Maximizing headroom and lowering current draw in a Graphic EQ

[Bill Mountain]

I built an EQ pedal based on the old MXR 6 band Graphic EQ's but using custom values from other classic EQ pedals.  I hesitate to actually refer to it as a Graphic EQ because I used rotary controls but its a gyrator based EQ.  I used this PCB: http://1776effects.com/wp-content/uploads/2015/03/Six-Shooter-3-19-15.pdf

I run it at 24 volts with my power supply but I want to be able to run it on a battery with a voltage inverter chip providing roughly the same voltage.  I've tried to run it at lower voltages but it's just not as clean sounding with my bass guitar.

My concern is the current draw when running 8 opamp stages at high voltage with an inverter.  My assumption is that if I'm tripling the voltage then I'm tripling the current draw for the circuit.  I'm currently using NE5532's for each stage and running all stages (buffer, gyrators, amplifier) at 24V.  I was thinking that maybe for the gyrators I could get away with using low current opamps at lower votages and possibly running only the final amplifier stage (the on the gyrators are tied to) at higher voltage with an higher quality audio opamp (like the NE5532 or something similar).

I don't understand gyrators enough to understand what is happening electrically and if having lower voltage and less "high end" opamps will make a difference.  I understand that the opamp is simply a buffer stage but does it need to be able to cleanly carry a signal to function properly?

I also understand that any of the design changes I'm referring to would require a new pcb so this is mainly theoretical discussion to help with future designs.

I would also appreciate any thoughts on minimizing noise when working with so many active stages.

[Transmogrifox]

I don't understand gyrators enough to understand what is happening electrically and if having lower voltage and less "high end" opamps will make a difference.  I understand that the opamp is simply a buffer stage but does it need to be able to cleanly carry a signal to function properly?

The positive feedback around a reactive network (R's AND C's) makes the assumption of unity gain go out the window.  The gyrator simulates a grounded inductor...except when it goes to the rails and the op amp is outside of its linear operating range.  A high-Q RLC resonant network can have amplitudes much greater than what is being fed into the network. 

Only with a gyrator it can only ring as high as the supply rails allow the op amp to go.  At that point it just reduces to an RC network and the resulting distortion is worse than clipping.  Changing to an op amp that does not cause phase reversal due to inputs exceeding input range might greatly improve sound quality -- not because it doesn't clip but because it doesn't turn so nasty when it goes to the rails.  You may not hear as much degradation due to a small amount of clipping on a transient as you hear when there is a phase reversal during time signal is out of bounds.  More headroom is the best solution if you want good solid bass.

Here's a thought experiment.  I'm working from the assumption most op amps don't change current consumption significantly with input voltage, but NE5532 datasheet only states current only at +/-15V.  I assume it's approximately the same from +/-5V to +/-15V.
P=V*I
@9V
P=9V*8mA*8 op amps/2 op amps per package = 288 mW
@24V
P = 24V*8mA*8 op amps/2 op amps per package = 768 mW
24V power translated to a 9V supply at 100% efficiency ==> 750mW/9V = 83 mA (compare to 36 mA when running at 9V supply)

Because of the wide range (max current) it could be as much as double, ~1.5 Watts when running at 24V so 9V supply now needs to be able to output this much to be safe:
I = P/V = 1.5/24 = 62.5 mA @ 24V.

Now a charge pump usually can't be expected to have better efficiency than what a linear regulator would have, which is roughly Vin/Vout = 9/24 = 37.5%. 

Now take worst case NE5532 op amp approximated above:
(Pout/Eff)/Vin = (1.5/37.5%)/9 = 444 mA

Now try an inductor-based boost converter at 80% efficiency:
(Pout/Eff)/Vin = (1.5/80%)/9 = 208 mA <--This is how much it should be designed to provide, but typical will be about 1/2 of this.

You don't need to change the PCB, to lower the power consumption and increase the headroom:
A)  Biggest improvement is choose a low-power op amp:  You'll notice TL072 is far better for power consumption if you compare the datasheets (like 1.5 mA).
...or if you don't mind paying $7.00/Dual opamp http://www.digikey.com/product-detail/en/LT1492CN8%23PBF/LT1492CN8%23PBF-ND/891781
B)  If you want to operate from a standard 9V pedal power source and you are concerned about power consumption, don't use a charge pump.  Use the 555 based SMPS or a purpose-built boost controller IC and an inductor.
C)  Final consideration is to experiment with a variable voltage supply to find out how much headroom you actually need for your bass.  Maybe you can use a 15V supply and still sound good.  The more you can decrease the power rails the less power you will use.
D)  From C, a rail-to-rail op amp can gain up to 3V headroom without changing the supply (1.5V from Voh, 1.5V from Vol).  Can you afford $7 to $10 opamps?

For headroom, unfortunately the 5 gyrators are where you can't skimp on headroom.  Gyrators act like a grounded simulated inductor.  As you turn the pot toward a lower resistance, you simultaneously increase Q of the resonant circuit (especially as you boost a band).  The amplitude of the signals seen on any one of those channels can ring pretty high, especially on transients, while the recombination of all the bands can cause some phase cancellation which reduces the effective output level on the through-signal path.

You can take into account the type of signal going in.  If Bass, you CAN reduce headroom on the higher frequency bands because the signal amplitudes will be comparatively small above 250 Hz.

As for noise, a gyrator is a noise circulator.  The positive feedback combined with the phase delay causes noise to regenerate itself moreso than other filter topologies.  The best way to keep noise down is to keep the simulated inductor Q down (increase R11,R13,R15...).  To lower the Q means fewer bands, or clumsy overlap between bands.  So if you can get the results you need with a 3-band parametric, then don't use 5 resonant gyrator stages...or maybe 2 peaking parametric bands combined with a low shelf and high shelf.

The only other way I know to reduce noise in gyrators is brute force: $10/device low noise op amps with <1 nV/rtHz equivalent input noise.  Also choose resistors carefully.  Unfortunately you don't typically get very high values with wire wound resistors.

My gut feeling is that noise on the 6-shooter is not too bad with TL072's and reasonable settings.  Your noise will get much more pronounced if you find you have to do an extreme boost or cut on a certain band.  If you have to do that then instead you may want to consider making a special function block for that purpose and choose a different filter topology.

Something that will significantly increase headroom and reduce noise AND power consumption all in one go is to replace all the gyrators with real inductors.  The down side is this reduces cost and physical size.

A DSP implementation starts looking pretty good after considering all of this...not a wonder most commercial units are going this direction.

[Mark Hammer]

5532s pull a lot of current.  If I'm not mistaken, the original, or at least one of the issues, used TL022s, which are much lower current draw.

[PRR]

> assumption most op amps don't change current consumption significantly with input voltage, but NE5532 datasheet states current only at +/-15V. I assume it's approximately the same from +/-5V to +/-15V.

[Transmogrifox]

Perfect.  Thanks.

[Bill Mountain]

Wow Transmogrifox!  Thanks for the extremely thorough answer!  This really helped me understand how I can tweak the EQ.  I'm gonna save this post and refer to it often.

I can't thank you enough for taking your time to type all of that out.

I chose the NE5532 because they are always billed as a good audio specific opamps so I thought it would benefit the EQ but I'm not married to them.

I'm gonna start out with low current opamps and see if the sound is effected at all.  I can also select 9, 12, and 18V with my power supply to I'll mess with those and give a critical listen.  My bench-top supply is variable up to 60V so I can also use that to fine tune the desired voltage.

As for the noise, it's not a huge problem but I notice it with certain setups so I'll need to narrow down the culprit.

Thanks everyone else as always for your help!