BOSS GE-7 demystified (SPICE Simulation of the whole circuit in LTspice XVII)

[paranoid_android]

Hi everyone,

It's my first post on this forum, so please forgive me for any misbehaviours of a newbie .
Besided, English is not my native language and I'd be grateful for any corrections.
Anyway, let's get to the point. I took some time to simulate a BOSS GE-7 in Spice. The only part I didn't simulate is BOSS' bypass circuit, since I own a clone of a GE-7, that has a true bypass.
I basically wanted to check whether bands' frequencies are labeled correctly and what is the real range of controls (BOSS states that each pot's range is -15dB to +15dB). I also wanted to check how does the pedal behave with strong signals (i.e. humbuckers or active pickups or outputs of other stompboxes), does it saturate and color the signal and by how much.

Schematic - stock pedal

So, here's the schematic I used for simulations:




Frequency Characteristics of a stock pedal

At first I check the frequency characteristic of a pedal with all pots sets to 0dB:



Then I did a parametric sweep of a volume pot, with all other pots being set to 0dB. Output volume adjustment pot's wiper (which in the case of this particular pedal is a linear pot) was sweeped from 0 to 1 in steps of 0.16667 which effectively means that I sweeped the pot from -15dB to +15dB in steps of 5dB. This is what the result looks like:



As you can see the pot does not change the volume in linear fashion (linear but in dBs), but the edges of the range are more sensitive. The second plot from the top should be at +10dB level (more or less) but it's at 7dB instead. The sensitivity curve is symmetric, so the same effect can be observed for negative settings of the vol pot.

Then I started doing pot value sweeps for each pot of the 7 bands of EQ. Once again - the gyrators on which GE-7's filters are based do not behave in a linear fashion. At the pot's range edges the sensitivity is much bigger:
















At last I did a sweep of all bands' pots at once. This is the result:



So, the band's Q is so wide, that boosting or cutting all bands at once results in the pedal behaving like a wide range booster (and of course it's used as a simple booster by some). It's true for all pot's settings except (once again) for ranges limits, since at these limits the Q of the bands is bigger (which is easily visible at the most upper and most lower plots of the sweep) and the bands do not overlap.

That's all for now.
I intend to do some more simulations (transient ones this time) as soon as I find the time.
For now, it'd would be great if anyone gave me a feedback regarding usability of such simulations. I'd also appreciate any hints about what should I simulate next.

[ElectricDruid]

Very interesting to see! I think sims are great for this sort of stuff.

I found the same issue with gyrator-based cut/boost controls when I was working on the Hard Bargain distortion pedal:

https://electricdruid.net/designing-the-hard-bargain-distortion-pedal/

As you've shown, most of the action is at the extreme ends of the pot, and the bit in the middle doesn't react as much. For some things, I guess that might be what you want, but for my purposes it wasn't and I finished up using a tweaked Baxandall stage instead.

[Rob Strand]

So, the band's Q is so wide, that boosting or cutting all bands at once results in the pedal behaving like a wide range booster (and of course it's used as a simple booster by some). It's true for all pot's settings except (once again) for ranges limits, since at these limits the Q of the bands is bigger (which is easily visible at the most upper and most lower plots of the sweep) and the bands do not overlap.

The common gyrator circuit, like the GE7, always shows low Q when the boost/cut it at low levels.  The circuits are said to have "proportional Q".   The Q and amount of boost/cut are linked, as the you increase the boost/cut the Q increases.    If you look at the Rane web-site you will find Dennis Bohn's papers on constant Q equalizer designs.

From a circuit perspective the Q is set by the resistor in series with the L+C network.   In most circuits the series resistor is part of the gyrator; it's the resistor at the output of the gyrator opamp.  Some circuit will add another resistor to the pot wiper.   Most articles calculate the Q using that resistor.  However, it's not the whole story.  When the the boost-cut pots are near the centre they add a considerable series resistance and that drops the Q.

For a multi-band EQ these a lot more going on.  The bands all interact and this affects the Q and the amount of boost/cut.   If you calculate the expected boost/cut from the gyrator series resistance you will find it is much higher than what you see in the complete circuit.

[Ben N]

A constant Q

[PRR]

> to check whether bands' frequencies are labeled correctly

Grumpy Old Man says:

There's no numbers on our ears, what use is numbers on the knobs?

If you must have numbers: simple math tells you faster than entering all that in a sim. R*R*C*C, SqRt, 1/x, /6.28. For the "3.2k" band my TI30 says 3305Hz, which is 3300Hz for any practical purpose. Can I repeat the trick? For the "100Hz" values I get 95.6Hz, close enough for polka.

The fact that slide-travel is not linear (whatever that means) is easily checked with the summing network and a 329r resistor (because the resonator reduces to 329r at center). You can use a sim but hand-calcs are not too lengthy.

Boost/cut also comes out of this exercise.

The fact that the Q is "proportional" (i.e. cheap) can be seen by the fact that LCR (as Rob says) must also include whatever parts of the pot are in circuit, so the effective R must rise from 329r toward infinity and Q must fall from near unity to zero.

[paranoid_android]

Thanks for all the replies,

The common gyrator circuit, like the GE7, always shows low Q when the boost/cut it at low levels.  The circuits are said to have "proportional Q".   The Q and amount of boost/cut are linked, as the you increase the boost/cut the Q increases.
[...] 
From a circuit perspective the Q is set by the resistor in series with the L+C network.   In most circuits the series resistor is part of the gyrator; it's the resistor at the output of the gyrator opamp.  Some circuit will add another resistor to the pot wiper.   Most articles calculate the Q using that resistor.  However, it's not the whole story.  When the the boost-cut pots are near the centre they add a considerable series resistance and that drops the Q.

Yes, you are right. But I have to admit I haven't analyzed this circuit before and I wasn't familiar with this effect. Now that I've taken a good look at the circuit I can see it has to be that way.
Nevertheless, I know I could've just done some simple calculations to estimate the center frequencies of all the filters, but you know what they say - a picture is worth a thousand words. That's why I did these parametric sweeps, which immediately revealed the truth about this circuit's limitations and side-effects.
Another thing about this circuit (and allmost all +9V stompboxes) is that it's rather impossible to have a clean +15dB booster that is supplied from 9V (it means -4.5V swing for the signal, theoretically, but in reality the opamps used are not rail-to-rail, so the achievable clean swing is much worse). +15dB for a 1V p-p input (Humbuckers? active pickups?) means +-5.62 voltage swing which means the opamps will heavily distort the signal, since they won't be able to produce output level this high.
However, I understand BOSS' concept of giving the user such a wide boost/cut ranges, since guitars are not created equal plus the signal going into the GE-7 can be attenuated thus allowing user to boost the signal by 15dB and still have a clean output signal.
I'll post some simulations showing this obvious distortion of this non-distortion effect

I also have a little off-topic question. How do I post an image as an image instead of a link to postimg.cc?

[PRR]

> impossible to have a clean +15dB booster that is supplied from 9V

"Common"(?) Sense says that you don't boost a zone 15dB unless it is pretty weak to start.

And that you only turn it up until it sounds bad, then back-off.

And experience says that the 15dB boosts are to make the potential buyer say "WOW! This really does something!", but in any sensible use 10dB even 6dB may be all you really want.

But in *music performance*, boosting 150Hz or 3KHz 15dB into gross distortion could be "that sound".

[Rob Strand]

Nevertheless, I know I could've just done some simple calculations to estimate the center frequencies of all the filters, but you know what they say - a picture is worth a thousand words. That's why I did these parametric sweeps, which immediately revealed the truth about this circuit's limitations and side-effects.

I totally agree.  I was just mentioning the effect.   In fact there so much interaction on this circuit the simple formulas aren't much good at all, they can only be used as a starting point.   I pretty much always use spice for these type of equalizers.  Another point is the outside bands are affected in a different way to the inside bands.   The bands skirts get skewed in that the left and right skirts aren't symmetrical.

There's more complex calculations, like in,
http://leachlegacy.ece.gatech.edu/ece4435/f02/equalizer.pdf
which are essentially doing what spice is doing but with explicit equations.   I've used numerical optimizers on these equations to select the parts.   Even then you run into problems because you have to decide if you want a good match on individual bands or groups of bands.

Try this in spice:  Find the settings pots for say xdB boost *individually* then set a number of bands to those settings.  You will often see the final response for the multiple band setting doesn't match what you expect.  So the "picture" of the pot positions doesn't show the shape of the frequency response very well.   Some of these issues are discussed in the Greiner paper.   Yet another thing with those equalizers is the boost/cut is cramped up at the ends of the boost/cut pots which is why you need to use G (or W) taper pots.

There's a lot of evils under the hood on those simple equalizers nonetheless that are the most common.

[R.G.]

Also look up Rane's paper on "constant Q equalizers".

The simple RLC graphic eq is a non-constant-Q variation, but falls in the range of (apologies to the UK) "cheap and cheerful".

[clintrubber]

Got an old GE-7 in today, the type still using the ACA adaptor.

Needs some work, as was announced - but surprisingly the freq-sliders don't have any freq-selective effect.

Looking inside, no messing around or after-fab soldering as it seems....
but I found the gyrator-opamps all three being the TI NE555P type ?!? Timer-ICs, not the dual opamps. Wow ! 

Now I wonder how _that_ would turn out in simulation 

[iainpunk]

that seems sketchy as h3ll

they should be double opamp IC's if i remember correctly.
i'd contact the seller and ask for explanation, i expect that someone with a lack of knowledge tried to repair it somehow.

cheers

[clintrubber]

Indeed weird!

Seller couldn't tell more, and in all fairness, pedal was sold as 'needs attention', and priced accordingly.

Was actually surprised it passed audio, and most faders accomplish level boost/cut.

It's not the first device/gear/thing I see 'populated' based on just mechanical considerations 

[iainpunk]

looking at the schematic, the actual part doing the frequency's reminds me of the tube screamer tone control, or the BOSS HM-2 and FZ-2 tone controls. really makes me think i can build a mod board for any of those pedals that turn the tone knobs in to a 7 band eq, probably house it in a bigger enclosure tho.

that would be a cool project

[clintrubber]

That'd indeed work, just slapping a bank of gyrator-sections to it.

Could even make it more fancy with freq-shift for each band (see for instance SWR SM-400 preamp schematic),
but then a graphic EQ can become a bit messy, easy to lose track what's actually the resulting EQ-shape when too many bands are present  ;-)

[Rob Strand]

Looking inside, no messing around or after-fab soldering as it seems....
but I found the gyrator-opamps all three being the TI NE555P type ?!? Timer-ICs, not the dual opamps. Wow ! 

Indeed

Obviously the NE555 is a cheaper version of the NE5532  .   One costs 5 units and the other 32 units.

[Radical CJ]

Looking inside, no messing around or after-fab soldering as it seems....
but I found the gyrator-opamps all three being the TI NE555P type ?!? Timer-ICs, not the dual opamps. Wow ! 

Indeed

Obviously the NE555 is a cheaper version of the NE5532  .   One costs 5 units and the other 32 units.

Haha. NE555s can be used in a heap of different applications, but I'm yet to get one working as a dual op amp.  I believe there is a configuration that will drive a small speaker, but the NE555 doesn't like it and overheats.

Put randomly in a circuit they could pose a risk to other components due to their voltage multiplying abilities.

[Rob Strand]

Haha. NE555s can be used in a heap of different applications, but I'm yet to get one working as a dual op amp. 

That would be a good hack.   Hard to get past the flip-flop.  The reset pin is the only hope there.

I believe there is a configuration that will drive a small speaker, but the NE555 doesn't like it and overheats.

Yes I've seen that PWM trick.   The raw PWM output has an average DC level which isn't great for driving the speaker.  If you connect the speaker via a series cap it will work a bit better.  Maybe add a series resistor to limit the current.   IIRC the tone changes when you add the extra parts.

[iainpunk]

didn't the original Atari Punk Console/Stepped Tone Generator directly drive a little speaker? its supposed to be able to put out up to 1/5th Ampere., i read somewhere it was mainly designed that way to drive gates of power mosfets ^{[citation needed]}
i might try that out with my 555 fuzz pedal directly in to my 16ohm cabinet, using a 30 ohm resistor to lower the current.
should give about 0.6W of power to the cabinet... 0.6 watts of pure squarewave/pwm/chaos goodness

cheers

[Vivek]

http://cushychicken.github.io/ltspice-boss-ge7-equalizer/

Is this from the same OP ?

[Radical CJ]

didn't the original Atari Punk Console/Stepped Tone Generator directly drive a little speaker? its supposed to be able to put out up to 1/5th Ampere., i read somewhere it was mainly designed that way to drive gates of power mosfets ^{[citation needed]}
i might try that out with my 555 fuzz pedal directly in to my 16ohm cabinet, using a 30 ohm resistor to lower the current.
should give about 0.6W of power to the cabinet... 0.6 watts of pure squarewave/pwm/chaos goodness

cheers

The 555 squarewave output from the Atari Punk Consol will drive a teeny tiny speaker at least. I've hooked one up to a 1.5 inch speaker. Not sure how large you could go.