Complex impedance load amp FR influence - WiiM Amp review by Erin

[ctrl]

Was pleasantly surprised today when I read the review of the WiiM Amp on Erin's website. Erin will be using two complex impedance profiles in all of his amplifier reviews in the future to measure their impact on the frequency response of the tested amplifiers.

My only criticism is that the impedance profiles, with two peaks around 70Hz and 1200Hz at 20 ohms impedance, seem a bit low:

To cover around 90% of speaker impedance, impedance profiles with peaks of around 60 ohms would actually be necessary - for example speaker with 67 Ohm peak at 62 Hz:

This would lead to significantly stronger effects on the frequency response of certain amplifiers than what is observed with only 20 ohms.

However, even with the impedance profiles used with 20 ohm peaks, the benefits of such measurements are evident. In the tested WiiM Amp. A 0.6dB SPL increase is observed in the presence range of 2-4kHz, which can result in an aggressive perceived sound:

When using speakers that have an impedance peak over 10 ohms in the frequency range of 1-4kHz (caused by the crossover), it would be advisable to perform impedance equalization (RLC circuit parallel to the speaker terminal) in this frequency range for roughly $20 when using such an amplifier to avoid this kind of FR change.

Such impedance equalization can be easily performed individually for each speaker model for which impedance measurements are available, using tools like VCAD. Which in turn leads to amplifier frequency responses as measured with the typical constant loads of 4 or 8 ohms. So the speaker can typically be used with the amplifier without significant loss of sound quality.

 

[Speedskater]

The 67 Ohm peak at 62 Hz would be the bass-reflex resonant frequency. It's not the peaks that are a problem, it's the dips.

A decade or so ago, there was a discuss about whether the Stereophile loudspeaker model was representative.

 

[ctrl]

The 67 Ohm peak at 62 Hz would be the bass-reflex resonant frequency. It's not the peaks that are a problem, it's the dips.

A decade or so ago, there was a discuss about whether the Stereophile loudspeaker model was representative.

A complex impedance load that covers 90% of the common speaker impedance variations at its extremes would also be preferable to me, as I mentioned before.

The most important insight for me from these initial measurements is that many current speakers cannot be used with inexpensive Class-D amplifiers without impedance correction, unless one is willing to accept a 1dB frequency response deviation below 10kHz - which is easily audible since this deviation only occurs in a specific frequency range.

On the other hand, when using an RLC circuit or a Zobel network, it should be possible to linearize the speaker impedance frequency response in such a way that the frequency response changes of the load-dependent amplifier are hardly noticeable.

Example 1 - ELAC Debut Reference DBR62

Impedance (source EAC):

This speaker has a 50 Ohm impedance peak at 2.3kHz that will spoil the FR of many load-dependent amps.

Correction with Zobel network (green curve):


Correction with RLC circuit (green curve):

Best results will provide a RLC circuit - from 0.5 - 30 kHz constant 10 Ohm.


Example 2 - JBL Studio 698

Impedance (source EAC):

This speaker has a 56 Ohm impedance peak at 6kHz.

Correction with Zobel network (green curve):

It's not perfect, but should give better results than without impedance correction.

The shown circuits, which are connected in parallel to the speaker terminal (thus before the crossover), incur costs of $5-12 per speaker (inexpensive electrolytic capacitors, ceramic resistors, and for the RLC circuit, 0.71mm air-core inductors) and thus avoid one of the biggest disadvantages of load-dependent amplifiers.

To decide on the use of impedance correction, one needs an impedance measurement of the speaker model and the frequency response measurement of the amplifier with a complex load - find it enriching that Erin provides both.

 

[Yorkshire Mouth]

How sure are we that +/- 1dB deviation is audible in normal usage?

Here is Amir’s estimated in-room response for the Revel M106, which I believe must be the flattest FR if all the speakers he’s tested, or very close. There are regular deviations from flat of +/- 1bD.

 

[Digital_Thor]

How sure are we that +/- 1dB deviation is audible in normal usage?

Please remember, the Q value
+/-1dB deviation actually just means that a band of frequencies can be 2dB higher than another band - like 2dB more level in the midrange than a tweeter. With a Q of 1, that difference would be huge and very audible - Q of 15 - not so much at all. So it all depends, unless we have - details

 

[Deleted member 48726]

Nice! I will be following his reviews with great interest. Thanks for informing us.

1 dB is indeed audible if it's broad as @Digital_Thor points out. I think too his high impedance peaks are a bit on the low side. Many class D amps without proper feedback may reproduce a heightened response here. But at least we finally have complex load testing and maybe the circuit will be changed sometime.

 

[Yorkshire Mouth]

Please remember, the Q value
+/-1dB deviation actually just means that a band of frequencies can be 2dB higher than another band - like 2dB more level in the midrange than a tweeter. With a Q of 1, that difference would be huge and very audible - Q of 15 - not so much at all. So it all depends, unless we have - details

In this instance it’s +1dB and you can take out the -1dB, apart from the blue line, where you can remove the “+1dB” and replace “-1dB” with “-0.5dB”.

Pretty much all deviations above 0.5dB are for 10kHz and above.

Just saying…

Please let’s remember this is a £300 streamer, DAC, PEQ-ing amp. And the above weakness is only applicable if you drive it very hard.

Personally, I’d still love to see a pre amp model of this costing £200 so you can add whatever amp you like.

Looking forward to the Ultra.

 

[Digital_Thor]

In this instance it’s +1dB and you can take out the -1dB, apart from the blue line, where you can remove the “+1dB” and replace “-1dB” with “-0.5dB”.

Pretty much all deviations above 0.5dB are for 10kHz and above.

Just saying…

Please let’s remember this is a £300 streamer, DAC, PEQ-ing amp. And the above weakness is only applicable if you drive it very hard.

Personally, I’d still love to see a pre amp model of this costing £200 so you can add whatever amp you like.

Looking forward to the Ultra.

Absolutely. Thing is, that when more than one deviance is combined with another, suddenly they can be greater than first anticipated, while other times counter-act each other, making the overall result better or worse. As initiated in the headline - it can be complex - especially if one brand/product deviate more than usual, from another that it is mated with.
I'm all for products that comply with a standard and live up to it properly - which is why we test things - so that cables, speakers, SPL and other factors, do not make that big of a difference to the end result when we enjoy our music.

That little Wiim amp is an awesome package of features. But I would be in no doubt that it'll crumble with difficult speakers. I have tried so many times to have people ask for help with speakers from known brand like Dynaudio, B&W and Focal, that simply pull the life out of their so called "beefy" amplifier, simply because they did not know that a powerful amplifier is something else than just weight, size and price - even though praised by 95% of users.
Proper knowledge can be quite hard to come by, when you really dig into detail. I only learned a little, by trying a much more powerful amplifier, with triple driver stage and playing with it on a set of B&W 800 Nautilus (the one with dual 10"). A huge set of Marantz mono-blocks and Lyngdorf could not do it. With my amplifier, the bass was suddenly way firmer and everything fell into place, with the speaker now sounding like it optimally could.
Did I measure this - no. But I saw the membrane of the woofers, suddenly move way more firmly and precisely to the music, making them play deeper and louder and with less "boom" and more "smack".
I believe that it also confuses many that even tiny speakers, can have a near short circuit in the impedance, ripping many amplifiers to shreds. But the audibility can be anything from "lose" or lack of bass to distinct distortion.
I remember a small set of DALI's back in the beginning of the 1980's, that would pull the teeth of any amplifier, except a "welding machine" from Mark Levinson.

Maybe this just boils down to the point, that if you paired the Wiim with a good power amp, it would shine more - no?

 

[voodooless]

Such impedance equalization can be easily performed individually for each speaker model for which impedance measurements are available, using tools like VCAD. Which in turn leads to amplifier frequency responses as measured with the typical constant loads of 4 or 8 ohms. So the speaker can typically be used with the amplifier without significant loss of sound quality.

Or you can just use the build-in EQ to flatten the response if needed. Costs nothing!

 

[Digital_Thor]

Or you can just use the build in EQ to flatten the response if needed. Costs nothing!

True - if the response is linear and not changing with load and volume, then that is always the way.

 

[ctrl]

Here is Amir’s estimated in-room response for the Revel M106, which I believe must be the flattest FR if all the speakers he’s tested, or very close. There are regular deviations from flat of +/- 1bD.

We should differentiate between the sound and behavior of a speaker as intended by the developer/manufacturer (regardless of whether this is "good or bad") and a significant change in frequency response caused by a playback device - which is the case with the WiiAmp with complex speaker load.

The changes in the frequency response of the load-dependent amplifier will be reflected as alterations in the PIR (predicted in-room response) of the speaker you displayed.
If the speaker then exhibits an additional boost in the presence range 2-4kHz (see blue FR change below), adding to an already non-optimal uniform frequency response behavior, it can lead, for example, to an aggressive sound impression from the speaker.

deviation is audible in normal usage?

As others have mentioned, a 1dB increase in the 2-4kHz range is easily perceptible. Even a +0.5dB boost in the presence range with Q <=1 should be noticeable to many - of course, this depends greatly on the music piece. If the speaker were to increase or decrease by one decibel across the entire human frequency range, it would be hardly audible, as it would then be at the threshold of perceptibility of 0.5-1 dB. However, if only specific frequency ranges are altered, the perceptibility is much better.

In this instance it’s +1dB and you can take out the -1dB, apart from the blue line, where you can remove the “+1dB” and replace “-1dB” with “-0.5dB”.
Pretty much all deviations above 0.5dB are for 10kHz and above.
Just saying…

However, you should consider that the complex load in Erin's review only covered the range of 3-20 ohms. For many speakers, the extremes are much more pronounced. Impedance below 2 ohms and above 60 ohms are quite possible.

In post #3, I showed a few examples of popular speakers that exhibit impedance profiles ranging from 4 to 56 ohms. In those cases, the ripple in the frequency response caused by load-dependent amplifiers is likely to be more pronounced, and might be reaching the +1dB threshold of frequency deviation below 10kHz.

Or you can just use the build in EQ to flatten the response if needed. Costs nothing!

But each speaker model causes a different frequency response change with load-dependent amplifiers and not all listener have access or want to use EQ.
Impedance linearization is a one-time process and works without any changes even when switching amplifiers.

 

[voodooless]

But each speaker model causes a different frequency response change with load-dependent amplifiers and not all listener have access or want to use EQ.

All users of this amp have EQ. Not wanting it is just silly.. we can stop the objectively good sound discussion right there… and yes, you have to measure, you should do that anyway. Passive impedance correction you have to measure as well

 

[Matt_Holland]

I really like Erin’s amp measurement format. Particularly I like the multi tone test with both channels driven to maximum output.

Hopefully he will review a few amplifiers that are high-SINAD on ASR so we can see how they fare with his tests. Maybe he could create his own amplifier scoring system that takes load tolerance and the multi tone both channels test into consideration.

I really like everything Erin is doing.

 

[voodooless]

Particularly I like the multi tone test with both channels driven to maximum output.

Why would you care what the amp does way into clipping? Best avoid getting there in the first place. It tells you absolutely zero about the performance of 1dB less gain.

 

[Matt_Holland]

Why would you care what the amp does way into clipping? Best avoid getting there in the first place. It tells you absolutely zero about the performance of 1dB less gain.

The test wasn’t way into clipping though was it?
By my logic if an amp cannot output the same power when both channels are driven with a complex signal compared to one channel with a single pure tone, then it is deficient in its ability to deliver power consistently. In addition if this metric shows worsening performance as impedance drops then it points to even more inconsistency of power delivery, particularly if the load is very complex in current demanding low frequencies.

I’m very much a layman in this area though so I am very happy to be educated. I presume Erin is doing these tests for good reasons?

 

[voodooless]

The test wasn’t way into clipping though was it?

It shows about 5% THD, so yes, this is well into clipping territory.

 

[Matt_Holland]

It shows about 5% THD, so yes, this is well into clipping territory.

I’m interested in the difference in power output comparing one channel single tone to both channels multi tone. If this test was done at 0.1% THD would it have more merit?

 

[voodooless]

I’m interested in the difference in power output comparing one channel single tone to both channels multi tone. If this test was done at 0.1% THD would it have more merit?

What is missing here is THD vs Frequency vs Power. The reason the multitone only gives 90W vs 120W can probably be found there. I would guess that it cannot deliver 120W in bass. But that is just a hunch, we don't have the measurements to back this up. This is indicative of the issue with the high-power multitone: you have no idea what to conclude from it.

 

[Yorkshire Mouth]

It'd be interesting to see the same with a subwoofer dialled in. The unit does, after all, have a subwoofer out and bass management.

 

[Matt_Holland]

What is missing here is THD vs Frequency vs Power. The reason the multitone only gives 90W vs 120W can probably be found there. I would guess that it cannot deliver 120W in bass. But that is just a hunch, we don't have the measurements to back this up. This is indicative of the issue with the high-power multitone: you have no idea what to conclude from it.