|
|
Post by unbeliever on Jun 14, 2017 12:15:54 GMT -5
Quick question: Why does the AP-report not back up the specs for the XPA Gen-3? Examples:
| Spec | AP | | BW | +/-0.1 (300W) | +/-0.17 (1W) | | BBFR | +0/-0.15 (less than BW?!?) | only graph, lot less than -0.15dB, more like -2dB! | | THD+N | <0.005% @ 100W | 0.0059% @ 500mV (~90W), ~0.07 @ 100W in stpd lvl sweep (THD without noise!)
|
Some discrepancies are rather large...
Also, these values are for one module; with all 7, crosstalk, IMD and SMPS-loading will only make matters worse.
BR,
Henrik
|
|
|
|
Post by garbulky on Jun 14, 2017 12:19:02 GMT -5
Quick question: Why does the AP-report not back up the specs for the XPA Gen-3? Examples:
| Spec | AP | | BW | +/-0.1 (300W) | +/-0.17 (1W) | | BBFR | +0/-0.15 (less than BW?!?) | only graph, lot less than -0.15dB, more like -2dB! | | THD+N | <0.005% @ 100W | 0.0059% @ 500mV (~90W), ~0.07 @ 100W in stpd lvl sweep (THD without noise!)
|
Some discrepancies are rather large...
Also, these values are for one module; with all 7, crosstalk, IMD and SMPS-loading will only make matters worse.
BR,
Henrik
I don't know the answer to your question. However I did find out the meaning of BBFR by googling it. According to Google it means "bare buns fun run." Now I know. www.allacronyms.com/BBFR |
|
|
|
Post by unbeliever on Jun 14, 2017 12:32:46 GMT -5
LOL  Well, that's what happens when you get lazy and abbreviate. Shame on me... BBFR = Broad Band Frequency Response (although google's suggestion sounds more fun). |
|
|
|
Post by unbeliever on Jun 14, 2017 12:34:42 GMT -5
BTW, same question applies to the mighty XPA-1, large discrepancies...
|
|
|
|
Post by garbulky on Jun 14, 2017 12:35:32 GMT -5
BTW, same question applies to the mighty XPA-1, large discrepancies... I didn't know. Like what? Probably the SNR right? I am interested. BTW, not contesting the discrepancies... but the XPA-1 gen 2 sounds phenomenal imo. The best amp I have heard...though some tube amps do come close. Near perfection in sound imo. |
|
|
|
Post by unbeliever on Jun 14, 2017 13:01:05 GMT -5
Don't get me wrong, measurement are notoriously bad representatives for sound quality!
Just look at tube amps (as you mention); THD easily reaches full percentage values. THD testing does NOT take 'sound flavour' into account!
I myself own both an XPA and XDA's from Emotiva and I'm very satisfied.
Regarding 'Spec vs AP' for the XPA, most properties differs but the AP THD/watt ends up att ~0.5% at 1kW which is 5 times out of spec.
Again, not complaining about neither specs nor AP-measurements, both are pretty good.
Just curious as to the reason for the differences!
Also, cudos to Emotiva for presenting AP-reports in the first place. Not many other companies dare...
|
|
|
|
Post by garbulky on Jun 14, 2017 13:12:54 GMT -5
Don't get me wrong, measurement are notoriously bad representatives for sound quality! Just look at tube amps (as you mention); THD easily reaches full percentage values. THD testing does NOT take 'sound flavour' into account! I myself own both an XPA and XDA's from Emotiva and I'm very satisfied. Regarding 'Spec vs AP' for the XPA, most properties differs but the AP THD/watt ends up att ~0.5% at 1kW which is 5 times out of spec. Again, not complaining about neither specs nor AP-measurements, both are pretty good. Just curious as to the reason for the differences! Also, cudos to Emotiva for presenting AP-reports in the first place. Not many other companies dare... I agree with it's not always obvious which is going to sound better when certain baseline specs have been met. For instance the XPA-1 gen 2 in class A mode has I believe a SNR in the 70's at 1 watt due to heat. Not as amazing as you'd probably expect a flagship amp. When I got it, I was worried it wouldn't do that great with detail. Only to have me hear significantly more detail than with the other amps I've heard. I have no problem with you taking issue with measurements and reporting. It is important to be accurate. Also I take this feedback as constructive because it helps the company from publishing errors. A sort of free review in a way. |
|
|
|
Post by Gary Cook on Jun 14, 2017 13:53:00 GMT -5
I don't see any discrepancies, the criteria for the measurements are different. For example BW is at 100 watts in the specs and at 1 watt in the AP report, so of course the results are going to different.
Cheers
Gary
|
|
|
|
Post by unbeliever on Jun 14, 2017 14:47:49 GMT -5
Yeah, I get the different criteria (which in itself is somewhat confusing). Unfortunately, for BW (your example), the error will most likely get worse as power output is increased.
'I don't see any discrepancies'...? Really? I think it´s quite obvious.
Another example is sensitivity: 1.5V specified for full power output (300W) but 1.67V is used in the AP-report.
|
|
KeithL
Administrator 
Posts: 9,929
|
Post by KeithL on Jun 14, 2017 15:20:54 GMT -5
The short answer to minor differences is that you often get slightly different readings when you run the test multiple times. (So, for example, the difference between 0.005% @ 100 watts and 0.0059% @ 90 watts is trivial - and well within experimental error or run-to-run variation.) We are literally talking about the kind of differences you get depending on whether your hand is resting on the bench when you run the test or not... or moving a wire a few inches. (We usually run individual tests to get the specs, then run a separate run to get a "pretty report for publication"..... ) The Broad Band Frequency Response looks like a typo to me........ I read it as 20 Hz to 80 kHz +0 / -2 dB......... I'll see about getting that corrected. Quick question: Why does the AP-report not back up the specs for the XPA Gen-3? Examples:
| Spec | AP | | BW | +/-0.1 (300W) | +/-0.17 (1W) | | BBFR | +0/-0.15 (less than BW?!?) | only graph, lot less than -0.15dB, more like -2dB! | | THD+N | <0.005% @ 100W | 0.0059% @ 500mV (~90W), ~0.07 @ 100W in stpd lvl sweep (THD without noise!)
|
Some discrepancies are rather large...
Also, these values are for one module; with all 7, crosstalk, IMD and SMPS-loading will only make matters worse.
BR,
Henrik
|
|
KeithL
Administrator
Posts: 9,929
|
Post by KeithL on Jun 14, 2017 15:53:00 GMT -5
In general, Broad Band Frequency Response is measured at low power, and is used as an indicator of "general flatness and wideness of frequency response". In the real world, an audio amplifier isn't expected to be able to deliver 80 kHz at full power (80 kHz is the limit of our test equipment). We specify how flat the amplifier is over the important audio band (20 Hz to 20 kHz). Then we give a wider broad band frequency response, with looser tolerances, to show that it can actually pass signals far outside the audio spectrum, although with slightly less accuracy. (The design of an amplifier that really couldn't pass signals above 20 kHz AT ALL would be seriously suspect.) Sensitivity is a different sort of spec..... it really refers to "minimum sensitivity". When we say that "the sensitivity is 1.5V" what we're really specifying is that any preamp that can deliver at least 1.5V will be able to drive the amp to full power. Having a sensitivity that is a tiny bit higher than specified is not considered to be inaccurate (we may simply use the nearest round number). If we spec the sensitivity of our amp at 1.5V, but it will really reach full output with 1.3V, then it DEFINITELY will reach full output "at or before" 1.5V, and so meets the spec. For example, all of the BasX amplifiers have the same gain, and we used the same sensitivity spec for all of them, even though some models will actually reach full output with slightly less input signal. (This means that they all "meet or exceed" the spec we gave.) The significance of "sensitivity" is that your preamp must be able to put out enough voltage (maximum output) to drive your amp to full power (input sensitivity). In general, since most preamps can deliver a LOT more output signal than is required to drive MOST amps to full power, this isn't especially important. All that really matters is that the sensitivity of your amplifier (as an input voltage) is LOWER than the maximum output level on your preamp. In the old days, there were valid reasons for trying to match output level and input sensitivity.... (mostly to minimize noise). However, with modern equipment, it isn't really important any more (except with super-low level signals like you see in microphone preamps). Yeah, I get the different criteria (which in itself is somewhat confusing). Unfortunately, for BW (your example), the error will most likely get worse as power output is increased. 'I don't see any discrepancies'...? Really? I think it´s quite obvious. Another example is sensitivity: 1.5V specified for full power output (300W) but 1.67V is used in the AP-report. |
|
|
|
Post by unbeliever on Jun 14, 2017 16:28:38 GMT -5
Yeah, I understand. I am responsible for large parts of the audio development at a 'world leading' surveillance and security company based out of Sweden (guess which ). I do indeed know about production tolerances, multiple source problems, software induced variations and just generally the impact of 'Murphy'. We, of course, do indoor testing on early prototypes for development purposes but final specs always comes from reliable late HW runs or preferably on volume batches. Since surveillance audio is a lot different than HiFi (most use cases doesn't even involve 'human ears' but use analytics run in large software centers, there can be no subjectivity involved), we do not have the luxury of using averaged or guesstimated values for specs. When for example audio is used for aggression detection and riot prevention at a prison, false triggers due to optimistics specs are kind of bad... We always spec based on worst case test results and add a margin, i.e. our products performs better than or at least identical to the stated values. This is somewhat of a problem since sometimes this difference is great and how do you tell the customer that statistically, all units will be better than what the datasheet says? Now for HiFi on the other hand, it seems more logical to present mean values, i.e. 'our products behaves close to that stated in the datasheet, sometimes better, sometimes worse'. Based on most values having a gaussian statistical distribution, my concern is when the standard deviation gets big. Some customers wil receive a golden sample while others get a nasty surprise. Not saying that this is the case for you of course, just ranting on a bit about general concerns with blown up glossies. Once again, I'm a happy customer and do indeed plan on more future purchases from you. Was mostly interested in how you guys spec your products. Your answer was satisfactory Thanx, Henrik |
|
