Good to see you went for 3A.
The "Sbooster" guy was commercially quite successful by adding a separate filter and a rather small buffer in comparison to yours at the end of the cable.
The "Sbooster" guy was commercially quite successful by adding a separate filter and a rather small buffer in comparison to yours at the end of the cable.
Yeap the new filter will have a "screw out" so you can have quite a small length cable (awg 16) going to audio device. Vin is by dc jack and usb (3A rated on all inputs and outputs)
But this new active filter is really something nice. -50db from 10Hz to 20Khz , add the inductor attenuation as well. Should make most (99%) smps quiet as a battery . (20uV)..at full 3A
But this new active filter is really something nice. -50db from 10Hz to 20Khz , add the inductor attenuation as well. Should make most (99%) smps quiet as a battery . (20uV)..at full 3A
Soundcheck , this is exactly like the Sbooster pcb at the end of a PSU ..except for the output we will have a screw connector and that we have both a passive and active filter + supercap
These are exciting developments, glad I chanced upon this thread. I was about to buy an ipower, but may just hold off for a short while... 🙂
Design and routing its over. We will send on Monday to PCB house. After assembling the samples we need to test and tweak .
But yeah...looks like a very interesting PSU indeed. No such PSU was made for 5V
But yeah...looks like a very interesting PSU indeed. No such PSU was made for 5V
As soon as we get the PCBs back we will start testing. After testing OK its usualy 6 weeks to mass production.
Realistically its at least 10 weeks away.
Pricing will be competitive well under 100$
Realistically its at least 10 weeks away.
Pricing will be competitive well under 100$
Will be looking forward to that. In the meantime, to go back to Soundchecks original question. Soekris is using Mean Well supply in his dac.
http://www.soekris.dk/images/dac1541_open_1200.jpg
Any comments to his solution?
Does this give any ideas to how the Mean Well could be "tamed"?
http://www.soekris.dk/images/dac1541_open_1200.jpg
Any comments to his solution?
Does this give any ideas to how the Mean Well could be "tamed"?
I have been trying Meanwells and MRAVLCAs TPS7A4700 modules. I also have several Sbooster supplies over here. I stayed with the iFi iPower @ $50.
Let see if Allo can beat that @ "well under" $100. 😉
To be fair I have to put my 47k Rifa buffer and some filter caps to the BOM around the iFi.
Let see if Allo can beat that @ "well under" $100. 😉
To be fair I have to put my 47k Rifa buffer and some filter caps to the BOM around the iFi.
We have been testing since Friday. Of course many problems...we got ripple of 1000mV 🙂
1000mV = 1V out of 5V !?!? 😱
Wow. You (or your supplier) managed to build the worst supply ever. 😀
I was actually hoping to play around with one of your devices before Christmas.
Christmas 2017... 😉
Let see how things are developing.
Down to 5mV...before the active noise filter . Still we need to get it down to 2mV. Its not easy..
From 2mV the active filter takes it down to 4uV (0-100Khz)
From 2mV the active filter takes it down to 4uV (0-100Khz)
Taming the beast.. different values for RC snubbers to quiet the emi and noise/ripple of the last diode (output side)
There is a significant difference between 1nf and 20nf *and much more whiteout any RC snubber)
There is a significant difference between 1nf and 20nf *and much more whiteout any RC snubber)
0-100Khz 275uV
1Mhz 875 uV at 3A
Last design of the PCB has started , we hope to decrease further
1Mhz 875 uV at 3A
Last design of the PCB has started , we hope to decrease further
Once your're done with the finetuning and related measurements...
...don't forget to run listening tests! 😉
While we're at it...
We already discussed that future Allo HATs should all have the same powering scheme (jacks and jumpers to disconnect power rails from lower and upper boards). Meaning. The user should be able to easily attach a single 5V supply to a complete tower or power a group of boards (e.g. Kali and DAC) or power each board separately.
Obviously the target behind it is primarily to achieve a little better performance and not boosting your product sales. (This would be a nice side-effect from your perspective though 😀)
Enjoy.
...don't forget to run listening tests! 😉
While we're at it...
We already discussed that future Allo HATs should all have the same powering scheme (jacks and jumpers to disconnect power rails from lower and upper boards). Meaning. The user should be able to easily attach a single 5V supply to a complete tower or power a group of boards (e.g. Kali and DAC) or power each board separately.
Obviously the target behind it is primarily to achieve a little better performance and not boosting your product sales. (This would be a nice side-effect from your perspective though 😀)
Enjoy.
Are you sure that 1uV of IFI is only ripple and no SW frequency and mains ? Today we have reached 220uV 0-100Khz (switching noise included) and 600 uV at 1Mhz (2.5A tested)
Check new Boss v1.2..
Check new Boss v1.2..
there are varous design choices related to the switching frequency vs the audio clock frequencies used in the system. one is to use a fixed dcdc frequency, like possible in the zvs resonant design, giving up some input range (although there are 3phasic options to combat that) and burn away some heat linearly to adapt to input voltage variations.
the fixed frequency can be set to lay either outside the audioband of interest, the harmonics will at some point conincide with clock harmonics. the next option is to sync the dcdc freq to a multiple of the audio word clock , for example 192kHz or 384khz .
the resonant designs have a considerably lower common mode noise due to weaker coupling between secondary and primary. and that is the noise that enters into the system, not the nosie across the terminal..
the fixed frequency can be set to lay either outside the audioband of interest, the harmonics will at some point conincide with clock harmonics. the next option is to sync the dcdc freq to a multiple of the audio word clock , for example 192kHz or 384khz .
the resonant designs have a considerably lower common mode noise due to weaker coupling between secondary and primary. and that is the noise that enters into the system, not the nosie across the terminal..