For the past year or so, I've been corresponding with Electrical Engineer Victor Friedman who has taken the barkstop CD concept to the next level and beyond. I was very impressed at what he has been doing and he has graciously given his permission to share his photos and ideas. For those interested, I have pasted some of our e-mail conversations in case somebody might be able to get some insight from them. He has also offered his e-mail address (vic and tj [geode@verizon.net]) in case anyone wishes to contact him directly regarding his projects.

----- Original Message -----

From: Michael Atwood

To: geode@verizon.net

Sent: Friday, May 01, 2009 5:01 PM

Subject: Re: FW: anti bark

 

Hey Vic,

I am so sorry for not responding sooner. I've been quite busy with Cirque and as you probably know, Barkstop is certainly not my main gig. It's something that kinda runs on auto-pilot. I sell on average about 2 or 3 discs a month so I guess it covers the cost of hosting the website but that's about it. I guess it's more about sharing what I've learned and maybe helping others who have or had the same problem.

So with that aside, thanks for the photos. I have great respect for your resourcefulness in a way that only a geek could understand.  I got 4 pictures but were there supposed to be more attachments with spectral analysis plots? I'm not sure I got them.

One area that still confuses me is the manner in which you are doing your analysis. What I'm trying to figure out is, are you analyzing what is actually being propagated through the air or you measuring a wav file, internally, that may have some high frequency artifacts such as harmonic distortion that appears to be valid frequency content in the high range.

Now first let me apologize in advanced if I'm explaining things you already know. I certainly don't want to sound like I'm talking down to anyone. To clarify some things, we know that in order for a CD to be "Red Book" standard, i.e., playable in a standard CD player, it must be sampled at 44.1k. Based on the Nyquist theorem, this puts the highest frequency capable of being reproduced around 22k (half the sampling rate) Since we have to filter out the Nyquist frequency to prevent aliasing, a brickwall filter is put on the output of the CD player to block this. But since there is no such thing as a perfect brickwall filter, the slope must start before the nyquist freq., say somewhere are 19 or 20k. That's why the CD player won't work. Even if there was no brickwall filter in place, you'd still be limited to a max reproducable freq. of around 22k, and it would probably sound very nasty due to aliasing.

So then we go to the Conexant HD sound card (which ironically, I bought stock in a few years ago that tanked. I hope I'm a better sound guy than stock trader.). I haven't been able to find specs on it but what I'm curious to know is; is the Digital to Analog conversion taking place at 96k? I'm somewhat familiar with high end converters such as Apogee, Prism, etc., And I've always thought that to get 96k conversion, it's usually quite expensive. So, if the sound card is converting D to A at 96k giving us a theoretical max freq of 48k, going to a car amp with a max freq range of 30k, going to piezo transducers capable reproducing that 30k, the next question is; what is the freq response of the measurement mic? Most microphones used in the music industry are limited in freq response to a top end of about 17 to 20k. In order to measure the freq response, you'd need a specialized measurement mic such as that offered by companies like Earthworks or DPA, etc. Without it, it's really impossible to do a spectral analysis of what the array is really doing.

Then, of course, we'd have to do the reverse process where we'd have to make sure that the conexant card is really doing a 96k Analog to Digital conversion, as opposed to some fancy sample rate conversion. Let me clarify this. If you were to take a sound file that was recorded at 44.1k (Say, from a CD), you could use a sample rate conversion program and convert it to a 48K or even 96k file, but it would not have the high end content.

So, after all of this, I guess I'm just saying that I am not sure that a consumer grade, laptop soundcard is capable of reproducing extended freq content. But what you got is obviously working I wonder if the dogs are responding more to the sheer power of what you are throwing at them and not necessarily any freq content about 20k or so. I hope I'm making sense. I'm gonna try to find more info on that soundcard to verify.

As far as manufacturing and marketing such a product; I unfortunately have no experience in that field either.

By the way, on my website someone did comment on your post. Check it out if you get a chance. I've also added notification capability to the site so you can elect to get an automatic email if somewhat comments or when there are changes to the site. Just sign in, go to your account settings and go to the notifications tab if you want to set that up. Also, if you decide against trying to market what you've got and want to help I'd love to post your ideas and pictures on the site. Of course I won't do that without your permission.

Take Care,
Mike

On Wed, Apr 22, 2009 at 9:12 AM, Michael Atwood <info@ihatebarkingdogs.com> wrote:

 

 

---

From: Victor and TJ Friedmann [mailto:geode@verizon.net]
Sent: Saturday, March 28, 2009 2:03 PM
To: info@ihatebarkingdogs.com
Subject: anit bark

 

Hi Mike,

 

Well, looks like we have a mission in common--a quiet backyard. Here are the answers to your questions:

1. The sound card. My control panel on the lap top says that I have a Conextant HD soundcard. I don't know what that means and I am not sure if this really answers your question.

2. Amplifier response. Sorry I confused you here. Yes, I used a car amplifier advertised to be good to 30khz because it was cheaper than a rack type amp, although it does require 12Vdc. Since my sampling rate was 96khz, I could only do filter analysis to 43khz. The amp. response obviously decreased after 30khz. I purchased the amp. I enclosed photographs of the amp,, 12vdc power supply, whole setup and speakers.

3. I did not do SPL measurements since I do not have the equpment; but my wife who has excellent hearing and the dog, who can hear even better, both attest that the sound carried at least 200 ft. Yes, increasing the frequency to 15khz or higher would make the sound less noticeable to humans. I found, though, that different dogs react to different frequencies (as you noticded) so my approach was to provide a range of frequencies. I might add that in my case this is not a training device because I cannot stand around all day waiting for the dog to bark. I use it only when my wife and I like to enjoy a quiet outdoors. Then, barking or not I activate the device for a few seonds, the dog hides in their garage, and stays there for a long time.

4. I really did not know about the brickwall filter until now. Apparently this filter is programmed in the CD player itself because no matter what sampling frequency I used, I always got the same max freq. out of the amp. So, that is when I thought about using my laptop audio output as an input to the high freq. amp. sure enough, I got a higher bandwidth. So, I concluded that the brickwall filter must be in the CD player itself or in the software that creates the audio CD--in my case Nero software.

I wired the speakers in parallel, six for the L and R outputs, for a total of 12. I thought about using a series resistor to protect the amplifier, but it turned out I did not need it. I use the amplifier at about 1/3 of maximum amplitude.

The spectral analysis. Here is the way I interpret the analysis; let me know if you agree with my conclusions.

Top Left. This is the specdtral analysis using Audacity freeware. The software allows different sampling rates (in my case 96khz), generation of signals and analysis. Bottom left is a short plot of the signal generated by Audacity.

Top right. I recorded the output played brom the CD player through the tweeters. Recording equip. was my latptop using audacity recording software set at 96khz. Note that the spectrum repeats itself every 18khz--18khz must the the brickwall filter in this case.

Bottom left. This is the spectral analysis of the same signal played from my laptop into the same amp. and speakers, recorded with Audacity software on my laptop using 96khz sampling rate. Note two things. First, there is no aliasing; the spectrum does not repeat itself. Two, there is a marked drop-off beyond 30khz--the amplifier's cut-off frequency.

My conclusion is that the PC soundcard must be putting out the full bandwidth as opposed to the CD which is bandlimited to 18khz. What do you think?

I think that we could design a better device than what is on the market right now. I am not a circuit designer but I would think that it must be possible to input a variety of signals in a programmable ROM, a high freq. amplifier, and at least four tweeters, all enclosed and house voltage operated because it would have a long range.

Any ideas?

Let's keep in touch

Best,

Vic