Measuring FR of an Amp alone (w/o speaker) may not be accurate. Your thoughts?

[dorakeg]

I have been thinking that measuring the FR of an amp under a constant simulated load (eg. 8ohms) may not be accurate. Not doubt we have seen that most amps are able to show constant ruler flat line for FR (I just use the Krell KAV-400XI as an example). What I wanted to highlight is the difference in level for each impedance. 4 ohm is around 0.25db while 2 ohm is ~0.75db. Important thing is they are not the same. I have already put the link for the Evo 600. ITs still flat but the difference between each impedence is much smaller (~0.12db for 4ohm and ~0.4db for 2 ohms).

Krell KAV-400xi integrated amplifier Measurements

Sidebar 2: Measurements

www.stereophile.com

Krell Evolution 202 preamplifier & 600 monoblock power amplifier Measurements

Sidebar 3: Measurements I don't have the necessary equipment to test the Krell components in their current-mode (CAST) operation. Therefore, I performed all the technical tests using conventional voltage-mode drive, in both balanced and unbalanced operation.

www.stereophile.com

Next look at the impedence of the Dynaudio Special 40 (just an exmaple). Its rated at 6 ohm but if you look at the graph, it varies alot depending on the FR. Such variations in impedance is typical for vast majority of the speakers.

Dynaudio Special Forty loudspeaker Measurements

Sidebar 3: Measurements I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Dynaudio Special Forty's frequency response in the farfield, and an Earthworks QTC-40 for the nearfield and spatially averaged room responses.

www.stereophile.com

So far, the tests done was for fixed impedance. We assume the impedance is constant at 8/4/2ohmes (most commonly tested). But, we know that this is not the case in the real world. Impedance varies at difference frequency.

Do you think the FR of an amp will vary due to the changes in the impedance of the speaker or will it still be a straight line? IF we use a varying simulated load to test the output of each amp, I feel we are not going to get a stragiht line and different amps will likely get different FR.

I would say that the only time we can still get a straight line is that there is 0 difference in sound level with regards to impedance.

Another thing is that the THD is likely to be different as well. IF you look at the Evo 600 THD into 8/4/2 ohm load, they are different. So, I am also thinking if the THD will vary if we use a varying simulated load (to mimic that of a speaker).

 

[GXAlan]

This is true.

My Marantz PM-10 didn't seem very different from the older PM-11s2 for test tones into a resistor. When I introduced a Studio 530 into the mix and was measuring things electrically not acoustically, the PM-10 was superior to PM-11s2 in performance in terms of THD (how the load increases THD compared to a resistor) and consistency of the frequency response.

The hard part is that there is no "great" simulated load. I just picked the Studio 530 because I could plant the whole thing face down so I wouldn't go deaf testing the speakers.

 

[mhardy6647]

Why not hook the DUT to an actual loudspeaker load and measure with a high impedance...umm... measuring thingy?
It would (could) certainly work with a VTVM (or a DMM, if you insist) or an oscilloscope.

 

[jae]

Aren't well-designed modern amplifiers such as purifi etc. more or less "load agnostic" in this fashion? Plus this is always modern measurement/dsp to fix issues. With many people moving to dsp controlled actives this is even less of a thing to worry about.

 

[dorakeg]

Aren't well-designed modern amplifiers such as purifi etc. more or less "load agnostic" in this fashion? Plus this is always modern measurement/dsp to fix issues. With many people moving to dsp controlled actives this is even less of a thing to worry about.

I don't think you will perform DSP adjustments when you perform a blind or double blind test.

 

[Beave]

Damping factor (related to output impedance) tells a lot about how an amp's frequency response will vary with load.

Most modern solid state amplifiers have sufficiently low output impedance (sufficiently high damping factor) that the frequency response won't vary by more than, what, a couple tenths of a dB?

Stereophile measures amp frequency response into purely resistive loads AND into a simulated "real world" loudspeaker load (one that varies in magnitude and phase over frequency). With most solid state amps, the difference is negligible.

 

[Blumlein 88]

Stereophile has a simulated loudspeaker load they use. Obviously every speaker is different, but it gives an idea of how load dependent an amp is vs FR.
Krell, then VTL monoblock, and Purifi based Nad power amps for example. 8,4, 2 ohms and simulated load in black.

https://www.stereophile.com/images/1119K300fig01.jpg

https://www.stereophile.com/images/414VTLfig01.jpg

https://www.stereophile.com/images/521NADfig01.jpg

 

[dorakeg]

Stereophile has a simulated loudspeaker load they use. Obviously every speaker is different, but it gives an idea of how load dependent an amp is vs FR.
Krell, then VTL monoblock, and Purifi based Nad power amps for example. 8,4, 2 ohms and simulated load in black.

https://www.stereophile.com/images/1119K300fig01.jpg

https://www.stereophile.com/images/414VTLfig01.jpg

https://www.stereophile.com/images/521NADfig01.jpg

Oh nice!! Thank you very much for the info, I didn't notice this!!

FRom the graphs, I am thinking that the listening volume will likely result in smaller/bigger FR difference?

 

[kongwee]

Aren't well-designed modern amplifiers such as purifi etc. more or less "load agnostic" in this fashion? Plus this is always modern measurement/dsp to fix issues. With many people moving to dsp controlled actives this is even less of a thing to worry about.

Even with DSP amplifier, you can't get as sufficient as crossover circuit. You are mincing the crossover point. VA power are likely to be the same toward crossover circuit equivalent. It is different issue with EQ or DRC.

 

[Axo1989]

Stereophile has a simulated loudspeaker load they use. Obviously every speaker is different, but it gives an idea of how load dependent an amp is vs FR.
Krell, then VTL monoblock, and Purifi based Nad power amps for example. 8,4, 2 ohms and simulated load in black.

https://www.stereophile.com/images/1119K300fig01.jpg

https://www.stereophile.com/images/414VTLfig01.jpg

https://www.stereophile.com/images/521NADfig01.jpg

Great examples. Can you link the measurements pages also not just the images? (both for Stereophile attribution and so I can see which amps specifically, that's neither of the Krells the OP linked, for example).

 

[Axo1989]

... Another thing is that the THD is likely to be different as well. IF you look at the Evo 600 THD into 8/4/2 ohm load, they are different. So, I am also thinking if the THD will vary if we use a varying simulated load (to mimic that of a speaker).

I've heard both those Krells (or near-identical stablemates). The baby KAV I owned, the Evolution monoblocs a few times driving big Focals (I tried the baby in the same store also). My subjective impression is that THD/IMD rising significantly with increasing output on that KAV was quite audible as sonic congestion at higher levels, whereas the big ones played very loud but very clear, so you could still hold a conversation (despite some exhilarating post-hardcore exceeding 105 dB).

Any FR differences were too subtle for me to recall via in-store comparison though. I didn't compare at home but may have noticed there, who knows.

 

[Willem]

The real basket cases are tube amplifiers, which may well explain their appeal for audiophools.

 

[fpitas]

The real basket cases are tube amplifiers, which may well explain their appeal for audiophools.

It may also explain the "matching" of components that audiophiles swear by. You have to find speakers that have the particular goofy response you prefer, once they're hooked to the tube amp.

 

[tomtoo]

Yeap tube amps enjoy to give you some more db at the impedance peaks of the speakers coused by there high inner resistance(bad damping factor).
A link in german language(sry) with good explaining.

Lets imagine you have a vented speaker with two impedance peaks at 70Hz and 40Hz. Then the automated tube eq, adds you 2 maybe 3db at each peak? Absolutly possible you like it. Or not.

 

[Blumlein 88]

Great examples. Can you link the measurements pages also not just the images? (both for Stereophile attribution and so I can see which amps specifically, that's neither of the Krells the OP linked, for example).

Krell K-300i integrated amplifier Measurements

Sidebar 4: Measurements

www.stereophile.com

VTL Siegfried Series II Reference monoblock power amplifier Measurements

Sidebar 3: Measurements

www.stereophile.com

NAD C 298 power amplifier Measurements

Sidebar 4: Measurements

www.stereophile.com

 

[Willem]

And as is obvious, only the VTL exceeds the limits of audible frequency response variation.

 

[fpitas]

And as is obvious, only the VTL exceeds the limits of audible frequency response variation.

But that's what makes it so good.

/

 

[RayDunzl]

Do you think the FR of an amp will vary due to the changes in the impedance of the speaker or will it still be a straight line?

Krell FPB 350 mcx at about 5 watts into MartinLogan reQuest electrostatic hybrid here in the audiotorium:

The vertical scale is 1/10 dB and the impedance of the speaker drops severely at the high frequencies.

And Stereophile's measurement of the speaker impedance - not the same model, but the prior version, and should be representative.

The FR dips above follow the impedance dips below

 

[Blumlein 88]

And as is obvious, only the VTL exceeds the limits of audible frequency response variation.

Devil is in the details. I had a smaller VTL for various electrostats I've owned. The response variation was just what was needed. The speaker's very high impedance at low frequencies goosed the lower couple or three octaves a bit and the near short at the high end smoothed a bit of brightness from the upper midrange, while a roughly 20 khz resonance with the output transformers added a bit of air in the upper octave. Now you could do all that with some DSP now, but hey serendipity and all.

 

[Chr1]

I use a VTL valve amp and a Lazarus hybrid on my "brighter" speakers and have yet to manage to make them sound better with DSP/EQ only. (Reckon most B&Ws would suit valves for this reason.)

Looks to me like if you have speakers with a widely varying impedance and low sensitivity then you need an amp with plenty of power to keep an even frequency response.
Just bought a pair of Neumann KH310s and looking at the power of the internal amplifiers (150W, 70W, 70W), makes me think that as long as there's no hiss or noise, you probably can't have too much power.