Output from phono stage too low for good digital recordings

[LateJunction]

I have a small problem of understanding terminology and technology in trying to find a solution to an audio issue I have – could I get some advice please?

First the issue: recordings I make from vinyl in Audacity are much too low in volume, so the histogram looks like a crinkly straight line with the average recording level (as shown by Audacity) being about -36 to -24 dB, with orchestral crescendos going to about -12 dB, instead of -3 to -1 dB that I am expecting. In contrast, when I’m recording from FM Tuner, CD or cassette, I see a recording histogram which looks like a set of large horizontal Christmas trees (that’s about as technical as I can get…), with average recording level of about -12 dB and frequent peaks at -3 dB.

Kit being used is Rega P7 + Dynavector DV20X2 Low + Cambridge Audio Azur 640P phono stage + Behringer UCA-202 + short USB cable to laptop with Linux Mint 18.1 XFCE + Audacity. (Adding an AudioLab 8000Q between the phono stage and the Behringer ADC, using the tape outputs of the 8000Q, makes little difference).

The problem is that the output from the Dynavector cartridge is only 0.3mv and the gain of the 640P phono stage is only 55dB. But I cannot work out, using these figures, what the voltage output from the 640P is and how that compares to the input required by potential replacements of the Behringer UCA-202. Most of these require about 500 mV rms. Can somebody explain to me how I can calculate the nominal output of the 640P?

As a secondary question, could I have some recommendations on a suitable ADC that will give me 24/96 or even 24/192 or even DSD? I have been looking at:

* Novation 2x4 AudioHub (basically a Focusrite Scarlett Solo re-engineered for domestic use rather than professional audio)
* Focusrite Scarlett 2i2
* M-Audio M-track 2x2 C series
* ADL Furutech GT40a
* Korg DS-DAC10-R

I have the impression that I will not be able to drive any of these devices to -1 dB recording level with my kit. Any recommendations?

[Stryder5]

Have a look here, see if it helps.

https://www.kabusa.com/pregain.htm

Looks like you need 61 dB of gain

From the 640p manual they state 0.5-1.0 input

https://techsupport.cambridgeaudio.com/hc/en-us/article_attachments/200154672/Azur_640P_User_Manual_-_English.pdf

[LateJunction]

Have a look here, see if it helps.

https://www.kabusa.com/pregain.htm

Looks like you need 61 dB of gain

From the 640p manual they state 0.5-1.0 input

https://techsupport.cambridgeaudio.com/hc/en-us/article_attachments/200154672/Azur_640P_User_Manual_-_English.pdf


Excellent reference - thanks. Confirms my suspicion that I should be looking for a phono stage with 65 dB gain (to give myself a little margin for error).

[Light Dependant Resistor]

I have the impression that I will not be able to drive any of these devices to -1 dB recording level with my kit. Any recommendations?

Digital levels are always held back because of risk of overload. However a read of this paper will show you technique to overcome that limitation
ftp://ftp.dbxpro.com/pub/pdfs/WhitePapers/Type%20IV.pdf

Cheers / Chris

[Infinitely Baffled]

Hello, Tony. Arkless Electronics, who have a trade room here on AoS, have developed a highly regarded phono stage which uses your Cambridge Audio stage as a "donor" unit. My understanding is that it is rather more than an upgrade of the Azur 640P. It is in effect a redesigned phono stage which uses the Azur as a basis. I don't know whether part of the Arkless spec involves a higher gain than the original donor unit, but my guess is that if you speak to Jez Arkless, he will be able to incorporate that specific requirement, if it is not a standard part of the redesign. I would have a word with him, if I were you. Good luck in sorting the problem.
IB

[Rothchild]

I wouldn't let it bother you, 0dBvu = -18dBFS ie when the record was cut and the needles on the meter went through 'zero' that would only make -18 on a 24bit digital meter.

If you're recording at 24bit you're capturing all the range on the record anyway, further analogue gain just risks pushing more analogue noise in to your recording. If you're really unhappy about the appearance of the file, or you want to have the files at a level where they match other sources just normalise them in audacity.

[mike1210]

Been recording vinyl for years and made several mistakes along the way. Soundcard with input-trim is essential to get the levels right. A drum a bass single running at 45rpm is way louder than a 33rpm album with 30 minutes of music on one side, getting a phono stage with more gain will only fix part of it if the input stage of the soundcard can't be set.

Many people have the opposite problem to what you have, the output from the phono stage has way too much beans on it. Some output up to 4VRMS.

Used a Focusrite Saffire for years with input trim on the left and right channels which is very useful. My Dino was overloading my previous card so had to get a new one.

Now with a RME Babyface Pro which isn't cheap but awesome.

Looking at your system, I'd highly recommend a better sound card to get decent quality recordings.

Mike :-)

[Arkless Electronics]

Hello, Tony. Arkless Electronics, who have a trade room here on AoS, have developed a highly regarded phono stage which uses your Cambridge Audio stage as a "donor" unit. My understanding is that it is rather more than an upgrade of the Azur 640P. It is in effect a redesigned phono stage which uses the Azur as a basis. I don't know whether part of the Arkless spec involves a higher gain than the original donor unit, but my guess is that if you speak to Jez Arkless, he will be able to incorporate that specific requirement, if it is not a standard part of the redesign. I would have a word with him, if I were you. Good luck in sorting the problem.
IB

Cheers Gary :)

The OP rang me yesterday actually for advice on this and we discussed this option.... I also suggested that he could take the output from his pre amp (the main outputs, not the tape outs which he was trying to use) and thereby, assuming the line stage has gain, boost the output from the phono stage to a suitable level.
The Arkless 640P has much higher output than the standard one (an AOS member with one on order has actually asked me to reduce the gain a bit on his!) at 64dB but I can customise for any gain at no extra cost.
This would vastly increase the quality of the sound as well of course! Much more so than a better sound card!

[cre009]

Some observations from my own experience.

I briefly had a go at doing transfers using the Peppermint Linux Distro (a variant of Mint) and also found the recording level was too low using an external Sound Blaster ADC. The Sound Blaster worked ok on Windows so suspect the problem may be within Linux. On Windows for two of my ADCs I can change the recording levels via control panel settings and would expect the same within Linux. However I am not an expert on Linux myself so would suggest placing a question with the Linux Mint Community forum.

For 24 bit recordings you really need to be using "Jack" in Linux or "ASIO" or "WASAPI" in Windows.

I have both the ADL GT40a and Novation Audiohub 4X4 out of the devices you mention. I use ASIO in Windows myself for 24/96 and both give me very good results as an ADC. There is a thread at Vinyl Engine on the ADL GT40a which I participated in. I really like the unit which is versatile though caution that the phono stage is adequate rather than great. There are 3 gain settings on the box itself but gain can also be controlled via the Windows Control panel.

For some reason the gain for the Audiohub can only be controlled via a switch on the box. Gain on the Windows Control panel is greyed out. The 2 settings are ok but not flexible enough for my liking so I use a Nakamichi Pre Amp to control the input gain for transfers via the Audiohub.

[LateJunction]

Hello, Tony. Arkless Electronics, who have a trade room here on AoS, have developed a highly regarded phono stage which uses your Cambridge Audio stage as a "donor" unit. My understanding is that it is rather more than an upgrade of the Azur 640P. It is in effect a redesigned phono stage which uses the Azur as a basis. I don't know whether part of the Arkless spec involves a higher gain than the original donor unit, but my guess is that if you speak to Jez Arkless, he will be able to incorporate that specific requirement, if it is not a standard part of the redesign. I would have a word with him, if I were you. Good luck in sorting the problem.
IB

Thank you for this recommendation. I have indeed spoken to Jez who was more helpful than I had a right to expect, given that my call was totally unsolicited. He made a suggestion (which to me is a measure of the man) to try a zero-cost alternative first before spending any money. I'll be making a further call to Jez soon I suspect!

[LateJunction]

Some observations from my own experience.
.

Ah. most interesting comment about Linux and the recommendation to use Jack - I hadn't thought of these factors before. Thanks.

I had read that the phono stage, especially for MC, is a bit 'entry level' on the GT40a so, having invested in a quite good MC cartridge, I think I'll give that one a miss. The Korg device sounds the most exciting - primarily because it's software appears to be able to identify the music being recorded and then do automatic looks-ups of recorded music databases to get the 'EXIF-like' data for the recording (Album Title, artists, composer, tracks etc) and package each 'track' (equivalent to movement I guess) for you. This saves a huge amount of work. Sadly none of this works under Linux though, which means that if I want to use this device I have to buy a refurbished MAC (being seriously concerned about cyber-crime I absolutely refuse to use that operating system apparently designed for and ideally meeting the needs of cyber criminals, aka Windows).

[LateJunction]

Cheers Gary :)

The OP rang me yesterday actually for advice on this and we discussed this option.... I also suggested that he could take the output from his pre amp (the main outputs, not the tape outs which he was trying to use) and thereby, assuming the line stage has gain, boost the output from the phono stage to a suitable level.
The Arkless 640P has much higher output than the standard one (an AOS member with one on order has actually asked me to reduce the gain a bit on his!) at 64dB but I can customise for any gain at no extra cost.
This would vastly increase the quality of the sound as well of course! Much more so than a better sound card!

Hello Jez, I guess I'll report back to you here on the forum rather than taking your valuable time on the phone. I couldn't try your suggestion until this morning when I collected my Rega P7 TT + Cartridge after being serviced by my local hifi adviser here in Winchester. My amplification consists of an AudioLab 8000Q driving an AudioLab 8000P. The 8000Q has two identical RCA line output socket pairs , so I was able to connect the Behringer UAC-202 to the 8000Q line output which was not being used, allowing me to continue to hear the music while digitising it via the Behringer.

The results have baffled me: the 'histogram' in Audacity was now an almost perfectly straight line. It could be adjusted using the volume control on the 8000Q (obviously) but only became a recognisable recording 'plot' when the volume from the speakers was at an uncomfortable level. Playing back this recording either through the Behringer to a tape input on the 8000Q or via the in-built speakers on the laptop or via the speaker output connector on the laptop into Tape-In on the 8000Q produced a barely audible output on my main speakers (Spendor SP1s). I swapped the connections on the 2 line-outs on the 8000Q - no difference on the volume on the main speakers and no difference to the recording levels on the laptop either - still just a straight line, much much quieter than connecting the Behringer UAC-202 to the output of the 640P. I tried by-passing the Behringer entirely, using the mic-in connector on the laptop connected to the output of the 640P. The recording volume was now higher - but no higher than when using the Behringer taking the signal into the laptop via USB (but of obviously poorer quality using the ADC backing the mic-in on the laptop),

Bottom line: I think I need your re-engineered quasi-640P running with about 64 dB gain!. But I'll talk to you outside this forum to agree the details - is that OK?

[LateJunction]

I wouldn't let it bother you, 0dBvu = -18dBFS ie when the record was cut and the needles on the meter went through 'zero' that would only make -18 on a 24bit digital meter.

If you're recording at 24bit you're capturing all the range on the record anyway, further analogue gain just risks pushing more analogue noise in to your recording. If you're really unhappy about the appearance of the file, or you want to have the files at a level where they match other sources just normalise them in audacity.

Actually, because of a long-track record of total ignorance of recording techniques, exacerbated by ill-thought out farting about, I'm not sure what it is that I'm worried about or trying to achieve. But I certainly feel quite dis-satisfied with what I am achieving. Aside from the (reasonable) limitation in the Behringer UAC-202 - 44.1kHz, 16 bit only - I think I'm enervated by the idea that the essentially flat plot in Audacity means that when I normalise it I'll be boosting the noise just as much as the 'real' signal (whatever that is). Everything tells me that the signal to noise ratio will be much improved if I'm driving the recording to the (once only) maximum of -1dB - which I take to mean dBvu. Or is my ignorance overwhelming my sense of judgement here?

[Rothchild]

Tony, you seem very keen to spend money, which is cool, but you're circling a rabbit hole unnecessarily - the levels you described in your OP are perfectly adequate for capturing a good recording at 24bit, if you really insist that the waveform has to be visibly bigger just use the 'normalize' tool in Audacity and set a peak of -0.2dBFS (this will be the same but much easier than faffing around with JACK etc). Pushing the signal level at the input just takes you closer to risking clipping and distorting the recording.

Bear in mind that 24bits gives you a dynamic range of 144dB, this means that if you had a suitable soundsystem a 24bit recording can contain data representing a sound quieter that we can hear and louder than we can tolerate, you don't really need the signal all stuffed up at the top end of this range.

[Rothchild]

Actually, because of a long-track record of total ignorance of recording techniques, exacerbated by ill-thought out farting about, I'm not sure what it is that I'm worried about or trying to achieve. But I certainly feel quite dis-satisfied with what I am achieving. Aside from the (reasonable) limitation in the Behringer UAC-202 - 44.1kHz, 16 bit only - I think I'm enervated by the idea that the essentially flat plot in Audacity means that when I normalise it I'll be boosting the noise just as much as the 'real' signal (whatever that is). Everything tells me that the signal to noise ratio will be much improved if I'm driving the recording to the (once only) maximum of -1dB - which I take to mean dBvu. Or is my ignorance overwhelming my sense of judgement here?

The 'fun' here is caused because there are a bunch of different standards. Analogue equipment tends to be rated in dBVU (Volume Unit) digital systems/meters tend to show dBFS (Full Scale) where 0dB is the point where you run out of bits (and if you hit that point you get distortion and clipping). In general terms if you were to feed a -18dBFS signal in to an analogue box with dBVU meters the meters would sit at 0dB, conversely, if you put a signal out of an analogue box with its meter sitting at 0dB (and don't do any other gain staging to the computer input) the meter in the computer will show -18dBFS.

So what you've described in your OP actually sounds about right levels wise. Yes if you turn the signal up digitally in audacity you'll turn the noise up too, but that would also be the case for adding gain electronically at the preamp stage.

By way of example, when I record bands I'll set the gains going in to the system so that the peaks hit between -18 and -12dB FS, this is plenty of signal to work with, means I'm generally safe from clipping the input if something gets momentarily loud and leaves me headroom for summing tracks together.

[Arkless Electronics]

Hello Jez, I guess I'll report back to you here on the forum rather than taking your valuable time on the phone. I couldn't try your suggestion until this morning when I collected my Rega P7 TT + Cartridge after being serviced by my local hifi adviser here in Winchester. My amplification consists of an AudioLab 8000Q driving an AudioLab 8000P. The 8000Q has two identical RCA line output socket pairs , so I was able to connect the Behringer UAC-202 to the 8000Q line output which was not being used, allowing me to continue to hear the music while digitising it via the Behringer.

The results have baffled me: the 'histogram' in Audacity was now an almost perfectly straight line. It could be adjusted using the volume control on the 8000Q (obviously) but only became a recognisable recording 'plot' when the volume from the speakers was at an uncomfortable level. Playing back this recording either through the Behringer to a tape input on the 8000Q or via the in-built speakers on the laptop or via the speaker output connector on the laptop into Tape-In on the 8000Q produced a barely audible output on my main speakers (Spendor SP1s). I swapped the connections on the 2 line-outs on the 8000Q - no difference on the volume on the main speakers and no difference to the recording levels on the laptop either - still just a straight line, much much quieter than connecting the Behringer UAC-202 to the output of the 640P. I tried by-passing the Behringer entirely, using the mic-in connector on the laptop connected to the output of the 640P. The recording volume was now higher - but no higher than when using the Behringer taking the signal into the laptop via USB (but of obviously poorer quality using the ADC backing the mic-in on the laptop),

Bottom line: I think I need your re-engineered quasi-640P running with about 64 dB gain!. But I'll talk to you outside this forum to agree the details - is that OK?

That's absolutely fine Tony. Nice speakers BTW!

[sq225917]

I do my rips on my sony pcm10, it's a little adc/dac digital recorder with line inputs, decent microphone amps and Microsd card and USB data outputs. It'll do 24/96 and has variable input gain with vu metering. It works fine with every phono stage I've tried.

Get the best phono stage you can and try a decent digital recorder the behringer is just a toy.

[cre009]

If the OP is using the Pulse Audio protocol on Linux then I believe that gets relatively low resolution rate even if trying to record off a 24 bit card. I read that Pulse has much the same issue as MME direct sound on Windows. A 16 bit recording will actually achieve far less resolution than 16 bit. My understanding is that for every 6db of attenuation below the clipping level there is a loss of 1 bit in the recording resolution. If Audacity is showing a flat line then that will be a low quality recording.

I still think he should enquire within the Linux community to establish how to get the best results.

[Rothchild]

There's a nice simple piece here: http://dbzeebee.blogspot.co.uk/2009/10/digital-recording-levels-rule-of-thumb.html explaining about levels and bitdepth (and why not trying to 'fill' the whole digital meter is going to save a bunch of issues down the line).

I know many folk don't like Pulse but I'd never heard anything about quality issues, got a link?

[cre009]

Nice link.

I am going off memory from researching the issue about 18 months ago. There is a lot of material out there but a lack of clarity and enough people knocking pulse that I preferred to steer clear of it. ASIO reads direct off the Soundcard and Jack is the Linux equivalent. Pulse is apparently similar to direct sound.

I was originally doing 16 bit recordings in Windows using direct sound but wanted 24 bit particularly when trying to fix issues using iZotope RX. It became quite clear to me that the 16 bit recordings I had been getting using a standard PC sound card were relatively mediocre. I also had a 24 bit capability with the SoundBlaster but eventually found that recordings using direct sound on that card were the same due to limitations in direct sound.

I explored 3 options. The first was Linux but I struggled to get it up and running on my old PC. I did not try very hard because I really needed things to be portable which the PC was not. The second option was an external digital recorder as suggested by sq225917 but was not sure which device would meet my needs. The third option was ASIO on a Windows laptop which I got going with the ADL GT40a device and have been very impressed with the clarity of the recordings over what I had before. Since then I have tried WASAPI with the Soundblaster which also gave me a good result.

[RothwellAudio]

There's nothing wrong with the phonostage - no need to get another.
Yes, if you want more signal into the Behringer you can use the main outputs of your Audiolab pre-amp but don't use the Audiolab for monitoring at the same time!
If anything needs upgrading I would suggest it's the Behringer.

BTW, I haven'y used Audacity for a while, but isn't there some kind of software control over the recording level?

[Arkless Electronics]

There's nothing wrong with the phonostage - no need to get another.
Yes, if you want more signal into the Behringer you can use the main outputs of your Audiolab pre-amp but don't use the Audiolab for monitoring at the same time!
If anything needs upgrading I would suggest it's the Behringer.

BTW, I haven'y used Audacity for a while, but isn't there some kind of software control over the recording level?

Precisely what I advised the OP on the phone.

As to whether to get a new phono stage well obviously the rips to hard drive can only be as good as the weakest link in the chain. The phono stage is holding it back without doubt as anyone who has compared a standard 640P with an Arkless 640P can attest.

The Behringer may or may not also be a weak link. If it is it's likely to be the analogue input section before the A2D converter that's the issue as Behringer will be using a proprietary bought in A2D chip the same as all other manufacturers... Probably from AKM.

[RothwellAudio]

Yes, I'm not saying the phonostage can't be bettered, but the OP isn't complaining that the quality isn't up to scratch, only that the level is too low.

[LateJunction]

<snip>

So what you've described in your OP actually sounds about right levels wise. Yes if you turn the signal up digitally in audacity you'll turn the noise up too, but that would also be the case for adding gain electronically at the preamp stage.

By way of example, when I record bands I'll set the gains going in to the system so that the peaks hit between -18 and -12dB FS, this is plenty of signal to work with, means I'm generally safe from clipping the input if something gets momentarily loud and leaves me headroom for summing tracks together.

Having taken an executive decision (which was to make a decision) I purchased a Focurite 2i2 USB Audio Interface (the ensuing fiasco with Focurite's dealer provides a rich store of material for possible future posts). With this device I find that I can make excellent recordings into Audacity under that security-challenged patch-deck of bloat-ware known as Windows and quite good ones under Linux, even though there appears to be no way of adjusting recording levels in either the operating system or the Application software. (Yes, both claim to be able to do so; neither can support the claim). So, relying on the gain control hardware in the Focusrite box, I find that recordings made following your level recommendations do not appear - to my ears/brain - to sound as 'effective' ( can't really find the apposite word here) as those where the recording peaks at -3 dB or so - which has been my naive way of working. And when I say 'peaking' there is, for example, only I place in a recording of a 45 minute performance of Beethoven's Op. 131 where the recording level reaches - 3dB and that is for less than a second.

[Rothchild]

Glad to hear you've got it up and running to your satisfaction

There is no way to control recording levels within the OS because the signal has already been converted by that point (so all you'd be doing is 'normalising' so we'd be back to step one).

If you're trying to compare a recording made with lower peak values to one with higher values they you need to be scrupulous about level matching the playback system (loud sounds better, experimentation has demonstrated that level differences of less that .5 of a dB can cause a listener to express a preference for the louder signal)

But, at the end of the day, if the signal isn't clipped and you're happy with it then this is somewhat academic.