We warn people about those English words. Stereo Line-In (blue) is very different from the mono Mic-In (pink). Very few sound cards have provision to switch between them allowing you to use one connection for both. The best sound cards that provide one connection have little “tick” sounds when they switch because they’re using mechanical relays to do the work. We are talking about managing 0.002 volt analog signals with a noise floor a thousands times lower than that. I need to drop by Tesco later and pick up some more zeros.
Chances are if you have a PC laptop, you have a Mic-In and many of these high-level, stereo production tricks are closed to you. If you have a much larger deskside, you probably have both pink and blue.
I have a dedicated soundcard inside my PC, it has only one input for both line-in and mic-in, although inside it has two separated circuits for each, which are selected (from the software) by an internal relay
Edit: and the plug is a 1/4" (6.3mm) female jack and it isn’t either blue or pink, it’s black and gold
The New Japan Radio NJM5532 and Texas Instruments R4850 op-amps are used for the line-in/microphone input. Asus chose to use miniature relays to do the switching between line-in and microphone modes. My theory is that the relays route the signal through the NJM5532s for the line-in, or the R4850s for the mic mode. The use of relays to switch the routing eliminates the signal loss that you would get if you used solid-state switching.
They also say that:
To the left of the AV100, you can see the Cirrus Logic CS5381 ADC used for the analog inputs. The DJ100 is apparently an AC’97 codec DAC (C-Media CMI9780), and therefore the microphone section of the AC’97 feature set is used as a pre-amp for the microphone (and front panel microphone). AC’97 is not synonymous with the highest quality, but as this soundcard is tailored toward high-end audio playback, it’s not really an issue. You could always get a proper microphone pre-amp if you wanted to do some production-level recording from a microphone. The microphone input cannot be configured as balanced anyway.
Nonetheless I’d still like to give it a try…
Connecting the mic to a preamp and then the preamp to the line-in on the soundcard (selecting the input as line-in and not mic) is another setup that I’d like to try… I’ve tried to connect my Marantz CD Player to the line-in and it seems to be quite noise free… so if I connect a decent mic preamp with line output I’m hoping to get a good signal too…
My quest here would be mostly to find out which ADC would be better… the ADC in a mic preamp + usb output VS. soundcard ADC + a preamp with 1/4" line out
There’s plenty of DIY circuits for that on the Internet. Really good regulation/smoothing is the key. At one time I was considering building a box for 5 pp3 batteries and a cap across the output - 45v should be close enough, but the cost of batteries put me off. For solid state microphones the current draw is really tiny.
If you get round to making one, good quality capacitors are essential as many off-the-shelf electrolytics will produce an unacceptable amount of leakage noise.
Black Gate caps are probably OTT and are hard to get hold of. Elna Cerafine and Nichicon FG are good and more reasonably priced, but again are difficult to get hold of in Europe, but Panasonic FC are good and readily available (CPC, Farnell, RS, …)
LOL - I spotted that yesterday.
Finally got to hear and play with your latest audio sample. This is a pleasure to work with
I’ve had a go at tackling the noise problem.
In this sample I first ran a 6dB/octave high pass filter at 20Hz to remove the DC off-set and a bit of sub-sonic rumble (was that a car going past near the end of the track?) The HP filter seems to work a bit better than the “Normalize” effect for removing DC off-set on this sample as the DC off-set amount drifts a bit (sub-sonic). Then Amplified close to 0dB.
Then I used a subtle amount of noise reduction using Gnome WaveCleaner. The amount of noise reduction was set very low (0.25 on a scale of 0 to 1.0) and the “gamma” setting dropped down to 0.5 (this helps to preserve transients).
Finally I used the noise gate with a fairly slow attack/decay.
Definitely useful doing this as it appears that there is an undocumented change in Nyquist which means that the Noise Gate effect is not compatible with Audacity 1.3.11! (I’ll have to investigate this further).
The result is very close to the sample from Koz (23-sample-with-silences-Keypex). You can hear a slight difference if you listen carefully to the decay on the final note. In the Keypex version you can hear the noise level creeping in as the note decays (before the gate kicks in). Also, in the original track there is a slight “bump” sound just after the note stops (your hand on the guitar?) which has disappeared in the Keypex version. In this version the noise reduction has removed the hiss from the last note and the “bump” is still audible (though on a production job I’d probably remove it). http://audacity.easyspacepro.com/examples/noise_removal/23-sample-with-silences-NR.flac
That will probably work pretty well. A lot of the poor reputation of AC’97 comes from the very poor quality analogue components that are typically used in the analogue end of on-board sound cards. Also the high gain required for microphone pre-amplification is much more demanding than dealing with line-level.
With a reasonable mic pre-amp I’d guess that the difference will be subtle. Getting the 50dB or so of gain to boost the mic level signal up to line level without introducing noise and distortion is the hardest part.
Most probably… I was recording in the living room (which is where I temporarily have my PC), where the cars passing in the front road can be heard more than I wished for…
The 10m cable I got is a couple of meters short to allow me to take the mic into the free room on the other side of the appartment… (option here is to use the laptop instead). Other place I tried to record the acoustics where terrible and the sound was too boomy…
You can hear a slight difference if you listen carefully to the decay on the final note. In the Keypex version you can hear the noise level creeping in as the note decays (before the gate kicks in). Also, in the original track there is a slight “bump” sound just after the note stops (your hand on the guitar?) which has disappeared in the Keypex version. In this version the noise reduction has removed the hiss from the last note and the “bump” is still audible (though on a production job I’d probably remove it).
The “bump” is probably my fingers going back to the strings after the final chord… You can also hear some breathing during the play
If you want another project to hang around for several years, you may be interested in this snippet that I stumbled upon recently: viewtopic.php?f=27&t=27426
That looks interesting… and it’s available for less than 5 euros at farnell… I may as well add it to the order I’m been thinking about placing at farnell for the last year or so… (still trying to finish that never-finished list of components…)
Although the problem here, as in any “sensitive” electronic gear, is to get/make a good power supply…
I plan on posting an update on the initial post of this thread once I got everything sorted out and settled… So that the newcomers may just quickly read the short version of a very long thread…
If your circuit has no coils at the input, then you get to fool around with not only balancing the 6k8 DC feed resistors very carefully, but balancing the coupling capacitors as well. Remember, they have to be large enough to pass the lowest frequency of interest plus extra to avoid low frequency current noise, and well behaved enough to work with no DC bias on them in the event you never turn the +48 volts on.
The worst curse of phantom power is not what happens when it’s running, although that can be fun. When you turn it off, it dumps a sudden 48 volt charge into the amplifier system. This would be the delicate, noise tuned, super low distortion amplifier system. I can’t find it immediately, but someone made a phantom system that faded the battery supply on and off to avoid that sort of damage.
<<<Also, in the original track there is a slight “bump” sound just after the note stops (your hand on the guitar?) which has disappeared in the Keypex version.>>>
But not all of the versions. The one before this one had a different threshold of noise sensing and the bump came through. I changed the threshold to get rid of it because it came through not as a polished-wood/string/musician activity, but a “pish” of noise.
Delicate events near the noise threshold are very important when messing with this process.
I guessed that, and it helped to identify the problem. Since Nyquist has been updated it is now (correctly) more fussy about whether values are integers or floats. Just adding a dot has fixed that problem. However, I’ve clearly learned a thing or two since I wrote that (almost a year ago) and I’ve identified a couple of other bits that should really be improved. Apart from the coding being rather inelegant, there is virtually no error checking in it and the attack/release times are totally inaccurate. I’ve added some error checking to prevent silly values from breaking it so I just need to sort out the timing issue and tidy it up. Hopefully get that done this week.
@Koz, this would be a good time to say if you want a feature for measuring the noise level, or are you now happy with using the Amplify effect to do that?
As a matter of taste, I’d do that as well. Creative use of noise gating.
Here’s one such circuit, though I’m not convinced that the voltage regulation is really adequate for high quality recording, though that will depend on how forgiving the microphone is:
That one with the LM317 sounds interesting… I’m already starting to cook something in my mind using those schematics and the phantom power example on the SSM2019 datasheet… I think the most expensive components there would probably be the capacitors (if I want to get decent ones), but maybe I should start a new thread to discuss that
In other news… I was playing around with gnome wave cleaner and I liked its noise removal. I didn’t know this app but I think it performs pretty well. The gui has some bugs but it’s usable.
It seems overkill to me, and I don’t see how they’re getting rid of rapid charge and discharge. Although that circuit would do it. Initially, I couldn’t see where they were getting enough DC to run everything, the monolithic regulator needs at least three volts across it to work, then I saw they were using two power supplies one + and one -.
But they’re both half-wave rectifiers. Wouldn’t you get better and stiffer raw DC with bridges or full wave? Then you might not need lab grade regulation or such large caps.
I didn’t look up the international spec on 48v phantom. I’m pretty sure there is one. It’s scary that people are not just using the juice to run a small amplifier inside a microphone. I have a new microphone that has three LEDs on the case as well as direction management and filter generation. Given all that, I’m sure they regulate inside the case. They were firm that they couldn’t do all those jobs on less than 48v.
I’m still not sure what I’ll use for source of power… I might go for a DC-DC converter. I have plenty of old power supplies ranging from 5V to 12V (DC) so I might use one of those and just make the voltage conversion to 48V plus some caps for filtering…
I can also try to use the 5V or 12V from the PC’s power supply, or even the USB’s 5V…
If I had a dollar for every time I’ve seen (or heard) a microphone being plugged in or unplugged with the phantom power on and the gain turned up on the desk I’d be moderately wealthy. Equipment needs, whenever possible, to be designed to withstand such abuse. The LM317 data sheet provides example schematics for how to protect the PSU from damage due to output short circuit. It also gives an example for a “soft start” regulator.