Notch filter: headphone and master output differ?

Getting Stuff Done How do I .... ?
#1

Hi folks~

I notice that when I implement a notch filter on either an instrument bus or the Master output, the target frequencies completely disappear from the headphone mix, but the input/output levels on the bus and Master bus continue to show the presence of the hiss.

I tried an experiment in recording 60 seconds of silence except for this low-freq. hiss (@ 55 kHz) and exporting to an audio file with/without the filter, but both files ended up being empty.

Is the filter actually removing these frequencies from the busses, or simply from the headphone mix (which is identical to the Master output)??

Thanks!

#2
...silence except for this low-freq. hiss (@ 55 kHz)...

low freq his at 55 kHz? I think a lot of people (especially those of us who are maybe not as young as we used to be…) would be glad (or amazed) if they could hear something, anything, at 55kHz :slight_smile:

#3

have you accidentally set your buses meters to pre instead of post fader? any channel or bus meters can be changed to different tap points. either that or its a bug.

do you mean low frequency hum instead of his. his is gennerally in the 10khz upwards and hum 100hz and below

#4

DOH! Er… yes… I meant 50 hz, not kHz. My bad!!

I checked that when I was initially seeing this, veda - the meters are all post / out.

#5

Going back over this, I’m noticing a bunch of things (I’m shaking down A 3.5.403 on my system), and maybe it might be more appropriate to write a new thread. But JACK is no longer able to perform monitoring? It’s either hardware or Ardour? This came to my attention because the signal route from the Master bus into the headphone hardware outs does not seem to replicate what is being shown on the Master metering strip in terms sound levels and such.

This whole process is an attempt to solve a VERY long-running issue with my set-up: input levels are extremely low, and bringing them up to a decent level brings up the noise floor (of course) to really obscene levels. There may indeed be some secret switch somewhere I need to flip to solve this, but I’ve not had much luck. And even more odd is the fact that every now and then, the issue vanishes, sound levels are fine, and things run smoothly. (I’ve checked power supplies and turned off everything else in the house, fuse box included, when trying to track this down).

My signal chain is pretty basic: mic (either AKG CS1000 for guitars, or Shure KSM44 for vocals) >> Vintech Dual 72 pre-amp (input = -35, output gain = max; Mogami cables) >> Presonus FP10 interface >>firewire>> DAW (U. St. Linux, Sandy Bridge i7, 16 GB memory, three drives). I generally set input track meters to around -25 / -30 to give myself some headroom. As I said - once in a while all of this works fine, but it usually doesn’t. Weirdness.

If I had money, I’d pay for someone to fly up and spend a few hours correcting all my issues once and for all. :-p

#6

Check the cables.
“once in a while all of this works fine, but it usually doesn’t”
Sounds like a bad connection in a cable, or where a connector connects to a printed circuit board. Have you ever tried wiggling the cables around at every connector, to see if moving the wires near one particular connector makes the problem get suddenly better or worse?

#7

“JACK hardware monitoring” was always limited to only a very small number of devices, and even on those devices, it turned out in practice not to be the most desirable of useful way of using the capabilities of those devices for hardware monitoring. You should basically forget JACK hardware monitoring entirely. If you have a device capable of providing a zero-latency monitoring path, you should use some alternate method for controlling it, since the JACK model was too simplistic to be useful.

But in your first paragraph you use monitoring in a way that is entirely different from the sense of it I’ve just used. You’re using it as “listening to something”, specifically the signal from the master bus. This has absolutely nothing to do with “JACK hardware monitoring”, and is just ordinary audio playback from the perspective of JACK.

(General Questions) Hardware Monitoring on Linux
#8

The gain staging for your setup looks strange.
Why have the mic preamp set for -35dB sensitivity, but the output gain at max? Best noise performance comes from having the highest gain in the earliest stage, and decreasing levels of gain in each subsequent stage.
The Presonus has mic preamps. Are you using the Vintech because it adds some particular kind of distortion you find desirable?
Which Presonus inputs are you using? With an external preamp you should be using the back panel line inputs if you are using channels 1 and 2, preferably with a balanced connection. The Vintech web site does not have a manual available, does that preamp have a balanced output, or is it single ended?
If it is balanced output, it may be transformer coupled, which would mean that connecting to a single ended input should be done differently than for most direct connected solid state outputs.
I assume you are using channels 1 and 2, because it doesn’t look like you can get any kind of sensible gain setup using an external preamp and the other channels. If you are using any of channels 3-8 then we can revisit that.

#9

Paul - you are absolutely correct - I’m just using it in the sense of “listening to the playback from the Master bus”.

CC - I was advised a while back by some folks at a large recording studio in San Diego (due to these same gain staging issues- input signal too quiet when using only the Presonus) to get a decent pre-amp for the signal, which I did (the Vintech). Setting the output at max and the input sensitivity at -35 dB is per the manufacturer directions, which advised maximum output and backing that down only if first setting the sensitivity to it’s limit of -20dB still produced a signal that was too hot.

Now, I am going to confess I have no idea how to handle two conflicting inputs - I had posted this similar question on the Gearslutz forums when I was addressing it this summer (I’ve not posted a single new song in well over a year due to this issue), and the reply I got was exactly the opposite - use channels 3 - 8. This was the reply:

[[Unlike the PreSonus Firebox, the FireStudio Project FP-10 has all of its inputs on the front panel. The rear panel TRS jacks are all interface outputs and should never be connected to an external mic pre output (only to line-level inputs).

On a FP-10 the correct way to connect your Vintech Dual 72 is still with the XLR to 1/4-inch TRS cable connected to any of the front-panel “Line” inputs on channels 3, 4, 5, 6, 7 or 8 by using the center (TRS) section of any of those channel’s “combo” jacks, never the XLR section. Don’t use Channel 1 or 2 because those are fixed “Inst” inputs which are not really correct for a balanced line signal from a mic pre.]]

Now, that particular thread solved a mismatch I had made between cables, but has yet to solve what must be a fundamental misunderstanding on my part regarding the basics of my own gear (mea culpa mea culpa, mea maxima culpa). I am indeed using Presonus channels 3 - 8, not 1 or 2. I experimented with those yesterday, thinking that a phantom power mic doesn’t qualify as an “active instrument”, but in doing so the result was quite a few snaps and pops and such. From what I understand, the Vintech has balanced outputs.

I have found a number discussions regarding removal or reduction of the sound floor for the Presonus FP10 ONLY by switching to a DC power supply, as it can run on either AC or DC; I’ll experiment with that this evening, but that won’t address this issue of signal attenuation somewhere in the chain.

Thanks so much!!!

#10

Here’s the original thread on Gearslutz about this:

#11

Switching to the DC power supply completely removed a hum at around 120 hz, which was very nice indeed. However, I have found that even with nothing at all connected to the Presonus FP10 except the power and the headphones, with the gains turned to max there is far too much noise floor hiss, which, when combined with the faint signal coming from the AKG C1000, is the reason I am a wee bit stuck.

#12

The C1000 is a fairly noisy microphone and isn’t one I’d choose to record a quiet signal source. Large diaphragm capacitor mics have better SNR than small diaphragm ones. The Rode LDCs are the quietest reasonably priced mics I’ve found so far, if you’re looking for maximum SNR.

#13

I have found that even with nothing at all connected to the Presonus FP10 except the power and the headphones, with the gains turned to max there is far too much noise floor hiss

If you have nothing plugged into the input of the mic preamp it will be very noisy, as the thermal noise produced by the input stage bias resistors isn’t shunted out by the relatively low source impedance provided by the mic. When testing preamp noise you need to put a suitable resistor across the hot and cold pins of the input XLR (200 ohms would be typical) to simulate the impedance of the mic.

#14

Ah - I did not know that, jrigg! Thanks for that. The mic itself implements an impedance over the Presonus circuits?

By interesting coincidence, there just happened to be a small Presonus Inspiration on Craiglist last night - I went out and got it to run the same sorts of tests on it and determine if maybe I simply have a bad FP10 for whatever reason. I’ve not gotten JACK to run with it yet, but we’ll see if it yields some new data.

#15

It seems like there are several conflicting assumptions and attempts at solving the issue by changing lots of things at the same time which is rarely a method for success. As I understand it - and I don’t have the exact hardware so I can’t be too specific, but, from the information available:

  1. The AKG C1000 can be either powered from phantom power (e.g. supplied via the mic cable by the pre-amp) or from its own internal batteries. Be clear which option you are using.
  2. The FP10 has ‘combi’ XLR and 1/4" TRS connectors on the front panel (that is, you can plug either an XLR into a socket in the normal way or you can plug a 1/4" TRS connector into the centre part of the socket. Obviously not at the same time, not that it is physically possible…
  3. If you use XLR connections on the front panel input it seems that all the inputs have mic pre’s built in, you adjust these with the gain controls, all the way from line level down to mic sensitivity?
  4. If you use the TRS connections on the front panel, inputs 1 and 2 are designed as instrument inputs, the rest are line level.
  5. The simplest option would seem to be, XLR to XLR from Mic to an input on the FP10, phantom power, adjust the gain control. That should “just work”
  6. The Vintech is designed to take the low level output from the Mic, and convert it to line level. You can adjust the sensitivity from -35dB (most sensitive, therefore highest gain) to -20dB (least sensitive, therefore lowest gain) and also use the “output” control for (extra?) gain adjustment. It should be possible to get a good line level signal from the AKG using the Vintech.
  7. If you use the Vintech, make sure its (balanced) ouput is connected balanced to a line level input on the FP10. If you go unbalanced with this you will lose probably 6dB of level and possibly incur noise / hum problems.
  8. This talk of DC vs AC supplies for the FP10 is worrying. Did it not come with a suitable adaptor? If so, use that.
  9. Running the FP10 with input(s) at full gain as a test will likely be noisy, but normally only some hiss (not hum). It is correct that “unterminated” inputs will pickup noise, these are balanced though, so that should be reduced (e.g. anything getting in will tend to be common mode). When you connect a device to the input, it’s output will normally present a low impedance - the exception being guitar pickups etc - but dynamic mics, or in the case of condenser mics the output stage of their internal electronics will normally present a low enough impedance to reduce some of the unterminated pickup. Impedance matching seems unlikely to be the problem here - just a diversion.
    I would recommend a methodical approach - e.g. vary one thing at a time, and concentrate of fixing one problem at a time (perhaps it would be best to ignore the F/W interface if you can monitor locally through the headphones connection on the FP10, until you are sure you have a good signal path to / through the FP10, then trace the signal into the digital domain through JACK etc and all the linux audio nastiness.

(Disclaimer: This may or may not be any help, and once upon a time I used to know what I was talking about, but if any of this contains glaringly obvious / embarassing errors, I apologise in advance, and its probably safe to assume I’ve already realised my mistake, but this forum won’t ever let me edit anything I’ve written…)

#16

Thanks for all of this, DSP.

  1. I’ve experimented with both phantom and battery power for the AKG C1000 - I did not notice any difference;
    2 - 7. Using the XLR input into the FP10 from the Vintech produces a ton of hiss; The Vintech Dual 72 has a transformer coupled XLR balanced output. That’s intended to drive a “Line-level” balanced input. The FP10 does not have a balanced “line-level” input on an XLR connector, only the 1/4 TRS, which I am using on Channels 3 - 8;
  2. The AC - DC discussion was limited to the FP10, which (apparently) is unique among the Presonus gear in that it can accept DC as well as AC. I found a number of discussions on various forums regarding the switch, which apparently works quite well;
  3. Can you suggest a good set of diagnostic tools to use when tracking the signal path from the hardware through the digital domain? My day job is writing probability models, so I’m not afraid of code, but I confess to limited OS mucking about and I’m a bit lost when it comes to analyzing this sort of thing.

Many thanks!!

#17

Can you suggest a good set of diagnostic tools to use when tracking the signal path from the hardware through the digital domain?

jaaa (JACK/ALSA Audio Analyser) is pretty good if you know how to use a spectrum analyser. You can look at noise levels coming from your preamp or send a sine wave (via the DAC) through the preamps to analyse the distortion. The analyser will also show noise and distortion produced by your ADC and DAC of course, but that’s usually quite a bit lower than the preamp under test.

If you want to look at the waveforms, x42-plugins includes some useful oscilloscope plugins (with the caveat that a scope plugin will only have a bandwidth of half the sample rate, so if there are RF problems you won’t be able to see them).

#18
Using the XLR input into the FP10 from the Vintech produces a ton of hiss; The Vintech Dual 72 has a transformer coupled XLR balanced output. That's intended to drive a "Line-level" balanced input..
My understanding of the FP10 from its manual is that the XLR inputs on the front panel are designed for Mic inputs, in which case, you either connect your AKG directly to the XLR input on the FP10 (as per the hookup diagram on page 15 of the owners manual) and adjust the gain to taste. Or, you need to connect the AKG to the Vintech, then the vintech's (balanced) output to the FP10 via the 1/4" TRS front panel input connection(s) on the FP10. The 1/4" TRS on the FP10 is a line level input (the Vintech converts the AKGs output to line level instead of using the inbuilt mic-pre in the FP10). If the 1/4" TRS input on the FP10 is not designed to accept a balanced signal, then you need to convert from the XLR balanced output of the Vintech to unbalanced for the FP10. This requires some care, to avoid causing more noise and hum problems. I would initially keep it simple. Connect the AKG to the FP10 via an XLR - as per the manual. If you cannot get enough level, try other signal sources suggested in the manual, and see if they work as expected, and / or try to verify the AKG is working as expected by e.g. connecting it to some other suitable equipment (like a mixer with a known working mic input which you can listen to). If you connect the Vintech output to the FP10 via XLR, you are most likely trying to feed the line level output from the Vintech into a Mic sensitivity input on the FP10, effectively two mic pre's in series, which amounts to huge gain, and not suprisingly a lot of hiss.
#19

The whole starting premise just feels wrong to me. That is, needing an external pre-amp because the internal pre-amps in the Presonus could not provide enough gain. Presonus microphone preamps seem well thought of from what I have heard, and the FP10 manual specs the preamp gain as adjustable from 14dB through 54dB. My experience with small diaphram mics is that 20dB is usually adequate when close miking, and even distant miking of e.g. choral or chamber ensembles is usually OK with 40dB gain if you are using condenser mics (as opposed to say ribbon mics).
I’m not a huge fan of the C1000, but there is nothing bad that can be said about a KSM44. People can argue about whether they prefer the sound compared to different microphones, but technically the KSM series cannot be faulted, they are quiet, low distortion, no major frequency response anomalies other than the high frequency rise that all large diaphragm condensers have. Connect the KSM44 directly to the FP10 with a good XLR cable (good just in the sense that it doesn’t have any broken connections), turn the mix control all the way to input, connect the headphones, set the channel gain control to something appropriate for whatever you are recording, and use that as a starting point. Make sure the 15dB pad is not set on the KSM, that would decrease the output of the microphone by 15dB and shouldn’t be needed unless you are close mic’ing drums. How noisy the interface is when all the gain controls are set to maximum is completely irrelevant, with condenser microphones you would never set the gain anywhere near that high unless you were trying to record mice conversations. What you care about is how much noise there is when the audio you are trying to record is at a reasonable level. The c1000 is about 20dB noisier than the KSM, so keep that in mind if you are making the same comparison with the AKG instead of the Shure.

Also, can you clarify your earlier comment that you “generally set input track meters to around -25 / -30.” What style metering is that, and do you mean -25 dB FS? That is pretty low, if you then have to boost the signal 20 dB in Ardour when mixing you will be raising the level of noise in the original recording, so depending on how the gain staging works in the FP10 that might cause higher noise, or it may not matter. I would usually play whatever you are recording as loud as you expect to play in the loudest part you are recording, and set the peaks at around -6dB to -10dB, depending on how consistently you play. 10dB headroom should usually be enough unless you really go crazy once you start recording and play twice as loud as you rehearsed.

#20

I think folks are right when they state this has become a hash of different issues; that’s my bad - sorry!

Addressing sound levels, I had always understood to set the meters around -25 to -15 dB to avoid clipping yet still have room to raise it during export so that your final .WAV file levels were not too low (part of my problem).

Okay - some tests from this evening. These use the Presonus FP10 ONLY - no Vintech pre-amp; signal chain: mic - XLR - FP10 - firewire - Ardour.

  1. AKG mic, battery power, Channel 1 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -25 dB FS
  2. AKG mic, battery power, Channel 3 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -25 dB FS
  3. AKG mic, phantom power, Channel 3 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -25 dB FS
  4. AKG mic, phantom power, Channel 1 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -25 dB FS
  5. KSM mic, phantom power, Channel 1 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -25 dB FS
  6. KSM mic, phantom power, Channel 1 FP10, FP10 outputs set to unity (straight up middle), track/bus/Master bus meters set to -15 dB FS
  7. KSM mic, phantom power, Channel 1 FP10, FP10 outputs set to 3/4 max, track/bus/Master bus meters set to -25 dB FS
  8. KSM mic, phantom power, Channel 1 FP10, FP10 outputs set to just under max, track/bus/Master bus meters set to -10 dB FS

Of course the KSM is more sensitive, so after the initial baselines I used that instead of the AKG.

  1. Clean signal (no hiss) BUT - needed to run amplifier plug in to 70 to get track meters to hit @ -10 dB during play (kept track meters at -25 dB as control)
  2. Empty (signal there but pretty much negligible)
  3. Empty "
  4. Good signal; needed amp. around 55 to hit -10 dB during playback
  5. Stronger signal than AKG (expected); still needed amp boost @ 50 to hit -10 dB
  6. As 5, but needed only 42 amp to hit -10 dB
  7. As 5; needed around 47 amp to hit -10 dB
  8. First level (turned FP10 gain to max and back off by 5 ‘clicks’ of the gain dial - clipped pretty much everything. Backed off another three ‘clicks’ and signal was clean and needed a few amps (5 or so) to hit -10 dB.

I did not notice any of the hiss that has been plaguing me. I am also thinking that some of my earlier issues a few years ago were from working with carbon fiber guitars which have a different wave form than wood (no absorption of any frequencies) and I was confusing those with something in the signal chain.

So along the lines of RTFM, and admitting that I can’t recall what I was doing three years ago when I was told I needed the Vintech, does it seem like am I correct in assuming the following:

a. It’s perfectly fine to set the FP10 output gain near max;
b. It’s perfectly fine to record with the track/bus meters set anywhere between -15 and -5 dB FS
c. Channels 1 and 2 are the only appropriate ones for direct XLR mic input;
d. It appears… sigh… that the Vintech isn’t really necessary (??) and in fact MAY be part of the responsible party for all my hiss crap;
e. Hopefully, all of this will translate into perfectly acceptable sound levels without fuzz when exporting to .WAV for CD manufacture.

Now - I am curious why I still needed an amplifier plug in set to 5-6 to reach -10dB output on the bus meter when the FP10 gain was set to max - 8 clicks (sorry - no idea how better to describe it!) and the track/bus meters all set to -10dB. Shouldn’t this already be sufficiently hot to hit that? Ccaudle, when you state " Connect the KSM44 directly to the FP10 with a good XLR cable (good just in the sense that it doesn’t have any broken connections), turn the mix control all the way to input, connect the headphones, set the channel gain control to something appropriate for whatever you are recording, and use that as a starting point", to what are you referring when you say “mix control all the way to input” (FP10 output to near max??), and channel gain control (I’ve been calling this the track/bus peak meter?).

You folks are being an absolute wonderful help - I live in Alaska with no road access and without much direct access to good recording help, so I greatly appreciate it!

And I might just stick with the KSM for both vocals and 6-string guitar; the 12-string is sufficiently loud and rich that the AKG seems to do quite well with it.