Impact of AC Distortion & Noise on Audio Equipment

[Pdxwayne]

@Pdxwayne Topping E30, Topping L30

You do not describe how they are powered up.

> Furman Elite 15 Power Factor i
Seems this is mainly an EMI/RF filter (-40db from 10kHz, -80dB from 100kHz). Plus additionnal surges protections.

If I would have very expensive audio stuff, I may buy such a conditionner, hoping he would burn on lightnings, and save so the remaining.

> Here are the captured files with orig
From here, this links to a single file only

FYI, I just checked further regarding the download. If using my phone, I see 9 files and also zip file option. If I use my laptop to click the link, I only get a choice for one zip file. Once you download that zip file and unzip, you will see 9 files total.

 

[b4nt]

FYI, I just checked further regarding the download. If using my phone, I see 9 files and also zip file option. If I use my laptop to click the link, I only get a choice for one zip file. Once you download that zip file and unzip, you will see 9 files total.

DL is ongoing. Don't know when I'll have a look. Is a WT display bug, proposes me one file, if I click, I get a zip...

May you perhaps also plot 32 tones using REW, to see how that looks like?

 

[b4nt]

@MrPeabody Amir made the choice to plot TDH measurements, I assume he knows what he did. pma mentionned that that way, he could identify harmonics due to PSU poor PSR. More noise from PSU side, not able to filter mains noises out, this might cause effects in audio stages.

I just asked @Pdxwayne if he could also plot 32 tones, to see. For what difference we could see in harmonics, in the bottom sides on such plots:

 

[b4nt]

@Pdxwayne I think the levels you where using are not the best. I beleive you add so a lot of quantization noise... With a low level, you get also closer to your noise floor.

The better I could match the levels so, during recording, the better where my DW results.

 

[Pdxwayne]

@Pdxwayne I think the levels you where using are not the best. I beleive you add so a lot of quantization noise... With a low level, you get also closer to your noise floor.

The better I could match the levels so, during recording, the better where my DW results.

View attachment 147491

I purposely limited the range to not exceed the limit of my chain.....

 

[b4nt]

I purposely limited the range to not exceed the limit of my chain.....

Yes, at some point, you may clip. But you are currently loosing a huge 10dB margin (for your ADC).

I see you took 4 captures per set. May allow you to make comparaisons over time (to see statistical differences).

I opened and compared cap 1/2 files. Either I'm again wrong with DW options (v1.0.70, no filter, autotrim, level + phase EQ), or both sets seem to be close, at -96dB for PKM. Huger peak being for wall jack, below -86dB? If I'm right, pkane would add "inaudible"...

Main noticeable diff I can see is in spectrum of diff. But with all at <100dB to <110dB. Your Forte should have a big impact there, in the outputs.

 

[Pdxwayne]

Yes, at some point, you may clip. But you are currently loosing a huge 10dB margin (for your ADC).

I see you took 4 captures per set. May allow you to make comparaisons over time (to see statistical differences).

I opened and compared cap 1/2 files. Either I'm again wrong with DW options (v1.0.70, no filter, autotrim, level + phase EQ), or both sets seem to be close, at -96dB for PKM. Huger peak being for wall jack, below -86dB? If I'm right, pkane would add "inaudible"...

Main noticeable diff I can see is in spectrum of diff. But with all at <100dB to <110dB. Your Forte should have a big impact there, in the outputs.

Yeah, I should have limited the max a bit less. Maybe minus 3 instead of minus 10.

Yes, between the captured files, nothing really significant observed.

When compared against original file, there is when something interesting showing up......Like pkmetric numbers and Delta of spectra. I wonder if those difference could be due to AC noise.....regardless of using wall jack or not.

 

[amirm]

A couple of quick points:

1. The dashboard is not a THD test. The THD are the two numbers. The rest shows full spectrum of the output of an audio device or the AC signal. All of that information is lost in the single number THD values. So please, don't call the dashboard a THD test.

2. THD is actually by far the most common quality specification for AC generation products including such things as power generators you buy for your home and UPS'. It is usually a rough spec though. Here is the spec for the BK 9801 I use for example:

Went looking for data on generator THD and found this guy showing it and comparing it to his home AC:

 

[b4nt]

When compared against original file, there is when something interesting showing up......Like pkmetric numbers and Delta of spectra. I wonder if those difference could be due to AC noise.....regardless of using wall jack or not.

Main diff is in spectrum of delta. But this points to very low levels, out of audible ranges(<100 <110dB). if i would have seen a huge diff, I could have been convinced. not for a conditionner, but for an additionnal EMI/RF filter.

Here, compared to original, wall_jack_capture3.wav seems better than Furman_capture1.wav. But I won't conclude anything from that as you where at -10dB.

From my own recordings and tests, I noticed an concluded results are the best when you raise the level for recording. This having an impact on residual jitter after DW math.

 

[GeorgeBynum]

I own a ferroressonant filter that is rated to 2kVA and is able to deliver squeaky clean power. Used it in a server room to protect sensitive equipment.

I don't know if they have changes since the 1970s when I played with them ... well, a ferroressonant REGULATOR by Sola; RMS output voltage was within a few % over a 25% input variation, but a sine wave it was not.

 

[Pdxwayne]

Main diff is in spectrum of delta. But this points to very low levels, out of audible ranges(<100 <110dB). if i would have seen a huge diff, I could have been convinced. not for a conditionner, but for an additionnal EMI/RF filter.

Here, compared to original, wall_jack_capture3.wav seems better than Furman_capture1.wav. But I won't conclude anything from that as you where at -10dB.

From my own recordings and tests, I noticed an concluded results are the best when you raise the level for recording. This having an impact on residual jitter after DW math.

Yup....Need to do another round with higher recording level......

Do check left channel. Seems off by a lot compared to right.

 

[b4nt]

A couple of quick points:

1. The dashboard is not a THD test. The THD are the two numbers. The rest shows full spectrum of the output of an audio device or the AC signal. All of that information is lost in the single number THD values. So please, don't call the dashboard a THD test.

2. THD is actually by far the most common quality specification for AC generation products including such things as power generators you buy for your home and UPS'. It is usually a rough spec though. Here is the spec for the BK 9801 I use for example:

View attachment 147511

Went looking for data on generator THD and found this guy showing it and comparing it to his home AC:

Are you now promoting honda generators instead of mains conditionners? Do they come with a noise killer option?

I wasn't considering only the THD number, but, yes, the full plot from 0 to 20k, and all the details at lower levels.

The doc I've mentioned earlier seems exhaustive, presents also such mains analyzers. It includes also some warnings:

https://www.emcstandards.co.uk/files/61000-4-13_immunity_to_ac_supply_distorted_waveforms.pdf

 

[b4nt]

Yup....Need to do another round with higher recording level......

Do check left channel. Seems off by a lot compared to right.

First raise the levels, and check how DW results get, jitter shall be better and similar.

There is no reason to get differences in L/R simultaneous records as the units get powered by the same mains and PSUs. If you notice, cause may be different (cables, audio cables routing...).

 

[Pdxwayne]

First raise the levels, and check how DW results get, jitter shall be better and similar.

There is no reason to get differences in L/R simultaneous records as the units get powered by the same mains and PSUs. If you notice, cause may be different (cables, audio cables routing...).

Likely cable caused the issue for left channel. Not sure why.... left Channel level when started was very low and won't improve until I replug all cables....

 

[b4nt]

Likely cable caused the issue for left channel. Not sure why.... left Channel level when started was very low and won't improve until I replug all cables....

The best is to make some DW checks at recording time, to check if they are clean, or to restart untill the bench is optimal.

I didn't use REW. Maybe 32 tones tests plots could be fast to run for bench checks.

For the cables, I would avoid a mix of audio and mains or power cables, to avoid capacitive couplings. Maybe see how you can organize this for audio/data/power cables in separate areas (cause any coupling is not what you are willing to investigate). Plus boxes separed as possible.

 

[GeorgeBynum]

Maybe he can simply improve the supply by adding $100 worth of cap at the amps DC supply.

Be wary of the initial longer, high current, charge of those caps. The rectifier MAY not like it. It's not an issue once steady state is reached. This is analogous to soft starters for induction motors. Higher rotating inertia increases starting energy required.

 

[amirm]

The doc I've mentioned earlier seems exhaustive, presents also such mains analyzers. It includes also some warnings:

And valid warnings they are. One mistake and you may not live to regret it!

A little story. I had the electric company come and install a surge protector in our meter. A nice technician showed up and as he was opening the meter, we started to talk. As we were speaking, I was getting a little tired and reached to lean with one hand on the post where the meter was. He stops and says: "you know the meter is live, yes?" Shoot, I had not even realized I was a) leaning and b) could have put my hand right into the open meter! Of course I acted cool and said: "sure."

I have been shocked many times but fortunately survived them all but it could been otherwise. No one should try such testing unless they are fully qualified and know the risks.

I took many precautions in running these tests and checked and double checked every time I energized a line. Watches and rings were taken off hands and I stayed insulated form ground.

 

[Pdxwayne]

The best is to make some DW checks at recording time, to check if they are clean, or to restart untill the bench is optimal.

I didn't use REW. Maybe 32 tones tests plots could be fast to run for bench checks.

For the cables, I would avoid a mix of audio and mains or power cables, to avoid capacitive couplings. Maybe see how you can organize this for audio/data/power cables in separate areas (cause any coupling is not what you are willing to investigate). Plus boxes separed as possible.

I already spent hours on this and don't want to spend more time to recapture. I just got a new toy and will check that instead.
; )

I would say, at least for the Node2i, E30, and L30 combo, using headphones, they sounded just fine using regular wall jacks.

The only curious thing I still have is stereo power amp. I won't spend time checking either.
: P

 

[b4nt]

I would say, at least for the Node2i, E30, and L30 combo, using headphones, sounded just fine using regular wall jacks.

As I already wrote, I would consider a "conditionner" for additionnal EMI/RF filtering, if one ow my boxes would be sensistive to. This can be audible, you get up to pops in your speakers if something is wrong. An old but good box may be improved so? At one point, I would take the decision to spend either the money for a EMI/RF filter or a new box.

Other aspect being surge/lightenings protection. A "conditionner" MAY protect coslty hardware, he shall burn the first.

 

[GeorgeBynum]

Can't say anything about the current because the neutral to ground resistance is unknown.

I don't know your wire size, but an assumption of 1 ohm/300 meters will be within an order of magnitude. Wire size and wire tables will give you a better number. Then an estimate of length to your power source where neutral and ground are bonded should get you within +/- 50%.