Faking a center tap on 3 terminal pot?

[Gustav]

I've been spending some quality time with the NetEQ schematic, trying to understand, redraw etc.

I am looking into implementing digitally controlled pots. I haven't come across any digitally controlled pots with a center tap, so I was wondering if I could get around it, and ended up trying this idea. (See attached).

For a "first try", its a  22K pot and 2 x 11K resistors.  On one hand, it just seems to make sense, while on the other, the throw and values are obviously off  (I can work on figuring out how and why I am getting the values I am getting, and how to adjust them to my need)

Ignoring the throw for now, and just focusing on circuit functionality, would D function as a center tap?

AC = 11K4
BA "range" = 11K4
BC "range" = 11K4
AD = 8K3
CD = 8K3
BD ccw = 11K4, center = 8K3, cw = 11K4

Gustav

 

[JohnRoberts]

To make a center tapped pot with digitally controlled pots (I ASSume you mean IC DPOTs not motorized pots), you can mimic a center-tap by stacking two DPOTs in series, Then you would need to use a transfer gate or some soft switch to select between which wiper you want to read depending on which half of the combined DPOT you are on .

If your DPOT allows you to set the wiper to High Z you could short both wipers together and alternately set one or the other to high Z making the other active.

This is a bunch of work and expense to mimic a center tapped pot (like used in loudness circuits and such), but about the only way I know to do it.

JR

 

[Gustav]

you can mimic a center-tap by stacking two DPOTs in series, Then you would need to use a transfer gate or some soft switch to select between which wiper you want to read depending on which half of the combined DPOT you are on .

If your DPOT allows you to set the wiper to High Z you could short both wipers together and alternately set one or the other to high Z making the other active.

I had been looking into stacking pots to do a center tap, but I just couldn't wrap my head around it - thanks for the input on that, even if its not really a practical solution.

Why does the fake work, though?

I spent some more time with it again today to understand it better, and for this application, I can't get it to "not-work" (see attached cut-out of schematic, P14 on the lower right is the pot in question).

Measured 24K end-to-end (AtoC) on my pot, and calculated 6K5 in parallel to get an end-to-end resistance of 5K1 (AtoC)

Soldered two 3K3 resistors in series (AtoC), and drew a center tap between them.

I am getting about 13dBU boost/cut with that pot on the circuit, and it seems to work perfectly...!?

Gustav

 

[PRR]

> seems to work perfectly...!?

I say your plan "does not work".

The main idea of this center-tap is so when you set the knob "center", the boost/cut gives *exactly flat* response. When the wiper is ON the center tap, it is not influenced by the boost/cut circuits, only the Ground. Dead Zero.

Your external bridge resistors do not do this. (And may be skewing the curves.)

> getting about 13dBU boost/cut with that pot on the circuit

The CT is not about max boost/cut. Proves nothing.

Put the pot dead-center. Is the response exactly "flat"? Most pots, mechanical center is not *exactly* electrical center. Fussy users want a known-FLAT position. You can't spend a whole day lifting knobs, measuring, and setting knobs so 0 is true-flat.

If you can't find/afford a true center-tap pot, add a switch to break the wiper from the pot and ground the stuff which goes to the wiper.

If you are very DIY, crack open the pot. Get "automotive window defroster paint", put a small stripe at the mechanical center of the pot tracks, and bring a wire out.

 

[Bo Deadly]

Ignoring the throw for now, and just focusing on circuit functionality, would D function as a center tap?

No. It would not function as a center tap.

A real center tap makes the contribution of the band totally zero by grounding the signal when the pot is in the center position. The point is probably to reduce common mode signal on the virtual summing amps or reduce noise or something. Or maybe it has to do with the taper. Not sure.

The reason it still works is because in the center position the two signals still cancel each other out without the center tap.

I'm not really sure I like this topology because if you connect something to the virtual ground of the op am, you have to be conscientious about potentially high noise gain. It might be that increased noise gain is /why/ they used a center tapped pot in this circuit. I would connect it to an audio analyzer (or PC audio input with the software like DSSF3 from ymec.com) and see what the noise difference is with and without the center tap (but not your fake one - use two equal resistors in place of the pot).

The question is, is the noise performance of this circuit with the center tap ground really better than other circuits that use conventional potentiometers? If not, you might want to choose a different design entirely.

Radio Shack in the US used to sell a center tapped pot. But I think it was dual audio.

 

[Gustav]

> seems to work perfectly...!?

I say your plan "does not work".

The main idea of this center-tap is so when you set the knob "center", the boost/cut gives *exactly flat* response. When the wiper is ON the center tap, it is not influenced by the boost/cut circuits, only the Ground. Dead Zero.

If you can't find/afford a true center-tap pot, add a switch to break the wiper from the pot and ground the stuff which goes to the wiper.

I am messing around with digital pots, so the "center detent is" in itself is not so important -  but according to Barry Porters notes.

"The amplitude pots do require to have a centre tap, which is grounded. If untapped pots
are used, the sections will interact quite severely, and the constant Q characteristic
will be lost."

I have to change around a few values from the original schematic to make everything "fit", but from the experiments I have done so far, the solution avoids the problems mentioned.

I'll mess around with it some more. If nothing else, I'll have learned  that it doesn't work.

Gustav

 

[Bo Deadly]

but according to Barry Porters notes.

"The amplitude pots do require to have a centre tap, which is grounded. If untapped pots
are used, the sections will interact quite severely, and the constant Q characteristic
will be lost."

So this is because the same virtual summing inputs are used for all bands (cutoff in your schematic). So without the center tap, it will absolutely not work.

A digipot would not work well in this particular case because they have significant wiper resistance. Even a high voltage analog switch like ADG1414 has 10 ohms of gate on resistance.

Use a stepped switch like Grayhill 71BD30-01 so that you can get the center tap. If you do 10 steps of 3 dB that's +-15 dB boost cut. Actually I would do 11 steps and make it cut a little more than boost.

 

[john12ax7]

I would check the band interaction with your circuit. For me that is the big selling point of the center tap. If you can minimize/eliminate the interaction then it would be a success imo.

I thought about this for another project and didn't find a good solution, ended up adding more active stages to fix things.

 

[JohnRoberts]

you can mimic a center-tap by stacking two DPOTs in series, Then you would need to use a transfer gate or some soft switch to select between which wiper you want to read depending on which half of the combined DPOT you are on .

If your DPOT allows you to set the wiper to High Z you could short both wipers together and alternately set one or the other to high Z making the other active.

I had been looking into stacking pots to do a center tap, but I just couldn't wrap my head around it - thanks for the input on that, even if its not really a practical solution.

Why does the fake work, though?

I spent some more time with it again today to understand it better, and for this application, I can't get it to "not-work" (see attached cut-out of schematic, P14 on the lower right is the pot in question).

Measured 24K end-to-end (AtoC) on my pot, and calculated 6K5 in parallel to get an end-to-end resistance of 5K1 (AtoC)

Soldered two 3K3 resistors in series (AtoC), and drew a center tap between them.

I am getting about 13dBU boost/cut with that pot on the circuit, and it seems to work perfectly...!?

Gustav

If the fake works for you as expected  in the actual circuit then use it and be happy.

Your fake is NOT a center tapped pot...  There are several different applications for real center-tapped pots.

Seeing the actual circuit it will be used in might help me give a more intelligent answer, or then again maybe not.  8)

JR

 

[Gustav]

Thanks for indulging me.

It does seem to follow the behaviour of the unit I built when I simulate it in spice - even the slightly lower boost on the high/low band compared to the mid bands.

Theres also a similar roll-off (very slight) over 10K when its set flat. It seems to be dependent on the setting of the high shelf/or adding a resistor going to the high shelf.

Gustav

 

[Martin Griffith]

I was just looking at Doug Self's book(googlebooks) on J111 fets and was thinking of using a dual gang linear with the second track feeding a comparator  to change the operation of the audio track to mangle the fets switching., it was just a little thought experiment that didn't get anywhere. I think opto fets could also be contendors. I then thought of using silver/conductive glue and wirewrap wire to  make a CT, but I dont have an Omeg type pot to play around with.  I was sitting down fortunately when I checked the price http://uk.rs-online.com/web/c/adhesives-sealants-tapes/adhesives-glues/conductive-adhesives/?sra=p

 

[JohnRoberts]

Thanks for indulging me.

It does seem to follow the behaviour of the unit I built when I simulate it in spice - even the slightly lower boost on the high/low band compared to the mid bands.

What is this "spice" thing you speak of?  I'll bet that requires a schematic.

Theres also a similar roll-off (very slight) over 10K when its set flat. It seems to be dependent on the setting of the high shelf/or adding a resistor going to the high shelf.

Gustav

Perhaps put a real center tapped pot model(?) into your "spice" simulation and see if it behaves differently... that should answer your questions (if spice is accurate). I don't know I haven't tried it in decades and way back when it wasn't worth the trouble (IMO). 

If you visualize the currents flowing inside a center tapped pot, and through the tap, you should see the difference between that and hanging some external components in parallel. 

JR 

 

[Michael Tibes]

Hi Gustav,

just curious: what do you want to use as a digitally controlled pot? An IC or a relay circuit like the ones used in relay-based volume controllers?

Also I wonder if you really need to replace the pot as such or if you could think differently: I might have misunderstood the circuit, but can't you directly feed either the inverting or non-inverting bus input via an adjustable resistor - which could be realized a lot easier, being only a 2 pole device?

Michael 

 

[joaquins]

For the circuit you attached in reply nº2 you don't need a CT pot, you could do one of two things, calibrate your spots for a dead center, maybe one or two steps out of the expected value... Or easier and better, add a soft switch to kill the band when it should be in the center, from the wiper to ground. That would act as your CT at the center while is most important, but won't affect once it's out of center.

JS

 

[Gustav]

Hi Gustav,

just curious: what do you want to use as a digitally controlled pot? An IC or a relay circuit like the ones used in relay-based volume controllers?

Also I wonder if you really need to replace the pot as such or if you could think differently: I might have misunderstood the circuit, but can't you directly feed either the inverting or non-inverting bus input via an adjustable resistor - which could be realized a lot easier, being only a 2 pole device?

Michael

Have to move both at once. The circuit isn't picky about the total resistance of the pot (AC), but AB/BC has to match the resisters programming the signal path opamps and lowering the resistance on AB, while raising it on BC  (or vice versa) symmetrically is essential.

I realise the pot I drew up is not a real center tap, but I figured that the focus on stability here would only demand a steady, symmetrical centre reference, not necessarily a real ground reference. It also goes without saying, that I have more resistance to the tap than at either side, not a "short-to-ground" at center position. Even for the application, it seems my experiment is a dud, even if I seem to have some sort of stability

Original schematic is here - https://pcbgrinder.com/download/NetEQ_Schematic.pdf

Ill attach a spice model with the pseudo pots drawn in, a modification to the Q and a few other changes to accommodate building the same resistor ladder for every switch (or just using 10K pos).  I find that this alternative way of doing the Q makes the circuit a lot more stable, and the center tap less essential for stability. I just keep screwing up small details cleaning up the mess I ended up with while trying out various things, so may not be perfect

I  ditched the shelving filters, since they really mess with the bells, even set flat/not switched in. This is also consistent with the unit I actually built. On the spice model, theres also a phase shift of about 5-8 degrees with everything set flat if the shelves are drawn in. I have no real-world reference to the actuality or relevance of that shift.

I plan to work on getting just one AD529x moving and seeing how it works as a direct pot replacements, if the "outside zero crossing" switching noise is significant etc. I am not even at the starting gate yet, though.

Gustav

 

[Michael Tibes]

I tried to scribble down how I understand it:

The grounded center tap turns the pot into two halves: one with a pot function and one which is just a resistor with half the pots value going to gnd. Depending on whether the pot is turned left or right either the left or right side is the 'active' one (top picture).
So I believe first of all you could replace the pot with two individual pots if you switch the wipers as indicated in the last picture.
Then you could probably go further and replace the pots with just variable resistors as indicated in the bottom left picture. If I get things right then the 'passive' part of the pot is not really doing anything but connecting the inverting input of the opamp to gnd, which seems unnecessary to me without the pot.
I'm curious what you think, I don't have the time to sim the whole thing right now and I also haven't understood yet, why the bands might interact an lose the constant q. So I might be on the wrong path here, but maybe not 

Michael

 

[JohnRoberts]

OK, the grounded center tap in that topology completely isolates the boost side from the cut side. More commonly used in graphic EQs and has nothing to do with Q which is defined in the state variable filter section.

If you want to realize the exact same gain law as Porter did, I repeat my two DPOT suggestion to mimic a center tapped pot, with wipers alternately selected.

Your extra parallel shunt resistors just increase the noise gain of the two opamps, and do nothing for isolation.

Perhaps TMI but the - input of a virtual earth op amp is not 0V, but will be 0V +/- the op amp output voltage divided by the Op amps open loop gain. Arguably these - input error voltages will be very small and signal related so crosstalk will likewise be small and mostly inoocuous, but real. Since open loop gain falls with increasing frequency crosstalk from this mechanism is expected to increase with frequency.

The Porter design is well executed, actually ahead of it's time (IMO).

With two DPOTS you can mimic it almost exactly, or change the topology since you are no longer locked into using a mechanical pot.  An alternate topology could reduce the noise gain of the connected op amps, while they may not dominate the noise floor of the EQ.

I don't know that you need to improve upon Porter's design, but some potential for lower noise possibly exists, but with different boost/cut  laws. However if you are digitally controlling a DPOT you can make the boost/cut law whatever you want.

JR 
 

 

[john12ax7]

I tried to scribble down how I understand it:

The grounded center tap turns the pot into two halves: one with a pot function and one which is just a resistor with half the pots value going to gnd. Depending on whether the pot is turned left or right either the left or right side is the 'active' one (top picture).
So I believe first of all you could replace the pot with two individual pots if you switch the wipers as indicated in the last picture.
Then you could probably go further and replace the pots with just variable resistors as indicated in the bottom left picture. If I get things right then the 'passive' part of the pot is not really doing anything but connecting the inverting input of the opamp to gnd, which seems unnecessary to me without the pot.
I'm curious what you think, I don't have the time to sim the whole thing right now and I also haven't understood yet, why the bands might interact an lose the constant q. So I might be on the wrong path here, but maybe not 

Michael

You could use a  dual gang linear pot, wired correctly the bottom right circuit would work. The switch would select boost or cut function.

The bottom left wouldn't work  though.

 

[Michael Tibes]

The bottom left wouldn't work  though.

Why not? I don't want to insist on it, but it would be interesting to know why it wouldn't 

 

[JohnRoberts]

The bottom left wouldn't work  though.

Why not? I don't want to insist on it, but it would be interesting to know why it wouldn't 

The bottom left suggestion will not have good hop-off (on?) behavior. The rheostat value would have to be pretty large value to reduct boost/cut near zero but would still have a step. 

JR

BTW the OP is talking about using a Digital pot,,, using two there are more options available.