I agree , thats why I am very careful with my statements on the competitors devices. Still , IFI is testing devices in environments that have nothing to do with how they are used . In fact you can see independent measurements all over the internet (audiosciencereview and other forums) that clearly shows (unless every device is galvanically isolated) that noise on the GND is pretty high.
Our testing is done in real world connection scheme . Also I will submit our PSUs for independent testing. I am very confident in our data, noise is lower than all our competitors including Gnd noise
I will say it for the last time , Nirvana SMPS is testing in a class of its own . To compare , its very close to the very best LPSs on all noise measurements .
Just had another look at the LPS image.
I do not see any more connection options then these two flimsy cables !?!?
Ony key element of powering a HAT tower would be to avoid long low voltage and low quality cable runs. (I run a custom wire @ 3inches max from PS->HAT)
Not to forget. Bypassing poor power sockets such as USB-C/uUSB is another
key element. (I bypass all of them) And that requires custom wiring as well.
What sense does it make to take an extraordinary well done PS and then you loose it on the connectivity part!?!?
Especially these connectivity questions impact the "transient" discussion heavily. Of course it goes on with ground loops, RFI/EMI pickup asf. in this context.
I do recall that Allo had shown us screw terminals in the early days.
Decisions and compromises .
I have been always a fanatic when it comes to audio quality . I always try to push the envelop and not to make any concessions in the pursuit of perfection .
Still , like most of us learn , like in my marriage , I realize now that concessions are necessary for the greater good. Same goes for my work.
Nirvana and LPS have both cables of 18AWG with a regular barrel connector at the end . You will need adapters depending on your end device.
This is not ideal. Screw in is best (even though last connection will always be an adaptor). So with screw in , you will save one adapter connection . (or if your end device has correct barrel its equal)
However the complexity and price of the device will increase. Now most of people will be perfectly happy with a moderate priced device that will pass through the 2 adapters "easy peasy" . Fanatics (including me) might find that suboptimal. Not to worry , unscrew the front plate(Nirvana or LPS) , use the cables that you want and solder them to the PCB.
As I said , compromise . We have 2 types of customers , die hards and people that enjoy this hobby and appreciate "ease of use" and good pricing.
We need to design for both.
Right now , the LPS MSRP will be less than 160$ and Nirvana less than 65$
Will both PS have 5V output only or will they be available with other voltages too?
Thanks
Matt
That'd be the same with noise. What noise? What measurement!?!?
It's not impossible to measure. Comparing the measurements is impossible.
To get around it you also define a reference load to run your noise measurements.
If noise would be the only concern, all powers supplies with a certain
SNR would have to perform the same way. They don't.
I do understand very well that this is non of your concerns. You can't prep
for all kind of loads out there.
You could just take your reference load as you do for noise.
The IFI noise document seems to be part factual and part obfuscation. There is going to be common mode and differential mode line noise in the real world that will vary from time to time and by locale. It isn't noise generated internal to a power supply, but how well a power supply rejects or attenuates such noise from input to output can be of great importance. IFI skirts around the subject because they don't use something like R-cores or equivalent technology to help attenuate line noise incursion (at least not to the extend they want to brag about it). However, it is not right to measure line noise or ground noise as absolute noise, because it depends on how much line noise there is to begin with. Better to describe it as attenuation vs frequency for both common and differential modes. In other words, there should be a graph with two curves, one for each mode, showing line noise attenuation vs frequency. The above entirely IMHO, of course.
Last edited:
We are gearing for production run of 500 units to start on LPS and another 500 3 weeks after. We might open for preorder 1 week before shipping .
Noise on Gnd comes from 2 sources , AC noise (less) and leakage current (more). Once you test with AP (or other test equipment) leakage current will transform into noise . This is why IFI is using isolation transformers and grounding schemes. In fact they only measure differential noise and clean the AC side with isolation . Of course this is why noise appears on all independent testing of the IFI .
Nirvana is tested under real world scenarios , where one of your devices will be earthed . Leakage , Differential and common mode noise was tested .
For the moment we are planning on the 5V only , might have a version with 9V in the future
There is no regulator but rather a modified CM with excellent performance and very low dropout voltage (0.55v). Noise was reduced before reaching the filter by using multiple caps , inductor on DC. Still , the main reduction of noise comes from snubbers and RDC networks on primary side.
At last we used CMC (ac line filter) and multiple caps on AC side.
So what I am trying to say is that noise reduction cannot be achived only thinking of the output filters but rather noise reduction has to be implemented at every stage , preferably reduced at the noise source
Last edited:
A slightly twisted view at least on the common mode part.
iFi gets rid of common mode noise altogether.
"Less is (can be) More"And that at close to no cost for them. This way you can land at 49$ or below.
They simply follow the twisted-view logic and hype of getting everything isolated. Nowadays pretty much all Audio manufacturers offers "isolators/isolation" as one of the key features.
"Everything being isolated !?!?" Nope! Unfortunately not.
And that's the main issue here. Most people simply don't manage to run a full isolated (floating) system.
And... ...there's nothing like a half isolated system.
You can't just close your front door and leave the back door open.Most people IMO simply don't manage to get a proper grounding scheme in place.
Let's have a look at the options for us users.
At least from my perspective that leads to two approaches.
1. full isolated - no common mode noise with e.g. ifi iPower or batteries
which requires all other devices being powered the same way - isolated.
And an e.g. starground scheme inside the float.
2. non isolated/heavily filtered e.g. Nirvana
which also requires heavily filtered supplies on the other devices, the
"backdoors" and of course also a proper grounding inside the system.
With all the experts around, I'd really like to see what'd be the best approach
to get to best results for a "system" and not just for a single device.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.