To what extent do the level of these measured output voltage drops correlate with AVR sound quality?

[The Wolf]

Hi, having been a lurker for many years, this is my first post on the forum.

I recall Gene saying several times how ideally an amplifier acts as a perfect voltage source in that the output voltage fed to the speaker will hold up under any load a speaker can throw at it. With this in mind, I recently performed a small experiment to try and test the respective capability of the three AVRs I currently own (Denon AVR-3802, Denon AVR-3312 and Yamaha RX-V3900) and am seeking your opinions on the results. I should add that I don't have an electronics/scientific background at all so this was very much an amateur experiment so am well prepared for a possible roasting from the techies among you LOL.

Anyway, here's what I did first. Using the AVR-3802, REW and a True RMS 6000 Count Multimeter I fed sine wave tones from 60Hz to 250Hz at a fixed voltage (-20dB volume) to an old bookshelf speaker (B&W 686) and measured the AC current flow to determine the easiest and most difficult frequencies the speaker's woofer presented in terms of load. Using Ohm's Law (V=IR), these impedance loads were derived as being 31Ω at 85Hz and 4.9Ω at 225Hz. I then found a frequency that represented a load closer to the speaker's 8Ω nominal rating which was 7.2Ω at 125Hz. These three frequencies were then used for my voltage test for the three AVRs.

So now to the test. I fed four different voltages (at 10dB volume intervals from circa -5dB to -35dB) ranging approximately from 0.06V to 1.7V from each AVR and measured the unloaded voltage (i.e. open circuit) and the voltages under load at the three test frequencies. I dropped the measurements into a spreadsheet to calculate the voltage drops from the open circuit voltage and these are the results when fed 1.7V:

Using such low voltages, this was never intended to be any form of maximum stress test like an ACD test so I believe the results will have more to do with the AVRs' output devices and not amplifier/PSU power. What I was really surprised about though is that the level of voltage drops did not change much at the different voltage levels e.g. the AVR-3802 still recorded a 6.0% voltage drop at 225Hz even when fed a tiny 0.0565V (-40dB on the volume).

Now, I know we don't listen to sine waves but my question to you is are these findings relevant/significant and to what extent should they correlate with sound quality? A 6% voltage drop must surely translate as being audible compression. The RX-V3900 recorded the lowest voltage drop and I should add that I've always thought it sounded better with music than the two Denons. It was a far more expensive model than the Denons and I believe it has the same amp section as the DSP-Z7.

If it makes a difference, this is the test meter I used:

https://uk.astroai.com/product/Astroai-Digital-Multimeter-Trms-6000-Counts-Auto-Ranging-Voltage-Tester-Voltmeter

Any thoughts/comments guys?

 

[TLS Guy]

Hi, having been a lurker for many years, this is my first post on the forum.

I recall Gene saying several times how ideally an amplifier acts as a perfect voltage source in that the output voltage fed to the speaker will hold up under any load a speaker can throw at it. With this in mind, I recently performed a small experiment to try and test the respective capability of the three AVRs I currently own (Denon AVR-3802, Denon AVR-3312 and Yamaha RX-V3900) and am seeking your opinions on the results. I should add that I don't have an electronics/scientific background at all so this was very much an amateur experiment so am well prepared for a possible roasting from the techies among you LOL.

Anyway, here's what I did first. Using the AVR-3802, REW and a True RMS 6000 Count Multimeter I fed sine wave tones from 60Hz to 250Hz at a fixed voltage (-20dB volume) to an old bookshelf speaker (B&W 686) and measured the AC current flow to determine the easiest and most difficult frequencies the speaker's woofer presented in terms of load. Using Ohm's Law (V=IR), these impedance loads were derived as being 31Ω at 85Hz and 4.9Ω at 225Hz. I then found a frequency that represented a load closer to the speaker's 8Ω nominal rating which was 7.2Ω at 125Hz. These three frequencies were then used for my voltage test for the three AVRs.

So now to the test. I fed four different voltages (at 10dB volume intervals from circa -5dB to -35dB) ranging approximately from 0.06V to 1.7V from each AVR and measured the unloaded voltage (i.e. open circuit) and the voltages under load at the three test frequencies. I dropped the measurements into a spreadsheet to calculate the voltage drops from the open circuit voltage and these are the results when fed 1.7V:

View attachment 53939

Using such low voltages, this was never intended to be any form of maximum stress test like an ACD test so I believe the results will have more to do with the AVRs' output devices and not amplifier/PSU power. What I was really surprised about though is that the level of voltage drops did not change much at the different voltage levels e.g. the AVR-3802 still recorded a 6.0% voltage drop at 225Hz even when fed a tiny 0.0565V (-40dB on the volume).

Now, I know we don't listen to sine waves but my question to you is are these findings relevant/significant and to what extent should they correlate with sound quality? A 6% voltage drop must surely translate as being audible compression. The RX-V3900 recorded the lowest voltage drop and I should add that I've always thought it sounded better with music than the two Denons. It was a far more expensive model than the Denons and I believe it has the same amp section as the DSP-Z7.

If it makes a difference, this is the test meter I used:

https://uk.astroai.com/product/Astroai-Digital-Multimeter-Trms-6000-Counts-Auto-Ranging-Voltage-Tester-Voltmeter

Any thoughts/comments guys?

Well what you have shown is that all amps have a source impedance. What you did demonstrate is that those amps have different source impedances. The higher the variation as you alter the load impedance the higher the source impedance of that amp.

Now tube amps have a high source (output impedance), and it is well known that the frequency response of tube amps has a high propensity to follow the the impedance curve of the speaker. So generally the impedance curve of speakers tends to be higher at the higher frequencies. Therefore there is every chance that a speaker that is a bit fierce on the top end, will sound less fierce (warmer) when being driven by a tube amp.

However you really need to calculate the power delivered by those receivers driving those loads, and then recalculate it to the db scale and I bet you will find that the differences in db are not detectable by the human ear.

The formula for the power delivered is the square if the voltage divided by the resistance/impedance.

Once you know that actual power delivered then you can go on line and find a calculator the convert the absolute power difference to differences in db.

Now you have your weekend home work.

 

[PENG]

Anyway, here's what I did first. Using the AVR-3802, REW and a True RMS 6000 Count Multimeter I fed sine wave tones from 60Hz to 250Hz at a fixed voltage (-20dB volume) to an old bookshelf speaker (B&W 686) and measured the AC current flow to determine the easiest and most difficult frequencies the speaker's woofer presented in terms of load. Using Ohm's Law (V=IR), these impedance loads were derived as being 31Ω at 85Hz and 4.9Ω at 225Hz. I then found a frequency that represented a load closer to the speaker's 8Ω nominal rating which was 7.2Ω at 125Hz. These three frequencies were then used for my voltage test for the three AVRs.

If you use V=IR, you would have to measure the current and voltage. If you use that multimeter to measure AC current from 60-150 Hz, what would be the maximum range and accuracy, according to the manual?

So now to the test. I fed four different voltages (at 10dB volume intervals from circa -5dB to -35dB) ranging approximately from 0.06V to 1.7V from each AVR and measured the unloaded voltage (i.e. open circuit) and the voltages under load at the three test frequencies.

Is that 0.06 to 1.7 V the input voltage to the AVR, or the output voltage at the binding posts?
If it is at the binding posts (where the speakers are connected to), the voltage at -5 dB should be much higher than that at -35 dB, so did you get it the other way around?

The confusion is, it sounds like you are referring to the output voltage but you said "fed", that would imply input, but I guess you mean "fed" to the speakers.

I dropped the measurements into a spreadsheet to calculate the voltage drops from the open circuit voltage and these are the results when fed 1.7V:

View attachment 53939

Assuming the 1.7 V was measured at the speaker binding posts (outputs), those results did not look good at all. Even the V3900, with 2.2% voltage drop at 4.9 ohm load impedance, the frequency response won't be flat. Yet all those AVRs except the 3802 (couldn't find any bench tests on it) you listed have been measured to have FR virtually flat 20 to 20,000 Hz into 8 ohm. I would think that all of the 3 AVRs listed should have output impedance lower than 0.25 ohm.

Using such low voltages, this was never intended to be any form of maximum stress test like an ACD test so I believe the results will have more to do with the AVRs' output devices and not amplifier/PSU power. What I was really surprised about though is that the level of voltage drops did not change much at the different voltage levels e.g. the AVR-3802 still recorded a 6.0% voltage drop at 225Hz even when fed a tiny 0.0565V (-40dB on the volume).

It wouldn't have anything to do with the output devices either, not at 1.7 V output. As TLSGuy said, it would have to do with the output impedance of the power amplifiers, but most likely the accuracy of your measurements has a large part too. Also, please double check your formula in your Excel spreadsheet in case of a typo or something.

Now, I know we don't listen to sine waves but my question to you is are these findings relevant/significant and to what extent should they correlate with sound quality? A 6% voltage drop must surely translate as being audible compression. The RX-V3900 recorded the lowest voltage drop and I should add that I've always thought it sounded better with music than the two Denons. It was a far more expensive model than the Denons and I believe it has the same amp section as the DSP-Z7.

I doubt it, compression would have more to do with weak power supply, and/or output devices but much more so the power supply. Output impedance being high in this would affect the FR most, but again I don't your measurements are accurate.

 

[PENG]

Just check with the Yamaha's RX-V3900 because I remember they do specify DF in the manual.

It says 150 or more, 20-20,000 Hz, 8 ohm load.

So the calculated voltage drop at 1.7 V, 4.9 ohm load should be about 0.54%, even if you allow an additional load of 0.05 ohm due to the resistance of the cable, the voltage drop would still be just 1.54%.

I suggest the OP double check the resistance of the cabling and connection/contact resistance. Contact resistances between the 3 measurements could vary a lot if those typical probes that came with the cheap multimeters are used, instead of some crocodile clips or solid connections.

 

[The Wolf]

Thanks very much guys for your input here, I really appreciate it.

Once you know that actual power delivered then you can go on line and find a calculator the convert the absolute power difference to differences in db.

Very good point, I hadn't stopped to consider relating these differences to audibility.

So looking at the 6% voltage drop of the AVR-3802 as it's the largest. That's a 11.6% power drop which equates to about 0.5dB which is a circa. 4% change in perceived loudness. Not sure how audible that would be, probably critical music listening only, to my ears at least which are not that "golden".

If you use V=IR, you would have to measure the current and voltage. If you use that multimeter to measure AC current from 60-150 Hz, what would be the maximum range and accuracy, according to the manual?

The meter's manual states its operating frequency range for AC voltage and AC current measurements to be 40Hz-1KHz and AC voltage accuracy to be +/-(0.8% rdg +5 dgts) and AC current accuracy to be +/-(1.5% rdg +3 dgts).

My meter is an AstroAI M6KOR which I understand (from independent Youtube reviews) is a rebadged KAIWEETS HT118A which due its popularity has been the subject of several independent bench test reviews, all of which I believe have found its accuracy to be pretty good which is amazing considering its low price.

If you're interested, the meter's manual can be found here: https://astroai-user-manual.s3.us-west-1.amazonaws.com/sku/user-manual/ASIM6K0R-EN-v1.pdf?version=1644224283

Is that 0.06 to 1.7 V the input voltage to the AVR, or the output voltage at the binding posts?

Yes, voltage readings were taken across the speaker binding posts. I connected the meter to the speaker via a 4Ft 13AWG (2.5mm2) OFC speaker cable with dual clamped gold plated banana plugs at each end. The meter conveniently uses standard 4mm female sockets so the banana plugs went straight into it and at the speaker end I just plugged them into the unused bi-amp sockets. These sockets are connected to the amplifier connected terminals via a standard B&W gold plated bridging strap. The resistance of the cable I used is too low to register a reading using the meter's resistance test function but that only has a resolution of 0.1Ω and accurate to +/-(1.5% rdg + 3 dgts). Obviously, the current readings were taken with the meter in series and I did this using one lead of the same cable from the meter to one of the speaker terminals and the other meter terminal being connected to the AVR.

If it is at the binding posts (where the speakers are connected to), the voltage at -5 dB should be much higher than that at -35 dB, so did you get it the other way around?

When taking the measurements I left the REW signal generator set at its default -12dBFS gain setting which would reduce the output voltage level below what you might have expected. Also, as the AVR-3802 lacks an HDMI input I had to feed its signal from my desktop PC's analogue audio line out (at max PC volume) which no doubt would be effected by the gain structure of the sound card. The other two AVRs' REW signals were fed via to them via HDMI from a laptop.

These are the actual readings and spreadsheet calcs.

 

[PENG]

Thanks very much guys for your input here, I really appreciate it.


Very good point, I hadn't stopped to consider relating these differences to audibility.

So looking at the 6% voltage drop of the AVR-3802 as it's the largest. That's a 11.6% power drop which equates to about 0.5dB which is a circa. 4% change in perceived loudness. Not sure how audible that would be, probably critical music listening only, to my ears at least which are not that "golden".


The meter's manual states its operating frequency range for AC voltage and AC current measurements to be 40Hz-1KHz and AC voltage accuracy to be +/-(0.8% rdg +5 dgts) and AC current accuracy to be +/-(1.5% rdg +3 dgts).

My meter is an AstroAI M6KOR which I understand (from independent Youtube reviews) is a rebadged KAIWEETS HT118A which due its popularity has been the subject of several independent bench test reviews, all of which I believe have found its accuracy to be pretty good which is amazing considering its low price.

If you're interested, the meter's manual can be found here: https://astroai-user-manual.s3.us-west-1.amazonaws.com/sku/user-manual/ASIM6K0R-EN-v1.pdf?version=1644224283


Yes, voltage readings were taken across the speaker binding posts. I connected the meter to the speaker via a 4Ft 13AWG (2.5mm2) OFC speaker cable with dual clamped gold plated banana plugs at each end. The meter conveniently uses standard 4mm female sockets so the banana plugs went straight into it and at the speaker end I just plugged them into the unused bi-amp sockets. These sockets are connected to the amplifier connected terminals via a standard B&W gold plated bridging strap. The resistance of the cable I used is too low to register a reading using the meter's resistance test function but that only has a resolution of 0.1Ω and accurate to +/-(1.5% rdg + 3 dgts). Obviously, the current readings were taken with the meter in series and I did this using one lead of the same cable from the meter to one of the speaker terminals and the other meter terminal being connected to the AVR.


When taking the measurements I left the REW signal generator set at its default -12dBFS gain setting which would reduce the output voltage level below what you might have expected. Also, as the AVR-3802 lacks an HDMI input I had to feed its signal from my desktop PC's analogue audio line out (at max PC volume) which no doubt would be effected by the gain structure of the sound card. The other two AVRs' REW signals were fed via to them via HDMI from a laptop.

These are the actual readings and spreadsheet calcs.

View attachment 53972

It looks like you know what you are doing, good job!

If you are sure the way you measured the current flow in each test would have introduced negligible contact resistance in the series loop then all I can say is that the measured output impedance of the AVRs output impedance are much higher than what they should be, especially the AVR-3802 and the RX-V3900

Regarding the RX-V3900,

The bench test results below confirmed Yamaha's specs.

RX-V3900 Report (milleraudioresearch.com) Note: you have to register to read it, once in, look to year 2009.

AVR-3805 Report (milleraudioresearch.com) Note: no need to register to read, just go to year 2004

The Denon does have a little higher output impedance, based on the AVR-3805 that most likely have the same power amp section, because I know the 3803 and 3805 do have the same amp circuitry.

Based on the 3805's, output impedance at the frequencies you conducted your test, the output impedance would be about 0.065 to 0.083 ohm.

You obviously know the math so you can do the calculations and see for yourself that the voltage drop for the Yamaha in % should be less than 1 and the Denon (assuming the 3802's amp section is the same or very similar to the 3803 and 3805's, would be less than 1.5.

I have no idea why your results are so much higher than expected other than measurement accuracy related reasons due to inconsistent readings from REW (fluctuating readings, for example, making it hard to read), contact resistance variations between measurements, test voltage too low etc.. Again, the way you described how you did it looks impressive, for someone who claimed having no electronics background.

 

[Pogre]

What an interesting experiment, and you got both Peng and TLS involved. Tagged!

 

[The Wolf]

It looks like you know what you are doing, good job!

If you are sure the way you measured the current flow in each test would have introduced negligible contact resistance in the series loop then all I can say is that the measured output impedance of the AVRs output impedance are much higher than what they should be, especially the AVR-3802 and the RX-V3900

Regarding the RX-V3900,

View attachment 53982

The bench test results below confirmed Yamaha's specs.

RX-V3900 Report (milleraudioresearch.com) Note: you have to register to read it, once in, look to year 2009.

AVR-3805 Report (milleraudioresearch.com) Note: no need to register to read, just go to year 2004

The Denon does have a little higher output impedance, based on the AVR-3805 that most likely have the same power amp section, because I know the 3803 and 3805 do have the same amp circuitry.

Based on the 3805's, output impedance at the frequencies you conducted your test, the output impedance would be about 0.065 to 0.083 ohm.

You obviously know the math so you can do the calculations and see for yourself that the voltage drop for the Yamaha in % should be less than 1 and the Denon (assuming the 3802's amp section is the same or very similar to the 3803 and 3805's, would be less than 1.5.

I have no idea why your results are so much higher than expected other than measurement accuracy related reasons due to inconsistent readings from REW (fluctuating readings, for example, making it hard to read), contact resistance variations between measurements, test voltage too low etc.. Again, the way you described how you did it looks impressive, for someone who claimed having no electronics background.

Thanks Peng. Just to be clear, the only time I actually performed AC current flow tests was at the very start to find my three test frequencies that would represent easy, difficult and typical speaker loads. I used the AVR-3802 only for this process. Once I had decided on these frequencies, I simply performed AC voltage tests at those frequencies (and unloaded) on all three AVRs at four similar voltage levels at 10dB intervals.

Regarding the greater than expected voltage drops, could speaker phase angle have anything to do with this? TBH, the whole speaker phase angle subject confuses me somewhat but, from what I've read, I understand that speakers do not present amps with simple resistive loads and that extreme (e.g. 45°) phase angles can double the current requirement. I guess the question I'm asking you is this - is the extra stress on an amp caused by the speaker's phase angles baked into the impedance figures I derived from the AC current flow? And if not, how does it manifest itself in terms of impacting an amp's output voltage to the speaker?

Thanks also for the compliments on my testing method. I can assure you that other than what I've read/watched online from being in this hobby for 28 years, my electronics education stopped at age 16 when I last studied physics at high school many moons ago. I'm actually a corporate restructuring finance professional who's very good at spreadsheets but I really do enjoy learning about this stuff as it's out of my comfort zone.

To be fair, it was actually one of your own posts some time ago when you measured peak centre speaker voltages at different subwoofer crossover levels that prompted me to buy a test meter to see what I could do with it. IMO, the most useful thing I've used it for so far has been to measure the AC voltages of 75dB pink noise test tone signals to deduce the actual in-room voltage sensitivity of my speakers at the MLP. I've used this to produce what I believe is a very accurate estimate of their maximum possible peak power requirements at my maximum listening level. If you're interested, that could be the subject of another thread.

 

[TLS Guy]

Thanks Peng. Just to be clear, the only time I actually performed AC current flow tests was at the very start to find my three test frequencies that would represent easy, difficult and typical speaker loads. I used the AVR-3802 only for this process. Once I had decided on these frequencies, I simply performed AC voltage tests at those frequencies (and unloaded) on all three AVRs at four similar voltage levels at 10dB intervals.

Regarding the greater than expected voltage drops, could speaker phase angle have anything to do with this? TBH, the whole speaker phase angle subject confuses me somewhat but, from what I've read, I understand that speakers do not present amps with simple resistive loads and that extreme (e.g. 45°) phase angles can double the current requirement. I guess the question I'm asking you is this - is the extra stress on an amp caused by the speaker's phase angles baked into the impedance figures I derived from the AC current flow? And if not, how does it manifest itself in terms of impacting an amp's output voltage to the speaker?

Thanks also for the compliments on my testing method. I can assure you that other than what I've read/watched online from being in this hobby for 28 years, my electronics education stopped at age 16 when I last studied physics at high school many moons ago. I'm actually a corporate restructuring finance professional who's very good at spreadsheets but I really do enjoy learning about this stuff as it's out of my comfort zone.

To be fair, it was actually one of your own posts some time ago when you measured peak centre speaker voltages at different subwoofer crossover levels that prompted me to buy a test meter to see what I could do with it. IMO, the most useful thing I've used it for so far has been to measure the AC voltages of 75dB pink noise test tone signals to deduce the actual in-room voltage sensitivity of my speakers at the MLP. I've used this to produce what I believe is a very accurate estimate of their maximum possible peak power requirements at my maximum listening level. If you're interested, that could be the subject of another thread.

You have done an interesting experiment. To answer your question about phase angles, you it will have a bearing on all this.

As you get increasing negative phase angles you get an increasing discrepancy between instantaneous power and actual power. The issue is that the current has to be provided for the instantaneous power, but is given back. The more negative the phase angle at a given frequency the greater the current required for the instantaneous power.

So the bottom line is that as the output impedance of an amp increases then the more these phase angles will limit current and upset the frequency response.

So you have touched on a topic where I have long maintained that bench testing of amps with a resistor does not give you a complete picture of that amp driving pretty much any speaker.

I think this generally gets hidden because of the extent of most speaker aberrations. What I will say is that the better the speaker the more it will benefit from better amplification. So low source (output impedance) is a very important power amp parameter.

This is nothing new, and we have mentioned it here previously. So that is just one of many reasons I use Quad amps which have a very low source impedance.

I'm personally not interested in driving any of my speakers from receiver amps.

 

[Swerd]

Very good point, I hadn't stopped to consider relating these [voltage & power] differences to audibility.

So looking at the 6% voltage drop of the AVR-3802 as it's the largest. That's a 11.6% power drop which equates to about 0.5dB which is a circa. 4% change in perceived loudness. Not sure how audible that would be, probably critical music listening only, to my ears at least which are not that "golden".

What is the smallest change in loudness that humans can hear? It depends on the frequency of the sound. In the middle of the audio spectrum, the so-called voice discrimination range, most humans can reliably hear smaller changes than at the extremes of the audio. In the mid range, changes of ±1 dB can be heard. At the extremes, it's more like ±3 dB.

This is a broad generalization, as there will be differences among individuals. But I think it's safe to say that a drop of 0.5 dB is too small to be audible.

 

[PENG]

Thanks Peng. Just to be clear, the only time I actually performed AC current flow tests was at the very start to find my three test frequencies that would represent easy, difficult and typical speaker loads. I used the AVR-3802 only for this process. Once I had decided on these frequencies, I simply performed AC voltage tests at those frequencies (and unloaded) on all three AVRs at four similar voltage levels at 10dB intervals.

Regarding the greater than expected voltage drops, could speaker phase angle have anything to do with this? TBH, the whole speaker phase angle subject confuses me somewhat but, from what I've read, I understand that speakers do not present amps with simple resistive loads and that extreme (e.g. 45°) phase angles can double the current requirement. I guess the question I'm asking you is this - is the extra stress on an amp caused by the speaker's phase angles baked into the impedance figures I derived from the AC current flow? And if not, how does it manifest itself in terms of impacting an amp's output voltage to the speaker?

Thanks also for the compliments on my testing method. I can assure you that other than what I've read/watched online from being in this hobby for 28 years, my electronics education stopped at age 16 when I last studied physics at high school many moons ago. I'm actually a corporate restructuring finance professional who's very good at spreadsheets but I really do enjoy learning about this stuff as it's out of my comfort zone.

To be fair, it was actually one of your own posts some time ago when you measured peak centre speaker voltages at different subwoofer crossover levels that prompted me to buy a test meter to see what I could do with it. IMO, the most useful thing I've used it for so far has been to measure the AC voltages of 75dB pink noise test tone signals to deduce the actual in-room voltage sensitivity of my speakers at the MLP. I've used this to produce what I believe is a very accurate estimate of their maximum possible peak power requirements at my maximum listening level. If you're interested, that could be the subject of another thread.

For your experiment phase angle is irrelevant. Phase angle would affect heat dissipation in the out devices and the calculated power out into the speakers.

Take a look of the formula you will see what I mean.

 

[PENG]

You have done an interesting experiment. To answer your question about phase angles, you it will have a bearing on all this.

As you get increasing negative phase angles you get an increasing discrepancy between instantaneous power and actual power. The issue is that the current has to be provided for the instantaneous power, but is given back. The more negative the phase angle at a given frequency the greater the current required for the instantaneous power.

So the bottom line is that as the output impedance of an amp increases then the more these phase angles will limit current and upset the frequency response.

So you have touched on a topic where I have long maintained that bench testing of amps with a resistor does not give you a complete picture of that amp driving pretty much any speaker.

I think this generally gets hidden because of the extent of most speaker aberrations. What I will say is that the better the speaker the more it will benefit from better amplification. So low source (output impedance) is a very important power amp parameter.

This is nothing new, and we have mentioned it here previously. So that is just one of many reasons I use Quad amps which have a very low source impedance.

I'm personally not interested in driving any of my speakers from receiver amps.

Okay, agreed using a resistor load for power output tests is not very good. It isn't a relevant point thought, in his experiment. Regardless of the relationship between VA and Watt, here we are talking about less than 2 Vrms, into a 4 ohm resistor, inductor, capacitor or a combination of RLC, the current is less than 2/4 = 0.5 A, and the power is V*I = 1 VA, that would be 1 W if the load is a resistor and 0.5 W if the load is RLC with phase angle of 60 degrees as we all know cosine 60 degrees is 0.5.

The much increased heat dissipation in the output devices due to the 60 degree phase angle would be insignificant under such low output level and would not affect his measurements.

 

[TLS Guy]

Okay, agreed using a resistor load for power output tests is not very good. It isn't a relevant point thought, in his experiment. Regardless of the relationship between VA and Watt, here we are talking about less than 2 Vrms, into a 4 ohm resistor, inductor, capacitor or a combination of RLC, the current is less than 2/4 = 0.5 A, and the power is V*I = 1 VA, that would be 1 W if the load is a resistor and 0.5 W if the load is RLC with phase angle of 60 degrees as we all know cosine 60 degrees is 0.5.

The much increased heat dissipation in the output devices due to the 60 degree phase angle would be insignificant under such low output level and would not affect his measurements.

I understand that, but we don't all operate our systems under low level conditions. The point I am making is that bench testing is far removed form real world conditions, as far as power amps are concerned. I was just pointing out the limitations. Adverse phase angles, just add an extra dimension as to why there are at least a few grains of salt in the view that all amps are not created equal when driving loads other than purely resistive ones.

It just adds to the case for designing speakers and power amps as a unit. So we really do need to hasten going to active speakers with DSP. It is time to be looking at the end for passive speakers. That will allow all to get much more consistent results.

Lastly, as you know I have used Quad amps exclusively for fifty years now. So it is certainly a possibility, that this fact has in a subtle way and possibly not such a subtle way influenced my design practices. So in effect I have deigned my speakers and and amps as a unit, without making the conscious effort. That is certainly a possibility.

 

[PENG]

Thanks Peng. Just to be clear, the only time I actually performed AC current flow tests was at the very start to find my three test frequencies that would represent easy, difficult and typical speaker loads. I used the AVR-3802 only for this process. Once I had decided on these frequencies, I simply performed AC voltage tests at those frequencies (and unloaded) on all three AVRs at four similar voltage levels at 10dB intervals.

Excellent point, I should have guessed because as I said earlier, clearly you know what you are doing.

Regarding the greater than expected voltage drops, could speaker phase angle have anything to do with this?

As I commented earlier, phase angle is not relevant in this case. All that is involved is Ohm's law for your measurements. V=IZ, or I = V/Z, Z will be a complex number so there is a phase angle component, but for your measurements, you are only concerned with the magnitude part. And to calculate the magnitude part only, the same I = V/Z formula would apply, and it is often written as : I = V/|Z|, |Z| is known as the modulus or absolute value.

Understanding Impedances (mit.edu)

I guess the question I'm asking you is this - is the extra stress on an amp caused by the speaker's phase angles baked into the impedance figures I derived from the AC current flow? And if not, how does it manifest itself in terms of impacting an amp's output voltage to the speaker?

The extra stress would be the heat that has to be dissipated in the amp as the speaker would absorb less "power". For example, at 60 degrees phase angle, lagging or leading, only half of the "power" output would be dissipated/consumed/absorbed by the speaker, the rest of it would have to be dissipated in the amp as heat. In extreme cases, a highly reactive speaker could cause instability issues in some amps, in addition to the extra heat.

when I last studied physics at high school many moons ago.

That explains some of it, high school physics do cover the principles of electricity, not just heat, light and sound..

I mentioned earlier that based on your measurements, the voltage drop should be less than 1%, and if you want to see the calculations:

Again the only electrical formula needed is V = I*Z

Let's use the Yamaha's numbers based on volume -16.5:

Voltage at binding posts......................................................................................................1.696 V
Impedance of the speaker as calculated............................................................................4.9 ohms

Amplifier output impedance as calculated from DF = 150, Zs = Zin/DF = 8/150=...........0.05333

Current (I) = V/|Z| = 1.696/(4.9+0.05333) = ............................................................................0.3434 A
Voltage drop due to Zs (output impedance = I*4.95333 = .................................................... 0.01826 V
Voltage drop in % = 0.01826/1.696 = ..................................................................................... 1.0767 V

So 1.077 V vs 2.2 V as shown in your measurements. That's not that far off in the grand scheme of things at least for the Yamaha. The numbers for the AVR-3802 does look way too too much, but again I don't know what would have caused the huge discrepancy. I can see it might actually has 2X the amp output impedance, but hard to imagine much more than that, yet your measurements show >6% voltage drop that's 3X higher than I would expect.

Note: That's for load impedance of 4.9 V, if you calculate it for the 7.2 V case, you will get less than 1 V.

Feel free to check my calculations in case I made some typos..