If I power from a Pi USB PSU my DAC is quiet and also so is the boot.
I am trying to power from a 24v PSU with a 5v step down on GPIO and tried all sorts of EMI filters, Wide Input SHIM from Pimoroni. Its EMI in fact you can almost hear the SoC at work with a background noise that often sounds like an old analogue modem.
Is it because on the same 24v DC that powers a 100 watt amp it creates a ground loop as at whatever stage I add EMI filters to the DC supply I can not seem to get rid of noise that seems to be generated by the Pi unless I use a seperate PSU for amp and Pi. Which sort of negates my idea of a neat single 24v DC supply to a small enclosure for the Amp and a step down to power the Pi. Tried linear regs, a couple of different bucks and the Pimoroni psu shim with emi filters from input to the dac to pi power input and still noise.
Take out the buck and with a standard Pi PSU powering the PI with the same 24vdc on the amp and all is lovely.
I am scratching my head really but thinking maybe the GND in the audio out of the DAC is creating a possible ground loop that maybe with the 2x PSU isolation doesn't exist. Anyone got any ideas?
I am trying to power from a 24v PSU with a 5v step down on GPIO and tried all sorts of EMI filters, Wide Input SHIM from Pimoroni. Its EMI in fact you can almost hear the SoC at work with a background noise that often sounds like an old analogue modem.
Is it because on the same 24v DC that powers a 100 watt amp it creates a ground loop as at whatever stage I add EMI filters to the DC supply I can not seem to get rid of noise that seems to be generated by the Pi unless I use a seperate PSU for amp and Pi. Which sort of negates my idea of a neat single 24v DC supply to a small enclosure for the Amp and a step down to power the Pi. Tried linear regs, a couple of different bucks and the Pimoroni psu shim with emi filters from input to the dac to pi power input and still noise.
Take out the buck and with a standard Pi PSU powering the PI with the same 24vdc on the amp and all is lovely.
I am scratching my head really but thinking maybe the GND in the audio out of the DAC is creating a possible ground loop that maybe with the 2x PSU isolation doesn't exist. Anyone got any ideas?
Best sketch out your wiring schematic so we can analyze where the problem is. Or you could just try adding isolation to bypass the likely problem - try a trafo on your amp input for example to isolate it from GND.
Powering a computer and an amp from the same PSU does often create a ground loop. If your amp has balanced inputs (often the amp does, but only unbalanced input is formed on the input terminals), you can use fake balanced output which worked perfectly in my case https://www.diyaudio.com/community/threads/adding-an-amp-into-thin-client-pc.320812/
The amp is https://store.sure-electronics.com/product/AA-AB31184
DAC https://www.aliexpress.com/item/4000373056838.html
Which on the Pi3b with a Pi USB PSU works great apart from from the power on crackle as haven't done anything with gpio standby yet.
I actually have 2x zero2 lined up but just been using a Pi3b as for now using bluetooth on 1 with snapcast to link them as a L/R combo on 2x CATON 2nd user bookshelf speakers I got from ebay, but stopped testing on the single Pi3 due to the noise.
Haven't really decided yet as the Bluetooth Le auracast sounds interesting but may go spotify premium and use there cast, but with things sounding like they do, no.
Actually things sound great but definately don't want 2x separate supplies for each speaker... Same simple default circuit but unplug the PI psu and put the buck to GPIO and oh my!
I have a 3.5mm jack to phono with on the phono side just a single jack to the wondom board that yeah its just a TPA3116 but those Wondom boards are pretty nice for $
Wiring is default Vcc/Gnd & the 3x I2S wires from pi to dac, 3.5mm to single phono on the anp.
24vdc SMPS powering the amp that feeds a 5v buck that give 5v/Gnd to the gpio via dupont jumpers that just use the nearest alternative 5v/gnd pi gpio pins for power to the dac.
The difference in noise of the setup from using a Pi PSU & 24dc to 24dc with a 5v step down is absolutely drastic and its if it suddenly picks up all EMI from the PI as if hearing it at work and the power on crackle becomes a cacophony whilst back to a Pi USB PSU its just that initial cap charge that gives any noticeable noise.
Single ended not differential and nothing special about the wiring its just how you would expect with the gear listed.
I don't think its the DC supply as rather than just hum its like listening to as said an old analogue modem of the pi in action 🙂
The only thing I haven't tried due to a prefab 3.5mm to phono is disconnecting that ground which maybe I should as a test.
Thnx for the quick replies.
DAC https://www.aliexpress.com/item/4000373056838.html
Which on the Pi3b with a Pi USB PSU works great apart from from the power on crackle as haven't done anything with gpio standby yet.
I actually have 2x zero2 lined up but just been using a Pi3b as for now using bluetooth on 1 with snapcast to link them as a L/R combo on 2x CATON 2nd user bookshelf speakers I got from ebay, but stopped testing on the single Pi3 due to the noise.
Haven't really decided yet as the Bluetooth Le auracast sounds interesting but may go spotify premium and use there cast, but with things sounding like they do, no.
Actually things sound great but definately don't want 2x separate supplies for each speaker... Same simple default circuit but unplug the PI psu and put the buck to GPIO and oh my!
I have a 3.5mm jack to phono with on the phono side just a single jack to the wondom board that yeah its just a TPA3116 but those Wondom boards are pretty nice for $
Wiring is default Vcc/Gnd & the 3x I2S wires from pi to dac, 3.5mm to single phono on the anp.
24vdc SMPS powering the amp that feeds a 5v buck that give 5v/Gnd to the gpio via dupont jumpers that just use the nearest alternative 5v/gnd pi gpio pins for power to the dac.
The difference in noise of the setup from using a Pi PSU & 24dc to 24dc with a 5v step down is absolutely drastic and its if it suddenly picks up all EMI from the PI as if hearing it at work and the power on crackle becomes a cacophony whilst back to a Pi USB PSU its just that initial cap charge that gives any noticeable noise.
Single ended not differential and nothing special about the wiring its just how you would expect with the gear listed.
I don't think its the DC supply as rather than just hum its like listening to as said an old analogue modem of the pi in action 🙂
The only thing I haven't tried due to a prefab 3.5mm to phono is disconnecting that ground which maybe I should as a test.
Thnx for the quick replies.
Last edited:
That amp uses the same chip as mine in the link above - TPA3116 which has balanced input. I am almost sure if you rewire your DAC -> amp connection to use the balanced input of that amp chip (like in my post) the problem will be gone.
Might not be possible with the board I got which is essentially same as your but in bridged node with on board single ended input.
Sure are pretty good that they do give some info but not that much to go on really https://store.sure-electronics.com/images/documents/Mid-power Audio Amplifier Board Series-Mono.pdf
The 100watt as all specs is never going to be 100watt but pretty well matched to the caton 75 watt speakers I have, I guess I could get 2x stereo like you have and halve the wattage as they are only £10 on ebay in fact cheaper than that last time I looked.
The way the board is implemented is just standard unbalanced marked up as gnd/signal...
With the same input but different psu setup, noise is gone and wondering if there is a way to get similar with what I have?
I might try 2x ac-dc psu and see if I can squeeze them into the enclosures I got and do away with the 24v the buck.
https://www.aliexpress.com/item/33010736780.html
Last edited:
IMO it's not about the PSU, but about wiring the ground leads for RPi and your amp. Just make a drawing depicting the large currents which flow on RPi and to your PSU. The wire and RPi traces have non-zero resistance which results in the ground currents flowing also through your RPi -> DAC -> amp 0V line. Since this line also has non-zero resistance, the extra noise current creates a noise voltage which adds up to your DAC output signal voltage. You can also play with small resistors in your RPi -> DAC and/or DAC->amp ground wires, to reduce the extra currents flowing through the ground loop. But the higher resistance generates higher noise voltage, so you results will vary.
The technical/EMC treatment of your problem is here : https://incompliancemag.com/article...rcuit, sharing the ground with other circuits.
Has something to do with the PSU as totally silent via USB PSU apart from capacitor charge. The cables from rpi -> dac and dac -> amp never changes but when the usb psu is swapped with what are fairly flimsy dupont jumpers to a 5v buck then we can hear the Pi in action 🙂
Cables are so short that I didn't really think about resistance but guessing maybe the Pi PSU
Cables are so short that I didn't really think about resistance but guessing maybe the Pi PSU
Attachments
Also draw a picture where it works OK and compare the ground connections (note your step-down 24/5V DC/DC has the IN/OUT GND terminal connected internally).
Also do not forget to draw the ground analog line connection between the dac and amp - ground-loop currents through the ground return line of that connection cause the noise you hear.
Lols you don't know how painful that was but its exactly the same circuit with the 5v buck gone replaced by USB PSU and that works great.
So the 24v input and 5v output to gpio is gone as 5v is by a usb wall charger and that confuses me the most as you would maybe think the ground loop would still exist.
3.5MM stereo jack -> 2x phono (L/R) using just a channel on a premade cable that also never changes.
I guess both supplies are isolated via the transformers?
So the 24v input and 5v output to gpio is gone as 5v is by a usb wall charger and that confuses me the most as you would maybe think the ground loop would still exist.
3.5MM stereo jack -> 2x phono (L/R) using just a channel on a premade cable that also never changes.
I guess both supplies are isolated via the transformers?
Yeah I am just going to remove the gnd between Dac & Amp as did so with the Pi and dac but was still the same noise level, but that is likely the long run on what is a premade 1M 3.5mm to phono.
Ordered a few 3.5mm & phono solder connectors to make a signal only connection the gnd should still be common but will just test to see
Ordered a few 3.5mm & phono solder connectors to make a signal only connection the gnd should still be common but will just test to see
You cannot remove the gnd between dac& as every signal transfer takes two wires. If you cut the GND wire in your 3.5mm connection, for the first case the signal return wire will go around through the 24/5V DC/DC and you will have all the RPi return current part of your signal. For the second case no signal will be transfered to the amp as there will be no return path for the signal current.
Yes you can as its common elsewhere as hence it has a loop, I know this because I took the gnd away from the dac and it still worked as the gnd in the 3.5mm to phono is likely common.
The 24vdc gnd & 5v buck gnd I guess are common Amp is connected to 24vdc & Pi to 5v buck and prob why I am getting a horrendous gnd loop of noise.
I am guessing it will also continue to work as if the gnd isn't common the dac should of stopped working.
The 24vdc gnd & 5v buck gnd I guess are common Amp is connected to 24vdc & Pi to 5v buck and prob why I am getting a horrendous gnd loop of noise.
I am guessing it will also continue to work as if the gnd isn't common the dac should of stopped working.
So you still have the GND connected through your 3.5 cable (the shield of the two coaxes) and the ground loop noise currents flow through that, adding noise to your signal. You have not removed the DAC - amp gnd connection if you did not cut the shield in the signal cable.
You can put a transformer in there instead or use balanced connection, to make the signal path avoid the wire with the ground loop currents.