BBD delays and distorted repeats

[zeta55]

So, I just fired up a BBD delay circuit I've been working on. I've got no previous experience of BBD delays so I'm not shure what to expect.

My question is about distortion in the repeated delays, should I always expect them to some amount?

Here's a short soundclip I recorded http://www.mediafire.com/?8bx6pmbciz2 
Hmm, does not sound as distorted as I first thought, now when I listen to it

/Krister

[gez]

BBDs only have limited headroom.  Companders were used in the higher end stuff to get round this problem, but they're not widely available on these shores (might be different in the US/elsewhere).

A year or so ago, I modded a delay for someone who posted here asking for help.  I divided down the signal going into the wet chain and boosted it in the mixer stage (op-amp mixer).  It worked well enough (though was a tiny bit noiser).

[zeta55]

Thanks for the reply.

So I guess a non distorted repeats circuit does not exist in the analog world of delays then.

Actually I put in an SA571 circuit in there with the 3205:s, yes I glanced at other designs

I'm gonna take a closer look at the signals in an osciloscope, and try to tweak all them trimers.

/Krister

[Mark Hammer]

The audio quality of the repeats will depend on the sample rate, and also which repeat you're talking about.

Consider that the BBD relies on efficiently and faithfully passing a sampled signal voltage from stage to stage, storing it in a teeny capacitor each stage.  The capacitors leak a wee bit, and the fideltity of the stored voltage being passed on depends on the length of time it has remained in the previous storage cell.  The consequence of that is that the longer it has been since the stored voltage was passed from stage/cap to stage/cap, the less faithfully it replicates what was sampled at the input of the BBD.

So, expect less distortion at short delays rather than longer delays, simply because at short delays the storage cap is not permitted to leak quite as long before the sample is handed off to the next stage.  Keep in mind that it is not the delay time, per se, but how fast the clocking is, that is required to produce that delay.  Clocking a pair of MN3205s at double speed to produce 350ms delay will yield less distortion than clocking a single MN3205 at normal speed to produce the 350ms typically attempted in "classic" delay pedals, and half-clocking an MN3208 to produce the same delay time will sound pretty grungy.

Even where the signal is passed from stage to stage fairly quickly, there is a tiny bit of error in each of those handoffs.  Every time the signal is recirculated, it is as if it has passed through the same number of stages again.  So, a signal that has passed through a 4096-stage MN3205, and been recirculated twice (i.e., the first repeat and two additional repeats), has been handed off from storage cell to storage cell 12,288 times, each handoff resulted in a tiny bit of error.  If the delay time is set for max, then that's even bigger error x12288.

The typical strategy to address this is to use lots of lowpass filtering.  Ironically, it is this filtering that earns analog delays their reputation as having a "warm" sound.

I should also point out that cleanest sound from the BBD is obtained by having an ideal bias.  Being slightly higher or lower in the bias voltage reduces signal quality.  This is one of the reasons why the MN32xx series of chips were developed.  Because they could be run off +5v, it became possible to power the pedals with a 9v battery that could be regulated down to 5v, and provide the ideal bias voltage to the BBD for a longer period of time.  Previous generations of BBDs could be biased properly and run off 9v, but what happens when your fresh 9v battery declines to 8.4vdc?  Would your bias be ideal?  The capacity to run off 5v and the capacity for a 5v regulator to be adequately powered by 7vdc or more made it possible.

Though it is often the case that you can set ONE dc bias voltage and feed that to multiple BBDs, there is no guarantee that the same DC bias will be perfect for every chip.  And the more BBDs you use to produce the delay, the greater the chance of that happening.  If you check out the inside pic of the Maxon AD-999 here - http://filters.muziq.be/model/maxon/vintage/ad999 - you will see there are 8 MN3007 chips (with a different house number), each with their own bias trimpot, individually adjusted.  Chances are pretty good that attempting to use one bias voltage for all 8 chips would not have yielded as clean and distortion-free a sound.  From what I gather, the current issue uses 4 MN3008 equivalents, instead of eight MN3007s, to reduce the parts count and also the labour costs involved in tweaking 4 separate bias voltages.

So, while it is not an unavoidable fate to have distortion with an analog delay,there is risk of distortion.  Using more stages than you need, fast clocking, keeping the delay time short and the number of repeats modest, making sure every BBD used has the best individual bias voltage possible, and using lots of lowpass filtering at a lower cutoff frequency, will move you in the direction of cleaner sound.

[zeta55]

Thanks for the reply Mark.

That helps a bit in understanding things.
I'll get my focus on proper biasing and output balance for my three 3205:s, then pay some attention to the lowpass filtering.

Tweaking time

/Krister

[zeta55]

Pictures are fun
Thought I'd update this thread and post some pictures of my build. I've tweaked the circuit a bit, hopefully i'll get some new soundclipps up later this week.





/Krister

[Mark Hammer]

As my older son says, "Wow.  Simply wow."

VERY nice build.  You should be very proud.

[zeta55]

As my older son says, "Wow.  Simply wow."

VERY nice build.  You should be very proud.

Thanks.
I think it also sounds better now, a slightly warmer delay (filtering tweaks)   And having the circuit properly boxed up made it sound even better

/Krister

[Robin]

Nice post (and clip and photos).
Care to share schem? I have a few mn3***'s kicking around on little reverb daughter boards from cb radios -- looking for a build involving stringing them together.

[zeta55]

Nice post (and clip and photos).
Care to share schem? I have a few mn3***'s kicking around on little reverb daughter boards from cb radios -- looking for a build involving stringing them together.

Thanks.
If there's interest, maybe i could get a project posted.

/Krister

[greigoroth]

There is interest! Sounds awesome!