[sdiy] Frequency shifted from BBD?

[brianw]

Thanks! Very nice to have that Eventide article.

One advantage of the H910 is the RAM. The glitch cannot be avoided, but it's much shorter. In the H910, the sample rate being written into RAM is constant. The sample rate being read out of RAM is also constant, but different, so when it wraps around the circular buffer there is a glitch, but that glitch only lasts for one sample.

A BBD cannot do the same. Since the input sample rate changes with the same clock as the output sample rate, the glitch lasts until all of the samples have been cleared. That's why the smallest BBD is best.

Note that the H910 is effectively changing delay time, and the delay time ranges from 0 msec to about 20 msec, but ideally not much more than 30 msec. The BBD technique changes pitch by changing sample rate (not delay time). The delay time of the BBD has a minimum, because each sample must still pass through every bucket. Changing sample rate does effectively change delay time, but it's different between a DDL and a BBD. At the very least, the H910 can go all the way to 0 msec.

Brian


On Oct 4, 2024, at 3:37 PM, Adam wrote:
> Not very technical but they mention the glitch and the "ideal" 20 msec delay..
> https://www.eventideaudio.com/50th-flashback-4-2-h910-harmonizer-the-product/<H910-4.2-Thumb-1024x512.png>
> Flashback #4.2: H910 Harmonizer® Pt. 2
> eventideaudio.co
> 
> 
> On 5 Oct 2024, at 8:15 AM, brian wrote:
>> I recall seeing it written up as using a triangle wave for the fade. I have never built one, though, and I think you're right: A trapezoid would be best. I also think that the smallest BBD would be best so that the glitch would clear out quicker than, say, a 1024-tap BBD.
>> 
>> On Oct 4, 2024, at 2:07 PM, Mattias Rickardsson wrote:
>>> Don't you mean that you need a trapezoid wave for the crossfade? Keeping one delay quiet for a while while emptying its period of accumulated control glitch... which I expect to be slightly longer than an instant... No? :-)
>>> 
>>> /mr
>>> 
>>> Den fre 4 okt. 2024 22:41brian> skrev:
>>> On Oct 4, 2024, at 12:26 PM, Ingo Debus wrote:
>>>> Am 04.10.2024 um 19:49 schrieb Gordon:
>>>>>> I recently re-read the section in Barry Klein’s Electronic Music Circuits on bucket brigade devices and noticed a brief mention of using them as a frequency shifter by modulating with a ramp wave.
>>>>>> He acknowledges that there will always be an audible glitch and some noise issues. 
>>>>>> I experimented with this idea a little to minimal success, but curious if anyone else has tried this or seen it used in interesting ways.
>>>>>> 
>>>>> 
>>>>> Didn't the Eventide harmoniser basically do this by ramping two delays, and crossfading from one to the other when one was about to "jump" and the other was mid-sweep?
>>>> 
>>>> Surely this would result in a pitch shift, not a frequency shift.
>>>> 
>>>> Ingo
>>> 
>>> Yes, this would be a pitch shift.
>>> 
>>> The BBD uses a single clock for input and output. There is no way to clock voltages in at a different rate than they're clocked out, nor can the sample "address" be random access.
>>> 
>>> The pitch shift works because the clock is constantly increasing (or decreasing) so that the output rate is effectively different from the input rate that was active when that particular sample was taken. But the caveat is that new samples are already coming in at the new rate.
>>> 
>>> As Gordon mentioned, having two parallel setups and crossfading between them helps. You'll need a triangle wave (for the crossfade) and two ramps offset from each other. At the top of the triangle, one ramp will be in the middle (no glitch), while the ramp that's completely faded out is recycling (glitch, but fully attenuated). Then, at the bottom of the triangle, the reverse is true. I can't remember seeing the oscillator design that would create these three synchronized signals. The two ramps would have to be at exactly the same slope, but their reset point would be offset so that they alternate in sync with the triangle.
>>> 
>>> Brian