Develop ultra capacitor power supply and LiFePO4 battery power supply

Jump to Latest
Y

yunyun

Banned
Joined 2019
andrea_mori said:
Not really, unless the noise corrupts I2S signals, the noise incoming from the Raspberry can only modulate bck and ws that means higher phase noise/jitter.

The purpose of the FIFO is just to remove this jitter providing a clean I2S.
If this high jitter is reflected to the output there is a problem in the FIFO, that’s not enough decoupled from the incoming signals.
I would compare the input and the output of the FIFO with and without super regulator to understand.
Click to expand...

Noise should only affect the analog circuit. Although the digital circuit will not be affected by the noise, the digital circuit and the digital circuit are connected through the ground wire. The noise from the power supply or external electromagnetic waves will eventually be transmitted to the analog circuit through the ground wire and cause impact . What I have said is confirmed by experiments.

In fact, the effect of the ground box should be equivalent to the ultra capacitor, that is, Ultra capacitor actually absorbs the ground noise. I used to connect the grounding box to the digital circuit, but he didn't bring any improvement, but when I connected the grounding box to the analog circuit, and even included speakers, he would bring a positive sound improvement.

I have not received the UcConditioner yet. I will compare the effect of the UcConditioner and the grounding box on the system, and I will be able to confirm my current conjecture.

Like the picture below from MSB, they clearly tells us the existence of ground noise. However, the UFL wire we use has a ground wire. Even if it is an I2S wire, the UFL wire will still transmit ground noise. Optical has no ground, but optical will have additional jitter generated by the photoelectric conversion. This part of the MSB is wrong.

Pro-I2S-diagram.jpg
 
Last edited:
R

Raj1

Member
Joined 2003
yunyun said:
My problem is also for LinearPi. Since UcConditioner has a charging current limit of 700mA~1A, even LinearPi can output a current exceeding 1A, UcConditioner cannot receive and use it to charge UC. Therefore, I cannot understand why the output current of LinearPi is designed to reach 3A.

LinearPi's output current can reach 3A, which means that it still maintains the best statue to power UcConditioner when outputting 1A current?
Click to expand...

Why do you assume LinearPi is completely designed around UcCondsitioner? It doesn’t have to be used that way. One user asked for the current limit of the supply for a specific purpose. Doesn’t mean everybody needs to use the supply that way. But, if you feel so strongly about these things, perhaps you can bring something to market that reflects all of your preferred design philosophies?
 
andrea_mori

andrea_mori

Banned
Joined 2005
yunyun said:
Noise should only affect the analog circuit. Although the digital circuit will not be affected by the noise, the digital circuit and the digital circuit are connected through the ground wire. The noise from the power supply or external electromagnetic waves will eventually be transmitted to the analog circuit through the ground wire and cause impact . What I have said is confirmed by experiments.

In fact, the effect of the ground box should be equivalent to the ultra capacitor, that is, Ultra capacitor actually absorbs the ground noise. I used to connect the grounding box to the digital circuit, but he didn't bring any improvement, but when I connected the grounding box to the analog circuit, and even included speakers, he would bring a positive sound improvement.

I have not received the UcConditioner yet. I will compare the effect of the UcConditioner and the grounding box on the system, and I will be able to confirm my current conjecture.

Like the picture below from MSB, they clearly tells us the existence of ground noise. However, the UFL wire we use has a ground wire. Even if it is an I2S wire, the UFL wire will still transmit ground noise. Optical has no ground, but optical will have additional jitter generated by the photoelectric conversion. This part of the MSB is wrong.

View attachment 859471
Click to expand...

We are talking about the digital domain, the only analog part is the end section of the DAC (switches).
If the noise coming from the power supply of the source affects the analog section of the DAC there is something wrong in the chain.

Word clock (and bit clock for the DAC latching on bck) should not share the same ground of the FIFO/Source.
But I believe there are other issues.
The Raspberry generates a lot of RFI/EMI, so it's not a good practice to stack FIFO/Master clock close to it.
You should keep the FIFO as far as possible from the Raspberry to avoid RF and EM interferences.

Moreover I suspect that a part of the huge jitter coming from the Raspberry crosses the FIFO and affects the DAC.
If so, any improvement of the source powe supply is useless.
The designer should measure and compare the phase noise of Word Clock and Bit Clock at the input and at the output of the FIFO.
The above plots would be very useful to understand.
 
R

Raj1

Member
Joined 2003
yunyun said:
Raj1,

This is not my assumption, the fact that it exists. The digital circuit generally only needs 1A to be large enough. I'm curious, as far as you are concerned, besides using LinearPi to charge UcConditioner, what else can you do for LinearPi?
Click to expand...

I don't use one. But if I did, I'd use it to power a circuit that is impractical for use with super capacitors or Lipo batteries. Im sure that's not too difficult for someone like you to work out. You first posed this question, ignoring a post by a user that specifically asked Ian if the supply could handle more than 1A. Ian subsequently tested and responded to that specific user. If the usage model doesn't apply to you, pick or design a product that does.
 
Last edited:
S

Studley

Member
Joined 2005
iancanada said:
@ clsidxxl

Thanks of the information of the wire. It looks very good. I'm interested in buying some.

UcConditioner can be looked upon as a kind of passive power supply. To keep the best possible low noise and dynamic performance(such as ultra capacitor or battery), all passive power supplies need connection resistance as low as possible. Silver Plated wire can have even lower resistance and better high frequency response than the pure copper wire with the same size. So it would be a very good option, and the 16 AWG... should be big enough :).

Ian
Click to expand...

Suggestion: how about offering a suitable USB C - USB C cable for use with the UC Conditioner ?
 
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
@Studley,

USB-C connectors make is very easy to getting start to power RPi or other board with UCB type C power connector, or from a USB power adatper. Then you can change to good power wires at any thing later.

UcConnector 5V has the USB-C connector for both input and output. But UcConditioner 3.3 doesn't have USB-C connector. You know the reason.

Maybe somebody knows audiophile grade USB-C cables. Mine was bought from Amazon. It's good and I'm happy with it. But I think there could something else better than.


Regards,
Ian
 
S

Studley

Member
Joined 2005
USB C cable

“Something better . . . “. That’s what I was thinking when I made the suggestion. My guess is that off the shelf cables are not designed to handle say 3A and will have too much resistance. The USBridge Sig that I want to use only has USB C DC input, no terminals for bare wire :(
 
andrea_mori

andrea_mori

Banned
Joined 2005
RaspBerry and FifiPi phase noise measurements

Hi Ian,

I could do the measurements for you.
With the Timepod I can measure the phase noise of the clock signals at the input and at the output of the FifoPi after reclocking.

I also own a sniffer to do a spectrum analysis of the RFI/EMI close to the RaspBerry.

Since I don't own your devices, if you want me to do the measurements you should send me a full setup including RaspBerry and oscillators (22/24 MHz because the Timepod has the upper limit of 30 MHz). After the measurements I will send them you back.

I believe we are a community, so I'm very glad to help when possible.

Andrea
 
Y

yunyun

Banned
Joined 2019
iancanada said:
@Studley,

Theoretically UcConditioner can turn it to a perfect ideal power supply. But we live in a real world, we have to face the reality. The ESR of an ultra capacitor is not constant. It increases along the frequency. Though the ultra capacitor can still remove the high frequency noise(based on real test), but will be not as perfect as low frequency. In this case, good low noise linear power supplies still make difference.

Besides the ultra low noise performance, LinearPi has very good dynamic load response. That will be really great to work with UcConditioner to set up a best possible power supply for ROi and other application. I highly recommend using the on-board on/off control with keeping the AC input always connected for your audio applications.

Ian
Click to expand...

iancanada said:
@yunyun

Ultra capacitors in UcConditioner are storage components. They don't consume energy. During conditioning stage(Conditioning LED ON), UcConditioner doesn't take additional current from LinearPi (in dynamic only). LinearPi only need to deliver current as much as the load is demand.

UcConditioner only takes current to charge at first time when empty. Even at that time, LinearPi is still enough to power both.

Ian
Click to expand...

Since During conditioning stage (Conditioning LED ON), UcConditioner doesn't take additional current from LinearPi, even if LinearPi has super good performance, what outstanding value can LinearPi have? . Everyone knows that the discharge quality of the super capacitor is not related to the LDO for charging, because the super capacitor is originally an independent discharge element.

Supercapacitors are similar to batteries, except that they charge and discharge longer and have a longer life. In other words, UcConditioner can replace expensive battery-powered modules, it really does not need to provide power to UcConditioner through the battery. This principle is also the super capacitor is originally an independent discharge element.
 
Last edited:
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
yunyun said:
Since During conditioning stage (Conditioning LED ON), UcConditioner doesn't take additional current from LinearPi, even if LinearPi has super good performance, what outstanding value can LinearPi have? . Everyone knows that the discharge quality of the super capacitor is not related to the LDO for charging, because the super capacitor is originally an independent discharge element.

Supercapacitors are similar to batteries, except that they charge and discharge longer and have a longer life. In other words, UcConditioner can replace expensive battery-powered modules, it really does not need to provide power to UcConditioner through the battery. This principle is also the super capacitor is originally an independent discharge element.
Click to expand...

@yunyun
UcConditioner takes care of dynamic demand current while LinearPi takes care of continuous current.

You can think of it in this way. UcConditioner is a big pond. LinearPi is the pump to supply the pond with water. And at same time you are using water to wash your car at the other side of the pound, the amount of water you are using keeps changing dynamically.

No matter how much water you are using, the level of the pond doesn't change because it's too big. But to keep the pond level no change for long term, the pump has to supply a continuous water flow into the pond. The continuous water flow equals to the averaged demand water that you are using.

If the pump did a bad job with making too much wave on the surface of the water, the wave can still be transferred to the other side of the pond (high frequency noise). So a good pump with very smooth (low noise) and stable (good response) output is also very significant.

I hope you can do a circuit simulation by yourself so that you can understand the principle more.

Ian
 
Last edited:
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
NicMac said:
Is this true Ian? Is LinearPi not continuously replenishing the UCs on UcConditioner (i.e. powered and connected)?
Click to expand...

Under conditioning state, the current that the UcConditioner takes from LinearPi equals to the averaged dynamic current it delivered to the load. So, overall UcConditioner doesn't take additional current from LinearPi . The output current of the LinearPi (attached to the UcConditioner) is always equal to the averaged demand current of the load.

Ian
 
S

Studley

Member
Joined 2005
iancanada said:
@yunyun
UcConditioner takes care of dynamic demand current while LinearPi takes care of continuous current.

You can think of it in this way. UcConditioner is a big pond. LinearPi is the pump to supply the pond with water. And at same time you are using water to wash your car at the other side of the pound, the amount of water you are using keeps changing dynamically.

No matter how much water you are using, the level of the pond doesn't change because it's too big. But to keep the pond level no change for long term, the pump has to supply a continuous water flow into the pond. The continuous water flow equals to the averaged demand water that you are using.

If the pump did a bad job with making too much wave on the surface of the water, the wave can still be transferred to the other side of the pond (high frequency noise). So a good pump with very smooth (low noise) and stable (good response) output is also very significant.

I hope you can do a circuit simulation by yourself so that you can understand the principle more.

Ian
Click to expand...
Excellent analogy Ian
 
You must log in or register to reply here.
Top Bottom