How to Build or Have Built a Low-Pass Filter Based On A Custom Schematic

[Ken Reeser]

I am looking for advice on how to either build a device myself, or find someone to build it for me, or some combination thereof.

I have a schematic for a low-pass filter designed to help me reduce the signal from an ultrasonic motion detector, which is emitting a very loud "noise" at exactly 25kHz. My microphones are receiving it, and my mixer is reacting badly to it. The subject of the signal, workarounds, other microphones, etc, can be found in this thread here:

http://www.dvinfo.net/forum/all-thin...detectors.html

In this thread, though, I want to leave behind other possible solutions and focus on this one:

I have a design and schematic for a low-pass filter created for me by an enthusiast in Germany who took up my cause. Unfortunately, the firm he recommended to build it wants far more than is practical for me to pay, especially considering that I have to get at least six of these, maybe more.

An alternative to getting six would be to install the same LPF into the proper place of my Shure M367 mixer, thus having all signals go through it before going on to cause havoc. This would be ideal, in fact. One LPF inside would be worth six externally. This kind of work adds a whole new level of complexity, but not out of the question!

So now I am looking for some help with either:
a) Building this for me at a rate I can afford, or
b) Helping me turn this schematic into a plan I can follow and build myself.

In my research, it seems that one can find low-pass filter designs and projects for DIY all over the web. So I must conclude that this is basic stuff. I am quite handy with a soldering gun, I used to work in electronic assembly, and can follow instructions quite well.

I have uploaded the schematic file as an attachment to this post.

Any volunteers, offers or recommendations?

[Roger Shore]

As this is to be implemented at mic level, you will have to take care that the inductors are well screened, so the construction may well be the most difficult aspect of the project.

What specific inductors has your expert recommended for this project?......

[Greg Miller]

As I stated long ago (in the other thread), proper design of a passive filter depends on knowing the source impedance and load impedance. Assuming that your designer got accurate numbers, these filters may work as hoped if they're inserted in your mic inputs (whether they're in an outboard box, or whether you connect them in line with the mic input jacks inside the mixer.

However, this design will not work correctly if you try to install it at some other point in the mixer circuitry (for example in the mix buss) because that is unbalanced, while the filter are balanced, and also because impedances at the mix buss will be significantly different.

A further problem is that if the 25 kHz signal overloads the mic preamps, it can cause all sorts of bad effects including significant intermodulation distortion, clipping, and god knows what else. If this is the case, then inserting one "master" filter further down the chain (e.g. at the mix buss) will not clean up the audio.

As Mr. Shore points out, the inductors need good electrostatic screening, and also good electromagnetic shielding. That, combined with your need for either 24 or 36 identical inductors, will make this project an expensive one. We don't even know whether an appropriate inductor exists in an off-the-shelf part. What mfr. and part number did your designer specify?

Note that this is a lot of circuitry to mount inside the present mixer. You will almost surely need to build it all in an external project box.

Also note that these filters are flat to only 6 or 7 kHz, so you will be losing a bit of the very top end of your audio.

IMHO you would be much better off to either (a.) record in a room that does not have these ultrasonic sensors running, or (b.) just use a microphone that doesn't pick up 25 kHz in the first place.

[Allan Black]

^ Option A. The problems will never end, every time you hear 'something' untoward, you'll be wondering if it's somehow caused by that motion detector.

Cheers.

[Roger Shore]

Or, perhaps option 'C'...?..
The previous thread on this topic has already stated that the requirement to record in a different location is not an option. The second option of trying to find a mic, particularly an electret or conventional condenser capsule, that will guarantee to reject a 25KHz signal of unknown amplitude is at best hit and miss... and probably expensive!

Option 'C' might be to derive an active Butterworth low pass filter - probably 4th order - with a cut off at 7KHz, and insert that into the signal chain at line level (not at the mic input).

The circuitry is simple enough... the attached schematic is directly from a free Texas Instruments filter app, which I can confirm gives some useful results!

That would give an attenuation at 25KHz of some -43dB, which is at least as good as the 3rd order passive schematic already presented, and, using standard off the shelf component values, simpler to make.
It is also much more tolerant of unknown input (and output) impedances than the passive version shown, and will give more predictable results. It would require a small amount of DC power though... hopefully that could be derived from the mixer ?...

A dual opamp circuit constructed around such a design would surely not be beyond the capabilities of many folk in the OP's Californian location?.... Silicon Valley isn't that far away!!......although to keep cost down, you might need to find keen youngster who could knock it up from stripboard, to avoid one off PCB charges.
Unlike the passive version, the line level circuitry would be much more flexible in layout design...

It does not of course address the problem mentioned by Mr. Miller above - that of the ultrasonic content overloading the mic pre-amps. Hopefully, there is not enough energy to actually do that. If there is, then even the passive schematic already posted may not stop the internal capsule pre-amps from possibly overloading. However, if it is THAT extreme, then you probably do need to find a new venue......

[David W. Jones]

Or since you already know where the problem occurs, just take it out in post and move on down the road.

[Ken Reeser]

This is the topic: I am looking for a person, persons or company that can build the device whose schematic is attached. Or a plan or advice on how I can build it myself.

You can assume that the person who designed this took EVERYTHING in my circumstances into account, and that, of course, I am going to review it all with the person who helps me complete this project.

Also note that I did get a quote from a firm in Germany which is just too costly. So I have at least a max price I'm willing to pay.

Also, I am willing to entertain other ideas - about the specific topic of building a low pass filter that will bring down the noise in the 25kHz range sufficiently for my purposes. That is my goal here: A low-pass filter, built for me, built by me, or maybe there's one available commercially that I just have not found.

So any suggestions along this line are great! Can someone reading this do this? Do you know of someone who can do it? Do you know of a better place for me to look for someone who can do this? Or do you have an alternate design or idea for a low-pass filter that will drop off the noise at the 25kHz level that you can build or that you can show me how to build?

I also asked about the possibility of inserting such a LPF into the mixer itself, a Shure M367, whereby I would only have to build one, not one for each channel. This would be a tremendous cost savings. This would, in all likelihood, involve slight modifications or maybe an entirely new design for a LPF. I and the person or persons who want to do this project will undoubtedly cover this in great detail.

The following are off-topic: Where I do my recording; removing or disabling the motion detector; fixing it in post, using different microphones; any other solution to the root problem, especially those that begin with the word "just"; Whether the mics are being "overloaded" by the ultrasonic signal (sheesh); what you think it will cost; what you think of my posting this question here in the first place;... I think I've made the point.

It's not that I am against looking at problems from a new perspective, its just that I've passed the point of diminishing returns on that tack a long time ago. This is the avenue I want to pursue for now. It's not uncommon for people who have nothing positive to add but who still want to hear themselves type to simply reach further and further afield to come up with some alternate universe scenario, mysterious certain dangers to the current trend, etc., etc.. I would like to put an end to that for now. It is why I started this thread.

I have tried a couple commercially available low-pass filters (The PSC ALPF), and did achieve noticeable, positive results, just not enough to solve my problem. This is how I know a low-pass filter will work.

Also, I have tried another mic, just not a lavaliere: an AKG C-1000 S. This microphone does not pick up the ultrasonic signal at all. Please don't even ask why I don't "just" use those mics.

[Ken Reeser]

Roger!

Yours is an excellent post! In fact, I wish you had joined the prior thread with this design configurator thingy before I contacted the guy in Germany. Maybe I wasted his time. After all, my quest all along has been for a LPF that would be cost-effective to build! (or buy.)

So, if I'm reading that graph right: your Butterscotch filter seems to be a very, very close match in results to the design I have posted. Plus I like those words you use like, "... simpler to make... much more tolerant... more predictable." They comfort my poor, unscientific mind.

However, the finding of someone to make it for me is just as elusive as it ever was. Silicon Valley outlawed sidewalk circuit fab vendors ten years ago, so I can't find one by walking down Oakmeade Drive anymore. (It was a blatant market protection law, if you ask me.) Regardless, they are also not found on the internet, Craigslist, nor anywhere else I have looked.

So my original quest stands: Who, oh, who will help me build this? Any positive suggestions welcome. Is there a particular name for a type of business I should be Googling? Or maybe another web forum?

By the way, having no clue as to how to read this or source the parts, how much do you think the parts for this would cost?

[Richard Crowley]

Yes, an internal filter was what I suggested last year (response #31 to the previous thread.)

Have you posted a sample of the audio? Did I miss it?

I would be willing to insert such a filter into a mixer, but ANYBODY attempting such a project really needs a sample of the audio to test whether the solution is effective or not.

Remember also that you can use a notch filter since the interference is a fixed frequency. It would be easy enough to make it "tunable" so you can adjust it for maximum rejection of the interfering frequency. And it would have far less traumatic effect on the desired audio signal.

PS: It is a "ButterWORTH" filter, not a "Butter SCOTCH" filter. But the butterscotch sounds more tasty.

[Roger Shore]

I can see the idea of a tunable notch filter being less intrusive to the full audio spectrum, but it does have the disadvantage of being more complex.
It is also unusual to find motion sensors operating at 25KHz. The vast majority of ultrasonic motion sensors operate at 40 KHz, where there are a large number of cheap transducers for manufactureres to choose from.
A steep low pass filter, with a cutoff around 7KHz, should solve the problem at either frequency.
Although, theoretically, that is quite intrusive into the upper 2 octaves of the audio spectrum, in reality there is little (if any?) useful speech data above 7KHz.

In terms of design, then a line level single channel low pass filter should be within the realms of most electronic hobbyists. Circuit wise, there would only be a few extra components to be added to the Texas schematic I posted earlier. I could post a sketch of something if that would be helpful?
Construction wise?... for a 'one off', then stripboard would be the cheapest solution. (An idea of what stripboard is here , if you're not familiar with it :Stripboard )

Simple project for a young hobbyist..... surely there must be some electronics hobbyists left in California?..:-)

[Roger Shore]

Originally Posted by Ken Reeser

By the way, having no clue as to how to read this or source the parts, how much do you think the parts for this would cost?

Being in the UK, most of my obvious parts sources are not relevant.....My understanding of the US market suggests names like Mouser, Digikey, Radioshack --although I have had no dealings with any of them.

Cost?.. electronic parts...max $20. Housing and connectors?...maybe another $20

Payment to your stripboard designer/ assembler/ tester?...... hmmm.... pick a figure. This is where it gets difficult......

As I say in my other post, the sort of project for a young hobbyist. Here in the UK, they're a disappearing breed. Not sure about California. Posting a request on an electronics forum like this one might help:

All About Circuits Forum

Edit: further thoughts.

For this line level approach to work, you need to insert the low pass filter into the signal chain after the summing amp, but before the limiter and the metering. That almost certainly means inside the M367 itself.
Sadly, Shure don't publish the circuit of the M367 in the user manual (although you may get Shure to email it to you if you ask here: http://shure.custhelp.com/app/answers/detail/a_id/1552 )

They did however publish the circuit of the similar Shure M267 mixer.
To perform this task in that mixer would mean inserting the filter into the signal chain between the summing amplifier output and the master opamp input. Unfortunately the limiter opto isloator LDR is across the summing amplifier feedback resistor. It probably wouldn't make much difference, although it might change the limiter characteristics a bit.
Now, whether the M367 schematic is anything like that at all?....

So you're really limited to two options. Build the passive mic input filters as suggested by your German contact, with the limitations and expense that would bring, or - assuming the M367 is similar to the M267 - find someone who is brave enough - and competent enough - to dive into and modify your mixer.
Probably not a difficult task - although it's never good idea to say that out loud!......

The 3rd idea of introducing an external line level filter into signal chain, without modifying the mixer is probably not possible, unless you are prepared to put it between the mixer and the recorder, and ignore the 'haywire' metering ... although seeing the circuit schematic of the M367 would allow you to confirm that....

[Richard Crowley]

Actually, a notch filter is much easier and simpler than a low-pass filter of any type. It has the added advantage of having almost no effect on the signal of interest (the speech). Furthermore, it can be easily scaled to work at microphone impedances and has much less dependence on source and load impedances.

Here is an online calculator for a twin-T notch filter. (Sample)Twin-T Notch Filter Design Tool - Result -

If you enter the value of 25000 Hz (25KHz) and capacitor value of .0000001 F (0.1uF) here is what it says:

R1 = 91Ω
R2 = 91Ω
R3 = 22Ω
C1 = 0.1uF
C2 = 0.1uF
C3 = 0.1uF

These resistors and capacitors are almost literally a dime a dozen. Even at Radio Shack prices you should be able to do that for < $5

If I had this problem, this notch filter is the first thing I would try. If it doesn't work then the other more complex and expensive options could be considered.

[Roger Shore]

Certainly a passive notch filter is simpler. Your suggestion was for a tunable notch filter, which I assumed would be active, and thus more complex.
Biggest problem with passive notch filters is the accuracy of the components. To maintain a usefully steep notch at the specified frequency you'd need to use at least 1% components throughout - including the caps. It would mean that all these motion sensors really are exactly 25 KHz - and not some at, say, 24.5KHz.

That of course is not a problem with low pass filters. And I still think that speech is not likely to be seriously compromised with a 7KHz cut off, although that clearly might not be acceptable for music.

[Richard Crowley]

Tunable means that one of the resistors is a variable pot. It also makes the absolute accuracy of the components not as critical. As I said, it would be my first choice to solve this problem.
Remember also that the exact frequency is also unknown here. It would be silly to assume that it was exactly 25.000 KHz.

[Roger Shore]

Do you think the 'notch' depth could be maintained at a useful level, with one of the resistors being variable from the ideal (calculated) value, so as to allow the filter to be tuned to match the specific motion sensor frequency?
One problem I have found with notch filters is just how easily the 'notch' can become just a gentle 'dip' once the 'Q' is reduced by changing one component value. And the notch needs to be at least - 40dB to come anywhere near the effectiveness of a 4th order low pass filter

As you say though, worth a try.

It would still mean placing the filter inside the mixer at the point I mentioned earlier though. The mics used here have balanced inputs, so a simple unbalanced passive notch is not going to be suitable for insertion into the mic input line....