Bucket Brigade Delay ICs (BBDs)

BBDs

Many analog delay pedals are designed around a type of IC called a "bucket brigade delay", or "BBD".  These are an interesting type of IC.  They are driven by a "clock" circuit, which causes them to sample an audio signal at a particular clock rate.  The clock rates are typically in the thousands to tens of thousands of times per second range (one thousand times per second is 1 kilohertz, or 1K Hz), but can vary an order of magnitude or so either direction.  When the IC samples the audio signal, it puts a copy of the signal voltage into storage in a "bucket".  The IC contains a long line of these buckets, connected from one to the next, from the input to the output of the IC.  Each time the clock ticks, all the buckets dump their contents into the next bucket down the line, and get refilled from the bucket before them in the line.  Once the clock has ticked enough times, the signal that was stored in the first bucket finally makes it to the last bucket and to the output.  The number of clock ticks this takes, along with how fast the clock ticks, determines how long it takes for an incoming signal sample to make it all the way through the IC.  That's the delay time for that IC.  That's really about all a BBD does - take a sample and copy it many times until it comes out, delayed.

The design is interesting because it is sort of analog, but also sort of digital.  Digital technology samples inputs at some clock speed.  Then the value sampled is converted to a number that represents a range of actual values, which is why digital sampling can't be generally guaranteed to be 100% accurate.  But BBDs don't generate a number to represent the sampled value.  They actually store the sampled value.  Using the actual values is what analog technology does.  BBDs might be described as "analog sampling".  But something else happens as a side effect of this sample-and-copy design.  Each time the stored value is moved from one bucket to the next, there is the possibility that the value won't be copied and stored exactly.  Tiny differences in the buckets and circuitry connecting them cause slight inaccuracies.  In addition, the voltage can degrade during the brief time it is stored in each bucket, again causing inaccuracies in the samples.  The more buckets in the chain, the more inaccuracies creep into the signal sample.  After enough copies, the final sample may not sound much like the original sample at all.  It's analogous to the childhood game of "Telephone", where the message gets mangled as it passes from one person to the next.

 

This growing inaccuracy in the sample gives rise to one of the major characteristics of analog delay, loved by many, hated by many.  The inaccuracies mount up and make the repeats of the signal sound fuzzy and noisy.  It sounds like each repeat comes through a little distorted, and like it is riding on top of a little cloud of noise.  In an IC with fewer buckets and a faster clock, this many be less audible and the repeats may sound clear and clean.  With more buckets and a slower clock, they can become garbled.  Of course, the number of buckets can be controlled by using different models of ICs, and the clocks can be adjusted to a speed that lowers the amount of generated noise due to signal decay.  When the IC selection and clock tuning are done properly, repeats may be clear or they may sound warmer with that little cloud of noise.  Some players really like the warm, realistic repeats, particularly when the noise is well-managed.  Others can't stand the tone coloration or the noise and prefer the sound of a crisp, clean digital delay.  There's nothing wrong with a good digital delay, but we'll stay on topic here and continue with our analog delays and BBDs.

BBD Analog Delays

There are many sources of detailed information about the design of BBD-based analog delays.  We aren't going to go into the details of all that, but we do need to touch on it.  The BBD may be the key component in an analog delay that actually causes the delay effect, but there are many other components in those circuits.  In fact, among effects circuits, analog effects that use BBD technology such as delay, chorus, flanger, and others are among the more complex effects circuits to build and tune.  Each circuit is designed to have its own characteristic tone, beyond just the delay time.  The circuits often contain buffers.  All of them contain filters that manipulate specific frequency ranges.  Some have additional controls and effects added for wet/dry signal mixing, modulation, and other features.  Most contain considerable manipulation of the signal to lessen the impact of noise.  Different brands and models of BBDs have their own tone coloration.  There are many factors that all contribute to the final sound of the repeats that you hear.

What is the point?  The point is that all BBDs do the same thing, pretty much the same way.  Yes, they are all designed and manufactured a little bit differently, but they are all very similar in their overall design and manufacturing.  Their designs will also allow for different ranges of clock speeds to drive them, and they will operate at different voltage levels.  And they all are surrounded by specific analog delay effect circuits that are manipulating the signal both before and after the signal goes through the BBD.  The point is that when you hear an analog delay, you cannot tell how much the BBD is responsible for the final tone.  Many other factors have big impacts on the signal.  But it is also true that the different brands and models of BBDs, even individual specific BBD chips, do contribute to the result.  It is very difficult, bordering on impossible, to generally judge BBDs and how they sound by listening to them in a particular analog delay circuit.  Of course, that doesn't stop people from trying!  A good deal of opinion gets mixed in and sometimes facts are a bit thin.  Let's look at some facts about what affects the sound of a BBD in an analog delay circuit.

Brands of BBDs

Let's get the least useful and most controversial out of the way first - brand.  There were never that many BBD manufacturers, and some of the brands were really the same devices.  For many years, the market was dominated by BBDs from Reticon and a small group of Japanese manufacturers we'll refer to as "Panasonic" since that was the company that developed the BBDs sold by all the members of that group.  Excellent products were designed around both brands of BBDs, and around several different BBD models from each brand.  There are preferences and opinions about both brands - we won't go there.  But, it is fair to say that even considering all the other factors that influence the sound, effects made with Reticon BBDs often sound a bit different from effects made with Panasonic BBDs.  Reticon chips probably enjoy an edge in opinion as the better of the two brands. 

Reticon ceased production of BBDs many years ago.  Due to their popularity and dwindling supply, today these chips are almost completely unavailable.  Occasionally you may find one for sale online, generally with a price of well over $100 per chip.  That makes them a prime candidate for counterfeits, used chips selling as new, used chips that are dead sold as "untested", and all sorts of shenanigans.  At GT we don't even attempt to source these any longer.  The chances of being ripped off are very high, and the supply is virtually non-existent.  The time involved and high cost would make any modules using a Reticon chip extremely expensive.  It just isn't worth it, and we do not offer any modules with Reticon chips.

Panasonic continued manufacturing BBDs for much longer, and in much larger quantities.  But even so, the demand caught up with the available supply.  The Panasonic chips are now rising in price and becoming less available.  Counterfeits are even more common with Panasonic chips than Reticon, making it difficult to source good BBDs from reliable suppliers.  But we are still able to get them and we generally have the ones in stock that we use in our modules.

The Reticon and Panasonic designs are not pin-compatible.  You cannot substitute one for another without making changes to the effect circuit.  In fact, even different models of BBDs from the same brands are not generally pin-compatible.  As a result, many of the vintage effects that use these BBDs are now very expensive and becoming harder and harder to find in good working condition.  That's a shame, because some of those are outstanding effects.  So at GT we have taken some Reticon BBD based effects and made the circuit changes necessary to build that effect with equivalent Panasonic BBDs.  Some of those modified effects sound almost identical to the originals.  Others sound great, but slightly different.  We've only done that with a small number of effects, because usually there is a Panasonic BBD based effect of that type that also sounds great.  We note these changes in our module descriptions.

Number of Buckets In A BBD

We've already discussed how the number of buckets affects the sound.  The more buckets, the longer maximum delay you can have.  The more buckets, the more noise and signal degradation you get.  BBDs come in a variety of "sizes", as different applications need different amounts of delay.  Common sizes include 256, 512, 1024, 2048, and 4096 buckets per chip.  Short delay effects like flangers and chorus normally use a single chip with 256 or 512 buckets, or maybe 1024 with a fast clock speed.  Delays normally use from 1 to 8 chips in sequence, with 1024, 2048, or 4096 buckets each.  The majority of delays use only 1 or 2 chips, though.

Clock Speed of a BBD

The clock isn't usually part of the BBD.  Some BBDs have special clock chips that are meant to drive specific BBDs.  Others use clock signals generated by other types of circuits and chips.  Regardless of how the clock signal is generated, it has a particular frequency which determines how fast the buckets in the BBD pass along their signal samples.  Adjusting the clock speed between the lowest and highest speeds supported by a BBD model gives you the range of delay that is possible with that BBD.  Typically, the BBD gets noisy near the slowest and fastest settings, so there is an actual range that is narrower and sounds best.  The range can be very wide, and some BBDs can easily be overclocked outside their specs with no problem.  So the first impact of the clock is the actual amount of delay.

Without getting into technical details, we'll just say that clocks can also affect the BBD's sound in another way.  Clock pulses are a signal that changes from "hi" to "lo" at the clock's frequency.  Ideally, these changes should be perfect and immediate, but they aren't.  There is some rise and fall time.  This can adversely affect the performance of the BBD.  Clocks are normally selected and clock circuits are designed to minimize this, but it still happens to lesser or greater degree, and is often accentuated at higher clock speeds.

Clock and BBD Noise

We already discussed that you can get some BBD noise by the accumulation of the inaccuracies that occur when the buckets of signal are shifted with each clock tick.  There's another source of BBD noise, too.  BBD chips have an operating voltage range (described below).  As long as you stay in that range of voltage, your signal doesn't get clipped or distorted by the BBD chip.  Due to manufacturing technicalities, that voltage range varies a little bit from one chip to the next, even of the same brand and model of BBD made on the same manufacturing equipment in the same manufacturing runs.  In the circuits that use BBDs, there are usually some trim pots that are used to adjust the signal to those little manufacturing differences in the BBDs.  This is called "biasing".  It centers your signal in the operating range of that exact chip.  Mostly you adjust the pots by ear.  At extreme settings of the pot, there may be no sound at all.  At other settings, the sound will be distorted or muffled.  It is distorted because it is falling partially outside the best operating range of the BBD.  At some very touchy setting of the trim pot, you will get the incoming signal "centered" in the sweet spot of the BBD's operating range and the distortion is completely gone or minimized, depending on how well the circuit is designed and how well the BBD is behaving.  Some BBDs may need to be discarded if they can't be dialed in nicely.  This type of noise is usually hiss or dark distortion in the sound.  With patience, you should be able to adjust the trim pot (need to know which one it is) to a better setting.  It should come already set to a good sound, as the manufacturer should set it and test it to make sure the BBD is good and working properly.  Those trimmers can be really, really touchy and difficult to dial in, so have patience.  Before you adjust it, note the current setting so you can get back to it if you get lost.

If you have used an effect with BBD technology, you have probably heard clock noise.  Clock noise can sound like a very annoying ticking sound or a high-pitched whine.  What's going on with that?!?!  It's a little messy, but it's good to understand it. When voltages are changed abruptly, there can be an audible "pop".  We discussed one of those situations in our Switching discussion, where two different voltages get connected to each other.  But with BBDs, there is a more annoying type of popping, and it is difficult to avoid.  The BBD needs a clock to turn off and on to cause it to shift the contents of all the buckets.  The clock may be a special chip designed to be used with the BBD, or a circuit that has a similar effect, called an LFO (low frequency oscillator).  In both cases, a voltage is produced and a clock "tick", is a sudden change in the voltage from "off" to "on".  If you looked at that voltage with an oscilloscope, you would see a square wave that goes from 0V to the "on" voltage very suddenly, stays there very briefly, then drops suddenly to 0V again.  The "clock rate" is how many times that happens in a second.  Each of those voltage changes can potentially be heard.  At slower clock rates, you may almost be able to hear each one, producing a "ticking" or "clicking" sound.  At higher clock rates, they tend to blur together and cause a high-pitched whining sound.  You've probably heard them, and like everyone else, hated them.  Yup, tone suckiness of the worst type.  In addition to a trimmer for biasing the BBD, there is often another trimmer that helps limit the clock signal in ways that help manage the ticking.  With a properly designed and built BBD circuit, with properly functioning BBD and clock, with the trimmers properly set, you won't hear this noise.  But few BBD circuits and components are that perfect, so you might get stuck with some.  That's part of the "charm" of analog BBD sound!

What can make the clock noise worse is the length and routing of the wire(s) that carry that signal.  Normally those wires are made as short as possible and kept as far away from the rest of the circuit as possible.  Those voltage changes on that wire can act like little radio transmissions, transmitting those clicks into other parts of the circuit.  That clock wire is not part of the audio path.  It just controls the BBD, which is in the audio path.  Any ticking noise is coming as a result of the broadcast of that noise into the audio portion of the circuit, which acts as an antenna to pick it up.  Good board design is critical, as that often determines the routing of the signal.  If that signal actually does go into a wire, such as the wires to and from your DELAY control, those wires should be shielded and grounded to prevent the wires from broadcasting that noise.  But the pot of that DELAY control probably isn't shielded and will broadcast some noise.

Some simple circuit design decisions can also have a big impact on clock noise.  Some circuits make the clock signal strength lower so that it can't broadcast so much.  Some circuits user components that operate at lower voltages so that the voltage differences are smaller and less audible.  It's virtually impossible to get this level of information about a BBD circuit, but it almost certainly contributes to different opinions about and experiences with different BBDs and different BBD effects.

It's very difficult to get rid of all the BBD and clock noise.  And it can vary in amount depending on your control settings and other aspects of your signal chain.  That's part of the love/hate relationship for these analog delay devices.  But two more points before we move on. 

First, digital circuits do voltage sampling and digital clocking like this all the time.  Why don't they have the same noise problems?  Without getting too wordy about it, it is because they don't actually manipulate the audio signal directly.  They use a numeric value instead of the actual voltages.  There is no noise in math.  Sophisticated digital circuits have crazy great ways of doing math and staying away from noise.  HOWEVER, those digital clocking circuits do generate noise just like the analog clocking circuits do.  And other devices, like your analog effects, can pick up that digital noise and sound awful.  In particular, digital switching power supplies can pollute the power of everything they are attached to with digital noise.  This can happen at an amazing scale.  If you plug a digital switching power supply into a wall outlet, everything else on that circuit or in the building can be affected.

Here's an example.  In a room at GT where we build our modules, we use some computers for design, email, and a million other things, just like everyone else.  To prevent damage to desktop computers, we plug them into UPS units so that they will have power to shut down gracefully if there is a power failure.  We also sometimes have laptops in the room, plugged into their power adapters.  And smartphones, sometimes plugged into their power chargers.  We had horrible noise problems when we first started building our effects.  We thought we were doing something wrong.  It took a long time and a lot of difficult work to find that the UPS units, laptop power adapters, and phone chargers were all using digital switching power supplies, which were making all of our power extremely dirty.  Unplug all that stuff and most of the noise goes away.  Yeah, bad deal.  It's actually even worse than that.  In figuring all this out, we found out that it isn't just audio devices like our effects and amps that are affected.  That digital power supply noise is really bad for motors, like those in your refrigerator, air conditioners, and lots of other appliances, perhaps reducing their service life by years.  Since that circuit with the noise is connected in your breaker box to other circuits in your house, that noise can go from one circuit to everything in the building.  There are ways to deal with that, but they can be very expensive and/or impractical.  The real learning is DON'T USE DIGITAL SWITCHING POWER SUPPLIES!  Some digital switching power supplies are being better designed, so maybe this problem will go away eventually.  But it is real now, especially with cheaper electronics.  Digital switching power supplies are much cheaper than the ones that use transformers, so beware.  The point of all that is that this isn't just a BBD problem.  It is mostly a digital problem, but BBDs happen to have that one digital aspect of their operation.

The second small point before moving on is not directly related to BBDs either, but can be more apparent in the types of effects that use BBDs.  We'll stick with delays for this discussion, but any sort of effect that adds copies of your signal to your signal, or even effects that add "motion" to your signal, like flangers or phasers, can have this problem.  A delay adds a copy (or multiple copies) of a note you played back to your signal, giving you the echoes of that note.  As we know with distortion or other gain effects, any time you add to a signal, you are probably going to add a little noise along with what you are trying to add.  As long as you are playing and adding new original signal, you don't really notice the extra noise being added along with those echoes.  But if you stop playing and let the echoes die away, you may hear some little blips of pure noise replacing the nice echoes.  That is just the base noise in your signal getting manipulated and turned into echoes just like your original notes.  But it can be annoying.  Reducing the number of repeats can help.  Adding noise gates or other noise reduction in you signal path can also fix that problem.  Many nice BBD circuits make use of some extra circuitry to help eliminate this noise, as well as the noise that comes along with your echoes.  This can get a bit complex, but let's visit some mostly-accurate examples to get a flavor for what might happen.  

Suppose you are worried about hiss, which is very common.  It is high-frequency noise and can creep into your circuits in various ways.  One way to get rid of that noise is to simply chop off all the high frequencies.  A bit drastic, perhaps, but it would get rid of the noise, along with the high-end treble and presence of your signal.  But what if you were clever and took all your signal and cut the frequencies to say, half of their original amounts before you went into some noisy circuit segment.  And when you come out of that noisy segment, you double the frequencies to restore your original sound.  What's the purpose?  The trick is that the noise would still be added at its original frequency, which would be higher than all original sound that is now reduced to half its frequency.  When you double everything to restore your original sound, you also double the frequency of the noise, putting it so high that human ears can no longer hear it.  Cool, but maybe your dog won't like it.  Or maybe just before you double it back to the original, you just chop off all the high frequency content, which is mostly just the noise, then restore your original sound?  Cool, and your dog will thank you.  Or maybe instead of chopping your original sound down and restoring it, you just give the high-end a big boost.  After the high-end noise gets added to your boosted highs, then you reduce (but not completely cut) the highs back to their original level.  The level of noise is reduced and the level of your original highs is dropped back to where it was.  Cool.  There are various such ways of manipulating your analog signal and handling reducing noise.  None of them are 100% effective, but some of them are pretty good.  Dolby and dbx and other hi-fi audio technologies use these types of tricks.  Filters may be built into the BBD circuit, or perhaps nice compandor chips will be employed to filter, squish, and restore your original signal while reducing the noise.

So what is the point?  It isn't just the BBD and clocking that make up that BBD effect.  There may be some other clever circuitry used with them that greatly impact how the overall effect sounds, particularly the noise level.  Just like with the clock noise noted earlier, it is nearly impossible to get this level of information about a BBD circuit before you select one.  But your ears won't lie to you.  The BBD and/or clock noise may be heard, and almost certainly will to some extent.  Since we're building these circuits for you, we actually do know which of the circuits have more or less clever circuitry or adjustment points that may help manage BBD effect noise levels - just ask us, if it is important to you.

BBD Voltage

BBDs run at some range of voltages, depending on the model of the BBD.  Let's say for example that we have one that runs at +9V with 0V ground.  That means your audio signal needs to be in the range of 0V to +9V.  Audio signals are AC, varying from some positive to some negative value.  The audio signal gets "biased" by adding a positive DC voltage to keep the signal from going "out of bounds" below 0 and being clipped off.  To try to optimize the performance, the DC bias will probably be +4.5V in our example.  That has the effect of shifting the audio signal to swinging above and below +4.5V instead of 0V.  If the original signal stayed within +4.5V to -4.5V, then the highs and low will be "in bounds" and won't get clipped or distorted to stay within the 0V to +9V range of the BBD.  Guitar pickups produce voltages that are typically less than +/-1V, so everything is OK, right?  Maybe.  The potential problem is that many effects change the voltage, some of them change it a lot.  It is possible to get a voltage that falls outside the +/-4.5V that the 0V to +9V BBD can support without clipping.  If that happens, there's nothing that can be done.  The signal will be clipped.  Since circuits aren't perfect, some 9V BBDs will clip to a narrower range, leaving you less headroom for your signal.

Different BBD models will run at different maximum voltages, with 9V and -15V being most common.  A 15V BBD will typically give you a cleaner sound with everything else held the same (size, clock speed, etc).  That's where we hit a problem.  The Reticon chips are gone, leaving only the Panasonic BBDs.  The early Panasonic BBD models ran at -15V and sound great.  However, it is generally desirable in the IC world to use less power, generate less heat, etc.  Panasonic replaced their -15V BBD product family with a +9V BBD product family.  That is why some older delays sound better than newer ones - the newer ones probably run at a lower voltage and are therefore more likely to clip and distort.

What BBDs Are Available?

 

 

 

 

 

 

 

As the rise of digital technology drove down the demand for analog circuits, Reticon and Panasonic both stopped producing BBDs.  Manufacturers of analog delays stocked up, but eventually their supplies ran out.  There was a time when it appeared that the days of analog delays were over.  As popular as analog delays are, the market reality is that not enough were sold to make it worthwhile for manufacturers to keep producing them.  Even if every player had to buy several, the quantities just weren't there.  Then economics and markets saved the day - twice.  The demand for analog delays was still strong, so the stocks of BBDs were depleted and BBD prices began rising dramatically in secondary IC markets.  The popular and much-loved Panasonic MN3005 BBD was fetching in excess of $50 apiece, and most of the Reticon BBDs were bringing $100 each or more.  Even in small quantities, those chips can be made for a lot less than that, there was a potentially lucrative business for someone that could make and sell these at some "reasonable" price.  So that's what happened.

A couple of companies jumped in and started producing BBDs again.  It appears that either the details for the Reticon BBDs are either lost or the owner of their rights isn't interested in having them made again.  Reticon or Reticon clones have not become available.  Some of the new BBDs are made to loosely match Panasonic models, at least by being pin-compatible and having the same sizes.  The performance of those BBDs is inconsistent.  Some sound reasonably good, others aren't nearly as good as the original Panasonics.  They are usable, but often don't sound quite the same.  For some years, that was the best you could get, other than scrounging up old BBDs.  Then something almost miraculous happened in the BBD world.  The MN3005 was brought back to life.  These new MN3005s are manufactured to be exactly the same as the original Panasonic MN3005 BBDs.  They aren't exactly identical in specs (it's a little doubtful that the originals were all in spec...), but we find them indistinguishable from the originals.  They are as much like an original as another original would be - within the same general manufacturing tolerance.  There are very few dealers, as most of the supply is going directly to the big pedal manufacturers.  They also aren't cheap, but the price is lower than that of an original Panasonic on the secondary market.  Now we can once again build great analog delays, although at a cost that is higher than we would like it to be.

Below is the list of BBDs that we use in our modules and keep in stock.  When we run out of original Panasonic MN3005, it is unlikely we will ever stock them again.  The XVive MN3005 sounds identical, is less expensive, and is more available - a no-brainer.

The "original" MNxxxx BBDs were made by Panasonic and other Japanese manufacturers.  XVive now produces the "new" MN3005.

The Vxxxx BBDs are made by Cool Audio.

The BLxxx BBDs were made by Shanghai Belling, though they seem to have stopped manufacturing them now

Note that the Max Voltage isn't the absolute maximum the device can use.  It is the maximum we use to avoid damage and increased distortion.  The actual maximums are a couple of volts more.

BBD	Size	Delay (ms)	Signal/Noise (typ)	THD (typ)	Max Voltage	Availability
MN3007	1024	5.12 - 51.2	80db	0.5%	-15V	out of production, reasonably available
MN3207	1024	2.56 - 51.2	73db	0.4%	+9V	out of production, reasonably available
V3207	1024	2.56 - 51.2	73db	0.4%	+9V	current production, limited distribution
MN3008	2048	10.24 - 102.4	78db	0.5%	-15V	out of production, difficult to source
MN3208	2048	10.24 - 102.4	71db	0.5%	+9V	out of production, difficult to source
V3208	2048	10.24 - 102.4	71db	2.5%	+9V	current production, limited distribution
BL3208A	2048	10.24 - 102.4	71db	2.5%	+9V	out of production, limited availability
BL3208B	2048	10.24 - 102.4	71db	2.5%	+9V	out of production, limited availability
new MN3005	4096	20.48 - 204.8	75db	2.5%	-15V	current production, limited distribution
original MN3005	4096	20.48 - 204.8	75db	1%	-15V	out of production, very difficult to source
MN3205	4096	20.48 - 204.8	67db	0.8%	+9V	out of production, difficult to source
V3205	4096	20.48 - 204.8	60db	0.8%	+9V	current production, limited distribution