Thought similar amps that test well here on ASR should sound the same?

[wavetrade]

After buying and discovering how good a PA5 II amp sounded I ordered up a Audiophonics NCx500 stereo amp about two years ago. No matter how much I wanted it to, I have never thought my NCx500 unit sounded as good as the PA5 II. I would listen a few months and eventually switch it out as the NCx500 unit sounded fatiguing and in my face compared to the Topping. Was afraid to mention it, figuring I had some sort of bias going.

I switched opamps to OPA1656 and thought I liked the sound a little better, although it still seemed brighter than my other amps. Eventually I bought a Audiophonics ET400 Purifi amp, another PA5 II and recently a E3 A7. All of these new purchases sound terrific and no way could I pick one from another in a blind test at reasonable listening levels. Yet the NCx500 amp continued to fatigue me.

So last night I decided to do some REW tests and see if I could see a difference. For this test I chose to compare the NCx500 amp against my new A7. I left the microphone in the same place (at 90 degrees at my usual listening position) and the only thing I changed in the setup was the amps. I also calibrated the volume level to 75db after doing the switch, there was only a 1db volume difference at the dac. I was using a SMSL DO100 for this and the speakers are Paradigm 700F.

Lo and behold there was a very visible difference from 4.5KHz to 16Khz that supports what I think I have been hearing. Hoping someone here can help me understand how or why this would happen? Here are the graphs:

STAFF NOTE: As this measurement was incorrect, may help to skip to here

 

[mglobe]

Maybe Amir or someone else here with the appropriate equipment could test it for you to see if something is wrong with it?

 

[kemmler3D]

Really interesting result here, at least what you are hearing is measured and more than plausibly audible, not typically the case when people hear differences between amps. Are you 100% sure you were standing in the same place relative to the speakers when you took these measurements? Were they done with both channels or just one channel? What is shown here matches what you describe hearing, but especially in the treble, small differences in how the measurements are taken can show up this way.

Otherwise agree with @mglobe that the next step would be to do some measurements on the amps themselves to run this down.

 

[voodooless]

Did you measure with the new opamp, or the stock one? How about if you bypass the buffer?

 

[kyuu]

The NCx500 measurement looks more like what I would expect than the A7. I wouldn't expect a rapid 5dB falloff from 3kHz to 10kHz that then shoots back up after 13kHz from a normal setup with a properly working amplifier.

I'm not a measurement expert, but the members who are would be able to help you better if you posted the entire .mdat rather than just a screenshot of a single graph.

 

[wavetrade]

Really interesting result here, at least what you are hearing is measured and more than plausibly audible, not typically the case when people hear differences between amps. Are you 100% sure you were standing in the same place relative to the speakers when you took these measurements? Were they done with both channels or just one channel? What is shown here matches what you describe hearing, but especially in the treble, small differences in how the measurements are taken can show up this way.

Otherwise agree with @mglobe that the next step would be to do some measurements on the amps themselves to run this down.

Yes, I sincerely don't understand this. But am happy to have some validation for what I have been hearing for two years. I was extremely careful with these measurements and waited until the house was silent, the mic never moved and I stood in the same spot about 5 feet behind the mic for all readings. I truly did not change anything except the amps and a 1db volume adjustment to calibrate to 75db. Subwoofer was switched off and no EQ or processing.

kyuu, I totally agree with you. To me the A7 dip does seem mysterious. I have previously done some less careful measurments in another room comparing the ncx500 to a pa5 II plus and seen something similar, but less dramatic difference. My subjective impression is that the A7 is the least bright of my class D amps. I primarily chose it for this comparison as it did not require me to change interconnects as both amps have xlr inputs. Tonight I think I will try this again, but this time comparing it to the PA5 II. I will also include the .mdat file.

 

[kemmler3D]

I couldn't find it for the 700F but here's Audioholics' measurement of the 800F impedance... doesn't obviously correspond to the measured SPL but there is a slight hint of load dependence with the A7 so... maybe?

 

[wavetrade]

Did you measure with the new opamp, or the stock one? How about if you bypass the buffer?

This is with the OPA1656 opamps with the buffer at 19dB gain. The dac in this room won't run the amp at full potential, but when I have some time this weekend I can switch out the dac for the ADI-2 and try no buffer. As a side note I have some OPA1612 opamps coming later this week and will certainly try them.

 

[oleg87]

Well, it's clearly not testing well for you - something seems screwy. It's not obvious to me what the issue would be from the data provided though - might be informative to test each channel in isolation.

 

[Ropeburn]

Very much looks like load dependency, only happens in exactly that upper range. NCx500 has none, therefore it's flatter.

Unless of course there's some kind of defect on the A7. If another one measured similarly, that's unlikely.

 

[wavetrade]

I couldn't find it for the 700F but here's Audioholics' measurement of the 800F impedance... doesn't obviously correspond to the measured SPL but there is a slight hint of load dependence with the A7 so... maybe?

View attachment 495133

Interesting. I can see I am likely to try this again tonight. Perhaps I will try it in another room that has LS50 M speakers and the RME dac. I am also curious to try a carefully measure comparison with the PA5 II plus. The Purifi amp is at my office and I don't really want to bring it home or measure it there.

For what it is worth, the A7 strikes me as the darkest sounding of my class D amps. But not significantly and I do really like the sound. It matches well with the Paradigms but leaves me feeling a little flat when used with the LS50's. But this is just a subjective opinion. The consitent bright impression of the NCx500 has been with me for years and I do think I could pick it out if A/B'd with any of my other amps.

I appreciate all of you checking this out, thank you.

 

[voodooless]

Load dependence is only fractions of dabs at the worst of times. So it seems, something else causes this.

 

[Sokel]

It would help to take your measurements one meter from your speaker or less so to also take some room out of them.

As long as both amps are not load dependent the only thing I can think of is power, maybe some scheme in the A7 chip does its thing out of stress up high?
Wild speculation but the result is strange, 5dB are no joke.

 

[Sokel]

Load dependence is only fractions of dabs at the worst of times. So it seems, something else causes this.

You mean for the specific amp? It has none according to Amir's tests.
Others can go way more a dB, at the thread about it with the load depended Wiim, Erin's tests and calculations with other speakers showed different and significant differences.

 

[dguillor]

A frequency response measurement of just the amplifiers would be easy to do and might shed some light on what’s going on. You could even make the measurements at the output of the amplifiers with speakers connected.

 

[kemmler3D]

Load dependence is only fractions of dabs at the worst of times. So it seems, something else causes this.

Yes, even at 16ohm I find it hard to believe the A7 would produce +/- 5dB deviation, based on Amir's review we should expect more like 0.2dB or something. I agree the cause is probably elsewhere. But it felt relevant to at least reference the impedance to see if anything crazy was going on.

 

[Juhazi]

Assumed that 700F has same mid and tweeter as 800F, it's treble is hot and room response of OP with NCore is realistic

As shown earlier, impedance and phase in treble are easy load for any amplifier unlike upper midrange with wild phase angle and lowish Z- so NCore is fine, other amps make the treble softer for some reason. This is interesting, I have not seen similar measured difference anywhere, not in my own tests or in the web or magazines.

 

[wavetrade]

It would help to take your measurements one meter from your speaker or less so to also take some room out of them.

As long as both amps are not load dependent the only thing I can think of is power, maybe some scheme in the A7 chip does its thing out of stress up high?
Wild speculation but the result is strange, 5dB are no joke.

I am going to give it another go tonight. In the scenario you suggest, should I measure each speaker separately from one meter directly in front? Should i put the mic head right at tweeter level? My previous REW use was to measure around listening postion so that I could create EQ files to make the room sound better. This will the first time of trying to take the room out of it. I will include the .mdat with the post this time as well.

 

[Ropeburn]

I am going to give it another go tonight. In the scenario you suggest, should I measure each speaker separately from one meter directly in front? Should i put the mic head right at tweeter level? My previous REW use was to measure around listening postion so that I could create EQ files to make the room sound better. This will the first time of trying to take the room out of it. I will include the .mdat with the post this time as well.

I would try this suggestion instead:

A frequency response measurement of just the amplifiers would be easy to do and might shed some light on what’s going on. You could even make the measurements at the output of the amplifiers with speakers connected.

You'd take the room out of the equation, removing all possible problems with measuring it, and measure electric behaviour directly. With speakers connected.

 

[wavetrade]

As an aside, not sure I didn't try something like this sooner. I had stuck with the common advice of avoiding EQ of the higher frequencies. Last night I used the NCX500 graph and focused on the zone in question. Created a simple 3 band filter that sounded really nice to my ears. No fatigue and got me to the same sound that I am used to with the other amps I own.