• Spectrum Filter Offset and Current

    Ghani Lawal5 days ago 0 comments

    • It was found than when a pull-down resistor is connect to pin 5 (signal input pin) of the LM3915 IC, the port would output ~0.14uA = 140nA, therefore limiting the value of the resistor to 1kohm so that a voltage of no higher than 0.14mV DC would develop
    • Therefore, for the high-pass RC filter to be appended to spectrum filters, R = 1 kohm
      • Sub-Bass and Bass: C = 100uF  
      • Low-Mid and Mid: C = 10uF 
      • Upper-Mid, Presence and Brilliance: C = 1uF
    • The capacitors I have at home that have a capacitance of greater than or equal to 10uF are polarized, but ceramic capacitors exists for up to 100uF, maybe even higher, but they get more expensive
      • 1u: $0.11 to $0.138 
      • 10u: $0.138 to $0.152 
      • 100u: $0.607 to $0.713
    • Calculated max current each filter and internal stage will output when amplitude of input voltage signal is 5V (output at final stage will be saturated to 3.5V):
    • Say signal was completely in the Brilliance spectrum, max outputted current would be 5.345mA.  The 3.6mA output of stage 3 is completely due to the 1k ohm resistor of the high-pass RC filter.  Assuming each op amp IC which contains 4 op-amps requires 1.5 mA, which there are 6 of -> 6*1.5 = 9 mA
      • 9mA + 5.345mA= 14.345 mA.  Negative Voltage multiplier can output a current of 17mA and still stay below 8V, so we are in the clear

  • Controlling Rhi and Input Offset Voltage

    Ghani Lawal06/27/2022 at 22:56 0 comments

    • 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

  • 555 Timer Negative Voltage Muliplier

    Ghani Lawal06/22/2022 at 23:01 0 comments

    • Constructed and tested Circuit on a breadboard
      • 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

  • Signal Input and KiCad Update

    Ghani Lawal06/09/2022 at 03:20 0 comments

    • 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

  • Decided to connect output of Filters directly to signal pin

    Ghani Lawal06/02/2022 at 18:24 0 comments

    • 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

  • Controlling Current Through LED's

    Ghani Lawal06/01/2022 at 04:26 0 comments

    • 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.

  • Sourcing Capacitors

    Ghani Lawal05/25/2022 at 00:56 0 comments

    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

  • Resistor Sourcing

    Ghani Lawal05/22/2022 at 15:10 0 comments

    • 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

  • Spectrum Filter Components

    Ghani Lawal05/16/2022 at 01:36 0 comments

    • 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

  • Git Push Problem

    Ghani Lawal05/16/2022 at 01:23 0 comments

    • 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