Music Spectrum and dB Visualizer

• Controlling Rhi:
  • The resistance between Rhi and Rlo is suppose to typically be 22kohm according to page 15 and 28kohm according to page 3
  • According to page 4, min is supposed to be 16 kohm, max 36 kohm
  • Measuring my LED drivers my self, they were all between 17.4 -17.93 kohm, average being 17.5 kohm, therefore the approximate equivalent resistance would be about 2.5 kohm
  • Will drive it with a voltage divider, where R2 is log potentiometer, the will be succeeded with a voltage buffer
  • Will need to use a high precision voltage buffer with an input offset voltage of 10 uV - NCS333ASQ3T2G
  • ***Simulated with LTC1250, it also has a max input offset voltage of 10 uV
    
    • *** LM324 has a max possible input offset voltage of |7mV|, if it is possible, then that limits lower bound of Rhi
• Input Offset Voltage in filters:
  • LM324 has input offset voltage in max absolute value of 7mV not good for filtering signals that are going to be displayed in a dB meter -> solution append high pass RC filter to output of active filters, the resistor will ground input when there is 0V input signal
  • when tested in hardware, there was a -0.1mV DC voltage across resistor
  • As a test, I replaced the LM324 op amps in the brilliance filter with LTC1250, output votlage when Vi=0 was never greater than |0.9fV|, whereas when LM324 was used Vo was -1.84mV

• Constructed and tested Circuit on a breadboard
  • https://www.eeweb.com/dual-supply-from-single-battery-source/
    
    • Input voltage to MC79L00 -5V regulator needs to be -7V or below, however when the circuit is loaded with 1 kohm, the voltage goes up to just under -7V, meaning that it us supplying 7mA, i.e circuit can't meet voltage and current needs
  • https://www.electronics-tutorials.ws/waveforms/555-circuits-part-2.html: The Negative Voltage 2X
    • When I physically tested the circuit under no load the voltage wasn't actually double, it's just a negative of the input voltage, 
    • ^confirmed that circuit simply creates negative of the input voltage, doesn't double
  • https://www.electronics-tutorials.ws/waveforms/555-circuits-part-2.html: Basic Multivibrator Circuit
    
    • Used the Same Ra, Rb and C values that were used on this webpage
    • At no load, Vo = -13.37; 1kohm, Vo = -11.14; 470ohm, Vo = -8.18, more than meets voltage and current needs
    • ****It seems that and Ra, Rb pair that gives low duty cycles allows for a more stable output voltage, since D = (Ra + Rb)/(Ra+2Rb), limit of Duty cycle is 50%
  • Used caps of 200uF in the multiplier circuit that can handle 10V, actual values were [120u:160u], switched out with caps of same value that can handle 50V, no difference in output voltage stability when loaded, though it was found that using larger caps reduced "triangle noise"
• Will use Ra = 1k, Rb = 10k, C = 50n
• Simulate 555 timer multiplier circuit in LTSpice
  • LTSpice has a 555 timer element called NE555, "NE" must be typed or won't get to the part
  • At no load, Vo = -16.25; 1kohm, Vo = -13.70; 470ohm, Vo = -12.20
  • Was found that for the first cap in the multiplier circuit, there are spikes of 1.5mA that last for less than 1u
    • max current that 555 timer can source or sink is 200mA
    • on my multimeter, amplitude of the ac current is 41.64mA, possible that it can't register the spikes
    • 555 timer being capped to 200mA may be why Vo in physical implementation is lower in absolute value than the simulation
      • Theory doesn't completely hold up when circuit isn't loaded, but this is good enough for now

• Got Case code and datasheet link for all resistors and caps
• using SOIC-14 footprint for LM324 op amp
• Drafted all filters onto KiCad schematic with correct footprints for all components
• Audio Input -  will use mono audio input, male & female connectors to get the audio signal, will use a switch to choose between the 2 sources

• Set out to test to test RC filtering on actual circuit, with a focus on using is for Sub-bass frequencies as for frequencies below 45 Hz it is flickering is clearly visible, however,
• When signal pin is not connected to anything (i.e. Is floating), Vsig = 0.13V
  • Same behaviour when the signal pin isolated from audio input by a diode
• When signal pin is connected to a grounded resistor, value of Vsig is none zero and is dependent on the resistor value
  • Hard to measure voltage of Vsig in this case because it is so low that my oscilloscope can't accurately measure it (resolution of my oscilloscope is 1mV, and the noise of the measurement exceeds that), and neither could my multimeter as when I applied leads it actually increased the voltage across the resistor (I know this because none of the LEDs would light up when the resistor I connected between Vsig and ground was 2.2kohm and Rhi was set to 300-450mV, then when I applied the leads at least one of the LED's would light up
  • When R = 100 kohm, two LED's would light up, Vsig was 0.013V
  • I set Rhi to  8.8mV (Using a voltage divider of 100 kohm and 100 ohm), connected Vsig to a grounded 1K resistor, voltage at Vsig = 0.1mV, ad yet all of the LED's were lighting up...
  • There may be some parasitic voltages or something going on, so I'm just LM3914 vs LM3915 vs LM3916 - What are the differences? was able to successfully use a diode and RC filter at signal input

• Don't have time to test this out, so I'm just gonna connect the output of all the filters directly to the Vsig of the IC's

• Sub-bass is more felt that heard, and when songs get mastered everything below 30Hz isn't really audible and song's rarely have stuff that goes below 40 Hz https://gearspace.com/board/mastering-forum/629542-low-frequency-response-rap-pop-rock-music.html https://www.reddit.com/r/hiphop101/comments/1oupjz/songs_with_bass_in_very_low_frequency_20_40_hz/

• Nominal value for pin 7 (Ref) is ~1.2V relative to pin 8 (Adj)
• The LM3915 model in LTSpice significantly breaks down when the current from pin 7 is greater than 2mA
• In the region of interest, current from pin 7 =[0:2mA] resulting in an LED current of [0:20mA], pin 8 can be shorted to ground to enable desired operation
• Iref = 1.1928/Resistance
• Final Design:
  • LED's are rated for 20 mA, will cap the max current at 17-18mA
  • Pin 7 of each LM3915 IC will be connected to a 680 ohm resistor rated at 0.5W, all of the resistors will be connected to an audio (logarithmic) 10k potentiometer.

Want capacitors of tolerance <= 5% for lowest possible cost

• Polymer caps have poor tolerance - lowest is 10%
• Niobium oxide capacitors only available for values above 2.2uF
• Mica capacitors are over-engineered, voltage rating of 500V, ~$80.00 for 47nF
• MLCC (Multi-layer ceramic capacitors) a bit too expensive, 47nF @ 5% tolerance is ~$1.00 at lowest end
• Tantalum Capacitors - lowest tolerance is 10%
• Will use film capacitors at 5%

Finished sourcing all capacitors and resistors for filter sub-circuits, total cost of components will be $28.39

• Attempted to match resistance in each part of the Sub-Bass Filter perfectly:
  • Number of resistors in filter: 9
  • Number of resistors needed for perfect value matching: 15
• Difficult to use exact resistance values because:
  • Exact resistor value doesn't exist, needs to be realized by a series and/or parallel combination of resistors
  • Resistor value too expensive:
    • costs more than $0.20 for one resistor
    • requires a minimum order of say 200 units
• More resistors -> Larger PCB -> more expensive product
• Ultimately, this is a personal project, a prototype, want to keep the cost down as much as possible, this is NOT a scientific measuring device, output of filters need just need to be within a certain threshold, care more about the dB visualization
• (Obviously) percentage difference between ideal and implemented resistor values matters more than absolute difference
• Differential between ideal and implemented values seems to balance itself out once all resistors have been selected
• Finished sourcing resistors for all filters

• All filters now using LM324 op-amp
• Using the LM324 op amp causes the filters passband to not be completely flat, there is a slop, most extreme for the Presence Filter
• Started sourcing the passive components - Resistors and capacitors
  • Seems that the site I was using the build the active Butterworth filters https://tools.analog.com/en/filterwizard/ takes into account what the discrete value of passive components are
    • All of the capacitor values seem to be standard
  • Using Metal Film Resistors that have tolerances of %1, as every percentage a passive component is off from its nominal value "It turns out that the total variation of the filter frequency then in percent is the sum of the tolerances of the components."  however, that is for RC filters, could be different for more advanced filter types
    • https://www.quora.com/For-an-RC-filter-taking-analog-signals-audio-what-is-the-effect-of-choosing-a-5-tolerance-resistor-versus-a-1-tolerance-resistor
  • Seems like it is standard for capacitors to have 20% tolerance... If the above is true that could be troubling
  • Will source from Mouser as shipping is free, Digi-key has an additional shipping costs of $20 for anything under $100

• Started getting this error when I tried to push my code a few days ago:
• error: RPC failed; curl 18 transfer closed with outstanding read data remaining
• So today I increased the buffer size
  • remote: error: File Misc/Villard_Cascade_Negtive_Voltage_Multiplier.raw is 207.98 MB; this exceeds GitHub's file size limit of 100.00 MB
• Seems that raw files just sky-rocket in size therefore...
• I copied the "Spectrum Filters"  and "Misc" folders to another location, did a hard reset to the last commit that was pushed, replaced the "Spectrum Filters" and "Misc" folders with the ones that were copied earlier, deleted .raw files that were over 1000kb, committed those changes and successfully pushed to the remote repository 

• Attempted dual voltage regulator design by simulating using 2 LT1521-5's (LTSpice doesn't have MC79L05), Did not work at all
• Then I looked up dual-rail voltage regulator IC's (very few are through hole), they can supply positive and negative voltage ranges **cost about $10 for one chip
• Max600: provides +/-10V output form a 5V input, max current it can output is 10mA, Vo = 2Vi
• LT1026: more or less identical operation to MAX600, simulation results are correct, can supply over 10mA, unfortunately LTSpice takes a long time to Simulate that IC, **cost about $10 for one chip
• *****Solved the problem:
  • https://www.eeweb.com/dual-supply-from-single-battery-source/
  • https://electromagneticworld.blogspot.com/2017/05/cascade-circuit-voltage-multiplier.html
  • Use MC79L05 to generate 5V dc
  • Connect 9V source to 555 timer as a free running oscillator (astable mode) then connect output to Villard Cascade (Negative) Voltage multiplier, connect output of that circuit to MC78L05 to generate stable -5Vdc
  • 555 timer can output up to 200mA, so we are in the clear