Is it "worth it" to try to develop an analog delay?

[acehobojoe]

With the ease of digital delays, and the ability to basically make them sound like an analog delay, is it worth it to make an analog circuit?

Are there any notable analog circuits that are reliable?

Ah yes, and then comes the taps. There are tap delays of many kind, even some analog. But how would you make a tap analog delay??

[Mark Hammer]

"Tap", meaning tap tempo, or tap, meaning multiple delay-times at once?

[Brisance]

How about tape delay?

EDIT: you could get these oldschool tape cassette decks, modify the cassette for a loop and use the mechanism

[Digital Larry]

I can settle this.  No, it's not worth it.   

[R.G.]

Let me add a nail to the coffin.

If you can't actually buy the chips to manufacture your design, you wasted your time. Analog delay chips are in much the same state as Mastodons - their descendents still exist, but are rare, and there are remarkably few of the originals still around.

[knutolai]

One argument for using analog delay chips is the chorus/flanger/comb-filter applications that the PT2399 cant handle

[R.G.]

I speak from experience - there is (well, almost nothing  ) nothing as frustrating as being ready to build something and finding that you have to scour the planet to find a magic chip that, like the unicorn, is reputed to have existed but can no longer be found.

[anchovie]

One argument for using analog delay chips is the chorus/flanger/comb-filter applications that the PT2399 cant handle

That's an argument for using DSP!

[anotherjim]

Currently working on a delay circuit using micro controllers. A little 8 pin DIL AVR AT Tiny85. 8 bit of course, but progress is encouraging. It's ADC cannot run as fast as a BBD, so Flanging range is done by switching to a very short delay buffer. The chip is cheap (actually found them cheaper than lower spec versions). Another Tiny85 provides the delay clock/lfo functions, but that could equally well be BBD style analog LFO with VCO clock generator. Usual BBD audio techniques necessary: Pre-emphasis/De-emphasis, Compress/Expand...

I'd agree that new BBD circuits are perhaps unwise, and I don't think the PT2399 is all that secure either.

There is the Spin Audio FV-1, but I'll always be doubtful about security of supply of specialist audio chips.

[knutolai]

That's an argument for using DSP!

true, but you would have to know programming, have the tools to do said programming and design a PCB for your SMD DSP chip. For a product to be mass-produced DSP is the way to go, but for a hobbyist making a one-off, searching for a BBD/unicorn might seem more tempting to some

[deadastronaut]

analog schnanalog pfft!...

i get by with 4x pt's...

https://www.youtube.com/watch?v=Kyn5q6-cFIs&list=PL6E5UJVYbhpAkdWbi0hSacGDBc7fUJmXt

[Luke51411]

Building an analog delay based on an existing design for your own experience/enjoyment could be worth it. Designing one from the ground up... Probably not.

[acehobojoe]

Ah, so you would say that the magic unicorn chip is sort of a marketing ploy?

and yes, tap meaning tap tempo. I just found it interesting that there was a tap tempo analog delay from JHS.

[Mark Hammer]

As many things as there are wrong with Dave Hunter's Guitar Effects Pedals book, the interviews with designers/builders in the back 1/3 of the book are fascinating and thought-provoking.  Roger Mayer, in particular, speaks at great length about BBDs vs digital delays.  His own position is that, with their essentially "infinite resolution", BBDs that sample the exact analog voltage, as opposed to the nearest digital value of the current analog voltage, have a more musical and listenable decay.  He proposes that analog-digital differences are negligible when the signal is near peak and the full resolution of the converters is applied, but that when the signal is decaying, one may only have 6-8 bits of resolution "left" to depict that decay, such that it sounds grainier than the same original signal delayed by a BBD or tape.

It's an interesting idea, although in the several years since the interview was conducted, subsequent changes in the technology may have made his point of view irrelevant.  Plus, there is the difference between auditioning a track through a studio system, under very favorable conditions, and a club gig or jam night, where the opportunity to listen to quiet fadeouts simply doesn't happen.  In other words, he may still be spot on, but only under conditions that don't apply to a great many of us.

Me, I'm pretty content with a PT2399, sensibly used.

[knutolai]

and yes, tap meaning tap tempo. I just found it interesting that there was a tap tempo analog delay from JHS.

Well a BBD devices delay time is controlled by a clock signal. Every clock speed has a corresponding delay-time. If you have a microcontroller (DSP chip) which reads the time interval between each tapping of the tap-tempo switch it could output a clocksignal with a frequency set by this time-interval. So in short every specific tapping speed generates a specific  clock frequency which generates a specific delay time.

[mykaitch]

Couple of years back I did a mad thing -- used some old cassette decks and made my own five head deck from scratch.
Yeah it worked. Absolute bastard making a tape loop...
I use the PT chip now.
But it was fun and all analogue too.

[anchovie]

That's an argument for using DSP!

true, but you would have to know programming, have the tools to do said programming and design a PCB for your SMD DSP chip. For a product to be mass-produced DSP is the way to go, but for a hobbyist making a one-off, searching for a BBD/unicorn might seem more tempting to some

For me, I know when to put a limit on combining the need for an effect with the hobby. I recently got a NUX Time Force - 11 delay types, tap tempo and a looper! - for £34 shipped through aliexpress. It's fun to tweak dirt circuits and tailor something to the sound I want, but when it comes to delays I really just want the thing to echo. I wouldn't even try to design a delay from the ground up as I can't imagine stumbling across anything special.

[R.G.]

Roger Mayer, in particular, speaks at great length about BBDs vs digital delays.  His own position is that, with their essentially "infinite resolution", BBDs that sample the exact analog voltage, as opposed to the nearest digital value of the current analog voltage, have a more musical and listenable decay.  He proposes that analog-digital differences are negligible when the signal is near peak and the full resolution of the converters is applied, but that when the signal is decaying, one may only have 6-8 bits of resolution "left" to depict that decay, such that it sounds grainier than the same original signal delayed by a BBD or tape.

I respect Roger's thoughts, but I would very much like to watch him do a double blind test of BBD versus analog-digital delays at a range of levels.

I have dealt with the vagaries of BBD sampling personally, and the idea that what a BBD gets is "infinite resolution" is a goal, not a realized result. In particular, the sampling process itself has a clock signal turn the "sample me now" FETs on and off, and also gating out the delayed signal, after it's washed into and out of the storage capacitor cells. There is clock noise from sampling, and imperfect transfer between cells. While the cells themselves may have a resolution down to the individual electron (or hole), the storage cells are by no means able to hold an infinite amount of charge, so the resolution, even if perfect, would be quantized.

Agreed, the quantization may be finer than is detectable, but then once there is an admission that there *exists* quantization finer than can be detectable, the argument then shifts from quantization levels to what is detectable again, and you can't then say that digitization per se makes the sound "grainier".

The addition of the inescapable clock feedthrough to each sample is another issue that adds to the question.

The claim of "only six bits left to carry soft signals" is an old one about digital audio in general.

[acehobojoe]

I'll get 4 pt2399's and call it the "it sounds like analog" delay. or the "most likely higher fidelity than old chips" delay

[ElectricDruid]

I'll get 4 pt2399's and call it the "it sounds like analog" delay. or the "most likely higher fidelity than old chips" delay

I would.

I agree with Digital Larry - it isn't worth it, except perhaps (as someone else noted) for chorus or flanger, where a little BBD is probably simpler than using a codec and DSP - even then though, an FV-1 would probably do more in less space. Maybe I'm just fond of analog flangers.

I bought four MN3205 4096-stage delay lines off eBay to use for an analog tap tempo delay project. The plan was to use a PIC to do the tap tempo and clock the delays. I got all that bit working fine. But running more than a couple of the delays in series just raised the noise floor so much it'd have been better named the noise ceiling. And two chips is only 400msecs - not even a long time!

The "sound" of a delay is much more to do with the input/output filtering and compounding (if used). You can use the same 1970's in/out filters around a modern digital delay, and you'll get the same sound.