Hello ladies and gents.
Music and linear power supply has been always a good match . The main problems of LPS is extremely inflated pricing and no data.
You see not all LPSes are equal . Just using a transformer will not automatically give you a better PSU. Worst is that most of manufacturers include no testing . no data that can help chose one.
We mean to change that.
First we started with the transformer . There are many types of transformer , some are worst than others (noise coupling , electromagnetic interference)
We chose one of the best transformers for noise ,r-core
The transformer has a dual output (galvanically isolated ) so you can power multiple devices (RPI + dac , Katana , etc) and we used independent (of course since they are isolated) filtering on each rail. First a capacitance bank (massive) to reduce the 50/60Hz noise , then a active filter , fallowed by an LDO to keep voltage stable . Since LDOs have poor transient and impedance , we used a last active filter (with good transient) and then capacitance but most importantly , super capacitors on the output . So all power comes in fact from supercaps..So yes , transients and impedance of the LPS is..good.
Of course , Shanti includes AC line filter and AC is earthed to casing.
A few words on what we achieved . Measuring noise (in real word scenario) we see about 80nV (0.08uV of noise from 0-20Khz at the floor of our AP machine. This is comparable to batteries from our internal testing .
As explained , Shanti is not only shining because of the ultra low differential noise, but also because of incredible low impedance / deep reservoir of electrons at the output of the PSU ( thx you soundcheck for your suggestions )
At last ... pricing . We are working very hard to make this LPS the lowest priced on the market . All components are on the way , it will take a few weeks to start accepting orders. Questions and comments are welcomed
Music and linear power supply has been always a good match . The main problems of LPS is extremely inflated pricing and no data.
You see not all LPSes are equal . Just using a transformer will not automatically give you a better PSU. Worst is that most of manufacturers include no testing . no data that can help chose one.
We mean to change that.
First we started with the transformer . There are many types of transformer , some are worst than others (noise coupling , electromagnetic interference)
We chose one of the best transformers for noise ,r-core
The transformer has a dual output (galvanically isolated ) so you can power multiple devices (RPI + dac , Katana , etc) and we used independent (of course since they are isolated) filtering on each rail. First a capacitance bank (massive) to reduce the 50/60Hz noise , then a active filter , fallowed by an LDO to keep voltage stable . Since LDOs have poor transient and impedance , we used a last active filter (with good transient) and then capacitance but most importantly , super capacitors on the output . So all power comes in fact from supercaps..So yes , transients and impedance of the LPS is..good.
Of course , Shanti includes AC line filter and AC is earthed to casing.
A few words on what we achieved . Measuring noise (in real word scenario) we see about 80nV (0.08uV of noise from 0-20Khz at the floor of our AP machine. This is comparable to batteries from our internal testing .
As explained , Shanti is not only shining because of the ultra low differential noise, but also because of incredible low impedance / deep reservoir of electrons at the output of the PSU ( thx you soundcheck for your suggestions )
At last ... pricing . We are working very hard to make this LPS the lowest priced on the market . All components are on the way , it will take a few weeks to start accepting orders. Questions and comments are welcomed
Last edited:
Internal testing was done with multiple DACs , some from allo some from topping and other companies . In all good DACs (thd+n) better than -100 , we have seen THD+N getting better by at least 1db , and some with 1.7db (max)
nice price.. any impression on its use with allo products.. Katana, digione sig, boss 1.2?
I can only comment on what AP machines test data shows. Katana has 1.4db better THD+N using Shanti. Noise is the dominating factor in Katanas THD+N and by reducing the noise it gets a better number.
As for SQ.. no comments on my subjective hearing.
As for SQ.. no comments on my subjective hearing.
cdsgames..
any safety certification being sought for this power source device.. UL, ICC et al?
how do you guarantee the safety of consumers from using this product?
take note that allo used to have issues with the battery adapter used by John Darko..
any safety certification being sought for this power source device.. UL, ICC et al?
how do you guarantee the safety of consumers from using this product?
take note that allo used to have issues with the battery adapter used by John Darko..
Certification under way FCC/CE. Marudo please note that battery holder had the wrong polarity on the units and despite our advanced warning the issue happened to Darko .
I looked up the PDF.
What hit me was...
3A and 1A ??
The recommended PS for a PI 3B+ should deliver about 2.5A.
OK. We know it can be run at less then 1A.
Still. 2.5A is the official recommendation.
That would leave the 2nd output of 1A for the Katana !?!?
That obviously wouldn't work! Right!?!?
What is your position on this?
What hit me was...
3A and 1A ??
The recommended PS for a PI 3B+ should deliver about 2.5A.
OK. We know it can be run at less then 1A.
Still. 2.5A is the official recommendation.
That would leave the 2nd output of 1A for the Katana !?!?
That obviously wouldn't work! Right!?!?
What is your position on this?
The output of the second line is in fact 1.5A and 1.8A on transients so yes enough power for Katana + RPI , or Digione Sig (clean and dirty side)
We are changing the markings on mass production units.
We are changing the markings on mass production units.
Ok. That looks better.
I usually account for about 20% headroom to be on the safe side.
PS are usually speced with continuous and peak power and overall wattage.
Would be nice to get all values listed for both outputs.
It would also be nice to see the output impedance.
Can you show us a datasheet?
Ripple, noise etc.
I usually account for about 20% headroom to be on the safe side.
PS are usually speced with continuous and peak power and overall wattage.
Would be nice to get all values listed for both outputs.
It would also be nice to see the output impedance.
Can you show us a datasheet?
Ripple, noise etc.
And.
The output voltage.
Is it straight 5V?
Many of us would appreciate a little headroom voltage to be able to feed
a 5V regulator. Something like 5.4V would normally do.
Can the outputs slightly be adjusted?
The output voltage.
Is it straight 5V?
Many of us would appreciate a little headroom voltage to be able to feed
a 5V regulator. Something like 5.4V would normally do.
Can the outputs slightly be adjusted?
The voltage is in fact 5.1V. Since the beginning we designed it with some headroom You can slightly go higher at 5.2V with no hit on performance (by adjusting a small POT)
In about a week we will publish all data. We never took impedance readings , so we need to study. Still with the ultracaps right next to the DC wire output, it should be good.
If really necessary 5.4V , its possible to adjust the LDO by changing the R divider , but some soldering will be required
Come on CDSGAMES. This is no DIY hack. 🙄
Some high quality devices clearly state impedance requirements.
Having the lowest possible ESR is not always a good thing.
The load heavily interacts with the supply and can even start oscillating.
This would be a nice read to see what I'm talking about.
We will measure.
However , we are talking impedance at around DC (low frequency) since the PSU is far from load and all HF decoupling must local (next to IC)
Impedance of our lytics + supercaps around DC is quite good (from datasheet)
This is why I am stating " "it should be good"
The output of Shanti is direct from supercaps + around 10.000uf of caps (a few in parallel). Further the active filter has a ESR of about 50mOhm and before that you have another 4500uF of capacitance .
However , we are talking impedance at around DC (low frequency) since the PSU is far from load and all HF decoupling must local (next to IC)
Impedance of our lytics + supercaps around DC is quite good (from datasheet)
This is why I am stating " "it should be good"
The output of Shanti is direct from supercaps + around 10.000uf of caps (a few in parallel). Further the active filter has a ESR of about 50mOhm and before that you have another 4500uF of capacitance .
Last edited:
Did you read the article?
It's not about good or bad. It's about matching the load.
As you've seen on Katana earlier. Alternating loads can modulate into the power rail,
causing severe noise through the backdoor.
It's always good to have a look at lab supplies e.g. this as reference.
You'll find a lot about transition times, slew rates asf asf. For a reason!
It's not about "good enough" or "trust me". It's about knowing what you're doing.
And there's a lot to gain for all of us.
Estimated 99% of the projects and findings over here at DIYA are based on trial and error experiences.
And that's IMO not good at all.
It's not about good or bad. It's about matching the load.
As you've seen on Katana earlier. Alternating loads can modulate into the power rail,
causing severe noise through the backdoor.
It's always good to have a look at lab supplies e.g. this as reference.
You'll find a lot about transition times, slew rates asf asf. For a reason!
It's not about "good enough" or "trust me". It's about knowing what you're doing.
And there's a lot to gain for all of us.
Estimated 99% of the projects and findings over here at DIYA are based on trial and error experiences.
And that's IMO not good at all.
The paper you are referring to is referenced to a system with feedback . Our Nirvana has a feedback loop , only feedback on Shanti is on the LDO thats well behind output.
Shanti has been tested extensively with different loads manually , but we don't have specific machines to test with variable load.
At last , the way that we designed Shanti was to have full power on output. Even a load that changed from 0.5A to 3A in 10ms , will receive all electrons from output of Shanti . 1F =1V at 1A for 1second
Rest of Shanti power will slowly ramp up as supercaps + cap bank discharge .
So to summarize , Shanti does not have a real feedback loop. Peak power requirements are taken care by output supercaps + cap banks
I am yet to see such architecture in any LPS, but I am sure others will fallow.
It just makes sense.
Shanti has been tested extensively with different loads manually , but we don't have specific machines to test with variable load.
At last , the way that we designed Shanti was to have full power on output. Even a load that changed from 0.5A to 3A in 10ms , will receive all electrons from output of Shanti . 1F =1V at 1A for 1second
Rest of Shanti power will slowly ramp up as supercaps + cap bank discharge .
So to summarize , Shanti does not have a real feedback loop. Peak power requirements are taken care by output supercaps + cap banks
I am yet to see such architecture in any LPS, but I am sure others will fallow.
It just makes sense.
Exacly . And how it was solved ? by adding capacitance on the last wire (after LDOs). Same for Shanti
Last edited: