Post by thread:[PIC]: Audio Ideas?

Greetings, I had an idea, and I wonder if it would work. Basically, I want to construct an 8 channel audio VU meter (level meter). This would be used on a mic/line level source rather than a speaker level source. I think there are dedicated chips for this, however I wondered if I could do it with a pic. If I used a 16f877, i'd have 8 analogue inputs, and I could use an 8 bit port for an 8 segment LED bar graph, and another 8 bit port to multiplex 8 of those bar graphs. I just am not sure that feeding the audio signal straight into the ADC would work. I know I would have to figure out the voltage range. Supposing this works, would it the be possible to make a small audio spectrum analyzer? Perhaps if I put a filter on each analogue input that only allowed one frequency, or a small frequency range through? Anyways, I was just brainstorming. Is any of this even possible? BTW James did I get the header right? Thanks in advance, Josh Koffman spam_OUTjoshyTakeThisOuTmb.sympatico.ca

Yes, I believe all this could be accomplished. As far as the audio being input directly into the PIC, I would convert it to DC first with a relatively fast filter (short time constant) to smooth the output. This would be better in my opinion than straight audio AC. As fas as filters, there are many many possibilities. But yes, this too should work. Unless you have some very good caps and resistors (or inductors) though, you're not going to be very accurate. Regards, Jim On Fri, 26 May 2000, Josh Koffman wrote: {Quote hidden}> > Greetings, > I had an idea, and I wonder if it would work. Basically, I want to > construct an 8 channel audio VU meter (level meter). This would be used > on a mic/line level source rather than a speaker level source. I think > there are dedicated chips for this, however I wondered if I could do it > with a pic. If I used a 16f877, i'd have 8 analogue inputs, and I could > use an 8 bit port for an 8 segment LED bar graph, and another 8 bit port > to multiplex 8 of those bar graphs. I just am not sure that feeding the > audio signal straight into the ADC would work. I know I would have to > figure out the voltage range. > > Supposing this works, would it the be possible to make a small audio > spectrum analyzer? Perhaps if I put a filter on each analogue input that > only allowed one frequency, or a small frequency range through? > > Anyways, I was just brainstorming. Is any of this even possible? BTW > James did I get the header right? > > Thanks in advance, > Josh Koffman > .....joshyKILLspam@spam@mb.sympatico.ca jimKILLspamjpes.com

This is something I've been meaning to make. But it will still be a month or two before I get to it, so... You have 1 analog input, and run the chip at 20MHz. Use a fast fourier transform to get the level in 10 different frequency bands, and drive a 10x10 matrix of LEDs. One could do an 8x8 or 16x16 or just about any size spectrum analyzer out of such a setup. A 16f876 would have the right number of I/O to do 10x10 with one audio input and an extra i/o pin. It may not be too much to ask for to have it drive two 8x8 displays with two audio inputs. At any rate, there have been several threads on FFT on a PIC in the past, look at http://www.piclist.com/ for the archive and search for FFT. You should probably start out doing a reguler VU meter first, then do a peak-holding VU meter, then decide whether a spectrum analyzer is worth it or feasable. By then the 18Fxxx chips will be common with their 8x8 multiply to speed your fft up quite a bit. -Adam Josh Koffman wrote: {Quote hidden}> > Greetings, > I had an idea, and I wonder if it would work. Basically, I want to > construct an 8 channel audio VU meter (level meter). This would be used > on a mic/line level source rather than a speaker level source. I think > there are dedicated chips for this, however I wondered if I could do it > with a pic. If I used a 16f877, i'd have 8 analogue inputs, and I could > use an 8 bit port for an 8 segment LED bar graph, and another 8 bit port > to multiplex 8 of those bar graphs. I just am not sure that feeding the > audio signal straight into the ADC would work. I know I would have to > figure out the voltage range. > > Supposing this works, would it the be possible to make a small audio > spectrum analyzer? Perhaps if I put a filter on each analogue input that > only allowed one frequency, or a small frequency range through? > > Anyways, I was just brainstorming. Is any of this even possible? BTW > James did I get the header right? > > Thanks in advance, > Josh Koffman > joshyspam_OUTmb.sympatico.ca

Josh Koffman wrote: >Greetings, >I had an idea, and I wonder if it would work. Basically, I want to >construct an 8 channel audio VU meter (level meter). This would be used >on a mic/line level source rather than a speaker level source. I think >there are dedicated chips for this, however I wondered if I could do it >with a pic. If I used a 16f877, i'd have 8 analogue inputs, and I could >use an 8 bit port for an 8 segment LED bar graph, and another 8 bit port >to multiplex 8 of those bar graphs. I just am not sure that feeding the >audio signal straight into the ADC would work. I know I would have to >figure out the voltage range. > >Supposing this works, would it the be possible to make a small audio >spectrum analyzer? Perhaps if I put a filter on each analogue input that >only allowed one frequency, or a small frequency range through? ......... Writing this after reading some of the other responses: - FFT on a PIC might be the worst way to go for this project. Digital bandpass filters might be better - much less computation intensive. Also, check a piclist thread a couple of months back called "Don't use Microchip's FFT". - many past audio systems have used h.w. bandpass filters, and this would cetinaly work here too. - if you really want VU meter functionality, you might need to consider a logarithmic scale - and remember the PIC only has 8-bit A/D which isn't gonna give you much dynamic range.

James Paul wrote: > > Yes, I believe all this could be accomplished. As far as the audio > being input directly into the PIC, I would convert it to DC first > with a relatively fast filter (short time constant) to smooth the > output. This would be better in my opinion than straight audio AC. > As fas as filters, there are many many possibilities. But yes, this > too should work. Unless you have some very good caps and resistors > (or inductors) though, you're not going to be very accurate. > > Regards, > > Jim Jim, Thanks for your response. I am actually most interested in just the VU meter section - the analyzer was just a "what if" question. What kind of filter would I need? And why would this be better than just straight audio AC. Thanks Josh Koffman @spam@joshyKILLspammb.sympatico.ca

"M. Adam Davis" wrote: > > This is something I've been meaning to make. But it will still be a month or > two before I get to it, so... > > You have 1 analog input, and run the chip at 20MHz. Use a fast fourier > transform to get the level in 10 different frequency bands, and drive a 10x10 > matrix of LEDs. One could do an 8x8 or 16x16 or just about any size spectrum > analyzer out of such a setup. A 16f876 would have the right number of I/O to do > 10x10 with one audio input and an extra i/o pin. It may not be too much to ask > for to have it drive two 8x8 displays with two audio inputs. Something like that (2 inputs, 8x8 display) sounds really interesting. I imagine the coding for this would be pretty hard though. This is probably a bit out of my league. > At any rate, there have been several threads on FFT on a PIC in the past, look > at http://www.piclist.com/ for the archive and search for FFT. > > You should probably start out doing a reguler VU meter first, then do a > peak-holding VU meter, then decide whether a spectrum analyzer is worth it or > feasable. By then the 18Fxxx chips will be common with their 8x8 multiply to > speed your fft up quite a bit. Right now all I really want is the VU meter. Do you think it is even feasible to do an 8 input, 8 bar graph out VU meter in a 16f877? Do you have any pointers as to how to even begin? I checked the archive, and I still am not sure how to process the input on adc to provide a proper output on the display. If I can only get one output per pic, it would probably be easier to use a dedicated linear chip. I only want to use the pic for the coolness factor, plus a reduced parts count, and eventually I can interface it with a computer. Thanks, Josh Koffman KILLspamjoshyKILLspammb.sympatico.ca

This sounds like a neat project. If you want to do the Vu meter thing, you might consider the following design: Take each analog input, run it thru an op-amp to buffer it (so you don't affect the signal in its normal use). Take the output of the op-amp and half-wave rectify it with a diode (actually, I think you have to include the diode in the op-amp's feedback somehow so that you don't have the diode drop affecting your output). Run the resulting DC signal thru a resistor to a cap. This forms a simple low-pass filter. Connect the cap to an A/D input on the PIC. In your PIC code, periodically reprogram the A/D input to be an output and drive it low. This will discharge the CAP. Then, go back to an A/D input, give the signal time to charge the cap up (this time interval should be very consistent), then read the voltage on the cap. Since you want a logarithmic display, you could actually just use each bit of the converted result to drive one of the LED's (each LED would then represent a factor of 2 (I think that is 3 dB!). Have fun. Bob Ammerman RAm Systems (high function, high performance, low level software)

Josh Koffman wrote: {Quote hidden}> > "M. Adam Davis" wrote: > > > > This is something I've been meaning to make. But it will still be a month or > > two before I get to it, so... > > > > You have 1 analog input, and run the chip at 20MHz. Use a fast fourier > > transform to get the level in 10 different frequency bands, and drive a 10x10 > > matrix of LEDs. One could do an 8x8 or 16x16 or just about any size spectrum > > analyzer out of such a setup. A 16f876 would have the right number of I/O to do > > 10x10 with one audio input and an extra i/o pin. It may not be too much to ask > > for to have it drive two 8x8 displays with two audio inputs. > > Something like that (2 inputs, 8x8 display) sounds really interesting. I > imagine the coding for this would be pretty hard though. This is > probably a bit out of my league. > > > At any rate, there have been several threads on FFT on a PIC in the past, look > > at http://www.piclist.com/ for the archive and search for FFT. > > > > You should probably start out doing a reguler VU meter first, then do a > > peak-holding VU meter, then decide whether a spectrum analyzer is worth it or > > feasable. By then the 18Fxxx chips will be common with their 8x8 multiply to > > speed your fft up quite a bit. > > Right now all I really want is the VU meter. Do you think it is even > feasible to do an 8 input, 8 bar graph out VU meter in a 16f877? Do you > have any pointers as to how to even begin? I checked the archive, and I > still am not sure how to process the input on adc to provide a proper > output on the display. If I can only get one output per pic, it would > probably be easier to use a dedicated linear chip. I only want to use > the pic for the coolness factor, plus a reduced parts count, and > eventually I can interface it with a computer. > > Thanks, > Josh Koffman > RemoveMEjoshyTakeThisOuTmb.sympatico.ca How about a compromise on the coolness factor? Use a PIC (or it may need the higher speed of an SX) to generate the clocks to several switched capacitor filters. Just a thought :-) -- Rich

Correction on the last formula: Voltage ratio: 20 * log_sub_base10 (V1/V2) = dB ----- Original Message ----- From: Jim P <spamBeGonejvpollspamBeGonedallas.net> To: <TakeThisOuTPICLISTEraseMEspam_OUTMITVMA.MIT.EDU> Sent: Saturday, May 27, 2000 1:36 PM Subject: Re: [PIC]: Audio Ideas? Oops! Did anyone else catch this? "of the converted result to drive one of the LED's (each LED would then represent a factor of 2 (I think that is 3 dB!)." - S/B "6" dB (Not meaning to beat a dead horse or anything - but here's a quick review of the dB anyway ...) Voltage factor of 2 = 6 dB Voltage factor of 1.414 = 3 dB Power ratio of 2 = 3 dB Power ratio of 4 = 6 dB Computing decibels: Power ratio: 10 * log_sub_base10 (P1/P2) = dB Voltage ratio: 20 * log_sub_base10 (P1/P2) = dB

... sawright ... that's what we're here for! Jim ----- Original Message ----- From: Bob Ammerman <RemoveMERAMMERMANTakeThisOuTPRODIGY.NET> To: <PICLISTEraseME.....MITVMA.MIT.EDU> Sent: Saturday, May 27, 2000 1:51 PM Subject: Re: [PIC]: Audio Ideas? Right you are of course. (What can I say, I'm a programmer Jim, not an Electrical Engineer!). I should have known better tho'. Bob Ammerman (high function, high performance, low level software) ----- Original Message ----- From: Jim P <EraseMEjvpollDALLAS.NET> To: <RemoveMEPICLISTEraseMEEraseMEMITVMA.MIT.EDU> Sent: Saturday, May 27, 2000 2:36 PM Subject: Re: [PIC]: Audio Ideas? {Quote hidden}> Oops! Did anyone else catch this? > > "of the converted result to drive one of the LED's (each > LED would then represent a factor of 2 (I think that is 3 dB!)." > > - S/B "6" dB > > (Not meaning to beat a dead horse or anything - but here's > a quick review of the dB anyway ...) > > Voltage factor of 2 = 6 dB > Voltage factor of 1.414 = 3 dB > > Power ratio of 2 = 3 dB > Power ratio of 4 = 6 dB > > Computing decibels: > Power ratio: 10 * log_sub_base10 (P1/P2) = dB > Voltage ratio: 20 * log_sub_base10 (P1/P2) = dB > > > ----- Original Message ----- > From: Bob Ammerman <RemoveMERAMMERMANspam_OUTKILLspamPRODIGY.NET> > To: <RemoveMEPICLISTTakeThisOuTspamMITVMA.MIT.EDU> > Sent: Saturday, May 27, 2000 6:05 AM > Subject: Re: [PIC]: Audio Ideas? > > > This sounds like a neat project. > > If you want to do the Vu meter thing, you might consider the following > design: > > Take each analog input, run it thru an op-amp to buffer it (so you don't > affect the signal in its normal use). > > Take the output of the op-amp and half-wave rectify it with a diode > (actually, I think you have to include the diode in the op-amp's feedback > somehow so that you don't have the diode drop affecting your output). > > Run the resulting DC signal thru a resistor to a cap. This forms a simple > low-pass filter. > > Connect the cap to an A/D input on the PIC. > > In your PIC code, periodically reprogram the A/D input to be an output and > drive it low. This will discharge the CAP. > > Then, go back to an A/D input, give the signal time to charge the cap up > (this time interval should be very consistent), then read the voltage on the {Quote hidden}> cap. > > Since you want a logarithmic display, you could actually just use each bit > of the converted result to drive one of the LED's (each LED would then > represent a factor of 2 (I think that is 3 dB!). > > Have fun. > > Bob Ammerman > RAm Systems > (high function, high performance, low level software)