iampoor1 said:
Maybe you will have some thoughts on this John....I have been contemplating what makes different state variable filter designs sound different. Part of me thinks that manufacturers may have slightly different potentiometer tapers, thus the control ranges on some eqs may "feel" better. I am also wondering if part of it comes down to ringing/pcb layout. I wonder if certain opamp types are more unstable being used as integrators. Any other ideas, outside of well, the circuits being different on paper? I know these could be completely off base, just trying to think slightly outside of the box. 8) ;D
In theory with full parameter control of center frequency and Q, they "should" sound the same, but as with everything the details matter.
#1- The SVF describes the BP filter topology but how that gets added to or subtracted from the full band audio signal makes a difference in control range and performance. The simple add or subtract at unity can deliver infinite cut, but only +6dB of boost, but with very low noise gain (quiet EQs are nice). The symmetrical boost/cut topology involves higher noise gain for large amounts of boost/cut but delivers more conventional control range.
#2- Q/Bandwidth adjustment. There are a number of ways to vary the Q in a SVF and I've seen most of them over the years. One of the slickest that is only available using DPOTs for the frequency control is to stagger or spread the poles apart for wider bandwidth, pretty much impossible to do with conventional pots. The Q can also be varied by the bandpass gain (I've seen this done with a second dual pot section used to scrub of the extra unwanted gain). Q can also be varied by R to ground increasing the noise gain of the first section (HF out).
#3- a subtle third concern is headroom... Different SVF implementations can use other than unity gain for different HP/BP/LP sections. The BP will always be scaled for unity, but a HP or LP with more than unity gain will clip before that BP section. How this matters in use, depends on how much extra gain it has, and the signals it sees. An unused HP or LP section clipping should not affect the BP output, but it always does so this can be a source of unexpected spurious distortion.
#4- frequency control. Besides the DPOTs I mentioned, I have seen this done a number of different ways. The name "potentiometer" reflects their strength as voltage (potential) dividers. They are not called dual rheostats as you would expect for variable resistances. Pots are best at delivering ratios of the input signals they see. The bulk resistance of pots can be 20% so their absolute value is not great, and adding end limit resistor to set their LF range will not be very repeatable. The most accurate way I have found to implement these to vary resistance is to connect the bottom of the pot to ground so when set for LF it is out of the circuit, then bypass with a large value resistor to set LF. This way you get 5% (or 1%) accuracy for both HF and LF endpoints, with decent tracking in between. IIRC I cap coupled the pots but DC coupled the LF resistance. The caps need to be large to prevent some odd LF instabilities.
#5- tricks, most popular implementations vary the Q with a variable resistance from either the + or - input of the HF output section (ignoring that one oddball that varied Q with BP gain), working again a fixed resistance in the other input. I found by spanning across both the + and - input with a pot using a resistor from wiper to ground delivers a wider range of Q adjustment with lower noise gain. For a trick that is perhaps too tricky for general use, in one parametric design I did, I spanned the Q pot with one end adding the BP feedback R to vary the BP gain, and the other end going to ground like typical Q pots. This gave me the unique intentional interaction between boost/cut (actually BP gain) and Q. This EQ was designed for Hifi use on wideband material, so intentionally dropped the amount of boost/cut as bandwidth got wider, and increased it for narrow Q. This gave a nice first order loudness correction when EQing full range source material.
JR
PS: The parametric kit I published back in the 70s used the trick boost/cut interaction, but not all the several tricks I mentioned that I picked up in later designs. In fact my professional EQ designs did not have any interaction between Q and boost/cut as you would expect for studio use on individual instruments.
