• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Amplifier measurements may require improvement

Djano

Member
Joined
Sep 10, 2023
Messages
62
Likes
52
Location
France
I apologize if this message should have been posted elsewhere. I couldn't find a "Complain about X measurements" thread, as we see with headphones or speakers.

The Fosi V3 measurements provided by Amir did not highlight a defect that was nevertheless observed in the measurements offered by Scientific Audiophile (video, from 20:00).

S.A. discovered that the Fosi V3, when heavily stimulated, exhibits irregular response across the frequency spectrum. The main issue is that it noticeably collapses below 145Hz, whereas this problem is not observed beyond that point. Here is his graph comparing the V3 to a more traditional class AB amplifier :
v3.JPG

I suggested to S.A. to conduct another test to verify his hypothesis that this result can be explained by the fact that low frequencies require more power. He seemed interested. These questions have been discussed here and in the comments section of S.A.'s video.

Therefore, it seems to me that tests should include a power test conducted across the entire audible spectrum. The frequency response appears to be subject, in addition to load dependency, to a "Watts dependency" (?).

I am clearly new to this field. Did Amir's measurements allow for the prediction of this defect in a way that I am unaware of?
I am already very grateful for the time Amir is willing to take to collect and make so much interesting information available to everyone. But we should not miss an opportunity to improve the protocols if they have indeed fallen short.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,513
Likes
10,529
Location
Prague
It would be good to see load complex impedance, modulus + phase, to see what forced Fosi to collapse.
 
OP
D

Djano

Member
Joined
Sep 10, 2023
Messages
62
Likes
52
Location
France
What? Something could happen differently outside of a controlled test environment?
The problem doesn't seem to me to be that the test yielded different results from those in the real environment. The tests simply omit an important aspect of the real use case, which appears entirely possible to test.

Even when tests measure something that doesn't correspond to what happens in real-life situations, it's important to understand why. This allows for improvements in testing protocols to better predict what will happen in actual use. That's the exact purpose of tests.

One cannot simply say, "tests are not real life, kid!" At least, I don't see the point in doing so. This implies that tests don't truly inform us about the real-world performance of devices (which is false) and don't bring any improvement.
 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,126
Location
Northern Virginia, USA
Therefore, it seems to me that tests should include a power test conducted across the entire audible spectrum.
Odd comment. He already does that; for example:

 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,126
Location
Northern Virginia, USA
OP
D

Djano

Member
Joined
Sep 10, 2023
Messages
62
Likes
52
Location
France
It clips, so what?
That could be an explanation, but tests should make it predictible to be exhaustive.
S.A. thinks (but is it true?) that the 89W power limit @8ohm before clipping does not suffice to explain what we are seeing.

Odd comment. He already does that; for example:
A graph in this looks like what I was thinking of, yes. However, it is quite rare to find this sort of graph.
And the graph talks about y=THD+N x=freq. response.
It does not talk about output power (SPL would be even better, but different methodology), but about output "quality". Both could cross with notions like clipping, but the culprit could be something else, maybe
 

voodooless

Master Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
9,962
Likes
17,225
Location
Netherlands
That could be an explanation, but tests should make it predictible to be exhaustive.
S.A. thinks (but is it true?) that the 89W power limit @8ohm before clipping does not suffice to explain what we are seeing.
Well, then he should investigate some more. Just look at the waveform of the audio. Clipping should be evident there, even if recorded by mic.
 

Holdt

Major Contributor
Joined
Jun 10, 2022
Messages
2,880
Likes
2,922
Should it be expected to clip though? And which test parameter would show the probability of clipping? We would need the load impedance / phase graph and at best a signal oscilloscope trace.
Most likely that's all it is.
 

IPunchCholla

Major Contributor
Forum Donor
Joined
Jan 15, 2022
Messages
1,072
Likes
1,315
I'm not going to spend 25 minutes to find out very basic information. This is why I hate videos. What is the testing protocol? It looks like we are talking about measuring an amplifier through in room response of SPL? So I need to know a few things to find out even if I should bother with the results.
 
OP
D

Djano

Member
Joined
Sep 10, 2023
Messages
62
Likes
52
Location
France
I'm not going to spend 25 minutes to find out very basic information. This is why I hate videos. What is the testing protocol? It looks like we are talking about measuring an amplifier through in room response of SPL? So I need to know a few things to find out even if I should bother with the results.
He uses a sweep signal. Yes, he measures speaker sound output (SPL) in room
 

IPunchCholla

Major Contributor
Forum Donor
Joined
Jan 15, 2022
Messages
1,072
Likes
1,315
He uses a sweep signal. Yes, he measures speaker sound output (SPL) in room
OK, so we need to know a lot of things then. What are the characteristics of the speakers? This could be an expected result if we just know the specs of the amp and the specs of the speakers. How does he control for variations in mic position (bass will be very susceptible to this)? What are the measurement to measurement variations?
 
OP
D

Djano

Member
Joined
Sep 10, 2023
Messages
62
Likes
52
Location
France
What are the characteristics of the speakers
Tannoy tower speakers 8ohm, but I don't remember anything else.
How does he control for variations in mic position (bass will be very susceptible to this)?
Non-issue I think. Just leave the microphone, do not touch the speakers, and plug the bananas into the other amp. I do not see why he would have done differently.
What are the measurement to measurement variations?
Good question. I guess slim as treble look alike on both amps. But maybe treble is less susceptible to variations ?

I have to go, and will read later when I can
 

SSS

Active Member
Joined
Jan 12, 2023
Messages
294
Likes
184
Location
Germany
Another Junk video without clear methodology. There are many non-controlled parameters when measuring an amplifier over loadspeaker via microphone. The result says nothing. Maybe the Yahama amp graph version is bad too.
 

Holdt

Major Contributor
Joined
Jun 10, 2022
Messages
2,880
Likes
2,922
Another Junk video without clear methodology. There are many non-controlled parameters when measuring an amplifier over loadspeaker via microphone. The result says nothing. Maybe the Yahama amp graph version is bad too.
It's not. You can literally see the graphs compared in the OP..

I can't see nothing wrong with switching amps and setting outputs to match and then doing sweeps.
 
Top Bottom