February 29, 2020
Determining audio quality is one of the most difficult questions I receive about consumer electronics. People will ask me “which headphones should I buy? What are the best speakers to get? Can you recommend a DAC (Digital-to-Analog Converter)?”. Honestly, I do not know how to answer these questions. Sound quality is determined by a combination of factors that are both objective and subjective. However, I am going to discuss the most under discussed parameter in sound reproduction that is Engineering Tolerance. More simply stated how consistent are the products that come off the assembly line?
Every audio product will provide you with a set of measurements that are descriptive of performance. None of these measurements matter without a defined tolerance! Here is a great hypothetical example.
You look and see a banner ad on the side of you favorite audio site with the following text
“Our speaker produces 60dBSPL at 1 meter at maximum continuous output! Each speaker is matched to each other with obsessive precision! Made from the finest old growth Italian hardwood! We have a passion for the highest fidelity sound reproduction on the market!”
All your friends go to the store and pick a pair up. There is a group text among friends to discuss the performance.
You go back to the store and wait in the line for the customer service department. You get to the front of the line and explain your problem. The technician whirls the speaker around looking for damage. The technician plugs the speaker in to see if they power on. The technician goes into the back room for a while and comes back and says
“Sorry,there is nothing I can do it is within manufacturer tolerance. The amplifier works, the output is within tolerance” – Technician.
“What is the tolerance? I can’t hear anything from the speaker” – You
“Plus or minus 80dB”-Technician
which means the speakers maximum output can be either above the threshold of pain (120dBSPL) or below the threshold of hearing (0dBSPL) . In this example your speakers could be functioning as designed and be completely inaudible to you.
Albeit, the hypothetical example was extreme but, notice that the speakers can be matched with each other and the consumer can still receives poor performance. Knowing the matching tolerance is very important to understanding the performance. People who purchase the same speaker have drastically different experiences. Depending on which set they had. This is an issues with recommending products without known tolerances. Information on tolerance is very useful for comparison shopping because you can see how a set of speakers compare to each other. Tolerances can tell you if speakers actually have the same performance or are very different. When tolerance is not provided by a manufacturer it is very expensive and time consuming to determine. I generally recommend staying away from products that do not reference their tolerances.
Here are some basic terms to look out for in speakers. Measurements can be tricky.
- Distortion (THD) – The ratio between a fundamental frequency and the harmonics amplitude generated by the devices. An example would be sending 1kHz into a speaker and getting 1kHz, 2kHz and 3kHz output for the speaker these harmonics amplitudes are used to calculate total harmonic distortion. Total Harmonic Distortion (THD) is a method of quantifying how much distortion a system has. THD is the most common method report distortion but, it is not the only way. Total harmonic distortion can be calculated including the fundamental or excluding the fundamental which gives slightly different numbers. The lower the better but, it does not tell you how a speaker is distorted so it is very difficult to compare two speakers without plots of the individual harmonics.
- Frequency Response – The range of reproducible frequencies within a specific tolerance range. The range is bounded by a specific roll off frequencies which are measured at specific reductions in peak amplitudes. So a speaker with a frequency range of 20-20kHz with the roll off defined at -3dB means that at 20Hz the level is 3dB below peak acoustic output. It is important to understand that to have an even frequency response your roll off regions should be outside the range of human hearing. This is due to the fact that humans can only detect a minimum differences in amplitude of 1dB so this roll off would be noticeable to the ear. Another important point to understand is that different roll off locations can not be converted to each other as they need to be re read off the graph. This means that a -6dB roll off cannot be convert -10dB roll off without reading from the frequency response plot! The frequency response of a speaker is roughly equivalent to the piston band or the region with flat acoustic output above the speakers unique resonance. (see John Eargle’s “Loudspeaker Handbook 2nd Edition” pg.10)
- Electrical Power Consumption – A useless measurement that should be ignored! Electrical power consumption is only useful if you know the speakers acoustic output efficiency or sensitivity. Aka Power consumption ≠ Acoustic power . Most electrical ratings are reporting total power consumption in watts. A smart speaker’s power consumption includes the WiFi and other microchips power consumption as well not just the audio portion!
- Acoustic or Sound Power – The mechanical work that is done on the air by a speaker each cycle which is also measured in watts. (See Marshall Long “Architectural Acoustics 2nd edition” pg.63)
- Sensitivity – The efficiency of converting electrical power into acoustic output at 1kHz or 250Hz for sub woofers. This efficiency is reported in decibels. It is the measured sound output generated with 1W of power input at 1m. An example sensitivity would be Speaker A has a sensitivity of 94dB at 1 meter. The 1W of input is always implied. A common reporting error is to report sensitivity a fixed voltage. A fixed voltage sensitivity is wrong because the wattage at input is being changed. For instance it is common to report sensitivity at 2.83V. If the speaker has an impedance of 8 ohms at 1kHz then sensitivity at 2.83V is a true efficiency. However, if a speaker has an impedance of 4 ohms at 1kHz the reported efficiency is double the true efficiency. The sensitivity is not constant across all frequencies. (see John Eargle’s “Loudspeaker Handbook 2nd Edition” pg.11-12)
- SPL (Sound Pressure Level)- this the is a decibel scale where 0dBSPL is the threshold of hearing at 1kHz. It is a measurement of acoustic output power. (See Marshall Long “Architectural Acoustics 2nd edition” pg.66)
Now using our understanding of tolerance we can compare speakers to each other.
The criteria for a speakers in this comparison is a single speaker that was rated most popular, highest priced and or lowest price on Zzounds an e-commerce website. The last speaker in this comparison is one that I own.
|JBL LSR305||Yamaha HS8||Genelec 8050B||JBL 305P MkII|
|ZZounds Filter Categories||Lowest||Most Popular||Highest||I own|
|Cutoff Location (dB)||?||-10||-6||-10dB|
|Frequency Response (Hz)||43-24k||38 – 30k||32-25k||43-20k|
|Frequency Response Tolerance (dB)||?||?||2||3|
|Max Peak SPL (dB)||108*||?||≥120||108|
|Max Continuous SPL (dB)||?||101||94|
|Distortion @ 90dB (THD (%))||?||?||0.5||?|
|Combined maximum driver Power (Watts)||82||120||270||82|
First thing to notice is that the Genelec’s frequency response is not directly comparable to any of the other speakers. Yamaha makes the most popular speaker but, doesn’t provide the user with much comparison data. This makes a comparison more difficult. With our data we can conclude the maximum output of the Genelec 8050B is higher than the JBL LSR305 and 305p MKII. Also the JBL 305P MkII is merely a refresh of the JBL LSR305 which explains why they have nearly identical measurements. We cannot assume however that the Yamaha has a higher output than both JBL models. Remember power consumption is not equal to acoustic output! Normalizing the acoustic output to 1 watt the JBL has a sensitivity 70dBSPL and the Genelec has a sensitivity 71dBSPL. In this example even if the speakers used the exact same amplifier putting the same amount of power out as the JBL they would have different max peak SPL. In fact the Genelec has approximately two decibels more output compared to the JBL. It should be noted that this comparison was done using peak acoustic output but, we do not know how much power each amplifier used at peak output or their efficiency at peak output.
In conclusion even if when we compare a $150 speaker to a $2000 speaker, the difference in output is explained by a 1dB difference in sensitivity and 1dB less variance in frequency response. Neither manufacturer has provided tolerances for their sensitivity so depending on the individual model they could easily have identical sensitivities. With out proper information on engineering tolerances it is almost impossible to say which speaker is the better performer and that is why engineering tolerance is the most under discussed acoustic parameter.