I would peek through here and see if it sounds familiar https://forum.pedalpcb.com/threads/...-im-getting-unusually-bad-noise-solved.21901/
EXAMPLE Developing a custom pcb for Daisy Seed - Funbox
- Thread starter keyth72
- Start date
Cwallmeier
New member
Thanks for the reply!
I'm actually using the Funbox PCB, not the Terrarium. To be honest, I had already checked out that thread, but the noise in those examples is quite different from what I’m experiencing.
The sound I’m getting is more of a high-pitched feedback with a bit of underlying noise, kind of like a hiss, which is why I described it that way. The Terrarium examples have a more typical kind of noise.
I’ve socketed all the ICs and tried swapping in three different ones for each position, so I’ve ruled out damaged chips. The feedback is still there regardless.
I’m planning to record a video of the sound tonight or tomorrow, maybe someone will recognize it and have some insight.
-Here’s what I’ve tried so far:
-Checked and replaced the jacks
-Reflowed all the solder joints
-Swapped out each IC one by one
-Scraped paint off the inside of the enclosure for a solid ground connection
-Searched multiple forums for similar issues
It could be a bad cap, but before I go tearing the whole thing apart, I was hoping someone might have a clue about what’s going on or even @keyth72 might have experienced this himself.
Oh I forgot, I don't have TRS cables at the moment to test both changes at the same time but I've wired the jacks multiple times to check both the Right and Left channels separately and both have the same issue.
I'm actually using the Funbox PCB, not the Terrarium. To be honest, I had already checked out that thread, but the noise in those examples is quite different from what I’m experiencing.
The sound I’m getting is more of a high-pitched feedback with a bit of underlying noise, kind of like a hiss, which is why I described it that way. The Terrarium examples have a more typical kind of noise.
I’ve socketed all the ICs and tried swapping in three different ones for each position, so I’ve ruled out damaged chips. The feedback is still there regardless.
I’m planning to record a video of the sound tonight or tomorrow, maybe someone will recognize it and have some insight.
-Here’s what I’ve tried so far:
-Checked and replaced the jacks
-Reflowed all the solder joints
-Swapped out each IC one by one
-Scraped paint off the inside of the enclosure for a solid ground connection
-Searched multiple forums for similar issues
It could be a bad cap, but before I go tearing the whole thing apart, I was hoping someone might have a clue about what’s going on or even @keyth72 might have experienced this himself.
Oh I forgot, I don't have TRS cables at the moment to test both changes at the same time but I've wired the jacks multiple times to check both the Right and Left channels separately and both have the same issue.
keyth72
Well-known member
@Cwallmeier Thanks for emailing me, I wanted to respond here as well in case anyone else is getting more noise than expected. Based on the recording you sent me, you are getting much louder noise than the typical daisy seed build, although it sounds similar to the 1kHz noise (based on samplerate/blocksize of the Daisy Seed), although I noticed your sample had a peak at 500Hz.
Anyway, it sent me on an investigation of noise levels on all my Daisy Seed based pedals, and some things that may help mitigate it. Get ready for a long post with alot of graphs!
To measure noise, I used the EQ plugin in Reaper to visualize the frequency spectrum, and I plugged in each device into an isolated power supply, with the guitar volume knob turned all the way down, and the pedal output going into my audio interface. First I measured the Funbox (running the Uranus module) in active and bypassed modes. Then I did the same thing for my bkshepard 125B oled screen pedal (running a similar granular delay effect), and I also measured true bypass mode to have a baseline. Finally, I measured my Terrarium pedal running the "Seed" software. This Terrarium is the v1 board (I noticed there is an official v2 Terrarium with the lowpass filter added), but I added the lowpass filter separately to work with the Daisy Seed rev 7.
Here is the Funbox, in digital bypass mode:
As you can see, the 1kHz noise is present at around -15dB, with harmonics continuing at 2kHz and up. You can predict this noise by calculating how often the daisy seed processed a block of audio. In the Uranus effect, I'm using 48kHz samplerate with a blocksize of 48. So you do samplerate/blocksize, 48000/48 = 1000 Hz. Unfortunately it is audible while not playing, and I think it's safe to say this is an expected baseline for the Funbox circuit. It's a through hole digital board, and the components are tightly packed. Not ideal, but I think that's the reality of the circuit I designed.
Next, here is the Funbox with the Uranus effect active (but no sound coming from the guitar, so this is only noise from the Funbox):
The noise profile is similar, but the peaks go above -12dB. Ouch!
Before moving to the other pedal hardware, I'll show some ways I was able to reduce the noise, but there is a trade off.
By changing the block size and/or samplerate, you can push the noise out of the audible spectrum (above 20kHz). First, I tried reducing the block size to 4. This changes the noise equation to 48000/4 = 12000Hz. Still audible, but interesting to see the result:
This is much better! But that 12kHz spike is still too loud. Also, as Uranus is currently written, the Daisy Seed was not able to keep up with processing, there were audio dropouts (clicks).
Next, I reduced the blocksize to 2, knowing that the Daisy Seed wouldn't be able to run the effect smoothly. 48000/2 = 24000Hz, and as you can see from the graph, the noise spike is not present.
There's another part to the equation, which is the samplerate. Unfortunately with the current libdaisy (unless I'm mistaken), the only samplerate above 48kHz available is 96kHz, which also makes the effect more difficult to process for the Daisy Seed (double the samples to process in the same amount of time). So this time I tried 96kHz samplerate with a blocksize of 4, so 96000/4 = 24000 (same as previous). The resulting noise profile was pretty much the same:
And as before, the Daisy Seed was not able to keep up without audio dropouts. However, it wasn't terrible, so I think if the Uranus code was reduced to use less granular particles, it would be able to run smoothly at either the 96kHz/4 blocksize, or the 48kHz/2 blocksize. Some of the other modules may have similar performance, but others (like the neural net based ones) would be impossible to run at those rates.
To summarize, the Funbox is a bit noisier than I realized. I think the component quality plays a factor. I'll also take more care when writing the algorithms to push any noise spikes out of the range of hearing, and maybe look back at planet modules to see if I can improve the noise profiles. If you get louder noise than what I'm showing above, there may be other issues with the build.
Now to compare with other pedals, first I'll show the 125B oled pedal. This one is a SMD board, 4 layer. I think the SMD components play a factor in noise reduction, and possibly the additional layers provide shielding. Any EE's out there feel free to comment!
Here is the 125b with the granular effect on, running at 48kHz with a blocksize of 48:
This is MUCH better. No noise spikes and very quiet background noise (digital bypass looked almost exactly the same). To compare, here is the same pedal in true bypass mode (the pedal is completely bypassed, direct connection to the guitar):
So the pedal adds some noise of course, but to my ears, a negligible amount.
Next, I'll show the Terrarium with the Seed software running:
You can see the noise spikes around 1kHz, 2kHz, and 3kHz, although the overall noise is less than the Funbox. Even though the Funbox uses a very similar circuit for the input/output/power, there are many more components in about the same amount of space, so I'm guessing that's part of the reason.
In summary, from my limited testing, the bkshep 125B oled pedal was the quietest, followed by the Terrarium, and Funbox was the noisiest. You can increase the samplerate or reduce the blocksize to help reduce Daisy Seed noise, but at the cost of processing power. With a well designed circuit (and possibly SMD as a requirement), you can have a Daisy Seed based pedal with almost no noise, as with the 125B oled pedal.
In case anyone's not familiar, this is the 125B kshep pedal I'm referring to, I recently built three of them: https://forum.pedalpcb.com/threads/interesting-daisy-project-on-github.18810/#post-238556
Anyway, it sent me on an investigation of noise levels on all my Daisy Seed based pedals, and some things that may help mitigate it. Get ready for a long post with alot of graphs!
To measure noise, I used the EQ plugin in Reaper to visualize the frequency spectrum, and I plugged in each device into an isolated power supply, with the guitar volume knob turned all the way down, and the pedal output going into my audio interface. First I measured the Funbox (running the Uranus module) in active and bypassed modes. Then I did the same thing for my bkshepard 125B oled screen pedal (running a similar granular delay effect), and I also measured true bypass mode to have a baseline. Finally, I measured my Terrarium pedal running the "Seed" software. This Terrarium is the v1 board (I noticed there is an official v2 Terrarium with the lowpass filter added), but I added the lowpass filter separately to work with the Daisy Seed rev 7.
Here is the Funbox, in digital bypass mode:
As you can see, the 1kHz noise is present at around -15dB, with harmonics continuing at 2kHz and up. You can predict this noise by calculating how often the daisy seed processed a block of audio. In the Uranus effect, I'm using 48kHz samplerate with a blocksize of 48. So you do samplerate/blocksize, 48000/48 = 1000 Hz. Unfortunately it is audible while not playing, and I think it's safe to say this is an expected baseline for the Funbox circuit. It's a through hole digital board, and the components are tightly packed. Not ideal, but I think that's the reality of the circuit I designed.
Next, here is the Funbox with the Uranus effect active (but no sound coming from the guitar, so this is only noise from the Funbox):
The noise profile is similar, but the peaks go above -12dB. Ouch!
Before moving to the other pedal hardware, I'll show some ways I was able to reduce the noise, but there is a trade off.
By changing the block size and/or samplerate, you can push the noise out of the audible spectrum (above 20kHz). First, I tried reducing the block size to 4. This changes the noise equation to 48000/4 = 12000Hz. Still audible, but interesting to see the result:
This is much better! But that 12kHz spike is still too loud. Also, as Uranus is currently written, the Daisy Seed was not able to keep up with processing, there were audio dropouts (clicks).
Next, I reduced the blocksize to 2, knowing that the Daisy Seed wouldn't be able to run the effect smoothly. 48000/2 = 24000Hz, and as you can see from the graph, the noise spike is not present.
There's another part to the equation, which is the samplerate. Unfortunately with the current libdaisy (unless I'm mistaken), the only samplerate above 48kHz available is 96kHz, which also makes the effect more difficult to process for the Daisy Seed (double the samples to process in the same amount of time). So this time I tried 96kHz samplerate with a blocksize of 4, so 96000/4 = 24000 (same as previous). The resulting noise profile was pretty much the same:
And as before, the Daisy Seed was not able to keep up without audio dropouts. However, it wasn't terrible, so I think if the Uranus code was reduced to use less granular particles, it would be able to run smoothly at either the 96kHz/4 blocksize, or the 48kHz/2 blocksize. Some of the other modules may have similar performance, but others (like the neural net based ones) would be impossible to run at those rates.
To summarize, the Funbox is a bit noisier than I realized. I think the component quality plays a factor. I'll also take more care when writing the algorithms to push any noise spikes out of the range of hearing, and maybe look back at planet modules to see if I can improve the noise profiles. If you get louder noise than what I'm showing above, there may be other issues with the build.
Now to compare with other pedals, first I'll show the 125B oled pedal. This one is a SMD board, 4 layer. I think the SMD components play a factor in noise reduction, and possibly the additional layers provide shielding. Any EE's out there feel free to comment!
Here is the 125b with the granular effect on, running at 48kHz with a blocksize of 48:
This is MUCH better. No noise spikes and very quiet background noise (digital bypass looked almost exactly the same). To compare, here is the same pedal in true bypass mode (the pedal is completely bypassed, direct connection to the guitar):
So the pedal adds some noise of course, but to my ears, a negligible amount.
Next, I'll show the Terrarium with the Seed software running:
You can see the noise spikes around 1kHz, 2kHz, and 3kHz, although the overall noise is less than the Funbox. Even though the Funbox uses a very similar circuit for the input/output/power, there are many more components in about the same amount of space, so I'm guessing that's part of the reason.
In summary, from my limited testing, the bkshep 125B oled pedal was the quietest, followed by the Terrarium, and Funbox was the noisiest. You can increase the samplerate or reduce the blocksize to help reduce Daisy Seed noise, but at the cost of processing power. With a well designed circuit (and possibly SMD as a requirement), you can have a Daisy Seed based pedal with almost no noise, as with the 125B oled pedal.
In case anyone's not familiar, this is the 125B kshep pedal I'm referring to, I recently built three of them: https://forum.pedalpcb.com/threads/interesting-daisy-project-on-github.18810/#post-238556
Attachments
Last edited:
Nostradoomus
Well-known member
keyth72
Well-known member
Yes, Pluto would actually be a really good one to try it on, because it's already running at 96kHz, and it's not doing super heavy processing for the effect. Try lowering the blocksize to 4.
Similar threads
