[Notch filter musing] Why can't pirates recite the entire alphabet?

[Rob Strand]

Presumably the adjustable resistor is used to adjust frequency,

https://booksite.elsevier.com/9780123914958/content/Notch_Filter_Calculator/notch_2.htm

[EBK]

The goal of the pot (in my schematic) is very small adjustment after getting close to the target frequencies by picking the other parts.  It should ideally remain quite close to its center position.

I'm going to recheck/rethink my math a bit later today in view of these simulation results.

[Rob Strand]

The goal of the pot (in my schematic) is very small adjustment after getting close to the target frequencies by picking the other parts.  It should ideally remain quite close to its center position.

That sounds fine.   Small  pots adjustments can affect the notch quite a bit.  It might be better to use say a 10k pot with 22k in series with the clockwise and counter-clockwise pins, effectively making a 54k pot with a narrow adjustment around the center position.

I'm going to recheck/rethink my math a bit later today in view of these simulation results.

I don't know about the notch frequency but the general values seemed OK.

I found this part summary:


I can confirm either or both Ro's tweak the frequency but the one connecting to the output has less effect on the notch depth when only one resistor is tweaked; so that site with the adjustable resistor is correct.

If you aren't getting any signal out in general there might be an error in the build.   A notch should mostly pass signal even if the notch frequencies are off.

One other point it might be wise to drive the circuit from a buffer instead of a guitar as the guitar impedance will surely mess-up the notch frequencies.

[anotherjim]

Eric, apart from all this filter theory malarky, what's the DC biases like? Vref all the way?

[PRR]

...I found this curious statement in a doc from TI ....: "Unfortunately, this circuit uses two op amps instead of one,....."

That IS curious. TI $ells op amps. Why would TI think using more op amps is 'unfortunate'?

[Rob Strand]

Why would TI think using more op amps is 'unfortunate'?

Filter designers always go for the single opamp filters (SAB = Single Amplifier Biquad).    When you want to adjust the filter parameters you end-up with more opamps like the state-variable filters where the frequency and Q can be tuned independently.  The Fliege is pretty good as far as adjustments go.

FWIW, the Fliege notch circuit is basically an LCR circuit where the L simulated with the two opamps (with a GIC).  It has a cool trick which takes the output at the opamp output.

It's more or less this LCR notch filter but with an R in parallel with the LC network to set the Q:

[EBK]

I definitely have a weird breadboarding glitch going on.  pins 2 and 6 of U1 are connected by a short jumper wire, but the voltages switch between about 1.4V and 4.7V, sometimes measuring differently from each other.  The jumper itself appears solid, so perhaps it is a bad spot on my breadboard.  The rest of the voltages are likewise sometimes 4.7V and sometimes 1.4V.  pins 4 and 8 measure fine (0V and 9.45V).  My green wire Vref also appears to be fine (4.73V).

I think I will try rebuilding on a different breadboard.

[anotherjim]

Worn breadboard is very frustrating. Mine need a refresh but I keep forgetting to add some to parts orders.

[EBK]

Worn breadboard is very frustrating. Mine need a refresh but I keep forgetting to add some to parts orders.

It's a brand new breadboard from Smallbear.  This is my very first build with it.   

[Rob Strand]

It's a brand new breadboard from Smallbear.  This is my very first build with it.

The Wishboard brand you have is the best, at least they used to be.  Inside the hole they have *two* springy metal rails on each side of the wire.   Is the wire diameter OK?  IIRC it's written on the packaging.  Too small can get problems and too large can splay the contact and disconnect something else (pretty rare though).

[Harry Muff]

looks like the walk away here is a BB vref gen

[EBK]

looks like the walk away here is a BB vref gen

I'm just happy that this thread has some value.   

[11-90-an]

[nonsense idea]

you can add a frequency tracker that detects the picked note and positions the notch filter there, which cancels that note out leaving the harmonics...

[/nonsense idea]

[EBK]

It's a brand new breadboard from Smallbear.  This is my very first build with it.

The Wishboard brand you have is the best, at least they used to be.  Inside the hole they have *two* springy metal rails on each side of the wire.   Is the wire diameter OK?  IIRC it's written on the packaging.  Too small can get problems and too large can splay the contact and disconnect something else (pretty rare though).

I'll slow down a bit and measure the U2 voltages, and if they look ok, then I'll rebuild the U1 circuit a bit further down on the same breadboard.

[EBK]

But as drawn, the amps have positive DC feedback via R1, R5 so the slightest DC offset in the amps and they are off to jamb into a power rail.

It is looking more and more like this is the problem.  My breadboard may be working just fine.  I'm still investigating though....

[anotherjim]

Once upon a time all opamps were single and had offset null trimmer connections which many designers fitted and used. Is this in this circuits history?

[Rob Strand]

But as drawn, the amps have positive DC feedback via R1, R5 so the slightest DC offset in the amps and they are off to jamb into a power rail.

It is looking more and more like this is the problem.  My breadboard may be working just fine.  I'm still investigating though....

You did well to spot that.

The "textbook" circuit works fine DC wise.

The root cause is the 1uF AC coupling caps on the inputs!!  You need to add a resistor from the 1uF AC coupling cap to ground (actually to Vref).    A value equal to the trimpot resistance should do it.    What that does is it shunts or divides down the DC loop gain around the outside loop ie. through the pot and back to the input.

It's not (just) the opamp offsets it's an outright positive feedback *by design* which makes the circuit go nuts DC wise. 

With a DC coupled input the low source impedance shunts away the DC positive feedback.

[EBK]

I hope I understood your suggestion correctly....

I've added resistors (approximately 200k) between the right-hand side (referring to the schematic) of C5 and Vref and between the right-hand side of C6 and Vref.  The DC voltages are now stable and around 4.7V on every pin (except for pins 4 and 8, of course). 

[Rob Strand]

I've added resistors (approximately 200k) between the right-hand side (referring to the schematic) of C5 and Vref and between the right-hand side of C6 and Vref.  The DC voltages are now stable and around 4.7V on every pin (except for pins 4 and 8, of course). 

Yes, that's it.   

FWIW,  you don't need to *match* the added resistors to the pot value.   It's only that something in the order of the pot value will work.   Lower resistor values will work for sure, for example 10k!

[EBK]

I made another Vref buffer board this morning as soldering therapy.  A bit more breadboard friendly this time (header pins instead of stranded hook-up wire):

I have a ton of 12k resistors, so I used them here.