OK, I found both. How do they connect to the Raspberry? It doesn't seem to be as straightforward as the Mezzanine, but I may be wrong.
How does it replace the on-board regulators?
Carlos
I have no clue about the technicalities in bypassing the RPi power supply, but I use this to power my RPi:
LiFePO4wered/Pi+ | Crowd Supply
cs-lifepo-04 Silicognition | Mouser
I only use A123 LiFePO4 battery power and glass optical in my system to reduce RFI exposure / Ground loops and overall clean crisp sound.
You can also use your own LiFePO4 battery. If you don't know how to DIY, you can order off Tindie without a battery compartment.
LiFePO4wered/Pi+ from Silicognition LLC on Tindie
LiFePO4wered/Pi+ | Crowd Supply
cs-lifepo-04 Silicognition | Mouser
I only use A123 LiFePO4 battery power and glass optical in my system to reduce RFI exposure / Ground loops and overall clean crisp sound.
You can also use your own LiFePO4 battery. If you don't know how to DIY, you can order off Tindie without a battery compartment.
LiFePO4wered/Pi+ from Silicognition LLC on Tindie
Last edited:
I like that solution. Very smart idea. But the problem for audio applications is that it's just a UPS, and the voltage rails are generated by switching DC-DC, so a lot of noise would be expected. But I could be wrong.
Regards,
Ian
^^^
No idea, but it's the best solution for my current needs now (transportable / portable). You can geek out:
No idea, but it's the best solution for my current needs now (transportable / portable). You can geek out:
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
But the ConditionerPi has regulators that bypass the RPi ones?
The idea is to eliminate all DC-DC regulation on the Rpi.
While bypassing single dc converters was an option in older RPIs, the latest iterations use a monolithic chip for all rails that also monitors them and doesn't even boot the device when it detects faults. So you will need to replace that chip and write the logic that controls the cpu power delivery in software. Here is some info on the PMIC in RPI 3b and 4: MaxLinear’s MxL7704 PMIC Powers the Raspberry Pi 4 - Bloomberg
No, of course not, its all empty speculation. No one has ever published test results showing how different rpi psu affect noise via a connevted dac
Amir at Audiosciencereview has shown a couple of tests where it makes no difference at all with usb connected dacs and a rpi file source.
I work in sales and marketing, whenever I have solid proof something works I use it to sell that product, I dont rely on conjecture...id even spend money to test a product to prove it works.
Amir at Audiosciencereview has shown a couple of tests where it makes no difference at all with usb connected dacs and a rpi file source.
I work in sales and marketing, whenever I have solid proof something works I use it to sell that product, I dont rely on conjecture...id even spend money to test a product to prove it works.
Amir at Audiosciencereview doesn't use listening tests on his reviews, as Stereophile or other magazines used to do.
So it's again a debate between the so called "objective" versus subjective tests. I wrote it that way because those that say "if it can't be measured, it doesn't exist", and do consider "speculation" as snake oil or similar. It happens that the human ear is an instrument as sophisticated or more than those used on tests.
Those magazines I mentioned did consider their subjective opinions as valuable and important as the measurements. They were not "speculation".
A more realistic answer might be that the instrument has to be developed yet that can quantify what the ear is listening. A trained ear can listen to many things that can't yet be measured. That is a fact and not speculation.
Well, the trust and respect you put on the reviewers on those magazines are for me a fact.
Of course they can be considered opinions, but they are not speculations.
I am a professional sound recordist, now retired, that worked for more than 45 years in film & TV audio. Myself and others similar to mine, professional audio engineers, you learn to train ear in the minimalist, many non measurable things you found working. That could be repeated and identified every time.
That's the part that has always been under discussion by those that rely only in instruments for their judgements.
I will not go into that discussion here either. If you do not believe that an ear can listen to things that can't YET be measured, that's your prerogative. But I learned different with experience, and knew people with trained ears that could listen and identify non measurable things 100% of the time.
Of course they can be considered opinions, but they are not speculations.
I am a professional sound recordist, now retired, that worked for more than 45 years in film & TV audio. Myself and others similar to mine, professional audio engineers, you learn to train ear in the minimalist, many non measurable things you found working. That could be repeated and identified every time.
That's the part that has always been under discussion by those that rely only in instruments for their judgements.
I will not go into that discussion here either. If you do not believe that an ear can listen to things that can't YET be measured, that's your prerogative. But I learned different with experience, and knew people with trained ears that could listen and identify non measurable things 100% of the time.
Indeed @wealas, I was going to mention the fact the linked pages and mods were obsolete; but stopped short of piling on about it.
the CPU core demands higher voltage when under higher load. the RP3b+ and 4 and the RP3 compute module require 5VDC, a few 3V3, 1V8 and 1V2. under load the CPU wants 1V4 IIRC. the sequencing flows from high to low voltages. the 5VDC is required first, then 3V3, 1V8 and the core voltage 1V2-1V4, but i'm uncertain if it will work if you just pin the core voltage at 1V4 all the time. the PSU cannot be completely dumb.
Certainly any efforts to design a base board similar to the USBSIG, would serve you much better than hacking at a standard RP3B+ or RP4. given the-at best-minutiae we would be talking about here, given any difference at all (unlikely) anything less than near perfect implementation would be counterproductive; for example, going against all EMC best practice, increasing loop area to the Nth degree and creating a new radio station in your lounge, while claiming lower RFI/EMI.
Even the compute module has the DC core voltage at the very least (not sure, the datasheet mentions the schematic omits the DC-DC section and I dont have one on hand) but designing a nice baseboard for the compute module seems the best way forward; for me, for reasons other than audibility (for me, lower power, extraneous features I dont need, cleaner build etc etc)
there is no getting around the fact that the board and CPU itself are switching devices and while local decoupling will certainly minimise the ground bounce etc, there will be uncorrelated switching noise regardless of what PSU is used. its a switching device; get used to it and if needed, isolate it.
the CPU core demands higher voltage when under higher load. the RP3b+ and 4 and the RP3 compute module require 5VDC, a few 3V3, 1V8 and 1V2. under load the CPU wants 1V4 IIRC. the sequencing flows from high to low voltages. the 5VDC is required first, then 3V3, 1V8 and the core voltage 1V2-1V4, but i'm uncertain if it will work if you just pin the core voltage at 1V4 all the time. the PSU cannot be completely dumb.
Certainly any efforts to design a base board similar to the USBSIG, would serve you much better than hacking at a standard RP3B+ or RP4. given the-at best-minutiae we would be talking about here, given any difference at all (unlikely) anything less than near perfect implementation would be counterproductive; for example, going against all EMC best practice, increasing loop area to the Nth degree and creating a new radio station in your lounge, while claiming lower RFI/EMI.
Even the compute module has the DC core voltage at the very least (not sure, the datasheet mentions the schematic omits the DC-DC section and I dont have one on hand) but designing a nice baseboard for the compute module seems the best way forward; for me, for reasons other than audibility (for me, lower power, extraneous features I dont need, cleaner build etc etc)
there is no getting around the fact that the board and CPU itself are switching devices and while local decoupling will certainly minimise the ground bounce etc, there will be uncorrelated switching noise regardless of what PSU is used. its a switching device; get used to it and if needed, isolate it.
How does one best measure nonstationary or quasi-nonstationary distortion? Obviously, the typical audio FFT works best for measuring distortion that is perfectly stationary relative to the acquisition time frame.
As I said, I DO NOT want to get into a discussion over this matter.
Whether people believe or not there are things the ear can listen that can't YET be measured I don't care anymore.
The only thing that matters is what I think will be better on a Raspberry setup for me, and for that I want a completely linear supply. And that's it. No further discussion.
Last edited:
Has anyone measured a Stradivarius from 1695 to prove that it has some sound that is 'better' than a modern violin that you can buy at K-Mart? Not sure they have been able to produce the science to reproduce the sound of these things. Yet even on YouTube you can clearly hear the difference between the two instruments in the hands of the same musician.
There have been so many reports of subjective improvement in sound of the Rpi and the FIFOPi power supplies that I think its safe to assume that it makes a difference by now. But as with all things, if you choose to claim that is impossible, then bless you. You can just plug in a cheap switching PS and enjoy audio bliss with a lot less hassle.
People are buying Allo usbridge to replace the Rpi. It has fixed the onboard voltage reg problem and separated the ethernet/usb circuits and owners report seem to value the sound impact when used with FIFO.
There have been so many reports of subjective improvement in sound of the Rpi and the FIFOPi power supplies that I think its safe to assume that it makes a difference by now. But as with all things, if you choose to claim that is impossible, then bless you. You can just plug in a cheap switching PS and enjoy audio bliss with a lot less hassle.
People are buying Allo usbridge to replace the Rpi. It has fixed the onboard voltage reg problem and separated the ethernet/usb circuits and owners report seem to value the sound impact when used with FIFO.
Last edited:
The usbsig is the board I was referencing. Pretty sure you'll find the compute module it's based on still has the smps for the core. Vbat supply feeds an smps for the core voltage. https://www.raspberrypi.org/documen...le/datasheets/rpi_DATA_CM3plus_1p0.pdf#page14
Last edited:
Their wording is pretty specific. They say there are no smps on the board. They aren't lying, it's on the rpi compute module. Just the one left for the core. Don't get me wrong, I think as solutions go it's pretty clean and it's not overly expensive given the complexity of such a design. I'm not convinced it's a problem that needs solving. But hey, have at it. I'll probably end up with something similar, but I do wish a compute module 4 was forthcoming, as I will be using some dsp