Artifacts created when using RIAA Equalization

[david_in_md]

Hi,

I’m trying to digitize my library of 33 RPM records and am using a Dual CS 508 turn table with a Grado GF3(E) cartridge running directly into an Alesis IO2 digitizer. From there I’m going to a Windows 7 PC running Audacity 2.0.5. The recording goes smoothly. No hum and no clipping, but it obviously needs RIAA equalization (my old amps with RIAA are terrible, hence the “software RIAA”). After making the raw recording, I normalize to remove DC bias and also adjust amplitude to -6DB (to give headroom for RIAA), and then I apply RIAA from the equalization menu.

The result sound good (tonally) but there is still a great deal of crackle and I have looked into its origin to find that correlated with the traces, some of the crackle (high-frequency distortion in the music) seems to arise from artifacts coming from the application of the RIAA. There are a number (~20+) incidents where the post RIAA trace makes a vertical rise to the maximum amplitude (and is clipped) and then the signal begins to slowly descend from there over a period of 1-2 seconds. It happens with both positive and with negative polarity relative to 0.

This seems to be correlated with incidents in the original (raw) recording where there are modest periods when a slight DC bias comes and then goes - either a positive or negative segment (perhaps only a second or so.) I guess the filtering operation is exaggerating these events (?).

From an electrical perspective, I cannot understand the origin of this bias or why the system isn’t fully decoupled. It looks like a big capacitor takes a hit and then renormalizes and may have its origin in the USB / Digization / Laptop setup. I feel like my setup (excluding the cartridge) is well shielded and grounded. I do see some random noise before the needle is down, like static(?), but it is quite far down in dB. The “remove DC bias” doesn’t take this “local bias” out because it seems to perform more of an “averaging” process. It works if applied locally, but it would be too much effort to take it out for all the incidents as they have to be done in small sections and that injects it’s own artifacts.

So, a few questions arise:
1.) Has anybody else seen this kind of “bias” come and go and what is the nature of it? (vinyl issue, electrical problem, static problem?)
2.) Is there an algorithm to take out these localized events automatically?
3.) Is there a way to prevent the RIAA equalization from creating these artifacts?
4.) Is there a way to remove the RIAA artifacts?

I’ve attached two images, one zoomed out after RIAA, one zoomed in with “raw” audio on bottom tracks

In advance, thanks for any help!

Best regards,

[steve]

I have an idea of what my be causing the problem but it’s difficult to tell from just a picture, Could you post a short sample of the affected audio before applying Equalization. The sample only needs to be a few seconds long but needs to be in WAV format. See here for how to post audio samples: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

[DVDdoug]

This seems to be correlated with incidents in the original (raw) recording where there are modest periods when a slight DC bias comes and then goes…

… I guess the filtering operation is exaggerating these events (?).

That actually makes sense, since RIAA equalization boosts the low frequencies and DC is zero Hz. If you suddenly introduce DC into a system that boosts low-frequencies, but the syatem can’t actually pass DC, that’s exactly what I’d expect to see.

So the BIG QUESTION is, why are you getting changing DC offset out of your interface??? It seems like there is a defect in your interface

\

A couple of unrealated things -

The microphone input on your Alesis interface has an input impedance of 600 Ohms to match a low-impedance XLR microphone. The standard load for a phono preamp is 47,000 Ohms. The low impedance is probably reducing the signal level and creating frequency response variations. You’ve got a decent setup with the Dual turntable and Grado cartridge, so you should probably get a proper phono preamp, or an interface with a proper phono preamp.

You probably need about 20dB of headroom for the RIAA equalization. But since Audacity uses floating-point and it won’t clip internally, as long as you normalize before saving (exporting), your output file shouldn’t clip.

So, 6dB of headroom is fine and it will prevent clipping of your analog-to-digital converter during recording. Just normalize after processing to make sure that your output-file (or digital-to-analog converter) doesn’t clip.

It’s probably also a good idea to remove DC offset after RIAA equalization.

[kozikowski]

That is a very serious mismanagement of the sound signal. Your system is crying out for a proper hardware phono preamp. One thing to look for is the separate ground connection, required on proper phono preamp systems. Note the extra grounding post on this preamp.

http://www.sweetwater.com/store/detail/UFO202

A note: Grado made two different types of cartridge, moving magnet and moving coil. It is required to match the preamp to the cartridge type, over and above getting the curve right.

Koz

[DVDdoug]

One more thought about this weird suddenly-changing offset problem -

The other day I noticed a “remove offset” setting for the built-in microphone on my laptop. I don’t have my laptop with me at the moment, so I don’t remember where that setting was, but it was a Windows or driver setting… It wasn’t anything in Audacity or an application.

Anyway… Open Windows Control Panel and make sure ALL of the “enhancements” for your USB interface are TURNED OFF. Something that’s supposed to correct offset could potentially make it worse, and you don’t want Windows (or the drivers) doing anything to the digital audio stream.

[david_in_md]

Many thanks for the suggestions / thoughts!

It isn’t lost on me that a phono pre-amp might be a quick fix!

I’ve got good grounding, which has basically eliminated the hum. I will say that those Grado cartridges “have issues with hum”, but of course it’s not hum I’m getting.

I’m coming into the Alesys “guitar” input (1/4" phone jack.) I can’t find the impedance for it in the literature (may need to look harder) but I assume it’s high-impedance, still the suggestion that I would need to re-normalize the RIAA curve is a good one. It sounded really heavy on the bass after normalization.

I think the observation that the RIAA equalization is somehow taking the derivative of the waveform and the seemingly very tiny bias is causing an unmerited spike. I may try the subsonic filtering to see if it would remove this bias as if it was a sound with a VERY LOW frequency component.

I thought I’d found the truth last night when I turned off the wireless on the laptop and shutdown any nearby gadgetry. The signal seemed cleaner, with fewer glitches, but the problem did remain. I may try with a completely different laptop / os tonight in the case this is a laptop issue. BTW, I am using Windows 8 not 7, as I inadvertently noted.

I’m going to post a short WAV patch up directly that demonstrates the issue.

Again, thanks for the help!

Best regards,

David

[david_in_md]

Please find attached two 2 second segments of the audio.

One labeled “Post RIAA” is after RIAA normalization and has the artifact.
The other, is raw with no processing for the same segment, between 0:40.0 and 0:42.0.

[steve]

As I thought, the “problem” is with your audio, not with the effect.
Here I’ve zoomed in on a small section :

Notice how the left (upper) channel is offset vertically. The waveform is not centred around the horizontal line that indicates the mid point, but is almost entirely above it.
This “off-set” is in effect, an extremely low frequency component of the sound.

Here is the same part of the waveform after filtering out frequencies above 100 Hz:

Because RIAA equalization boosts low frequencies a great deal (around +20 dB), this low frequency component becomes hugely amplified, resulting in the weird waveform that you experienced.

Needless to say, that extreme low frequency component should not be present.
Warped records can cause extreme low frequency components, but I don’t think that is the cause in this case, because there is a definite “step” in the offset. and it is only occurring in the left channel. My guess would be that it is a fault in the Alesis IO2. You can test to see if that is the case by reversing the leads that plug-into the left/right inputs of the Alesis (only change the ends that plug into the Alesis) so that the left channel of the turntable plugs into the right channel of the Alesis, and vice verse. If the fault then occurs in the right channel, then you will know that the fault lies before the Alesis (record, cartridge, turntable, leads). If the fault remains in the left channel, then the problem is probably with the Alesis, though there is a possibility that it could be a problem with the drivers.

As a temporary workaround, apply the “High Pass filter” with the frequency set to about 40 Hz and “roll-off” set to 36 dB per octave before applying the RIAA equalization.

[david_in_md]

Thanks Steve,

That is an excellent analysis and I think you’re spot on. FWIW, the “effect” occurs on both channels, usually separately, but occasionally together.

I agree that the RIAA is the “victim” of weird data. I saw the offsets and figured it must be giving rise to the issue.

I am extremely suspicious that the power for the Alesis innards is not stable (either coming from the laptop USB port or internal to the Alesis itself.) I’ve see strange things, like when you flip on the 48v phantom power the gadget shuts down for a few seconds. Or if there is a very loud, over driven passage, it’ll “wane,” and then slowly recover. I suspect they have some big capacitor to “moderate fluctuations” and it has about a 1/2 second capacity. Likely it may turn out to be an unwanted feature of my laptop like a runted USB output, or current limiting (I’m going to try another laptop tonight.) If that is no joy, then I’m going to try to pry open the Alesis and connect a stable power supply to see if that might cure it.

The low frequency filter is a good tip. If the problem traces back to a central equipment problems, like the power supply, I’ll probably re-record anyway to see if more of the “crackle” will leave, but it is a very nice next step, if even for testing purposes!

Again, thanks for your help and insight!

Best regards,

David

[kozikowski]

The cartridge and preamp are a matched system even though you buy them separately. There has to be an impedance match, resistance and capacitance to work right.

Very few turntables work without that ground lead properly connected.

Did you ever figure out which type of cartridge you have?

Koz

[david_in_md]

The cartridge is a Grado GF3(E) - at least the front says “Grado GF3” and the scant documentation I can find came from The Vinyl Engine" and indicates that this one only came in an elliptical configuration.

GRADO STEREO PHONO PICKUPS 

SPECIFICATIONS__GF3E+
Output—3 MV J 3.54 CMV—(45 degrees)
Frequency Response— 10-55,000 Hz
Channel Separation—Average 25 DB. 10-30,000 Hz
‘Stylus—Diamond

SPECIFICATIONS—ALL MODELS
Input Load—47,000 Ohms excellent—10K optimal
Inductance—55 millihenries
DC Resistance—700 Ohms
Mounting—i /2 inch centers
Pickup Weight—4.5 grams
‘Last letter in model designates stylus type; R=Spherical, E=
Elliptical Models GF2+, GF1-j-, utilizes a special Grado stylus. G+,
G1-l-. True Ellipsoid Stylus, G2+ Twin Tip Stylus.

The shields for the audio lines are isolated from the chassis ground on the turn table, and they are carried through to the Alesis that way, however I believe it joins the chassis ground and signal grounds at the input. I have connected the turntable chassis ground to the “frame” (such as it is) of the Alesis and it significantly suppressed the hum.

[DVDdoug]

I am extremely suspicious that the power for the Alesis innards is not stable (either coming from the laptop USB port or internal to the Alesis itself.)

Usually, a power supply problem would affect both channels at the same time. But as I always say, "You never know for sure what the problem was, 'till it’s solved.

The low frequency filter is a good tip.

That might help… It might even make it inaudible, which would be great! But, it’s NOT going to eliminate it. (You CAN filter-out constant DC completely.)

I’ll probably re-record anyway to see if more of the “crackle” will leave…

Crackle is usually vinyl surface noise. Sometimes cleaning the record will help, but it’s usually permanently on the record. Hiss is usually from the preamp, or it could be from the record too. I remember some records where I could hear hiss (presumably from the master tape) kick-in just before the music started. (I guess you know where hum comes from.)

Audacity has the normal Noise Removal effect and a Click Removal effect, or you can zoom-in and manually re-draw the waveform.

For low-level constant background noise, Noise Removal usually works very well. But sometimes you can get artifacts, so I’ll usually give it a try and if “the cure isn’t worse than the disease”, I’ll use it.

I haven’t used tried Click Removal. For clicks & pops I normally use a special-purpose program called Wave Repair ($30 USD) that has a few different repair algorithms,and it only “touches” the audio where you identify a defect. The downside is that it takes forever to clean-up a recording, so I plan on trying some more-automated software next time.

I’m coming into the Alesys “guitar” input (1/4" phone jack.) I can’t find the impedance for it in the literature (may need to look harder) but I assume it’s high-impedance.

Right. A guitar input is in the ballpark of 1 Megohm. Typically you wouldn’t get enough gain from an instrument input, but you are not having that problem.

1M would be preferable to 1K or 600 Ohms, but you may still get some frequency response variations.

It’s unusual that the specs say 10k is preferable to the 47k standard. I don’t know where you’d buy a phono preamp with 10K or variable input impedance… I suppose it’s an audiophile item. I have heard of audiophiles playing-around with the capacitance.

Since you are apparenly comfortable with modifying the Alesis, if you like doing that kind of thing you could add some 10k or 47k load resistors. Or better yet, built a little “load box” with RCA jacks that you could connect between the turntable and interface. That way, you wouldn’t have to make any permanent mods to your interface that would screw-up the guitar input.

[kozikowski]

47K with 100pF shunt capacitance. Without the capacitor, many cartridges have a noticeable peak at high frequencies. A cartridge is a resonant circuit and it has to be matched to the preamp for proper operation. It’s a given that left and right need to match, and no, it’s not “legal” to extend the RCA cables from the turntable to the preamp. The cable capacitive loading is taken into account.

Koz

[david_in_md]

Curiously, the “reference manual” for the IO2 doesn’t give the input impedance, but the “Spec” webpage says consistently,
“GUITAR IN (unbalanced) to SPDIF/USB OUT: 94dB A-weighted, input impedance at 1/4” jack = 600 Ohms, Input Gain = minimum." I’m not quite sure how a guitar would work with that low impedence. It might just be a mistake. I might be able to measure it using an ancient oscilloscope I have - at least to the extent to determine if it is 600 ohms or high impedence!

I did try a simple experiment last night (~2AM.)

My previous observations were made with my work laptop, a Lenovo running Windows 8. This configuration gives the strange spikes. I’ve also noted that if you flip certain buttons like switch on the phantom power, or crank the gain controls to max rapidly, the lights on the IO2 would seem to dim, the signal on Audacity face, and then over a few seconds it would revive - hence my suspicion of the power supply. I changed the USB cable to no joy.

Last night, I got a different PC, a Dell Latitude (?) running Windows 7 and hooked it up (Audacity 2.0.5). In this configuration the Alesis was basically inoperable due to “shutdown events” where it would spontaneously power down. The Alesis had a on/off duty cycle of about 20-25%. That is it would light up and begin to function perfectly for perhaps 1-2 minutes, then spontaneously power down and there perhaps for four or five minutes, then power back up and begin operation. This cycle would continue over and over.

Thinking this might be characteristic of a power problem, or some form of current limiting / overload protection, I put a powered USB hub inline. The same shutdown behavior was exhibited.

This made me think that rather than being an electrical problem, perhaps the computer was commanding the USB port to shutdown. I loaded the driver from the Alesis site (versus the generic microsoft driver) and that made no difference. I check into the Windows Event logs and could see the USB activation events, but with no errors, but this was not a extremely thorough check.

At last, I fetched another old Dell laptop running Windows XP and hooked it up and loaded Audacity 2.0.5. I connect the Alesis w/o Audacity it stayed online and from headphones on the Audacity all is good. Then I started Audacity and recorded a side of the record with no shutdown events at all. The waves looked to be centered around the center-line. Finally, I ran the RIAA normalization and this time it produced none of the curious glitches as before.

This experiment basically seems to raise more questions than answers, but I believe the strange coming and going bias seen on the Windows 8 PC are related to the “shutdown” issue that occurred on my Windows 7 PC. FWIW, all these PCs are pretty clean (off-the-shelf) with no strange configuration that would raise undue rationale for this.

I believe I’m going pursue things from this angle for a while. My father had a good suggestion, to make a “USB extender” that would expose the leads and then hang the oscilloscope on them (1 & 4 = power, 2 & 3 = data) to see if any strange electrical nonsense is going on. I am inclined to believe the problem may be “logical” rather than electrical however. It is hard to believe that two laptops and a powered USB hub have an electrical problem, however it could be that aging components in the Alesis have made it voltage sensitive.

At some point, I’ll have to give up on this and buy some new stuff. I really need a clean pre-amp and a digitizer - ideally a standalone digitizer like the Alesis. My budget is limited (2 kids in college.)

[waxcylinder]

@david_in_md

At some point, I’ll have to give up on this and buy some new stuff. I really need a clean pre-amp and a digitizer - ideally a standalone digitizer like the Alesis. My budget is limited (2 kids in college.)

I recommend having a good look at the ARTcessories USB Phono Plus - see: http://artproaudio.com/artcessories/turntable_preamps/product/usb_phono_plus-ps/

For converting my LPs I had an Edirol UA-1EX USB soundcard (no longer available) originally running through an ART DJ-Pre11 phono pre-amp. The USB Phono Plus is essentially the DJ-Pre11 with an ADC (soundcard) added with USB output (a device which essentially integrates the functions of my two pieces of kit; had that been available when I was purchasing I would almost certainly have bought it. In the UK they for for around UK£60-70 - Amazon US has them for US$80-90.

WC

[david_in_md]

Thanks for the reference! It sounds like a good “all in one” solution.

I was able to make a couple of rips using the Windows XP laptop this weekend and it didn’t have the previous issues, however I will need to assess the quality, which I haven’t had a chance to do. I did read up on the Alesis IO2 a little bit. Other people have had issues using it on the 64 bit operating systems and the normal fix is to install the provided driver, however there seems to be some controversy re: needing a driver or it being a USB standard appliance that should work with the Windows driver. All this points to a product that is becoming obsolete, IMHO. To exasperate things, my unit is the “Alesis IO2”, not the newer “Alesis IO2 Express”, so there is even less support for it. I think acquisition is the only practical path forward, and your reference is very helpful!

Best regards,