Question About Low & High Pass Filter On BBD Based Chorus

[Agung Kurniawan]

Hi eyeryone..
there is somethink that still disturb me since I have finish my first chorus pedal.
some chorus schematic show a Low Pass Filter And High Pass Filter before and after the BBD on it. I cant figure what does they do there.
can someone please help me to understand that one?
Thanks....

[teemuk]

Bucket brigade delay's operation is based on switching. The stored analogue signal sample is moved along the line of capacitors, one step at each clock cycle. Thus according to Nyquist-theorem there is an upper frequency limit for the input signal after which "aliasing" effect will begin to distort the output signal. A low-pass filter is therefore introduced to cut upper frequencies above the limit.

The same operating principle of switching causes output signal to contain high frequency switching artefacts, which must be filtered out to "reconstruct" the signal.

[ElectricDruid]

Some chorus effects use high pass filters to make the sound more "airy". The woozy detuning is less noticeable if you cut the low end out, and you keep the chorused harmonic content that gives the richness.

So the lowpass filters are probably there for technical reasons as Teemuk said, but a highpass would probably be for tone shaping.

HTH,
Tom

[Agung Kurniawan]

Bucket brigade delay's operation is based on switching. The stored analogue signal sample is moved along the line of capacitors, one step at each clock cycle. Thus according to Nyquist-theorem there is an upper frequency limit for the input signal after which "aliasing" effect will begin to distort the output signal. A low-pass filter is therefore introduced to cut upper frequencies above the limit.

The same operating principle of switching causes output signal to contain high frequency switching artefacts, which must be filtered out to "reconstruct" the signal.

Ok, I get it now
How if I ignore the low pass on the BBD input?
Since Im using distortion for my base sound, would leave the low pass be harmful for the BBD it self?
My own design chorus just leave the low pass section so I can get the natural distortion sound to get in to the BBD

[Mark Hammer]

The lowpass filtering used would depend on the clock frequency used which, in turn would depend on the delay time aimed for and the chip used.  If the clock is going to come down to 20khz at its slowest , the filtering is probably going to need to begin a relatively steep rolloff around 5khz or so to make sure that none of the clock whine is audible at the output and  aliasing is kept to a minimum.

As Druid notes, reducing bass component in the delay signal makes the pitch wobble less noticeable.  This is particularly important if using a chorus with a bass guitar, where the note fundamental needs to be clear and precise for other players to play relative to.  Rolling off the low end (highpass) of the delay signal still leaves the fundamental in the dry path untouched, while still providing the comb-filtering of the harmonics.  So, lots of swirl, none of the wobble.

I will also note that, among chorus pedals that do not use companding (which is most of them), the noise-control strategy used is a sort of poor-man's Dolby, where the treble is emphasized on the way to the BBD, and then de-emphasized in complementary fashion at the mixing stage, such that any noise coming from the BBD path is attenuated at the output. 

[ElectricDruid]

Ok, I get it now
How if I ignore the low pass on the BBD input?
Since Im using distortion for my base sound, would leave the low pass be harmful for the BBD it self?
My own design chorus just leave the low pass section so I can get the natural distortion sound to get in to the BBD

The "distortion" caused by aliasing isn't the same as the "distortion" that you're using. Aliasing causes a weird effect where frequencies that are above the Nyquist limit "bounce back" and become lower frequencies. Since they have moved, they're no longer harmonically related, and they sound metallic and unnatural. It's not a nice sound.

What I'm really saying is that aliasing distortion should definitely be avoided if at all possible. That means leaving out the anti-aliasing LPF isn't generally a good idea, unless perhaps you know the clock frequency will remain very high, in which case you might get away with it.

It's particularly an issue if you're using a rich sound like a distorted guitar as an input, since the signal has a lot of harmonic content that can potentially produce aliasing. Since it's caused by frequencies above the Nyquist limit, it will be worse for higher notes, and may not appear at all lower down.

HTH,
Tom

[teemuk]

the treble is emphasized on the way to the BBD, and then de-emphasized in complementary fashion at the mixing stage, such that any noise coming from the BBD path is attenuated at the output.

This makes sense since BBD chips are inherently noisy. Expanding the dynamic range before delay line, and compression after it, is one method to reduce noise. Enhacement of higher frequencies before delay line, and de-enhancement post it, is another.

If you look at it, many bucket brigade delay circuits also contain some sort of muting or "noise gate" circuit, which is toggled when no signal is detected. Roland Jazz Chorus series amps detect if any signal is present at their speaker terminals, CE-1 detects if any signal is present at BBD input terminals. Nevertheless, "no signal" -condition mutes the output, and therefore most of the noise as well. (Noise is harder to detect within other signal content but very apparent without it). As trivial as it may sound, I believe Roland was still able to patent the idea few decades ago when analog BBD's were still considered "hi-tech".

[Agung Kurniawan]

Ok, thanks for all of your great explanation guys..
So then in my case the final conclusion would be:
Leave the LPF is a mistake that may provide many disavantage to my signal, right?

When my first time plug down this chorus, I can hear a clock ticking up. But when i check the soldered side and resoldered it, the tick gone.
But i can hear a his(not so noisy) and when i strike the string to hard, i can hear my clean signal distort

[Rob Strand]

Leave the LPF is a mistake that may provide many disavantage to my signal, right?

You shouldn't remove the low-pass filter before or after the BBD.  You need both.  The cut-off frequency of the LPF depends on the lowest clock frequency going to the BBD.

When my first time plug down this chorus, I can hear a clock ticking up. But when i check the soldered side and resoldered it, the tick gone.

Clicking is often related to the LFO (the low frequency oscillator that produces the sweep).   Sometimes it is layout sometimes you need to add a bypass capacitor close to the LFO.  In the layout use separate power tracks for the LFO and audio.   An keep the LFO wires and tracks away from audio.  Another possible cause is using the same 4.5V (or whatever) bias divider for the audio and LFO. 

BBD distortion is often cause by incorrect biasing of the BBD chip.  On many designs you see a trimpot to set the BBD bias.

But i can hear a his(not so noisy) and when i strike the string to hard, i can hear my clean signal distort

Distortion on the clean signal is often a problem with too much gain in the clean signal path, or there is something wrong with the bias divider, the one that supplies 4.5V.

Disconnect the output from the BBD path to prove that the distortion is in fact in the clean signal path.

Look at the Boss pedals for tips as they tend to do things to correct way.

[Agung Kurniawan]

You shouldn't remove the low-pass filter before or after the BBD.  You need both.  The cut-off frequency of the LPF depends on the lowest clock frequency going to the BBD.

Is that any formula to count the lower clock freq?
Im using Boss-tipe LFO with CD4046

[PRR]

You need the LPF front and back or you get queer sick-robot sounds at low clock rates.

If need heavy LPF for low clock rates, the sound gets "dull" and "bass heavy". Delaying or reverbing bass makes little sense anyway. So a designer/tweaker may opt to shave the bass a bit for a more balanced (or less unbalanced) tonal sound.

You will find that most designers don't add "stuff" without a reason. True, your needs may be different, and experimentation is always good. "Queer sick robot" IS "a sound". But conversely, don't plan to discard "stuff" without thought or experimentation.

[thermionix]

"Queer sick robot"

And now you have a new band name!

[Rob Strand]

Equations:

For the BBD (serial mode):
f_sample = f_clock
f_nyquist = f_sample/2
t_delay = N / (2* f_clock)

For the filter, based on the minimum clock frequency,
f_filter < f_nyquist   ;  this is fairly useless limit because the filter will have no attenuation at f_nyqust

A better upper limit which will give some amount of attenuation at f_nyquist (with a 4th or 5th order filter) is,
f_filter  < 0.3 to 0.35 f_sample

You can also play games by using lower order filters and using a lower cut-off.

This estimate is usually fairly high and many flanger and chorus units will use a cut-off based to some extent on the sound of the unit.    Typically ending-up 6kHz to 8kHz.

For the VCO:
The boss chorus LFO produces a minimum voltage around 1 to 1.2V.
From the 4046 datasheet, and based on your "R1" and "R2" resistors and the timing cap of the 4046, you can estimate the minimum f_clock.   A better way would be to inject a DC voltage of 1 to1.2V then measure the 4046 frequency.

[ElectricDruid]

BBD distortion is often caused by incorrect biasing of the BBD chip.  On many designs you see a trimpot to set the BBD bias.

It's also often caused by people assuming that "headroom" equals "supply voltage". This is notably not true with BBDs. Running the MN3207 on a 5V supply, I found the maximum signal level through the BBD before distortion was 1.2V, or about a quarter of the supply. That'd give you a little over 2V signal level on a 9V supply. If you're not expecting that, it's easy to overdrive the BBD.

HTH,
Tom

[bloxstompboxes]

"Queer sick robot"

And now you have a new band name!

Nah, then you would have the gay rights and politically correct robot people after you. Even though PRR's use of it was in the "weird" definition of the word.

[Mark Hammer]

BBD distortion is often caused by incorrect biasing of the BBD chip.  On many designs you see a trimpot to set the BBD bias.

It's also often caused by people assuming that "headroom" equals "supply voltage". This is notably not true with BBDs. Running the MN3207 on a 5V supply, I found the maximum signal level through the BBD before distortion was 1.2V, or about a quarter of the supply. That'd give you a little over 2V signal level on a 9V supply. If you're not expecting that, it's easy to overdrive the BBD.

HTH,
Tom

True.  The BBD itself has much less headroom than the support circuitry around it.  So using a 12-15V supply may well improve the headroom of any op-amps prior to and after the BBD.  But even at 9v, those same op-amps probably had more headroom than the BBD at just about any supply voltage.  So jacking up the headroom of the op-amps is unlikely to achieve any noteworthy improvement in disortion-immunity in the BBD.

[Agung Kurniawan]

I found my chorus distort with 9V, but clean with 15V

[Mark Hammer]

Did you rebias when switching supply voltage?

[Agung Kurniawan]

Did you rebias when switching supply voltage?

No
Did i need to rebias?

[Agung Kurniawan]

How big the voltage did MN3002 can handle?