NEW CIRCUIT DESIGN: NZF Flanger

[armdnrdy]

Alex,

First let me say that I've been following this thread with great interest! This is quite an undertaking.
If I wore a hat..it would go off to you!

So Larry, when you say that changing a BBD can mean you need to re-bias, do you mean that you need to do that because the optimal bias level has changed (i.e. it's not an audible difference, but you can see the difference when you go through the bias procedure with a scope), or do you need to rebias because the new BBD bias point is so far from the old that you get noticeable clipping if you don't rebias?

I have experienced changing a BBD and the bias being so far off for the new device that no signal would pass. This was with different SAD1024s. My experience with the MN series (the pieces that I've tested) is that the bias requirement does seem to be closer.

I've got around a dozen MN3207 to experiment with and I'm hoping that once I've set a single bias level, changing ICs will still keep things working OK with no clipping.

You may find that this batch is similar in characteristics/bias requirements, but another group manufactured at a different time/place might be different. And let me add, if one were to use the Belling BL3207 or the Cool Audio V3207D, again, you might obtain different results.

I built an AD-3208 (with two 3205s) for a friend's son, Austin. The AD3208 was designed by Scott Swartz and is a mix of an Ibanez CD-10 and a Boss DM-3. The AD-3208 is powered by 9 volts and incorporates a 78L05 regulator to power the BBD. I had previously built a Ibanez AD-80 which is powered by 12 volts for myself.

When I finished the AD-3208 build, I calibrated it and was surprised at the amount of distortion I heard, but chalked it up to a lower headroom effect. I had never played through an original effect using low voltage BBDs, so I had nothing to gauge the distortion amount with except my AD-80.

About a year later, a friend of Austin's wanted me to build him an AD-3208. When I completed the build...I listened to the amount of distortion, and thought....this is not acceptable! this can't be right!

After some research on the net, It seemed that other builds exhibited this issue, so I started a thread on this forum.
Brian from Madbean had been dealing with this issue with members from his site, so with a few suggestions, I started to hack on the board, drilling holes, adding resistors and the like.

It turns out that the V3205Ds I was using output a "hotter" signal than the MN3205! The AD-3208 circuit didn't have any way to accommodate for that. The Ibanez AD-900 has a level adjustment trimmer in between it's two series BBDs.

JD Sleep from General Guitar Gadgets ran across the "fix" thread and PM'd me for the info. I sent him my files so that he could update his AD-3208 project with the "fix"

Sorry for the long story but....I wanted to outline that BBDs can exhibit different characteristics. They do not seem to be cookie cutter replicas of one another.

I've also seen that many pedals don't bother with any sort of BBD output trimpot (Original EM, MXR 117, Morley Flanger, Foxrox Paradox TZF, Boss CE-2, Boss DC-2) while other ones do (Deluxe EM, A/DA Flanger, Flanger Hoax).
I can see how manufacturing differences and ageing affects can change the bias requirements of one IC versus another, but is there really a big difference between the two sets of buckets within the same IC?  

I would imagine that production costs, including components and calibration time, has set a standard of what is acceptable.
It could also be that if a BBD device doesn't meet the criteria in a circuit that lacks adjustment, it is "tossed" in a box for a circuit that does include adjustment.

Example:
MXR M-117 no output trimmer.
MXR M118 (3) SAD1024s in series, output trimmer at last BBD output.

Another theory is: distortion is more acceptable (not as noticeable) in a flanger than in an analog delay. In fact, distortion may add to the "flavor" of a flanger.  

[puretube]

That one glitch you don`t cancel at the output of the first of a few cascaded BBDs,
will ride on top of the audio into the next one and appear delayed at the output of that one. Now cancel that...
(or will it be that synchronized that it really does disappear by balancing that last BBD only?)

And how about the accumulating variable DC-pedestal?

More details to be found in BBD-related patents...


BTW.: tube-filament supply can be grounded on either side,
or centered fix,
or may be balanced with an anti-hummer pot - it all worx...   

[puretube]

BTW.: there were experiences in this very thread, that the PS-voltage varied with the rate (in an LFO...)
same goals with CMOS-clocked (BBD-) circuits: current-demand rises at higher frequencies...

If now the Vdd for that BBD-circuit runs through a dropping series-R, the voltage will drop...
the bias will drop...
the headroom will drop coz Vdd is less, and coz the signal-room around the bias-voltage is less,
or less centered...

If now the modulation-LFO-rate is fast, the BBD-clock spends it`s time in the high-frequency region more often,
cauzing higher current-demand more often IYKWIM...

Now throw that reduced possible headroom into the next BBD, and experience above scenario in accumulated continuation...

[puretube]

B.T.W.: DrAlx,
I really dig your concept of "approaching modulation",
and that`s why I want you to get rid of the hindering hetero-crap
that is limiting your idea too much!
You gotta get over the (zero-) hump... 

"snap" seems to have anticipated a similar concept...

[DrAlx]

So it seems I could get away without the extra trimpots (at least for the 2 BBDs that I've got) but should really include them if I want to be able to handle possible variability in BBDs.
Given that I'm not going to be able to fit the vero in a 1590BB anymore   then I could let the board size increase to allow the option of biasing BBDs locally instead of via the audio path.   I'll probably still bias via the audio path on the next build because my main aim is to see if I can build this on vero (with no ground plane) and avoid heterodyning.
OK, I've got enough info to carry on with this.  Thank you all again.

[StephenGiles]

So it seems I could get away without the extra trimpots (at least for the 2 BBDs that I've got) but should really include them if I want to be able to handle possible variability in BBDs.
Given that I'm not going to be able to fit the vero in a 1590BB anymore   then I could let the board size increase to allow the option of biasing BBDs locally instead of via the audio path.   I'll probably still bias via the audio path on the next build because my main aim is to see if I can build this on vero (with no ground plane) and avoid heterodyning.
OK, I've got enough info to carry on with this.  Thank you all again.

Let not the 1590 police be sharpening their tongues!!!

[puretube]

Save on drill-bits: go SMD...
(spares dozens/hundreds of holes...)


apropos holes, Stephen: remember "Hole In My Shoe"?
[be it with, or without TZF, it was mind-blowing, at that time...]


[edit]: but then again, I`m just a little Tin Solderer
caught in Itchycoo Park

[DrAlx]

"snap" seems to have anticipated a similar concept...

It's not the same but I can see one similarity.  My LFO sine waves have the exact same rate and phase all of the time.  They just have different amplitudes.

"snap" talks about having two identical LFOs with adjustable LFO phase difference.   To me that suggests having fixed rate sine waves, but varying phase rather than amplitude.  The one similarity I see is that at some point you can have two LFOS that have exactly the same amplitude and phase, but its reaching that situation in a different way.

In any case,  the only reason I took the approach I did was because I was trying to find a simple reliable trim procedure to put the zero at the end of the sweep.
If I had gone the usual route of having a varying delay in one path (lets call is sine wave) and fixed delay in the other then the trim procedure would have gone like this...

1) Set the sine wave to minimum amplitude (i.e. depth pot at minimum)
2) Trim the fixed delay so that the zero occurs at the bottom of the sine wave (already a real PITA).
3) Increase the sine wave depth (lets say to maximum).
4) Oh no, the zero point isn't at the bottom of the sweep anymore !!!
   That means the sine wave has a different minimum value to before.
   That means the "offset" trimpot was set wrong.  
5) Try a new value of the "offset" trimpot and go back to 1).

Of course you could speed up this process by first trying to set the "offset" trimpot using a scope (i.e. try give the sine wave the same minimum value at both maximum and minimum depth) but I think that would be a real PITA too.
I then had a brainwave when I realized that I could use a sufficiently low amplitude sine wave as a reference in the other path (instead of a fixed delay) and that got rid of the whole mess of trying to "catch" the zero point at the end of the sweep, because you could trim things so that it was at the zero point for ALL of the sweep.

[DrAlx]

Save on drill-bits: go SMD...
(spares dozens/hundreds of holes...)


apropos holes, Stephen: remember "Hole In My Shoe"?
[be it with, or without TZF, it was mind-blowing, at that time...]


[edit]: but then again, I`m just a little Tin Solderer
caught in Itchycoo Park

Being a teenager in the 80s, I don't remember the original but do remember the cover in the UK by Neil from "The Young Ones".

[StephenGiles]

Dr

Save on drill-bits: go SMD...
(spares dozens/hundreds of holes...)


apropos holes, Stephen: remember "Hole In My Shoe"?
[be it with, or without TZF, it was mind-blowing, at that time...]


[edit]: but then again, I`m just a little Tin Solderer
caught in Itchycoo Park

Being a teenager in the 80s, I don't remember the original but do remember the cover in the UK by Neil from "The Young Ones".

And the drummer on Ichycoo Park lives near you DrAlx in Ewhurst Green - he owns Hurtwood Park Polo Club where we are members!

[StephenGiles]

A thought came to me whilst throwing a ball for our dog just now - Barberpole TZF - now that would be something!! Come on EH, you can do anything after the B9 pedal!!

[DrAlx]

A thought came to me whilst throwing a ball for our dog just now - Barberpole TZF - now that would be something!! Come on EH, you can do anything after the B9 pedal!!

Interesting idea.  First thought that comes to mind is what's the minimum number of delay lines needed to pull that off, and how do you make sure that you ignore the downward slopes of the sweeps and keep only the upwards ones.  i.e. the mix ratio between the various delays needs to smoothly vary too.

EDIT:  Actually having thought about it more I am not sure it is even possible.  
Let's say you have a fixed delay line of 1ms and 4 other sine wave LFOs in quadrature that each gives a varying delay from 1ms to 5 ms.
Each of those sine waves would give you a through zero effect at 1ms (when mixed with the fixed delay).

As I understand it, the barberpole thing involves the sweep only ever appearing to go in one direction (lets from the 5ms point to the 1ms point).
So you would need to fade out the flange effect from each sweep as it approaches its 1ms point and leave it faded-out before fading it back in again when it reaches its 5ms point and starts to move back towards 1ms.  i.e. you only want the bit of the sine wave that slopes downward from 5ms to 1ms.
So the problem is that the sweep will be faded out before it ever reaches the zero point at 1ms.  So you never get to hear the zero.

[DrAlx]

New vero layout almost done.  Just fussing now and trying to make things tidier.

I've been thinking about the 4046 as an alternative VCO to get the part count down.  Maybe not in the next vero build, but certainly if I do a new version of the flanger at some point in the future.
The 4046 has the wrong sort of voltage to frequency characteristic compared to what I have at the moment for two reasons.

1) In the 4046 , the CV causes frequency to increase rather than decrease.  So my touching sine wave bottoms would cause the zero point to occur when both delay times were at there largest.
I'd need to rework the CV circuit to give touching sine wave tops instead.

2) The 4046 causes frequency to change linearly with V rather than like (1/V) which is what I have at the moment, and what I would prefer since it gives me sinusoidal delay variation (better for the fast Leslie style effect).

I am looking at ways of warping the CV (I think the UltraFlanger does something along those lines with diodes at the LFO output) but am also wondering
if it is really necessary?  For small amplitude changes in CV, I may still get roughly sinusoidal variations in delay, and at the other extreme any big slow sweeps may not be worth worrying about.
So maybe I don't need to warp things.

There is one other potential advantage (apart from reduced part count) that I can see from using a 4046:

With my current VCO, the clock trim pot alters the charge rate of the clock capacitor.
This changes the gradient of the CV to frequency characteristic.  It may slightly alter the intercept too but I haven't checked.
So you only have one control parameter to play with when altering the characteristic of the VCO.

Using a 4046 would allow two dimensions of control via resistors on pins 11 and 12.
The resistor on pin 11 lets you alter the charge rate for the clock capacitor (i.e. gradient of the characteristic) and the resistor on pin 12 lets you control the intercept.
So although the 4046 CV to Frequency characteristic is different to the current LFO, it should be possible to do an even better job of matching the two VCO characteristics than I have at the moment.
Still not clear if both levels of control are needed, but its nice to know they are available.

Any advice/stories from people who've used one of these chips would be appreciated.  I  have read people have had difficulties with their UltraFlanger builds.

[armdnrdy]

Hey Alex,

I haven't experimented with the 4046 much, (yet) but I did produce a clone of the Mutron Flanger that uses this IC for the clock.

Scroll down for the project file PDF.
http://www.diystompboxes.com/smfforum/index.php?topic=104959.0

The Mutron switches the 4046 VCO input between the LFO and a LED/LDR pedal controlled voltage source.

The LFO is unique in having stop/start controls which move the sweep points up and down instead of the standard manual and depth controls.

One thing that I like about the 4046 is the ease of incorporating a CV input.

[DrAlx]

Cheers Larry.  Just took a quick look.  That's quite an LFO!!!  I don't understand what the block with the LED's above the main square/triangle LFO is doing, but the opamps on the right with the diodes look like log and anti-log amplifiers, so I'm guessing the wave is begin warped to give the CV.
BTW your Muton Flanger clone sounds great and the pedal control is VERY cool.

[armdnrdy]

I don't understand what the block with the LED's above the main square/triangle LFO is doingBTW your Muton Flanger clone sounds

Thanks for the compliment Alex. I may lack knowledge, but I have an overabundance of tenacity. 

The section that you were referring to is actually a smart design.
When the Pedal/LFO switch is in the LFO position, a rate LED is activated. When the switch is in the Pedal position, the rate LED goes out, a LED by the switch goes on indicating Pedal mode, and the LFO is stopped.

Great care was taken in this design to make sure that it was quiet. There is a noise gate, compander, and having the LFO shut off while in the Pedal mode ensures that no residual LFO noise will be introduced into the circuit.

Another thing about this circuit....this is one of the best sounding flangers that I've heard. This flanger doesn't produce that tell tale metallic sound that's inherent in most flanger designs. The Mutron has a very rich tone.

[DrAlx]

Larry, I was just flicking through those build instructions of yours and saw the nice scope pictures of the effect of trimming at the BBD outputs
I only have a simple hand-held, single-channel scope with LCD output, so I can't get anything like the detail in those pictures.
So I have quick question for you.  In the picture where the output is unbalanced, can you remember how things were set up to get that picture.
Did you put the trim pot as far away from balanced as possible?  And when you had optimally trimmed things, how close to a 50:50 ratio were the two halves of the trim pot?  I'm trying to get an idea of how bad things are if you just use a couple of equal resistors.

The new vero layout is basically done (33 strips with 39 holes per strip).  I still need to calculate/simulate some filter stuff so not all component values have been decided, and I need to do a thorough check of the layout again before I start the build.

I've allowed for the option of using separate BBD bias levels (i.e. I've put in the tracks for the extra components).

I haven't yet included the option to allow a trimpot at each BBD output though, which is partly why I've brought this up again.  I could include the trimpots by default, and there is enough space on the layout to give the user the option of removing those tripots and using a couple of resistors instead.  But they'd be standing resistors.  Yuck.

[armdnrdy]

I only have a simple hand-held, single-channel scope

Only one channel is needed to adjust the balance. Connect single probe to wiper of balance trimmer which is combined BBD output.

In the picture where the output is unbalanced, can you remember how things were set up to get that picture.
Did you put the trim pot as far away from balanced as possible? 

I honestly do not recall. When I set up a balance trimmer in this fashion, the image that I posted is normally what I first see. The trimmer is always in a random position before calibration.

And when you had optimally trimmed things, how close to a 50:50 ratio were the two halves of the trim pot? 

Once again....I do not recall but, the flanger is within eyeshot. Today I'll pop the bottom off and take some measurements.

But they'd be standing resistors.  Yuck.

Before the advent of SMD components, a group of companies had built their whole product lines based on designs including standing resistors. I use them when necessary.

The Psalm of DIY:

Though I walk through the valley of the shrinking PC board, I fear no resistor, for I know I have standing resistor options. With my pliers and soldering iron in hand I feel comfort......

[armdnrdy]

Okay...

I took a resistance reading from the "balance" trimmer (T2) of the Mutron Flanger:

lug 1 to the wiper..... 3.1KΩ
lug 3 to the wiper..... 5.7KΩ
Across lugs 1 & 3 ..... 7.9KΩ

T2 is a 10KΩ 20% trimmer.

Now you may wonder why the first two readings add up to more (8.8K) than the total resistance of the trimmer.

So did I.  

I took the readings again and looked at the schematic to try to find why I would be getting "false" readings. I kept getting the same readings and I didn't see anything in the drawing that would cause an increase of resistance.

Finally...just to verify...I grabbed a 10KB pot and took the same readings.

lug 1 to the wiper..... 5.4KΩ
lug 3 to the wiper..... 4.5KΩ
Across lugs 1 & 3 ..... 9.6KΩ

Added total of first two measurements 9.9KΩ

So...I got a higher reading adding the first two readings than the second reading with the pot as well. I would call the trimmer readings good.

Seeing that there is a 2.6K difference when the BBD outputs are properly balanced definitely has me sold on using a trimmer instead of resistors.

Edit:
While I was taking measurements, I took a shot of a typical Boss type circuit BBD output. The BBDs two outputs are tied together with a 56K resistor to ground, connected into a .047µf capacitor. Notice the "ghosting" of the sine wave. It looks similar to the Mutron BBD output before it was calibrated.

[DrAlx]

Thank you Larry for taking the time to do this.  I really appreciate it.
The discrepancy in the measurements is down to 2 things.
In the case when you measured an isolated pot (not connected to anything else) the difference must come purely from errors in the meter.

For the pot in the circuit there is actually another effect. Looking at the schematic, there is a second path for the current to go between your measurements points (i.e. the two 100k resistors with a common ground).

Here's a worked example when you have exact 100k resistors and a 10k trim pot split exactly into 5k and 5K.

Taking a measurement across half the pot, the probe sees parallel resistances of 5k and 205k = 4.88k.
Taking a measurement across the whole pot, the probe sees parallel resistances of 10k and 200k = 9.52k
(Notice that 4.88k + 4.88k = 9.76k).

What if that pot ratio is split into a 6k and 4k ?  If you plug the numbers in you get this...
5.83k, 3.92k and 9.52k .  So we have the same sort of situation where the sum of the two smaller numbers exceeds the big one.
Note that the ratio of 5.83k to 3.92k isn't too far off the ratio of 6 to 4.

So going back to your values, it suggests your trimpot is in the ratio of about 1:2, and definitely nowhere near 1:1.
I'm fully convinced now and I'll put those trim pots in the layout.
I am guessing lots of other circuits get away without a trimmer because they have heavier filtering or the clock noise doesn't cause them a problem.
When you have a circuit with potential heterodyning, I think it makes sense to kill off as much of that clock noise as soon as possible.

Thanks again.