|
|
Post by ashcatlt on Jun 27, 2009 13:11:48 GMT -5
I'm working on streamlining my recording rig while simultaneously trying to build a rack for my band's live performance. The idea there is to implement my "silent stage" idea by using nothing but amp simulators and electronic drums. We'll have a single rack that we drag into the venue, plug into whatever amplification is available, and be mixed and ready to go. That's the plan anyway.
Part of this involves my standard vocal chain: an ART Dual MP mic pre into a dbx 266 Project 1 compressor/expander. They both have balanced I/O, though I usually connect via TS using the unbalanced out on the pre.
What I want is a smallish box to go between these two objects. I want this box to accomplish one, maybe two things:
1) A passive high-pass filter with cutoff somewhere around 100Hz. This is the most important aspect, and I could live with just this.
The exact cutoff point isn't exactly critical. I want it to help reduce things like wind noise (our first show is going to be outdoors), stage rumble, bass/drum bleed, and handling noise. Get all that subharmonic noise out of the way so the compressor can work more efficiently on the actual program material. I'd actually prefer to have this between the mic and the pre itself, but different mics have different Zs, which complicates things beyond my capacity.
2) A passive all-pass filter for use as a phase rotator. I guess 100 Hz is a good place for a cutoff on this as well. Or should I say I'd like to rotate phase on most of the relevant vocal frequencies. This is my new secret weapon in mixing. It's used always on radio broadcasts to redistribute the energy of certain signals - like male voices - so they come through a bit more symmetrical. This sounds something like a more natural form of compression, and should allow the compressor to work a bit less.
The problem is, I've got no clue how to calculate values for the various components. Seems like a cap in line with the "hot" wire should accomplish the high-pass just fine. I've read somewhere about a T-filter involving a cap and an inductor that can be used for all-pass on unbalanced signals, but haven't found a schematic.
Anyway, I know the output Z of the mic pre and the input Z of the compressor, but I don't know which of these (or which combination - series or parallel) I should plug into the old filter cutoff equation to get my values. I could spice it up, I suppose, but I'm a little confused as to how best to draw the simulation.
I'm not asking anybody to do the math for me, just need a couple hints maybe. Any help will be greatly appreciated.
|
|
|
|
Post by sumgai on Jun 27, 2009 18:59:51 GMT -5
ash, Ideally you'd use your filter device to match the impedances on each end, but you want passive, so the only method of doing that would be with a transformer or two. Not elegant, not small, not light to carry, and not cheap. Ugh.  Here we're looking at the same thing as a guitar into an amp - we're transferring a voltage moreso than a current, thus an impedance mismatch of low-into-high will not (or at least, should not) hurt the quality of the signal. Thus, you'll want to maintain that relationship by making the new filter seem to be, for all practical purposes, a part of the input stage of the following unit or device. Incidentally, this is why our amps, stompboxes and what-have-you are designed with input impedances much higher than what our guitars are set at - maximum voltage transfer occurs with a low-to-high mismatch, which is more desirable than the other way around. By setting the impedance so much higher, it's much closer to a guarantee that the mismatch will exist in this direction - not many guitars exhibit an output impedance of 100KΩ or more. There are other factors involved, such as frequency (remember, impedance specifies that AC is involved), but for our needs, they can be safely ignored if the mismatch ratio is 1:10 or greater. Keep us posted on your results.  HTH sumgai |
|
|
|
Post by ChrisK on Jun 27, 2009 22:05:06 GMT -5
I do. It's complicated. It will cause massive aneurysms a'board. Let's talk. I use an 1982 edition of A. B. Williams "Electronic Filter Design Handbook". The passive stuff doesn't change much. www.amazon.com/Electronic-Filter-Design-Handbook-Fourth/dp/B000MAHC5O/ref=sr_1_1?ie=UTF8&s=books&qid=1246230908&sr=1-1It represents passive as well as active filters in normalized form that are easily transformed for driving impedances, corner frequencies, and order requirements. The 4th edition includes an Excel spreadsheet for ease in calculations. It also includes MATLAB routines as well. Knowing the cutoff frequency and slope of cutoff will determine, along with the driving impedances, the component values. A passive filter WILL have insertion loss (signal level loss). |
|
|
|
Post by ChrisK on Jun 28, 2009 18:13:41 GMT -5
|
|
|
|
Post by ashcatlt on Jun 28, 2009 23:59:59 GMT -5
Well, I certainly don't need any aneurysms. Maybe we could start with the slightly less complicated (dumbed down) version? The digital graphic is fancy, and way out of budget. I have a Behringer unit that I believe to be this one, which includes 18db/octave high- and low-pass filters. I'm not completely clear on how this would accomplish the phase rotation. Maybe if I boosted (or cut) all frequencies a tiny bit, it wouldn't have much effect on the sound, but there would be phase distortion between the band centers... Unfortunately, it has stopped functioning as a dual-mono device. I should run some more scientific (read: sober) tests to see if I can determine what exactly is wrong. I do know it's not working right. Anyway, it's 2 rack spaces I'd rather not have to deal with. I was hoping to have something the size of a typical DI, which could sit in the back of the rack. Better almost would be to build it into the cables themselves. Easy enough if we're talking a smallish cap value. Not so easy if inductors are involved. I did some 5spice on this. Don't know how accurate the results are for my actual application, but I think it's helped me answer my question. The circuit I used looks like this (R1 = the stated output Z of the preamp, can't find a manual for the 266 Project 1, but the manual for the 266xl says input Z is ">40K" ):  the response thereof:  I guess I should mention that I have previously determined through sweeping the C1 cap that .04uF gave me a -3db point pretty close to 100Hz. Actually, it's somewhere very near to 101.25. So, if I take that thing with the f = 1/(2 pi R C) and manipulate a little, I get R = 1/(2 pi f C) Plugging my known values for f (101.25) and C (.00000004) and Excel's version of pi, I end up with R = 39297, which is pretty darn close to the 40K AmpZ, especially considering that the closest I can get to -3db is only accurate down to the hundredths. Sweeping the value of R1 does nothing (please trust me, didn't see any point in posting a picture of nothing) until its value gets right around that 10:1 area with the AmpZ. At that point we start to see cable capacitance bring an LPF into the picture, as well as general attenuation across the spectrum. Sweeping the AmpZ, however, has a pretty significant effect, in this case it goes (top to bottom), from 40K down to 300:  I know this is over-simplifying everything. There about has to be AC coupling capacitors in series on either end. The extent to which these are very large (to avoid unwanted hpf action) will limit their contribution to the action of this filter, thanks to the way series capacitances combine. Or that's what I think, anyway. Any serious flaws here, or does this .04 uF sound pretty good? |
|
|
|
Post by JohnH on Jun 29, 2009 5:57:15 GMT -5
That all seems cool enough, if the slope of the simple passive filter is enough, which tends towards 6db per octave.
One thing that is helpful in your setup is there is a good factor of more than 100 between the output z and the input z. If you wished, you could use a higher cap, maybe 82nF, and put a variable resistor (and maybe a fixed resistor in series with it) across the input to the amp, to artificially sweep the effective input z and so vary the frequency of the -3db point of the filter.
John
|
|
|
|
Post by D2o on Jun 29, 2009 9:28:39 GMT -5
|
|
|
|
Post by ashcatlt on Jun 29, 2009 12:55:56 GMT -5
That's a pretty neat link. It seems to answer my basic question, and confirm my observations above. This leads to the idea that I could put this before the pre itself, since the mic Z doesn't much effect the cutoff. Needs 2 caps, which probably want to be matched as closely as possible. |
|
|
|
Post by ChrisK on Jun 30, 2009 21:19:16 GMT -5
Oh, I thought that this was something complicated. Of course, I read more into it than what it was.  I think that you're well on your way. |
|
|
|
Post by ashcatlt on Jul 9, 2009 23:42:17 GMT -5
Yes, this was a very simple project, but I read and write about these things much more than I actually do them. Also, I'm married with children and have plenty other crap to do. Oh, and I have a congenital hand tremor exacerbated by heavy caffeine use. So I'm still pretty impressed with myself when I manage to finish something like this. I decided to forgo the all-pass filter, since I still haven't found a good example to borrow from. I did get the high-pass filters built. I used the info from D2o's post, and built them to go between the mic and pre-amp. No point amplifying frequencies I don't want to hear. Haven't heard the filters yet, but my meter readings indicate that everything is wired as intended. Please indulge me while I bore you with details. I happened to have a pair of ART mic splitters which I haven't used in 3 years. For the time being, I have disconnected the transformer* isolated output and its associated ground-lift switch. I removed the input and the "parallel" output from the PCB that they shared. I determined that with a 2K resistor in parallel with the 2K input of the Dual MP, I needed exactly 10/π uF to hit a 100 Hz cutoff. I didn't have any 10/π uF caps. What I had was a number (like 6) of 2.2uF's and some 1uF's. These in parallel come close enough for my purposes. I got out the old meter and figured it out to where the combined parallel capacitance on each side of the filter (remember, there needs 2 caps for a balanced mic signal) matched as closely as possible. I managed to get accurate to the .001 (give or take the slop in my meter/readings). The two different filters didn't end up matching in capacitance, but... I didn't have any 2KΩ resistors, but I had several of the 1KΩ variety. I chose these so that they would add up in series in such a way that R*C was as close as possible between the 2 filters. You know, so both mics will roll off at the same point. I don't figure this is nearly as important as having the two "sides" of one filter match capacitance-wise, but while I had the meter out... So, here's a picture:  The brown wire goes off to a screw which connects to the chassis/shield. I'm going to try them at practice tomorrow. If they don't make a whole lot of noise or make everything suck, we'll use them at the show on saturday. *It has occurred to me that this transformer might be used as an inductor for the all-pass filter |
|
|
|
Post by sumgai on Jul 10, 2009 1:49:47 GMT -5
ash,
Doesn't the beer puddle on the concrete floor kinda exacerbate the caffeine thing? No wonder you hold the world record for most double-pick strokes per second!
;D
sumgai
|
|
|
|
Post by ashcatlt on Jul 10, 2009 10:02:26 GMT -5
 You guys are never going to let me live that down, are you?
|
|
|
|
Post by sumgai on Jul 10, 2009 17:22:12 GMT -5
ash,
You guys are never going to let me live that down, are you? Sure we are. I figure next year oughtta be soon enough, does that work for you? ;D Your friend, sumgai |
|
