Use filters with more stages (and possibly Chebychev) so that the slope of the transition bands will be sharper
Append an inverting amplifier, with R2 (or R1) as a trimmer potentiometer to the outputs of all the filters so that I can match all of their amplifications before they output signal gets to the LED drivers
This way I can sure all the filters have unity amplification at their midpoint frequency's
Add test-points
Prototype each filter on a breadboard, and possibly PCB beforehand
Seems that the issue with the sub-bass filter outputting a 343Hz signal is due to low gain margin which is exploited by the PCB
Should be done for all sub-circuits to be honest
For through hole devices, make sure there is enough room between holes for their footprints so there would be little to no chance of a short circuit between the terminals
Replaced Sub-Bass op-amps with one from the old PCB, still getting the same problem
According to online, this is possibly due to the third stage circuit being unstable to to a small phase margin, causing sustained oscillations
Will get metrics on accuracy of each filter tomorrow, taking how much attenuation exists at the midpoint, cut-off frequencies and 10 times/divide the cutoff frequencies
Will make notes on how improvements can be made (steeper cut-offs, individually controllable gains etc.)
Sub-Bass filter produces a 2V 343Hz output even when there is no signal input, that is why the LEDs are always activated
The Midrange filter only lights up a few of the blue LEDS for the same voltage that lights up all 10 LEDs for the other filters, that means it is attenuating the signal by at least 18dB, or at least 85%
***Found out there the +5V power and -5V power lines were shorted together
Location of the short circuit was at U18, the high precision op-amp in the Rhi-Control circuit that is being used as a voltage follower, removed it from the PCB, -5V regulator output is now -4.9V
Even when disconnected from external circuitry, the op amp power terminals are still shorted together
Soldered a new 5V regulator, now both +5V and -5V are delivered to the op amps
Took U18 from old PCB and soldered it onto the new one, ensured the power terminals were not shorted together
When I turned the device on, the power terminals of U18 shorted again...
Took it off the board
****** I removed both + and - 5V regulators earlier because they were overheating due to the short, and I assumed they were broken, turns out they have internal circuitry that protects them from damage due to shorts... Which I knew that earlier, should have read the data sheets
I have spare 2N2222 NPN transistors, made a voltage follower (CE) circuit out of it
Diode doesn't "turn on" until Vi is >= 0.658V (as modelled in LTspice), therefore max voltage sent to Rhi terminals is 2.842, didn't need the full 3.5V range anyway
The base is connected to pin2 (the output) of the potentiometer, the collector is connected to 3.5V and the Emitter is connected to the the Rhi pins of the LED drivers, which collectively have an equivalent resistance of 2.5kohms
Soldered the voltage range and LED brightness potentiometers to the board, the Voltage range pot is working properly
Soldered LED driver holders to the breadboards, then inserted the drivers as well
Soldered the plug source terminal, Dot/Bar header pins and source switch
Soldered the solo op-amp IC (U18) - this gave me trouble even with reflow soldering, nearly ripped off a pad
Soldered all of the power circuitry, -5V inverter regulator (U8) the 5V regulator (U9)
U9 footprint has terminals REALLY close together
The rocker switch isn't completely flush on the PCB
Noticed in the schematic that I had the polarized 1uF cap (C55) oriented the wrong way, will correct this
ISSUES:
When I powered the circuit on and flipped the switch, I wasn't getting 9Vfrom the drain of the pnp mosfet
I thought that maybe the reverse voltage polarity circuit (the mosfet being between the power source and the circuit) just wasn't working for whatever reason (this was a dumb thought-process) I bypassed it and removed the mosfet
After I bypassed it I measured the output of the -5V inverter and regulator and 5V regulator, I was only getting about +1.1V, turns out the terminals for the 9V and ground of the -5V reg and inverter were short circuited, and both IC's were getting really hot
I removed the -5V reg and inv, going the remove the 5V regulator since it is obviously broken as well
The PCB has in input resistance of [19.4kohm : 6.0kohm] and an input capacitive reactance of [4.2kohm : 499] over the audio range of 20 to 20kHZ
At DC, the input resistance is about 21kohm
The equivalent capacitance is [2.11uF : 6.17nF]
No issues with drawing too much current:
The audio jack was measured to have an output impedance of about 73 ohms << than the PCB input impedance
In worst case scenario, over 98% of vsig will still be maintained
Also, the headphones I have been used mere measured to have 0 ohms (resistance) headphones typically no more than 600 ohms
No issue with signal reflections
One only needs to worry with transmission line effects if the distance the signal will propagate is comparable to the wavelength, using c = 3x10^8, the wavelength of 20khz is 15km
Distance from audio output to furthest frequency filter = 26cm (cable) + 30cm (width of pcb) + 10cm (account not taking straightest path) = 66 cm
Negative instantaneous power i.e. power being delivered from PCB to the audio jack to to phase shift between
Worst case scenario, it will deliver under 100 uW to the audio jack, should be fine
Source - ChatGPT
In typical audio systems and applications, an audio source is not at risk of being damaged solely because it delivers a current out of phase with its voltage. In fact, phase shifts between voltage and current are quite common in audio circuits due to the presence of reactive components like capacitors and inductors in speakers and crossovers
Audio signals fall within the human audible range, typically from 20 Hz to 20 kHz. At these frequencies, the phase shifts introduced by typical audio components are small and generally do not result in damaging conditions
Audio sources, such as amplifiers, are designed to handle a range of loads, including those with phase shifts. They are typically equipped with protection circuits to prevent overcurrent and overheating, which are more common causes of damage
In many audio applications, phase shifts of less than 30 degrees are often negligible. These small phase shifts usually do not noticeably affect sound quality or system behavior.
Many audio devices incorporate protection circuits that prevent overcurrent or overvoltage conditions that could potentially damage the equipment. These circuits can handle variations in power flow due to phase shifts
Made version of each spectrum filter prepended with a voltage follower (buffer), then made a file with buffed filters connected to the same signal, another with all the non-buffed filters connected to the same signal, and finally one with both versions connected to equivalent signals
Then I compared their input impedances (measured between Vsig and ground)
The one with input buffers (green) had an input impedance of [28Mohms : 1.213Gohms]
The one with no input buffers (red) had an input impedance of [6kohms : 20 kohms ]
This was measured over the audio range - 20Hz : 20kHz
Compared the difference of the outputs of each spectrum filter
The differences grew from ~40uV (Sub-bass filter) to ~7.4mV (Presence filter), this is way less than I expected
Ghani Lawal
