The so-called bucked-brigade device (BBD) delay line generator is a somehow quirky and really unique technical design. Such devices are built upon analog components entirely, but being discrete in time they are halfway digital. Their analog input voltage samples are stored and moved through a line of capacitors one step after another and hence the name comes from analogy with the term bucket brigade: a line of people passing buckets full of water.
Some prominent representatives are the Ibanez AD202, the Dynacord VRS-23 and the Ace Tone EH-50 just to name a few. But why is their frequency bandwidth so much limited? Just as an example, the Ibanez AD202 specification states 2kHz as an upper frequency limit in it’s delay mode. Let’s have a closer look at the AD202 technical design. In the Ibanez AD202, there are 3 BBD chips utilized in series to obtain the delay line: Two Panasonic MN3005 and one Panasonic MN3007. The MN3005 is a 4096 stage long delay and by using a clock speed of 10kHz a maximum of 4096/(2*10kHz)=204.8ms delay time can be obtained. Similar, with the MN3007 – which is a 1024 stage device – there could be delay times from 5.12 up to 51.2ms achieved.
Since a BBD based device is discrete in time, the aliasing effect has to be dealt with. This means that frequencies above the Nyquist frequency (which is half the sampling frequency) needs to be eliminated. So, aliasing and reconstruction lowpass filtering (LPF) has to be used and in the digital domain one would use a rather steep digital LPF at equal or slightly below Nyquist frequency. But since we are in analog land with the AD202 to that time, a common practise was to use a 2- or 3- pole analog lowpass filter (e.g. in a Sallen-Key topology) which is not that steep by far. Therefore a good conservative design was that the clock frequency should be no less than 3 times the roll off frequency of the reconstruction filter which, in the case of the MN3005, immediately implies a corner frequency somewhere below 3.3 kHz.
This comes down to a tradeoff between sound quality and the length of BBD delay line. A longer delay time implies a further limited frequency response. To compensate this, two or even more BBD chips could be setup in series but this comes to a price too: The more buckets involved, the more gain loss (and other unwanted side effects like distortion) occurs which results in a higher noise floor just as an example. In case of the AD202 this results in a noise floor at around -85dB.