Hi! It is a topic that has been discussed a lot. BUT yesterday evening there was a discussion with gptchat and he suggested that I not change the well-known R18 150k resistor that affects both. But for the release potentiometer in R20 and for the attack potentiometer before C13 in series. I have not seen anything like this anywhere. He analyzed it in depth and justified what he mentioned to me. My schematic
(https://i.postimg.cc/MMyLK21f/IMG-5500.jpg) (https://postimg.cc/MMyLK21f)
is from a kit and I am attaching it. What opinions are there about this?
Quote from: Motifnick on April 21, 2025, 01:11:30 AMWhat opinions are there about this?
don't talk to ai - stick with us real people.
from your circuit diagram, R20 is in series w/ the out level pot, and therefore has no correlation to either attack or release. surprisingly, C13 is also in that series string, and also has nothing to do with attacking or releasing.
so you can tell old mate chat to stop talking bollocks.
Quote from: duck_arse on April 21, 2025, 11:17:32 AMQuote from: Motifnick on April 21, 2025, 01:11:30 AMWhat opinions are there about this?
don't talk to ai - stick with us real people.
from your circuit diagram, R20 is in series w/ the out level pot, and therefore has no correlation to either attack or release. surprisingly, C13 is also in that series string, and also has nothing to do with attacking or releasing.
so you can tell old mate chat to stop talking bollocks.
you are right... about attack i notice that the attack is being carried away by the R13 before the T2..about the release the R20 control the C10 before T3..maybe make sense to try pontesiometer there? forget the C13 spot this is the output cap...but the C10 make me wonder....for release time. is possible to control separate attack...and release with two pot's? 😥
On the subject of chatgbt, I had been playing with it myself recently and similarly, asked for some circuit and even code analysis.
I quickly found myself correcting the AI and also asking it to stop lying to me...
"I'll have that for you in a minute"
2 hours pass, I asked about it.
"Sure, here you go"
-Provides a fake link
I continue to extrapolate for proper information
-Timed out, must pay for subscription
While it's handy for using pooled resources, it won't look at any old datasheets or anything. Just a quick guess without any self-corrections.
Use it for entertainment and propose some surface-level questions only. It's handy & fun but totally unreliable for anything in-depth.
I've been using Claude.ai for some C coding, and that's not too bad. At least, my C is rusty, so it serves to remind me of stuff I knew once, and as soon as I see it I know whether it's right or nonsense. It's quicker than doing a lot of google searches to find an answer. However, there is definitely still a decent amount of nonsense, so you *really* need to know what you're looking at. Some friends recommended it as a better bet than ChatGPT for technical queries. You can also upload files for it as a reference, so you *can* get it to read old datasheets for you. Of course, that's no guarantee that it'll understand them any better than we would...;)
* Side question: How long will it be before there's a major accident caused by something badly designed by an AI getting past the humans that were supposed to be doing QA?
> is possible to control separate attack...and release with two pot's? 😥
Certainly! With a sufficiently complicated side-chain you can do anything.
That drawing of the DynaRoss is toplogically correct but is not the clearest part-order. This is maybe clearer?
(https://i.postimg.cc/NyZFDWDw/Dyna-Ross-Attack.gif) (https://postimg.cc/NyZFDWDw)
Go through the rectifiers T3 T4, through the attack resistor, past the decay resistor, to the timing cap and buffer T5.
In real life T3 T4 and associated parts do not provide an infinite BANG of attack power, so Rattack is not needed for most musical purposes. Pick C12 for happy attack then fiddle Rdecay to taste. When you change C12, change Rdecay. This is not intuitive or handy.
But if you really want slow distorted attacks, re-work the traces/jumpers as shown. Try 1k to 50k for Rattack. You may never get really happy because it all interacts. The CBS Volumax (many models) has enough 'brains' to give good separate controls.
I had been thinking along those ^ exact same lines for timing changes, but I couldn't quite reconsile the effect of the changes and the terms attack and decay. not that I have any idea of use or abuse of compressors, mind.
Quote from: ElectricDruid on April 22, 2025, 02:59:44 PMI've been using Claude.ai for some C coding, and that's not too bad.
In this moment I'm realizing that "Claude.ai" is not a virtual version of one of my coworkers.
Seriously, my pal Claude at the day job sent me a link last year and I disregarded it thinking it was a virtual "clone" of him loaded with conspiracy theories and outdated anecdotes.
Or perhaps it was, but now has taken on a life of it's own :icon_eek:
To answer your question, most definitely within the next decade.
I'm certainly not convinced enough to take any leap of faith with AI.
Sorry to derail. PRR has the question covered.
Though I would be more inclined to snub in a better envelope generator instead of working with an already minimalist design.
Turning a circuit into something it's not will almost always leave more to be desired.
-From my experience at least.
Quote from: Kevin Mitchell on April 24, 2025, 02:59:32 PMI would be more inclined to snub in a better envelope generator instead of working with an already minimalist design.
He was missing an Attack knob. I added one. He did not say it had to be precise, did not mention the tempco (a couple dB from frostbite to heat castration), gave no preference to log or lin knob response, or over-easy. The wish-list for limiter sidechain is infinite. That stupid 2-transistor job really does OK for its leg-count.
Merlin's Engineer's Thumb is a useful trade between function and complexity.
https://www.diystompboxes.com/smfforum/index.php?topic=97123.0
I have to ask what the purpose of seeking separate attack and release controls is. Studio compressors will have such controls because the compressor may be applied to any of a range of signal sources with very different dynamics, note spacing, and roles within the mix. For guitar purposes, simple variable gain-recovery time, and the addition of a clean/compressed blend control, does pretty much all one might want a guitar compressor to do.
Keep in mind that the equivalent of varying R18 has long been labelled as "Attack" on those pedals using it, because it affects the audibility of the initial attack, depending on how closely spaced any subsequently-played notes are. Play fast (e.g., "chicken-pickin") and rapid gain-recovery matters. Put some space between your notes, and the difference between the 5:00 and 7:00 positions on the Attack knob become largely inaudible.
Because slower gain-recovery CAN impact on the audibility and "bite" of pick attack, a clean/compressed blend allows for uncompressed pick attack to be heard clearly, as a slow gain-recovery provides for the illusion of longer sustain. I'm not sure what any guitar player would want or need more than that. If it was the case that one was trying to use the same compressor for recording snare or kick drum, as well as acoustic guitar, or piano, I imagine a more sophisticated sidechain might be appropos.
I highly recommend the overview article entitled "The Big Squeeze" in this old issue of Electronic Musician. https://www.worldradiohistory.com/Archive-All-Music/Electronic-Musician/2001/Electronic-Musician-2001-02.pdf It surveys a wide array of what was available at the time, but more importantly it describes why THIS or THAT particular compressor was great for this or that particular instrument or purpose.
This sounds a lot like the early synthesizers with the attack / decay / sustain / release controls that shaped the loudness profile of the sound. All of these controls were simple first-order time constants. In fact, emulating some instruments is more likely to be a second order attack and first order release with no need for decay or sustain, which were already part of the guitar input. This is easier to do from a digital representation of the loudness profile.