Are BBD chips considered Analog or Digital, or even neither or both?

[ElectricDruid]

BBDs are analog but operate in discreet time intervals. Digital also operates in discreet time, hence the confusion. But what makes digital digital is not the discreet time (although that is one of the prerequisites for it to work) but the encoding of signal levels in discreet values that can be represented in binary code.

+1 what Andy said. BBDs are definitely analog, but use discrete time (sample the signal), which is where the confusion arises since some people think "sampling = digital".

[bushidov]

Fair enough. It looks like I have a good enough understanding of why a BBD is not digital, as it doesn't fit the definition of a digital signal. However, does this statement carry any weight?

"Analog" means analogous to another continuous signal. In audio, this means analogous to sound pressure. Once you quantize the signal—in either time or voltage—it is no longer analogous to sound pressure.

The difference between discrete-time and digital is that discrete-time is quantized only in time, while digital is quantized in both time and amplitude. So, while discrete-time devices such as BBDs aren't digital, they're not analog either. This also doesn't mean that discrete-time devices are better than digital.

This seems a bit "matter-of-fact", but also doesn't provide any proof to such statements either.

[Digital Larry]

While you may argue that discrete-time breaks the association with "analog", it doesn't automatically make it digital either.  It may simply be the case that you need to describe such a system using two dimensions rather than forcing it into a single dimension where it doesn't completely fit.

Thank you, I'm here all week, tip your waitress, etc.

[ElectricDruid]

The difference between discrete-time and digital is that discrete-time is quantized only in time, while digital is quantized in both time and amplitude.

This part is accurate and I agree with.

So, while discrete-time devices such as BBDs aren't digital, they're not analog either. This also doesn't mean that discrete-time devices are better than digital.

This bit I don't agree with at all. For me, "digital versus analog" is one question, and "discrete versus continuous" is another question. One is about quantisation of amplitudes, and the other about quantisation of time. Those aren't the same thing, so I don't see why the two issues need to be conflated.

[Fancy Lime]

Fair enough. It looks like I have a good enough understanding of why a BBD is not digital, as it doesn't fit the definition of a digital signal. However, does this statement carry any weight?

"Analog" means analogous to another continuous signal. In audio, this means analogous to sound pressure. Once you quantize the signal—in either time or voltage—it is no longer analogous to sound pressure.

The difference between discrete-time and digital is that discrete-time is quantized only in time, while digital is quantized in both time and amplitude. So, while discrete-time devices such as BBDs aren't digital, they're not analog either. This also doesn't mean that discrete-time devices are better than digital.

This seems a bit "matter-of-fact", but also doesn't provide any proof to such statements either.

Nope. I don't know where this notion of continuity between input and output comes from but it is definitely not a prerequisite for an analog signal. If it was, then a deep tremolo would somehow not be analog. It is also not part of the definition of analog that the output has to map to the input in real time. If it was, then analog cassette tapes would not be analog, as Iain has already pointed out for tape delays. There would be no analog recording of any kind. So no, analog does not mean "analogous to another continuous signal". At all. Analog means that a variable (in our case voltage) changes in time in a continuous fashion, it does not say relative to what. That is totally the case with BBDs. The input signal is sampled at intervals by the clock but the resulting sampled signal is still continuous. You cannot freeze a BBD and the restart it later as you could with a digital device. The voltage in the BBD behaves completely analog.

Look, I get where the feeling comes from that BBDs are somehow "not quite analog". In fact, I used to have that feeling too before I really thought about it. I have come to the conclusion that by any reasonable definition I can think of, BBDs are analog. Period. If someone else wants to define analog differently, that's OK with me too. There is, as far as I know, no established definite authority on the matter. But I certainly wouldn't loose any sleep over it.

Cheers,
Andy

EDIT: The entire "analog sounds better than digital" discussion and related marketing claims are of course a whole other story. This understandably rubs some people the wrong way and it seems to get folded into the "are BBDs analog" discussion a lot. But it is a separate issue and I hope for the sake of our sanity that we do not have to start discussing that. Anyway, there are many threads on that in the archives from way back.

[iainpunk]

warning: off topic-ish

EDIT: The entire "analog sounds better than digital" discussion and related marketing claims are of course a whole other story.

as much as i would benefit form telling everyone analog is better, i don't think it is.
one of my favorite pedals ever is digital, and doing the same thing analog would be near impossible. i really like FZ-5 with digital COSM technology, it sounds great too! the fact it that 99% of people wouldn't notice any difference between digital and analog.
if i didn't have a strong aversion of programming, i would do everything digital, even the crazy weird fuzz stuff i do
looking at it that way, i really don't care if a delay chip is digital or analog, as long as it delays a given signal.

cheers

[Rob Strand]

I'd go so far to say that a quantized signal is not digital.    For a signal to be digital it needs to be converted to numbers.   You can do quantization in analog processing.   Digital signals are necessarily quantized but quantized signals don't have to be digital.

A discrete time system means you are taking samples.  You don't even need the samples to be regular.  The samples can taken on analog levels (ie. any level and in fact this is the BBD case) or be quantized in order to get them into a digital system for processing.

If you take switch mode power supply from an analysis point of view it has some parallels to sampled systems.  The output filter is similar to a reconstruction filter but it not technically a discrete time system (unless specifically implemented that way).  It's a modulator hence the name PWM.

A sampled system is a special case of modulation.  You are multiplying a train of spikes of height 1 with the input signal and the output signal is a sequence of samples.   Aliasing is a consequence of the modulation.

dougkerr.net/Pumpkin/articles/Sampling_as_Modulation.pdf

[DrAlx]

For a signal to be digital it needs to be converted to numbers.

Exactly this. Converted to something with numerical digits. The clue's in the name.

[ElectricDruid]

I'd go so far to say that a quantized signal is not digital.    For a signal to be digital it needs to be converted to numbers.   You can do quantization in analog processing.   Digital signals are necessarily quantized but quantized signals don't have to be digital.

Yes, this is a good point. Like we don't often think about analog sampling, we don't often think about quantized analog voltages, but there are a few examples; in analog-synth-land, you can quantise a Pitch CV to make sure the output is a semitone voltage. This is useful to turn random CVs into random notes, for example.

[Mark Hammer]

I suppose it requires a wee bit of mental gymnastics, but peak-limiting, itself, may be considered as a 2nd cousin of quantizing.  A wider range of analog signal above the peak is being shoehorned into a predefined amplitude.

[merlinb]

I consider it to be both. Or neither. Like wave-particle duality.

[merlinb]

Converted to something with numerical digits. The clue's in the name.

But any voltage is a numerical value, or way or another...

[Rob Strand]

But any voltage is a numerical value, or way or another...

It not the fact it has a value it's the fact it has been converted to digits (digitized).

[FiveseveN]

But digits are a construct of our minds, no matter where you look inside a CPU or hard drive you won't find any digits, only voltages. 

[Rob Strand]

But digits are a construct of our minds, no matter where you look inside a CPU or hard drive you won't find any digits, only voltages

The voltages are used to represent the digits.   I could easily use magnetic fields (hard disk) dots (CDs/ DVDs) to represent the digits.  I could even print them out on paper.

Digitizing is the *process* of converting an analog value to digits.

How the digits are abstracted *after* the analog value is converted to digits is not digitizing but storing.   If you think about printing out digits the paper and ink decay in time.   If you store the digits as voltages those voltage are not constant they have noise.  The storing only needs to be good enough to recover the digits.   However all this is storing,  the process of converting to digit is separate and done before storing.

[Rob Strand]

Ok how's this.

You take a tape measure.   You measure the width of a piece of wood.

The tape measure is analog and so is the width of the wood.

When you decide what value to assign to the measurement, doesn't matter if you make-up decimal points or only use the marked lines, that's the digitizing.     Someone else might do the same measurement and get different measurements that just means their analog to digital converter is different.  It's not about accuracy.

Once that number has been assigned to the measurement  then you can store it in your head or write it down.

   

[vigilante397]

You measure the width of a piece of wood.

Like anyone can afford wood these days

I'll see myself out.

[iainpunk]

You measure the width of a piece of wood.

Like anyone can afford wood these days

I'll see myself out.

i've got major wood!
my GF just came out of the shower

ill see myself out too
cheers, Iain

[Fancy Lime]

Ok how's this.

You take a tape measure.   You measure the width of a piece of wood.

The tape measure is analog and so is the width of the wood.

...

Of course it's analog if you use a tape measure. If you want digital, you need a CD measure.

Yeah, all right, I too shall see myself out

Andy

[amz-fx]

BBDs are not digital. I don't think anyone in this forum is making that argument.

BBDs are sample-and-hold circuits, with an odd twist or two.

Analog or not? It depends on how you define analog, i.e. if it must be continuous. It then depends on how you define continuous too. 

Ok how's this.   

How's this... see image below. This circuit could actually be built. We send a signal to a BBD; the MN3011 since it has taps along the delay line. The input signal is sampled and available at Tap1 after 396 delay stages... at this point, our FV-1 digital chip samples the output from the tap and delays it for the equivalent of the balance of the BBD (2932). The same clock is driving both chips and the outputs will be very similar.

Does Out1 have the analog mojo? Does Out2? If not, where did it lose the mojo, and where did the BBD get it in the first place?

The ADC samples an analog voltage that has already been quantized. If we are using a 3v peak-peak audio signal, then each sample taken by the 24-bit converter will be within plus or minus 0.000000089407 volt if we have adjusted it to make max use of the bit depth. An error, but a very tiny one. In fact, because of the degradation in the 2932 remaining BBD stages after the first tap, Out1 (bbd) will be less accurate than Out2 (digital). Does this make a difference? Can you hear the difference? 

If the degradation is making it analogish, could we program the DSP chip to give some similar quality loss and produce a BBD-type sound?



Best regards, Jack