24/192 spectra... what's wrong here?

[manisandher]

Following on from my thoughts in another thread, I thought I'd open a new one to explore things.

 

Have a look at the spectrum of a 24/192 needle drop below:

 

 

I'm trying to understand where the horizontal lines come from. My ADC is set to 24/192. It's hard getting any real info about the ADC, but here are a couple of quotes from its designers (coming on to 15 years ago now):

 

- "[At 192] we use a proprietary filter (non-oversampled) optimized to that sample rate."

- "Up conversion is not used at 176.4 kHz and 192 kHz sample rates. Instead, a passive analog filter with group delay correction removes alias products..."

 

So, can anyone explain where the horizontal lines come from?

 

Cheers, Mani.

[Synfreak]

What album is that Vinyl-rip made from?

 

A real time FFT analyser might bring some more insight too. (like the one I use from Steinbergs Wavelab)

[bdiament]

Hi Mani,

 

Could that be a resonance in the phono cartridge?

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

[manisandher]

Thanks for the ideas guys. I'm going to make a few more needle drops at different resolutions. And also at 24/192 with a different ADC.

 

I'll be back...

 

Mani.

[wkhanna]

This is quite interesting.

[esldude]

Well I have seen similar patterns doing needle drops at 96 khz.

 

It probably is a resonance in your phono. And could be more than one thing.

 

You can record some silence before you drop the needle in the groove and see if those are still there. Also were you using a phono pre-amp doing the RIAA or just going straight to a mic input on the ADC?

 

Anyway I have seen lines like that and doing an FFT even without being in a groove they show up. One I did was a 7500 hz resonance. And all the harmonics were there up to 37,500 hz. You also will often see the resonance shift just a few hundred hz. once you drop it in a groove before music starts. Also though pretty visible in the spectrograph view in an FFT they usually are in the noise floor of the vinyl other than perhaps the highest couple harmonics. The level of those is likely at or lower than -80 db.

 

I also have seen other lines show up once music starts. I take it those are cutter head resonances. Those were in the 8-12khz range in some cutting heads. Usually controlled with feedback, but they still show up with some LP's. Among many advantages one reason to do half speed mastering was it put the cutter head resonances an octave higher pretty much out of the way, and put 1/4 the energy into them.

 

I haven't investigated too much exactly where those come from or how to stop them. Once I saw how low in level they were, and that they get buried by surface noise I didn't worry about them too much. I suppose one could EQ out very narrow slices at the frequencies of the resonance though I have not bothered doing such. Would be better to get it out of the system before recording of course.

 

Changing the impedance the cartridge feeds into would alter the resonance. Lower impedance should dampen the resonance level. Also changing phono cables to some with higher or lower capacitance will change the resonances perhaps in level as well as at which frequency they occur. I believe higher capacitance or longer cable would lower the frequency of resonance, and shorter or lower cpacitance cable would increase the resonant frequency if I haven't gotten this backwards in my head.

 

SNARK WARNING:

 

Of course we all know how wonderful and how much higher fidelity analog LP is. That is why so many consider it the reference to judge other systems by.

 

Admittedly, the fact LP works as well as it does with such a primitive system is quite a testament to what people were able to get out of it.

[Miska]

Here's one result I have on my HDD from 10+ years ago of transfer I did with M-Audio Delta1010 (goes only up to 96k PCM).

 

There's some resonance/response peak around 18k with this record.

 

[manisandher]

Could that be a resonance in the phono cartridge?

 

It probably is a resonance in your phono. And could be more than one thing.

 

I managed to find some time today to do some tests... and it seems you were on the right track. But perhaps not resonance... more like severe harmonic distortion. Have a look below.

 

Here is a recording of a 3150Hz tone on the PMII:

 

 

Here's the same tone recorded on a Tascam HDD recorder:

 

 

(Note: the levels were not matched between the ADCs.) In both cases you can clearly see the multiple harmonics of the 3150Hz fundamental, but generally what a bloody mess! A few thoughts:

 

1. My cartridge is a London (Decca) Reference. It's a weird 'moving iron' design and is very low compliance. My arm is a short air-bearing linear tracker, based on Poul Ladegaard's design. Either there is something seriously wrong with my cartridge, or it's actually working as designed, but is just a very bad match for the arm. In any event, I will send it back to the manufacturer here in the UK to be inspected.

 

2. The reason why the lines in the spectrum aren't perfectly straight is because the test LP wasn't perfectly centred. My current cartridge/arm combo really struggles with these types of LPs.

 

3. I prefer the sound of my old PMII to the much newer Tascam (the latter using the PCM4222 ADC chip, considered one of the best currently available).

 

4. And the weirdest thing of all... My needle drops of real music (not test signals) sound bloody good to my ears!!! Perhaps the ear/brain adjusts for harmonic distortion in some way so as to make it benign. To me, my needle drops sound more lively and more real than the CD rips of the same music. So is it added harmonic distortion that makes some people prefer vinyl to digital, and tubes/valves to solid-state??? Dunno.

 

Seems I've got a lot of work to do to start getting anywhere near decent results.

 

Mani.

[esldude]

Are you using a phono pre-amp? It could be your cartridge is over-loading the phono pre-amp. Some phono pre-amps don't have a large overload margin. You are showing both odd and even harmonics as if overdriven. But dropping level just a bit, most of that may go away. The test groove is pretty high in level depending on the test disc you are using.

 

Also, you could put your cursor in the FFT over the spikes and it will read the nearest peak value. When using 65K on the FFT, often the actual peak is well above the graph as the lines are too small to show up in the screen shot (you are showing perhaps 65000 bins with a 1980 pixels wide display). Dropping to 8192 on the FFT would also let the peaks of the harmonics show up closer to their real level. But the spectrogram view does look like the level is quite high.

 

But it does look like you have some things to iron out in your matching or setup.

[tranz]

Another thing to look for is any other equipment that is nearby such as tv, lcd, switching PSU, even the tiny lcd screen in a handheld radio all will throw out lines like that at varying frequencies. At one point I was tearing my remaining 2 hairs out thinking multiple horizontal lines were coming from a speaker (using a 100kHz mic). Turned out that a silly am radio with a clock lcd and the plasma tv in standby mode were causing it!

[dc2bluelight]

How about a spectrum with no tone, just residual noise?

[manisandher]

Are you using a phono pre-amp?...

 

... Dropping to 8192 on the FFT would also let the peaks of the harmonics show up closer to their real level.

 

Yes, I'm using a phono pre-amp. I've got a few knocking around. Here are the 3150Hz spectra using two different phono pre-amps:

 

1. battery-operated, solid-state phono:

 

 

-15.3 dB Fundamental

-40.6dB 2^nd

-47.0dB 3^rd

-45.8dB 4^th

-57.6dB 5^th

 

2. tube phono:

 

 

-19.3dB Fundamental

-42.5dB 2^nd

-51.6dB 3^rd

-51.0dB 4^th

-65.8dB 5^th

 

Apart from the substantially higher noise floor with the tube phone, the spectra look pretty similar. I don't think there's much doubt that the cartridge and/or arm are not working properly, either because of a fault, or simply because of bad compatibility. I'll have a bit more time to look more deeply into things in a couple of weeks' time and will try a few different cartridge/arm combinations.

 

Thanks everyone for you ideas and suggestions.

 

Cheers, Mani.

[manisandher]

Another thing to look for is any other equipment that is nearby such as tv, lcd, switching PSU, even the tiny lcd screen in a handheld radio all will throw out lines like that at varying frequencies.

 

Good idea, but it's not the cause of the lines. I went to great lengths to minimize such things by building my 'vinyl studio' in a Faraday cage and having all power supplies sitting outside. I switch the PC monitor off during all recordings.

 

How about a spectrum with no tone, just residual noise?

 

I did this a while ago, but can't find the files now. I remember that the noise floor was sitting pretty well down using the SS phono pre-amp. (My tube phono pre-amp is much more noisy.) But I'll repeat this at some point.

 

Cheers, Mani.

[manisandher]

Just to finish things off for now, here are a couple of 20Hz-20KHz sweeps using the two different phono pre-amps:

 

 

 

Things look a total mess from 100Hz upwards.

 

But what amazes me is how 'good' my recordings of real music sound... with all this going on. I don't want to get drawn into the subjectivist vs. objectivist argument, but unless I had looked at these spectra, I would have merrily continued with my needle drops. I have to say though that both of the test LPs I've been using have slightly off-centre holes, which causes a total nightmare for my arm/cartridge. Most of my LPs are better in this respect and pose far less problems.

 

Ripping CDs is so much easier. Downloading easier still!

 

Mani.

[dc2bluelight]

It would be most helpful for analysis to post a spectrum without playing a signal, just the residual noise content only, like with the arm just sitting on the arm rest.

[dc2bluelight]

You could also post an audio file of the above too so we could have a "listen".