Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Class D Switching Power Amplifiers and Power D/A conversion

Phase shift in class D amplifiers. How it affects the sound?
Phase shift in class D amplifiers. How it affects the sound?
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 4th May 2018, 08:57 PM   #21
NielsMayer is offline NielsMayer
diyAudio Member
 
Join Date: Jun 2010
See also Hypex Ncore NC400 - input anti-alias filter?
Ferrite bead effect on Class-D amplifier audio quality
Quote:
As can be seen from Fig. 9, the ferrite bead deteriorates
the signal THD in the amplitude range [10 mV - 7 V ]. Ferrite
beads have a negative impact on the audio signals even though
the component current in the range of [1 mA - 737 mA] is
lower than the component rated current which is 2 A. However,
below 10 mV the amplifier output noise does not allow any
comparison. Moreover, Fig. 10 shows that the ferrite bead
increases the THD on the entire frequency range. Note that,
beyond 6.5 kHz the THD measurements decrease suddenly.
This is because some of the odd harmonics are already out of
the audio band.
As a result, the ferrite bead increases the THD of the signal
delivered by the amplifier at all frequencies and a wide range
of signal levels. The deterioration caused by the ferrite bead
can reach −35 dB. Note that, −40 dB of THD level can be
distinguished by the human ear. Thus, the ferrite bead reduces
the audio quality of the amplifier due to its nonlinear effect in
the audio band
Concluding:

Quote:
This paper showed that ferrite beads used for EMI suppression
in Class-D amplifier systems, have a non-negligible effect
on the amplifier audio quality. In the time domain, the ferrite
bead nonlinear behavior is seen as peaks apparition at voltage
zero crossing. This can be translated to the frequency domain
by an increase of the odd harmonics in the audio frequency
band.
Furthermore, the THD measurements showed the impact
of the ferrite bead on the audio quality for all the tested
frequencies and a wide range of signal amplitudes. The
THD measurements and FFT calculations proved, therefore,
the significant impact of these non-linearities on the audio
performance.
This paper highlights the need of a model taking into account
the ferrite bead non-linear behavior, to help the designers
choose their optimal ferrite bead.
  Reply With Quote
Old 4th May 2018, 09:04 PM   #22
maty tinman is offline maty tinman  Spain
diyAudio Member
 
maty tinman's Avatar
 
Join Date: Apr 2012
Location: Tarragona (Spain)
Where is the phase shift discussed?

************ *************

About the RF/EMI ferrites or filters and class D, some European stores add a chinese filter on the IcePower 1200AS2, but this is other story. Apollon and Rouge Audio.

I have a thread in Audiocircle/All Solid State:

Amplifiers with IcePower 1200AS2, class D module

Last edited by maty tinman; 4th May 2018 at 09:09 PM.
  Reply With Quote
Old 4th May 2018, 09:25 PM   #23
NielsMayer is offline NielsMayer
diyAudio Member
 
Join Date: Jun 2010
Quote:
Originally Posted by maty tinman View Post
Where is the phase shift discussed?
The paper doesn't directly mention phase shift, however, note the troubling zero-crossing "hysteresis" effect:
Quote:
It can be seen that voltage measured before
the ferrite bead Vin keeps it’s sinusoidal nature. However,
The voltage measured after the ferrite bead, presents some
nonlinear behavior near to zero. According to (??), the voltage
across the ferrite bead is computed and shown in Fig. 6.
As can be seen in Fig. 6, [s]pikes appear when the signal
passes through 0 V
https://hal.archives-ouvertes.fr/hal-01103584/document

Although phase not mentioned specifically, please note Figure 5 showing what appears to be an amplitude-dependent phase-shift from input to output across ferrite, with a discontinuity at zero.

Figure 6 shows the zero-crossing distortion.

Figure 8 shows frequency-domain distortions introduced.
  Reply With Quote
Old 4th May 2018, 09:50 PM   #24
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
I would like to contribute my experience in sorting this subject:

- Absolute phase (evaluating a single signal) can only become audible at mid and low frequencies. Neurons are not fast enough to characterize phase in high frequency transients.

- Relative phase (evaluating 2 signals) can become audible at any frequency. At high frequencies relative perception depends more on amplitude than on phase. At low frequencies relative perception depends mostly on phase.

- While designing a system there are 2 priorities into play: Good engineering practices (for the conservation of discipline), cost reduction practices (for the conservation budget). Conservation of discipline and budget are mutually opposing.

- There are many shortcuts possible in the design of class-D amplifiers. There is also a general solution. There is not much difference in the cost of the parts. Most of the difference comes from the degree of truth in everyday life required to understand technological solutions. General solution requires more truth to be understood, this has a cost. But past some point higher cost is not necessarily more truth.

- Sound processing without phase shift and/or delay is not possible.

- An amplifier with phase shift proportional to frequency across pass-band acts as a fixed delay. This is good engineering practice. This can be corrected *easily* with delay, for any application. Such a kind of amplifier is a reference design, or basic building block (see 2 pictures attached, showing my reference design).

- Shortcuts in the design of class D amplifiers lead to circuits whose phase shift is not proportional to frequency, and whose frequency response is not flat. This is not good engineering practice (such gear is of no help to conserve discipline). This is budget. This cannot be corrected easily.

- Relative phase is not affected when two amplifiers with same HF phase/amplitude artifacts are used, as long as tolerance in artifacts between amplifiers is satisfactory. Artifacts have to be characterized, with tolerances, to be able to engineer "more system" on top of them. "Amateurs" use shortcuts without tolerance analysis. No way to build professionally on top of that.

- Only two reasons can lead someone to meticulous analysis of phase shift in audio amplifiers. It can be a runaway from lack of knowledge about phase shift and other abnormalities in speaker response. It can be after learning everything about speaker physics.

- In multi-way speaker systems, due to unavoidable minor comb-filter phenomena (residual comb-filtering due to minor sound reflections), changes of just 0.1dB in amplifier+loudspeaker response can be heard.

- Voice coil resistance increases by 1dB (+12%) for each 35C temperature increase (copper), thus reducing response up to -1dB, but not as much at all frequencies, as a loudspeaker has a impedance plot characterized by (Rvc + major/minor resonances + inductance), so phase shift and box tuning are also affected. Even ambient temperature changes of 10C can perceptibly de-tune a precision DSP-tuned multi-way speaker system, at low volume. Listening at moderate volume on Hi-Fi speakers is also enough for detuning. Even worse: different types of music heat more one driver or another in multi-way systems, leading to a type of de-tuning dependent on type of signal!

- Voice coil resistance can change pretty fast in HF drivers. Primary thermal time constant of smaller voice coils, like 1" tweeters, can be in the range of 100ms. This means that the system can become de-tuned and recover as fast as 100ms + 100ms. De-tuning often means shifting of polar pattern, thus cancellations and peaks that go back and forth, and unsteady sound imaging.

The missing gap between amplifier and loudspeaker integration is voice-coil resistance vs. temperature change compensation. As temperature dependency is eliminated, any other artifact becomes constant vs. temperature, thus straight to correct with equalization and delay.
Attached Images
File Type: png class_D_phase_shift.png (3.5 KB, 284 views)
File Type: png class_D_phase_shift_2.png (48.8 KB, 278 views)
__________________
I use to feel like the small child in The Emperor's New Clothes tale

Last edited by Eva; 4th May 2018 at 10:19 PM.
  Reply With Quote
Old 4th May 2018, 10:26 PM   #25
maty tinman is offline maty tinman  Spain
diyAudio Member
 
maty tinman's Avatar
 
Join Date: Apr 2012
Location: Tarragona (Spain)
But both do not answer the question. Should not the curve be flat, at least < 20 kHz or in the amp bandwith (1200AS2: 50 kHz)?

Click the image to open in full size.

***********

When I commented about RF/EMI ferrites and filters I was referring to those used to clean the power, the dirty mains.
  Reply With Quote
Old 4th May 2018, 10:36 PM   #26
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
NO! Because Icepower is not High-End, many shortcuts against good engineering practice were taken in design. B&O is a company led by money discipline, not by engineering discipline.

But, are you going to start a chase for windmills just because they are noisy and look ugly?
__________________
I use to feel like the small child in The Emperor's New Clothes tale

Last edited by Eva; 4th May 2018 at 10:39 PM.
  Reply With Quote
Old 4th May 2018, 10:37 PM   #27
maty tinman is offline maty tinman  Spain
diyAudio Member
 
maty tinman's Avatar
 
Join Date: Apr 2012
Location: Tarragona (Spain)
These days, in a class A amplifier thread there are comments about the phase shift:

Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier

Hugh Dean (AKSA)

Quote:
4. At 1MHz, a typical amplifier would have a phase shift of -100 degrees from the input zero datum point. When this phase shift increases with rising frequency to the magic 180 degrees, somewhere around 2MHz, the negative feedback presented to the inverting input stage amplifier point has effectively now positive feedback...


IcePower 1200AS2
has 50 at 2 kHz. And about 70 at 20 kHz. In other technologies (A and AB) it would be unacceptable. Why not in class D?

I think it is a design flaw but class D is a bit mysterious to me. That is why I opened the thread.

But if it is such a clamorous flaw how is that the graph appears in the manual? -> I decided to open this thread.

Last edited by maty tinman; 4th May 2018 at 10:42 PM.
  Reply With Quote
Old 4th May 2018, 10:44 PM   #28
maty tinman is offline maty tinman  Spain
diyAudio Member
 
maty tinman's Avatar
 
Join Date: Apr 2012
Location: Tarragona (Spain)
Quote:
Originally Posted by Eva View Post
NO! Because Icepower is not High-End, many shortcuts against good engineering practice were taken in design. B&O is a company led by money discipline, not by engineering discipline...
AS Series - ICEpower

Quote:
ICEpower 1200AS2

The 1200AS2 combines a single stage PFC converter with a 2x 1200 W high performance, ICEedge based class D amplifier. Besides extreme audio performance and power it also features monitor outputs for amplifier temperature, voltage and current output. The DC hanger bus can be used for powering additional 300A1 amplifiers. Perfect for professional dual subwoofers, 2-way and 3-way speakers as well as high end consumer active speakers or high-end stereo amplifiers
  Reply With Quote
Old 4th May 2018, 10:48 PM   #29
maty tinman is offline maty tinman  Spain
diyAudio Member
 
maty tinman's Avatar
 
Join Date: Apr 2012
Location: Tarragona (Spain)
Which companies design high quality class D amplifiers?
  Reply With Quote
Old 4th May 2018, 10:49 PM   #30
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
Somehow you are focusing in a love-hate relationship with something that will not satisfy your expectations. Also running away from another love relationship (with speakers and physics) that could provide more certainty.

Phase shift wise, class D is different from linear because there is a delay involved in modulation, and sharp LPF is required above 20Khz for clean modulation. However, as I showed, there are approaches to make a class D amplifier act as a fixed delay through audio band.

General solution to class D modulation is phase-shift self-oscillating with post-filter feedback. This can be tuned for flat output impedance, delay and frequency response. This is currently made by Hypex and a few others. The technology I'm developing uses same modulation, but it is more oriented towards high-volume listening, and has been offered to complement their product range.
__________________
I use to feel like the small child in The Emperor's New Clothes tale

Last edited by Eva; 4th May 2018 at 11:07 PM.
  Reply With Quote

Reply


Phase shift in class D amplifiers. How it affects the sound?Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Is phase shift audible? Conrad Hoffman Everything Else 4 30th October 2014 07:52 PM
minimal phase shift (at audio frequencies) class d possible? peskywinnets Class D 19 14th June 2012 04:02 PM
Why Class A Guitar Amplifiers Sound So Good ........ poynton Instruments and Amps 16 18th March 2011 08:20 PM
Phase Shift PWM Andrew Eckhardt Class D 6 14th November 2010 08:56 AM
Zero phase-shift gareth Solid State 11 9th May 2008 05:45 PM


New To Site? Need Help?

All times are GMT. The time now is 08:18 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki