Clean & Simple Circuit for ~20hz Hipass and DC cut?

[Dakota]

Hey, lots of brilliant people on this board. A utility device that would be very useful to me (and others I am sure) is a simple clean rumble trash cutting hardware Hipass in the 10, 20, 30, 40 hz region.

I'm open minded about slope and trade offs with phase behavior. What is important is cleanest mastering grade path that has as little footprint on the overall audio quality as possible.

And something flexible for both balanced and unbalanced application. I'm sure impedance comes into play here as well. Passive would be great, but active clean also in consideration.

I'm reasonably capable in the DIY dept myself, and am also comfortable deferring to better techs than I. Or do any pieces for this already exist?

Thoughts and discussion much appreciated, thanks.

[The Scum]

Have you seen the Shure barrel hipass filters?

Ethan Winer published an article with some similar stuff here:
http://www.ethanwiner.com/gadgets.html

Wikipedia has some good articles on filters, too.

Any basic electronics book will contain some filter circuits and theory. H&H have a good filter chapter, and Don Lancaster has his Cookbook. Realistically, there are only a few filters that will be suitable for your criteria, and a lot of filter references will spend a lot of time doing imiginary math to teach theory, while designing 100 dB per octave brickwall filters.

For a simple, but functional filter, one cap and one resistor are all you need. Impedance & balanced vs unbalanced are indeed concerns.

I've got a prototype active & balanced HP/LP filter on the workbench right now, but getting it to market requires capital I don't have yet.

[millzners]

Filters are fun to build and interesting to study. I would recommend doing the legwork to learn about how they work and why they do what they do -- because you're building something that's not just a toy but a critical tool.

You should not only know that you want a Butterworth filter http://en.wikipedia.org/wiki/Butterworth_filter, but I would recommend studying the basics of filter design so you know why you want a Butterworth filter as oppose to a Chebyshev http://en.wikipedia.org/wiki/Chebyshev_filter.

There's imaginary numbers and intimidating math assuming you didn't have the coursework previously, but it's really just jargon and the extra work put in to understanding that will go a long way. Trust me, it's easy stuff once you get passed that initial shock of looking at those formulas and realize the frequency domain transfer functions are all you need to know.

[Dakota]

Thank you so much, The Scum and millzners for chipping in, very appreciated.

The Scum, you are one of the two guys at Square State Solid State, which one are you? I really want to try your EQ's. Go Colorado, I used to be in Gunnison & surrounds a lot, family zone. Slice of heaven, mountains and air, best stars at night anywhere.

So let's say the target is a passive racked hi pass, stereo and balanced, stepping switches to successive values of capacitors. Let's say this is going to be in application going to and from a mix of gear, some old and some new, the old stuff having that archaic 600 ohm impedance. What are the typical input impedance ranges on new gear? For example, the Pro VLA II I just got specifies 20k ohms on the XLR, 1meg on the 1/4", which is 1/4" balanced. That's quite a range to cover.

Not that I'm asking y'all to do my math for me, but what capacitor/resistor ballpark are we then talking to do say a 20hz cut going to 600 ohms, 20k, 1 meg? Does this seem right:

Using the little app downloaded from the Ethan Weiner page (great link, thanks!), for 20hz cut going to 600 ohms, I get C1 = 34.501 microfarads, R2 = 1000 ohms

20hz cut going to 20,000 ohm input, I get C1 = 8.758 microfarads, R2 = 1000 ohms

20hz cut going to 1,000,000 ohm input, C1 = 7.978 microfarads, R2 = 1000 ohms

Correct me if this is naive, but if then those two outlying values were then filled in with intermediate values at 1.618x each switch step, then I'd be no further than ~1/3 of an octave at any time from a desired cut, and then I could just run white noise through and see what the actual cut is in that gear configuration.

If that's too many steps, then 2x each step, octaves.

Then, what formulation of capacitors would be recommended for this application? Electrolytic, mylar, tantalum, there are so many types and I'm no expert on what the right choices are in what application.

And very importantly, what happens if the output hits an unbalanced input, and pin 3 (or ring) is either shorted to ground, or left hanging? Does the circuit still do the same frequency function? Does the 1k resistor bleed signal off to ground if shorted, dropping gain?

[The Scum]

The Scum, you are one of the two guys at Square State Solid State, which one are you? I really want to try your EQ's.

Yeah...I'm Byron, the skinny guy with the ponytail...the development & manufacturing side of the business. I'd love to be able to say "for a demo, simply click here..." but we're not there quite yet. All I can say is "watch this space."

So let's say the target is a passive racked hi pass, stereo and balanced, stepping switches to successive values of capacitors. Let's say this is going to be in application going to and from a mix of gear, some old and some new, the old stuff having that archaic 600 ohm impedance. What are the typical input impedance ranges on new gear? For example, the Pro VLA II I just got specifies 20k ohms on the XLR, 1meg on the 1/4", which is 1/4" balanced. That's quite a range to cover.

You're learning that the devil in the details. It's hard to make a passive filter that's good for balanced & unbalanced, with a wide range of source and load impedances.

But we can assess the impedances a little further.

Most line inputs are in the range of 10k to 50K...10K more common on pro stuff, 50K more common on consumer. And typical line outputs are in the 50 to 100 Ohm range...just a little something so that a dead short won't kill the output stage.

The 1Meg inputs are instrument inputs, intended to not load down passive guitar pickups.

Using the little app downloaded from the Ethan Weiner page (great link, thanks!), for 20hz cut going to 600 ohms, I get C1 = 34.501 microfarads, R2 = 1000 ohms
20hz cut going to 20,000 ohm input, I get C1 = 8.758 microfarads, R2 = 1000 ohms
20hz cut going to 1,000,000 ohm input, C1 = 7.978 microfarads, R2 = 1000 ohms

The concept to take away from this is that the 600 ohm input is a significant load on the 1K filter resistor, and it swamps the effect of the 1K in the filter...giving you 1K || 600 = 375 ohms (|| meaning literally "in parallel with"). The 20K and 1Meg are so much greater than the 1K that they're basically invisible (in general, we assume impedance loading won't be much of an issue when the mismatch is greater than 1:10).

So it'll be easier to do this if you eliminate the 600-Ohm requirement. Or build a second set of filters dedicated to 600-Ohm lines.

And very importantly, what happens if the output hits an unbalanced input, and pin 3 (or ring) is either shorted to ground, or left hanging? Does the circuit still do the same frequency function? Does the 1k resistor bleed signal off to ground if shorted, dropping gain?

Depending on the balanced/unbalanced combination, different things can happen. If you're building the balanced filters from Ethan's page, it possible that unbalancing it will effectively halve the capacitance...shifting the cutoff up an octave. It might be easier to build both balanced and unbalanced filters.

Correct me if this is naive, but if then those two outlying values were then filled in with intermediate values at 1.618x each switch step, then I'd be no further than ~1/3 of an octave at any time from a desired cut, and then I could just run white noise through and see what the actual cut is in that gear configuration.

Another dirty little secret...when a filter is labeled, for instance, "1K" that's the nearest round number...it's probably actually some odd number that nearby...985 or 1050 or something that's easy to get with off the shelf RC combos. Don't forget that the components are probably not exactly the stated value...they might be within +/- 5%...plenty close for audio work.

Then, what formulation of capacitors would be recommended for this application? Electrolytic, mylar, tantalum, there are so many types and I'm no expert on what the right choices are in what application.

Mylar, AKA Polyester film is a good choice...but you'll have trouble finding them much larger than 10 uF.

[ashcat_lt]

I've been watching this thread with interest since I'm considering something similar.

Jumped in here to thank The Scum for some really good clear answers, and also just to chime in that it's a good idea, when building the balanced filters, to measure your caps and try to match them as closely as possible to avoid leaving a "hole" for low frequency noise to sneak through.

I made a pair of these for insertion between microphone and pre-amp. For a number of reasons, I ended up having to use 2 caps in parallel on either side of the balanced line. I think this actually helped in matching the overall capacitance values. They work great in the live situation for which they were intended.