Interpreting spectrum analysis

[cre009]

Hopefully find attached some spectral views from a couple of transfers I did earlier this summer for comparison. The views are for a few seconds from the start including the lead in grooves to where the music starts. The LP was an old bootleg with a fair degree of surface noise.

The first image was for a transfer using an LP12/Ekos/AT-F7 into a Nakamichi CA5E preamp. The second was for an AR XA with Linn K9 into a Kinshaw Perception SA phono stage.

The LP12 version starts with just the phono stage hiss followed by a thump/clicks when the tonearm is lowered onto the record. There is then a degree of noise due to vinyl roar and crackle and towards the right the music starts to fade in. For the XA version there is some horizontal banding mixed with the hiss. This banding is a very low level hum that is only audible if I turn the volume to max but has proven annoyingly stubborn to remove. Rumble if present would be most pronounced and visible at the lower frequencies during the lead in.

[IMG][/IMG]

[IMG][/IMG]

[Barry]

Barry, I'm not sure what you mean, sorry! I must admit that Audacity is a little more complicated than I first thought and I think my preferences and Hz range were way off. I shall post some more, hopefully, meaningful plots over the next few days.

The spectrogram shows a noise/rumble level of better than -27dB. But what is 0dB? Is there a test tone of a known reference level available?

As it stands a rumble level of -27dB is not that good. One would expect an unweighted rumble figure of -40dB or more.

[Beobloke]

Sorry, but what you are measuring by playing an unmodulated groove is a combination of bearing noise, groove roar and the general dust and detritus contained therein. To measure bearing noise and rumble accurately you need something like a Thorens Rumplemesskoppler.

http://www.theanalogdept.com/thorens_dept_.htm

[Wakefield Turntables]

Im not too sure that the set up is optimal for measuring bearing noise...

Do you have some kind of solid puck that you could sit next to the platter and lower the stylus onto it so that any vibrations can be transferred from the body of the TT into the stylus?

Using a test record you would need to isolate the noise from the record surface and the phono stage and this will most likely be quite broadband. Audacity has filters that you can apply. You could try a 250 hz lowpass filter to start with. You should maybe also apply a reverse RIAA too.

Another option could be to place a mic on the TT close to the bearing. You can also buy stethoscopes on Amazon for a few quid. These actually work and are great for a bit of roll play as well

I agree. I don't honestly think that my preference are set up correctly. If you have any hints, tips or suggestions I would be eternally grateful. Well I don't think we can isolate noise from the records surface, that's always going to be there and using a filter might be a good alternative, but can you explain why I specifically need a 250Hz lowpass filter. In fact what is a loss pass filter? I understand what a reverse RIAA is and I can see where this is useful but at the moment I'd just like to keep things simple and then start more complex as my knowledge increases.

Hopefully find attached some spectral views from a couple of transfers I did earlier this summer for comparison. The views are for a few seconds from the start including the lead in grooves to where the music starts. The LP was an old bootleg with a fair degree of surface noise.

The first image was for a transfer using an LP12/Ekos/AT-F7 into a Nakamichi CA5E preamp. The second was for an AR XA with Linn K9 into a Kinshaw Perception SA phono stage.

The LP12 version starts with just the phono stage hiss followed by a thump/clicks when the tonearm is lowered onto the record. There is then a degree of noise due to vinyl roar and crackle and towards the right the music starts to fade in. For the XA version there is some horizontal banding mixed with the hiss. This banding is a very low level hum that is only audible if I turn the volume to max but has proven annoyingly stubborn to remove. Rumble if present would be most pronounced and visible at the lower frequencies during the lead in.

[IMG][/IMG]

[IMG][/IMG]

Many thanks. These are the sorts of things I'm interested in. And it explains quite a few things, if you have any others please post them.

The spectrogram shows a noise/rumble level of better than -27dB. But what is 0dB? Is there a test tone of a known reference level available?

As it stands a rumble level of -27dB is not that good. One would expect an unweighted rumble figure of -40dB or more.

Barry, as ever thanks for your input. My knowledge is limited but getting better. I now know that 23Hz is where the motor can be found 5-10 Hz is the tonearm and 50/60 Hz the mains supply. I'm also led to believe that 2nd harmonics from the mains supply can occur at 100-120 Hz range. So I now know that I have to produce a Spectral analysis of between 0-120Hz if I want to see information with respects to these turntable components. Do you know of any other specific Hz ranges which would be of interest? Does the bearing have a Hz range for example? Finally I'm a little confused with respects to dB ranges. Am I correct in saying that a (-) value i.e. -50db is better than a +50db for example?

I wonder if I could ask another questions (SORRY). Lets say if I wanted to assess the effects of a new upgrade to the turntable. Please correct my methodology if I'm wrong. I would need to first measure the old components Hz range and dB levels then compare the same Hz and dB range against that of the new component. I large -dB value would indicate an improvement in vibration reduction whilst an increase into the +db range would indicate the opposite?

[Primalsea]

The lowpass filter will just cut out everything above 250 hz. This will just get rid of frequencies that are not of interest. Have a look here

http://manual.audacityteam.org/man/low_pass_filter.html

250 hz is a bit of a guess but I think bearing rumble will be much lower in frequency than this, so you could potentially move the lowpass frequency lower.

[Wakefield Turntables]

Paul,

Seems anything of interest appears to be in the 0-160hz range so do you think it's wise to lower Hz band to this level?

[Primalsea]

Paul,

Seems anything of interest appears to be in the 0-160hz range so do you think it's wise to lower Hz band to this level?

Actually, you may want to go higher to 320 Hz and select 24dB / octave roll off 12dB if 24 dB is not available) . 320Hz will be the point where the attenuation wil be - 24 dB down, leaving 160 relatively untouched.

I was expecting the frequencies of interest to be lower, but like I said it was a bit of a guess (and I guessed badly).

Remember that your phono stage will be boosting low frequencies, so without reverse RIAA applied you will have distorted results. However, if you always keep to the same setup you will have results that are comparative.

[helma]

You can gain a lot of useful information, given you keep the reference same - so however you measure, make sure you keep the recording levels the same, use the same cartridge & phonostage (or microphone if you use one) etc. Ofc your results will not tell that much to other people because they don't follow a standard, but for your own experiments it's not a problem. If you record a lead-in or lead-out groove then obviously groove noise and what even tiny amount of rumble got into the master from the cutter is obviously going to be there. You can also put something on the top plate of the deck and rest the stylus on that, which will tell you something about what vibration gets into the top-plate one way or another, but again this adds variables like whatever you use resonating, but again for your own experiments can be useful.

0dB is the maximum level available, you cant go higher than that, the signal will clip at 0dB. So if you use an actual record, setting the recording level so that you get a healthy max level maybe in the 0 - -6dB range or so when playing music with peak loudness, so that when you record "silence" on the disc you get a rough idea how much below the noise is of actual music. But again not a big deal for comparative analysis as long as you keep the variables same between recordings, use the same record etc. It's a good idea to find a record that's as silent as possible in the portion of silence you plan to use, there's significant variation between discs and especially some 50s / 60s stuff will have built in rumble in significant amounts, even if it's not readily audible in normal listening. In the spectrum image you posted, that 100Hz level is probably 2nd harmonic motor vibration getting through motor isolation or whatever drive system the deck uses, but from the gentle slope and relatively even level to all the other noise present, it seems healthy. If there was isolation issues there should be sharper and higher peaking.

It's not rocket science, I think you'll figure your way through it with experience and perhaps surfing around the web a bit. When using 50Hz mains, 25, 50, 100, 150, 200 etc. become significant numbers because they are the harmonics of 50Hz. Like you if you measure a Lenco, idler wheel, plinth, top plate mounting method etc. will affect those levels, and sometimes changing something might bring one harmonic or the base frequency down but increase others, so beyond getting things as silent as possible it's also a matter of finding the right compromise. Like personally I think 100Hz is the nastiest frequency to have vibration come through and it's one where you can get significant amount of bleed through, while the upper harmonics are usually insignificant unless there are problems. At 100Hz you are well into the range where a lot loudspeakers and rooms have significant bass boost (or at least the bass is not sloping down yet), it's in the range of bass guitar and kick drum. With most setups 50Hz levels are not as noticeable and there's really not much music that low, except the deepest body of kick drum. Unless we're talking organ music or something other than pop/rock/jazz. Ok modern pop music has plenty of very deep bass, but vinyl is not the right medium for that stuff anyway. Anyway for example if a change I make reduces 100Hz while increasing 50Hz, it's usually a trade-off worth making, but that obviously depends on what the levels were before etc.

Sort of lost my train of thought but I hope there was something helpful there

Kai

[Wakefield Turntables]

Kai, yes thanks, I'm sort of caught up in a few other bits so I keep coming back to this when I have time in between projects.