means that any single-ended signal is first phase-split into two out of phase signals. Each signal gets modulated using one half-bridge leg of the full-bridge topology. The output filter and speaker is connected between the two half-bridge legs of the full-bridge power stage.

So if each half-bridge leg is modulated at a switching frequency of Fs, the full-bridge differential output will have an effective modulation frequency of exactly 2Fs

View attachment 45276View attachment 45277


As you can see in the attached images, not only that the effective switching frequency at the differential output doubles, the number of PWM levels increases from 3 levels to 3 effective levels.

Both doubling the switching frequency and increasing the number of output levels have positive effects on the sound quality, as the harmonic distortion decreases by more than 50%.


Instead of taking a single-ended audio signal and splitting it into differential, the best practice for full-bridge class-d is to feed it with a balanced audio signal, making the full-bridge class-d amp a native and true balanced amplifier


This is exactly how the Iceppwer AS series works. If Amirm could measure and review the Icepower 700AS1 modules, I'll bet that it will outperform even the purifi's 1ET400A , which BTW is also based on a single-ended topology.


Eli
You are talking about BD modulation, which is different from full bridge. Full bridge is needed for BD modulation. It has its own advantages and disadvantages.
Hypex/Purifi implement AD modulation. It can be implemented as half or full bridge, with it's own advantages and disavantages.

The scheme you link is for "classical" class D amplifiers, i.e. not self oscillating. The weakness of classical amplifiers is that they do not control what's happening at the filter, itself source of distortions and having an output impedance increasing with frequency. Moreover, the filter will resonate with the load at some frequencies.
The problem of the output filter is that at some point, it's phase is close to 180°. Controlling what's happening after the filter is thus quite difficult. Here comes the idea of using the filter to voluntary get the loop oscillating, and using these oscillations for the signal modulation. That's how UcD and it's successors NCore and Eigentakt are working.

Here are some papers explaining this much better than I do, with some math that should not be a problem for a PhD:
Simple Self-Oscillating Class D Amplifier with Full Output Filter Control
NCore technology whitepaper
Globally Modulated Self-Oscillating Amplifier with Improved Linearity
And a course on class D:
AES Class D master class

For best harmonic performance, two single-ended output stages should be bridged BEFORE the output low pass filter, not after. Nothing beats a true full-bridge topology.
Any data supporting your claims? People tend to like them here.
 

EB1000

Senior Member
Joined
Jan 8, 2020
Messages
484
Likes
579
Location
Israel
You are talking about BD modulation, which is different from full bridge. Full bridge is needed for BD modulation. It has its own advantages and disadvantages.
Hypex/Purifi implement AD modulation. It can be implemented as half or full bridge, with it's own advantages and disavantages.

The scheme you link is for "classical" class D amplifiers, i.e. not self oscillating. The weakness of classical amplifiers is that they do not control what's happening at the filter, itself source of distortions and having an output impedance increasing with frequency. Moreover, the filter will resonate with the load at some frequencies.
The problem of the output filter is that at some point, it's phase is close to 180°. Controlling what's happening after the filter is thus quite difficult. Here comes the idea of using the filter to voluntary get the loop oscillating, and using these oscillations for the signal modulation. That's how UcD and it's successors NCore and Eigentakt are working.

Here are some papers explaining this much better than I do, with some math that should not be a problem for a PhD:
Simple Self-Oscillating Class D Amplifier with Full Output Filter Control
NCore technology whitepaper
Globally Modulated Self-Oscillating Amplifier with Improved Linearity
And a course on class D:
AES Class D master class


Any data supporting your claims? People tend to like them here.


You can self-oscillate both half and full-bridge topologies. Both BD and AD approaches can be combined with self-oscillation:

https://core.ac.uk/download/pdf/9703959.pdf


As for my other claim about bridging before the filter, it is based on common sense and power converters therapy, which you can read here:

D. G. Holmes, Thomas A. Lipo, Pulse Width Modulation for Power Converters Principles and Practice, IEEE Press., 2003.


The higher the switching frequency, the lower the load's current THD. The higher the switching frequency fed to the low pass output filter, the lower the harmonic distortion at its output because the cutoff frequency of the filter shifted to a lower frequency with respect to the switching frequency. You can easily verify this claim by simulation. You can self-oscillate a full-bridge class-d amp by using two separate control feedback loops, one for each half-bridge leg with two input source signals (positive and negative balanced input).


The full-bridge class-d, when implemented correctly, should hade better performance compared to a half-bridge class-d with the same control strategy, same switching frequency range, and the same output filter.

BD-modulation (full-bridge only) could be either filter-free or filtered, while AD-modulation (half-bridge only) must be filtered. This is also an indication that full-bridge is alway superior, regardless of the modulation technique used (constant open loop PWM, or self-oscillating closed-loop).



file:///D:/Downloads/1-38.pdf
 
Top Bottom