heres some interseting reading on the subject courtesy of pedal guru Mark Hammer.
There are a few ways that clock noise is reduced in analog delays.
Most BBD chips have two complementary outputs, A and B. In some cases (e.g., MN3010, SAD1024), the chip will be in two halves, each with their own A and B output. These two outputs produce complementary signals. As you know (I hope), *two* opposite phase clock signals have to go to the BBD chip. Each clock signal is responsible for one of the BBD outputs. When the two outputs are mixed in a perfectly balanced way, the opposite-phase clock signals from each output cancel each other, in exactly the same way that the two coils of a humbucker pickup add their audio signals together but cancel their hum signals. Substitute clock noise for hum in the BBD and you have the same thing.
To produce the cancellation, manufacturers use a variety of methods. Some use a balance trimpot on the two BBD outputs. Others use two equal value fixed resistors. Others just tie the two outputs together (essentially zero ohm resistors).
There are problems with each approach. Trimpots gets dislodged over time. Equal value resistors may not be perfectly equal. Even when there is no resistor, the two halves of the BBD may not put out precisely equivalent signals, so some balancing may still be needed.
The ideal, of course, is to have a multi-turn trimpot and a scope, although you can tune it by ear if you have a good ear. I suppose thew other thing to assist in tuning by ear is to eliminate all the bass and midrange in whatever source you are listening through. The clock signal, even at its' lowest, is unlikely to be lower than 4khz or so you can even tune by ear through a piezo tweeter for that matter. If your unit lacks a trimpot, you can probably be safe using a linear trimpot of about 5-10k. I may be wrong about that, but after poring over dozens and dozens of BBD-based schematics, I've rarely seen one higher in value. If it is no problem to find, buy, and install a higher quality trimpot, even a multi-turn one, do it.
The other approach to reducing clock noise is to use lowpass filtering after the BBD. Just about every BBD-based effect does this, although some have more stages of lowpass filtering than others. Depending on the design, some manufacturers chose to let a little clock signal get through at very long delays so they could have more bandwidth at shorter delays. They *could* have chosen to have switches or pots to change the filter setting, but that would have increased cost. Instead, they chose to allow one flaw in order to create another virtue. If you know the model number and a schematic is circulating around for that model, we could identify the filter component values to change that would reduce the audible clock signal.
)
