How do these two charts come from amps with [essentially] the same THD+n measurement (1kHz @ 5W / 4 ohm)

[mike7877]

How can this:

And this:

Both come from amplifiers with:
PA3s: -87 & -85 dB THD+n
3700H: -89 & -87 dB THD+n



Edit: literally every PA3s line is above -70 and all but maybe 1% of the 3700H's are below -70!



I have the Denon X3700h, and for sure the charts above are a much better representation of the audible performance difference than the 2dB THD+n rounding error.

The chart above.. the PA3s is -65dB THD+n @5W and 1kHz.

Am I missing something? Or is the PA3s supposed to be rated 20dB worse than it is?

Mini review below from the 6 minutes of listening I did before deciding I'll be returning
(2 mins ea of 3 songs I'm very familiar with):

Compared to what's available at big box stores, the PA3s is good value for money, but considered as its own component, it's far from my favourite. The upper treble is too present and slightly harsh. Exacerbated by weak bass which is not only anemic, but lacks definition..
Sure, it probably sounds better than anything you can buy in a "real" store for the same money, but the bar there is pretty low...

If you live in a building and don't need a lot of power, and your speakers are lower to mid entry level with the following characteristics:
- a bit bass heavy
- a soft top end
(which are common in the category)
and your budget is $170 CDN max, I'd say go for it. Otherwise, keep looking!
(I don't know all that's out there - this could be all there is, so before following my advice to keep looking, make sure there's actually other contenders lol) don't forget the used market!

 

[Mnyb]

Conditions for the measurement are not stated when it’s just one number it can be at 1kHz only who knows ?
That’s why you are supposed to spec at some standard or clearly state what your conditions are for example 20-20kHz.

But the bigger issue is actually it’s load dependency. Frequency response in the treble is speaker impedance dependent due to the design , no post filter feedback in a class d amp does this due to high output impedance in the treble.

 

[Mnyb]

Frequency response problems in large part of spectrum is much more audible than distortion.

Especially THD at 15kHz where the pa3 is not so good . The second harmonic to 15kHz is 30kHz which you can’t hear .

And the measurement bandwidth 45 kHz which is necessary to measure this kind of class d amp migth not be enough to fully resolve this .

If you hear a difference .

I bet my dollar on the fr response problems which are obvious , I would not by this amp for serious listening .
Or the difference in power between the amps ( the most important spec of an power amp is actually power , that’s whats it for )

 

[mdsimon2]

You are comparing measurements of different bandwidth.

Michael

 

[Gruesome]

Maybe the PA3s was remeasured with a different bandwidth? I agree that those THD+N graphs do not at all look like better than 70 dB SINAD.
This is the earlier 1 kHz plot in the same review:

 

[Mnyb]

Sorry for rambling..

But load depending frequency response actually translates to different frequency response for different speakers. The pa3 migth work fine for some speakers and rubbish for other speakers and you can’t predict the outcome ?

Hence this is a property you don’t want in a power amp .

 

[mike7877]

You are comparing measurements of different bandwidth.

Michael

They both say 45kHz though...

 

[mike7877]

Conditions for the measurement are not stated when it’s just one number it can be at 1kHz only who knows ?
That’s why you are supposed to spec at some standard or clearly state what your conditions are for example 20-20kHz.

But the bigger issue is actually it’s load dependency. Frequency response in the treble is speaker impedance dependent due to the design , no post filter feedback in a class d amp does this due to high output impedance in the treble.

View attachment 292090

Oh.. my... how did I miss that? I got the PA3s though - is it different?

If not, that sure explains the bright top end... my speakers are 8 ohms!

 

[Gruesome]

But the 1 kHz curve from the later plot does not agree with the 'official' distortion vs power curve earlier in the review. So something changed.

 

[mdsimon2]

They both say 45kHz though...

Dashboard is 22 kHz bandwidth, this is why the PA3s looks better on the dashboard.

Michael

 

[Mnyb]

They both say 45kHz though...

We have to ask one of the experts, it may also be a filter involved when measuring class d amps ? I’m not sure

 

[Mnyb]

Dashboard is 22 kHz bandwidth, this is why the PA3s looks better on the dashboard.

Michael

Thankyou that’s explains a lot dashboard is also at 1kHz rigth ? Edit . Yes dashboard on amirs measurement are at 1kHz

 

[mdsimon2]

Thankyou that’s explains a lot dashboard is also at 1kHz rigth ?

Yes.

Michael

 

[Mnyb]

Oh.. my... how did I miss that? I got the PA3s though - is it different?

If not, that sure explains the bright top end... my speakers are 8 ohms!

Btw 8 ohm is just some average or marketing number ?
For the actual impedance in the treble you need to look at an impedance vs frequency plot of speakers

Here is a random specimen.

 

[mdsimon2]

Dashboard is 22 kHz bandwidth, this is why the PA3s looks better on the dashboard.

Michael

And should say audible differences are likely due frequency response as suggested by @Mnyb. This will likely be worse with a real speaker due to the load dependent response of the PA3s.

Michael

 

[Gruesome]

Ok, that makes me a whole lot happier with my A30!

Just kidding. As I remarked elsewhere, it would be nice to get a universally accepted replacement for SINAD that includes more than just harmonic distortions at a single frequency.

 

[Mnyb]

Ok, that makes me a whole lot happier with my A30!

Just kidding. As I remarked elsewhere, it would be nice to get a universally accepted replacement for SINAD that includes more than just harmonic distortions at a single frequency.

You can’t have it in one number , one must look at the total picture of all the measurement.
And put them in context.

For example the pa3 migth be just fine with a speaker with a very smooth impedance plot listened to at lower power levels .

Better amps are safer to predict ? If you have very low output impedance in all the audio spectrum you know it won’t change the frequency response.

And if it has a lot of power it can drive all speakers.

If it also handles low impedance it can really drive all speakers !

Then we can look at sinad , but also look at the plot is it distortion or noise dominated sinad .
Noise is more problematic than THD so you want to product to be quiet .

But if sinad is very high we know that both noise and THD are at very low levels so it’s fine

That’s how I think about this .

 

[mike7877]

Btw 8 ohm is just some average or marketing number ?
For the actual impedance in the treble you need to look at an impedance vs frequency plot of speakers

Here is a random specimen.

View attachment 292095

When my speakers were new I was interested in spec'ing them as much as I could. I wanted to make a custom crossover with a lower point.

Because I want these speakers to last until I'm dead (and I'm not even old yet), I need to treat them well. That means keeping temps down to no higher than 30-40C by limiting average power to around 5WRMS if it'll be playing for a few hours at a time. Also keeping excursion down at reasonable levels so no parts are unduly stressed. Although this tweeter has the longest travel capability I've ever seen on a tweeter (by 3x!!!!!), I still don't want to have it moving more than 0.3mm peak-to-peak for transients, and would be much happier if it were 0.12-0.15.

I used an online website tool you can find by typing into a search engine "piston excursion calculator". I set the piston to 1", and from there you can choose the frequency and SPL and it tells you how many mm the driver would need to move to create that SPL.

I basically discovered that 1.6kHz was the lowest I could cross the tweeter, and 1.8kHz was more comfortable.

I then went and found the tweeter's impedance curve - I needed to know how much the tweeter's impedance rose at resonance, and obviously what the resonant frequency is. I did this using my trusty Digital Multimeter. It's a good one - linear up to 2000Hz! Unfortunately, not linear past that (it's not bad, but...).
So first, what I needed to do was, use my plain 140W class A amplifier (linear through 100kHz), driven by my recording audio interface (professional RME device, definitely linear). I think it was 1kHz 10.00V I started from, and from there, in 3 or 4 steps per octave, I rose to 20.0k. By then I think my DMM was registering 6.85 volts or something. Anyway, I plotted and then graphed (manually on graph paper lmao).

Then I found the results for the tweeter, and adjusted the rdsults by multiplying the tweeter DMM readings by 10/6.85 for 20.0 etc, then used them to do the math for tweeter impedances at x frequencies.

Turns out, the peak was quite high at 25 ohms down around 800Hz, and quite wide. Basically, I found that the lowest reasonable crossover point with a passive crossover was in the 1.8-1.9 kHz range. Simce it's already 2.1, I scrapped the idea.

Oh, about its impedance. It was surprisingly flat! Obviously like all dome tweeters it has that characteristic impedance rise as frequency goes up, but it didn't rise very much. I think up near 20kHz was still in the single digits. The lowest part of the impedance curve was also quite high. If memory serves it was low 7s, (7.1, 7.2 I think? If not, no lower than high 6's, and I believe the trough was in the high 3 to low 4kHz range. It's been almost 4 years now! stupid pandemic....)

Now some time in the future I'm going to do active crossovers. Since everything can be adjusted and tuned on the fly, and phase can be perfected so perfectly it's almost unbelievable. For ultra-wide dispersion, I'm going to go for 1.2kHz (quieter listening obv- still quite loud, just 6dB-10dB less peak power. Crossing at this frequency will allow the 135mm midbass to radiate omnidirectionally to like 700hz, and then to at least 180 degrees until its output is down at least 12dB by 1.6-1.7kHz The in room power response, especially with it being 3rd order, is going to be phenominal! Closest thing to Magic that exists!
If you can't tell I'm excited

 

[pma]

How can this:

And this:

Both come from amplifiers with:
PA3s: -87 & -85 dB THD+n
3700H: -89 & -87 dB THD+n

The answer to your question is extremely simple. SINAD (THD+N) at 1kHz/4ohm/5W tells nothing and nothing else than the behaviour at 1kHz/5W/4ohm. Because amplifier non-linearity is frequency dependent and level dependent, you get nothing more from that single number than you already get. Unfortunately, laymen audience started to take this number as a measure of amplifier quality, as a result of Amir's decision to post the comparative SINAD 5W/1kHz chart.

 

[MaxwellsEq]

The answer to your question is extremely simple. SINAD (THD+N) at 1kHz/4ohm/5W tells nothing and nothing else than the behaviour at 1kHz/5W/4ohm. Because amplifier non-linearity is frequency dependent and level dependent, you get nothing more from that single number than you already get. Unfortunately, laymen audience started to take this number as a measure of amplifier quality, as a result of Amir's decision to post the comparative SINAD 5W/1kHz chart.

Hmm, I agree that a single measure can not describe what a power amplifier does and that the reader should study all of Amir's charts and explanations to really understand how an amplifier is likely to perform with real loads, but I think it's unfair to criticize Amir for posting comparison charts.

At least somebody is doing these measurements whilst not in the pay of advertisements! If a buyer only looks at the comparison charts they have themselves to blame for not reading the rest of the review. But at least they won't be buying truly bad-measuring kit!