more real world testing on load dependency of cheap class d
https://www.mtg-designs.com/tips-tr.../amp-frequency-response-variability-with-load
I find the issue to be so obnoxious that I stopped using those little amps and even recommending them to people. I've shared my own results before and the issue has made some speakers just totally unlistenable, the worst being a 10db rise from 6k to 15k. Id like to get something class ab as I can't afford the good class d like hypex. Audiosource amp100vs looks good but idk if it's class d or not. I've only had the previous non-vs model.
https://www.mtg-designs.com/tips-tr.../amp-frequency-response-variability-with-load
I find the issue to be so obnoxious that I stopped using those little amps and even recommending them to people. I've shared my own results before and the issue has made some speakers just totally unlistenable, the worst being a 10db rise from 6k to 15k. Id like to get something class ab as I can't afford the good class d like hypex. Audiosource amp100vs looks good but idk if it's class d or not. I've only had the previous non-vs model.
Thanks for the explanation. It is very helpful.
I wonder how the ICEpower amps would react in this situation? The ICEpower 100AS2 is a 2 channel 100W/Ch amp board with built in power supply. It seems ideal for this application, and it is priced at about $140.
The price is not a criterion here. There are different subtypes of Class D amplifiers, and in those amplifiers where the low-pass filter (LPF) is not covered by feedback, such a dependency will exist. However, the impact of the amplifier's LPF on the frequency response of the speakers is in a frequency range where the majority of the world's population won't be able to hear it. There are very few people who can audibly notice a decrease or increase in voltage level by 3 dB at 20 kHz. Despite all this, it's important to consider the non-ideal frequency response of the high-frequency driver, which is always uneven and varies with different listening angles. In general, such LPF behavior often does not lead to any serious consequences when listening to music through real speakers. Moreover, if we take into account speakers with DSP, it's not a problem at all, as compensating for the LPF's influence on the speakers is very easy.
I listened to this speaker both with, and without the additional + 3 dB lift above 9k. Honestly, the effect is very subtle, and I knew exactly what to listen for. Most of the time I would not notice the drop-off above 9k. My hearing stops at 12k, so I may not the right person to definitively judge this.
If there was no opportunity to fix this with DSP, such as in a purely passive speaker, I would say it would be a minor issue. However for a true high fidelity experience, it should be fixed, or a more high performance amp should be used.
j.
From the images in the datasheet for this amplifier, it appears that its low-pass filter (LPF) is not covered by feedback.
In general, this amplifier is relatively old, and as I understand it, there are many replicas that sometimes have worse characteristics than those stated by the manufacturer. It's necessary to verify the purchased amplifier to ensure it meets the declared specifications. As of today, I consider amplifiers based on the TPA3255 chip to be the most optimal solution. If control over the low-pass filter (LPF) is needed, then it's advisable to look for an amplifier based on TPA3255 where a so-called PFFB (Post Filter Feedback) is implemented. However, this is purely my opinion, and in each specific case, the choice should be made based on the set goals and requirements.
In reality, the rise in the frequency response (HF boost) of a Class D amplifier can be relatively easily addressed by introducing an RC circuit between the amplifier and the speakers. However, I suspect that the issue of poor sound quality you are experiencing is not necessarily due to the amplifier's frequency response boost, but rather in the level of nonlinear distortions at frequencies of 3 kHz and above. This is precisely the key problem with simple low-pass filters (LPFs) at the output of Class D amplifiers. If you were to measure the nonlinear distortions of those amplifiers, I believe you would find answers to why they sound unpleasant in the high frequencies.
Off-topic.
I'm a newcomer to this forum. There are many diverse topics here, and it's challenging for me to find a discussion that interests me. I'm interested in discussions on this forum about Class D amplifiers with entirely digital inputs that are tied to DSP, i.e., multichannel Class D amplifiers with DSP where there is no analog input signal? If anyone can point me to where I can find this topic, I would be grateful.
I'm a newcomer to this forum. There are many diverse topics here, and it's challenging for me to find a discussion that interests me. I'm interested in discussions on this forum about Class D amplifiers with entirely digital inputs that are tied to DSP, i.e., multichannel Class D amplifiers with DSP where there is no analog input signal? If anyone can point me to where I can find this topic, I would be grateful.
By the way, the frequency range from 10 kHz and above is quite individual; the sensitivity of hearing in this range can vary significantly among different individuals. This, in turn, imposes certain limitations on the claim that a flat frequency response above 10 kHz is an indicator of high sound quality. In reality, to achieve high sound quality from speakers, each person should adjust this range according to their hearing. To be more precise, the entire frequency range should be tailored to the individual characteristics of a person's hearing. Only then can it be said that they will experience high-quality sound.
A flat frequency response in speakers serves as a universal reference point, which, upon detailed and careful examination of individual hearing properties, is not absolute. I am leading to the point that a flat frequency response in speakers should not be the ultimate goal. If you feel that the high-frequency range is insufficient to your ears, feel free to add SPL in that range without adhering to a flat response. Personally, I advocate for tuning speakers to match one's hearing based on measurements, taking into account the influence of the room. This can yield the maximum result, although it is a complex path that requires at least a microphone and an understanding of what affects what and what needs adjustment according to one's hearing. The main idea behind my thoughts is that a flat frequency response in speakers is not an indicator of high sound quality. The perception of sound can be improved by making a correct correction to the frequency response in the listening area.
i will respectfully disagree with the notion that the speakers should be adjusted to compensate for individual hearing. We all adapt, to some extent, to the limitations of our hearing. If your hearing has reached the point where you need to adjust the frequency response to achieve intelligibility, then it is time for hearing aids. Speakers should have basically flat of slightly decreasing in room response (house curve).
I am familiar with your opinion, and I respect it. Yes, such an opinion exists, and it has every right to exist. However, from my own experience, I can assert that a flat on-axis frequency response is not the optimal solution for achieving high-quality sound perception. There are various listening conditions for speakers, and sometimes correcting the frequency response in the listening area leads to a distortion of the on-axis frequency response. In other words, to obtain a more even frequency response in the listening area, you may need to alter the on-axis frequency response.
Personally, I prioritize the frequency response in the listening area over the on-axis response, which, in turn, sometimes imposes limitations on the flatness of the on-axis frequency response. While auditory adaptability is always present, it is not all-powerful, and sometimes a slight increase or decrease in SPL is needed to balance the perceived tonal balance for each individual. It's like fine-tuning according to one's hearing.
Whether to add or reduce SPL in a particular frequency range based on one's hearing is a subjective decision that each person can make according to their feelings. I won't insist that a flat frequency response is the only correct path; in my view, it's not always the case. This is purely my subjective opinion, not claiming absolute truth.
Thank you @wchang.
How does this hearing test work?
In clinical settings, hearing tests are related to looking for hearing impairment (aka deafness) and the tools are very established (aka old). An audiogram taken by an audiologist only covers between 125HZ to 8KHz at one octave intervals; save for a few in the middle frequencies (1, 2, 3KHz).
You don’t “fail the test” unless you have more than 10dB of Hearing Loss. Why? Because it’s considered otherwise “within normal limits” - after all, we are talking about clinical conditions that impact speech intelligibility and a person’s everyday daily living.
(The most common causes of hearing loss include age related hearing loss and prolonged eg. occupational noise exposure, which causes losses at the highest frequencies, and all frequencies respectively. The most common is probably outer ear conditions such as earwax, middle ear conditions such at myringitis following a cold/flu or eustacian tube dysfunction following changes in altitude eg. flights. Some middle and inner ear conditions can cause long term problems eg. otitis media and Ménière’s disease (not uncommon) and will have different patterns of frequency dependant hearing loss; the usefulness of these audiograms is to assist in differentiating between them)
The individual variation in normal healthy young individuals can and do vary as much +5dB to -10dB between 125Hz and 8KHz, so audiograms or tests by audiologist are too blunt.
So we are not talking about perfect (musical) pitch perception or Hi-Fi hearing ability, because individual variations in hearing acuity can and do vary.
Furthermore , we have not even considered psycho acoustics, or even the human tendency to observe and interpret (listen, read, see) the same data differently.
So it is good enough? (Laptop speaker for speech)
Better? (11.1.4 Atmos soundbar for living room)
Perfect speakers for all listeners listening to recorded music: illusion
It’s simply not possible to have a single perfect solution that suits everyone.
Back to the topic at hand-
I’m in favour of fine tuning/equalisation the top octave (and bottom octaves) to listener preferences/taste…
How does this hearing test work?
In clinical settings, hearing tests are related to looking for hearing impairment (aka deafness) and the tools are very established (aka old). An audiogram taken by an audiologist only covers between 125HZ to 8KHz at one octave intervals; save for a few in the middle frequencies (1, 2, 3KHz).
You don’t “fail the test” unless you have more than 10dB of Hearing Loss. Why? Because it’s considered otherwise “within normal limits” - after all, we are talking about clinical conditions that impact speech intelligibility and a person’s everyday daily living.
(The most common causes of hearing loss include age related hearing loss and prolonged eg. occupational noise exposure, which causes losses at the highest frequencies, and all frequencies respectively. The most common is probably outer ear conditions such as earwax, middle ear conditions such at myringitis following a cold/flu or eustacian tube dysfunction following changes in altitude eg. flights. Some middle and inner ear conditions can cause long term problems eg. otitis media and Ménière’s disease (not uncommon) and will have different patterns of frequency dependant hearing loss; the usefulness of these audiograms is to assist in differentiating between them)
The individual variation in normal healthy young individuals can and do vary as much +5dB to -10dB between 125Hz and 8KHz, so audiograms or tests by audiologist are too blunt.
So we are not talking about perfect (musical) pitch perception or Hi-Fi hearing ability, because individual variations in hearing acuity can and do vary.
Furthermore , we have not even considered psycho acoustics, or even the human tendency to observe and interpret (listen, read, see) the same data differently.
So it is good enough? (Laptop speaker for speech)
Better? (11.1.4 Atmos soundbar for living room)
Perfect speakers for all listeners listening to recorded music: illusion
It’s simply not possible to have a single perfect solution that suits everyone.
Back to the topic at hand-
I’m in favour of fine tuning/equalisation the top octave (and bottom octaves) to listener preferences/taste…
Last edited:
Actually pretty sophisticated. First off, the former flagship Mi-9 phone offered preset EQ curves indexed by listener age, with MUSIC sound-clips for comparison. These curves tamed mid-high a bit and boosted very-high and low frequencies. And one could adjust them. To go a step further, the hearing test played a succession of randomized, very faint tones of various frequencies, to one ear/channel or the other, to determine (beyond) one's limit of hearing. A compensating EQ curve was then produced etc.
While I doubt many people would go through the trouble of a music hearing test, as palm-held Supercomputing and AI further improved, the process would only become more invisible and ubiquitous.
As for sound quality from a phone? My diy test apparatus is Huawei P10+ (the first truly high-end Chinese-branded phone, no expense spared) cloud-playing my own EAC-ripped files (sensitive enough to distinguish WAV from FLAC), output through OCC wire. My Chord Hugo might be a tad better, I'm not sure.
Last edited:
This stopped two days ago
https://www.diyaudio.com/community/threads/mp3-player-before-it-existed.407973/page-2#post-7573193
And I read trough the Clarion flyer...
Mostly discontinued, the speakers...
It reminds me of member Fast Eddie D
Who works in some innovative speakers factory
Ah ah! I read 'phono' and I imagined a \ shaped curve Phono EQ applied after BT transmission from linear( mic) amplification of the pick up.
Ok, a phone doesn't have an AD direct, I guess
I also disagree.
After all it is my goal as a developer to achieve the most natural reproduction of music possible (yes, I know what kind of things always have to fail in the end).
You can't adapt a real violin to your own hearing. So if I follow the above premise, why should I do this when the same violin is recorded and played back through a music system?
Best regards
Michael
Our brain adapts to our ears!
Every ear+headcurve is different, often even between left/right ear. But we are used to our curves and that's the "normal" for us. When our ears degrade over the years we are still used to this new curve cause it's happening slowly. We are not able to hear 15kHz any more but as long as you don't suffer from hearing damadge from an event (explosion etc) we are used to this.
So stick with reproduction when you want to hear natural sound.
Every ear+headcurve is different, often even between left/right ear. But we are used to our curves and that's the "normal" for us. When our ears degrade over the years we are still used to this new curve cause it's happening slowly. We are not able to hear 15kHz any more but as long as you don't suffer from hearing damadge from an event (explosion etc) we are used to this.
So stick with reproduction when you want to hear natural sound.
Hypex FA series are integrated dac-dsp-multichannel Ncore amps are most popular products among diyers in Europe - discussion thread started in 2017... https://www.diyaudio.com/community/threads/the-new-hypex-fusion-plate-amps.306815/
- Hypex FA are used in many projects that you can find here at diyaudio. I have done three stereo pairs, but I use analog input with FA.
https://www.hypex.nl/products/amplifier-families/fusion-amplifier-family/