Oversample to DSD128 Into DS SR via USB?


The power inlet, the power cord, the “wall”. in the diagrams I showed a battery, but really you’ll probably have a power transformer, fuse, maybe an AC switch, etc. but none of these really filter the hash that the power supply is kicking up upstream.


Cheers @tedsmith

One more thing, the hash you refer to is just voltage spikes and RF? Or hash is a broader term that includes leakage currents and other things? Sticking to SMPS’s only.

I’ve probably misunderstood stuff I’ve read from a couple other designers about hash being put back into mains (and being a problem), so I’ll re-read everything again and PM you to discuss at some point, to avoid things looking like a pissing contest.


If those experts also poo-poo better power cords, etc. then you should discount their statements about switching power supplies. Read the link I gave.

Here’s another simpler doc with some graphs showing SMPS’s causing equipment to not meet emissions standards. It shows some measurements at the outlet: https://www.murata.com/~/media/webrenewal/products/emc/emifil/case/household/k_3.ashx

The biggest issue is current spikes which may or may not make an appreciable voltage spike, but current spikes cause radiation at high frequencies - the power cord can radiate these to other parts of your system.

Here are some other papers discussing these kinds of problems:
The second talks about some mechanisms whereby RF can cause problems in analog circuits.


Ha cheers Ted. I have a lot of reading and re-reading to do.


Although I don’t take everything as truth, Audiosciencereview measured the LPS 1 and found it didn’t block leakage current. In an identical test in parallel, an Sbooster passed 0 leakage. This is why I’m unclear on the benefit of the LPS-1 when a good linear supply accomplishes no leakage and doesn’t emit RF.

Of course Alex from Uptone argued back and forth and claimed that what Amir thought he was observing wasn’t leakage but was RF emitted through the air and picked up by the cables. It certainly looked like bad leakage to me. Either way the DAC was significantly polluted with noise.

They have created a cult status and a bandwagon and people claiming it’s the best power supply out there based 100% on their claims.


Correct and John Swenson gave a simple trick to fix this a long while ago. I mentioned above, grounding the DC output of Mean Well powering the LPS-1. This shunts what he calls ‘high impedance leakage current’ to ground. The LPS-1 then blocks most of the remaining ‘low impedance leakage’.

He also recommends this same cheap trick with a network switch, after finding these high impedance leakage currents sail right through ethernet’s transformers. He found it works well with a NetGear GS105 or GS108.

He showed measurements too…


My issue with all of this is are the big claims that the ultra-capacitor banks provide 100% galvanic isolation. I’m not sure how that is true when you have to shunt the ground. It shouldn’t matter if the technology works the way they describe. If the second bank of ultra-capacitors provides the galvanic isolation then it should be immune to leakage upstream no matter what.

For instance, what would you think if you had a toslink link to your DAC and a manufacturer suggested that you had to shunt the ground of your blu-ray player to cut down on AC leakage? The math just doesn’t add up.


I understand and many of us LPS-1 owners were initially frustrated at John’s findings about these ‘high impedance leakage currents’ sailing right through even his LPS-1. Important to also note that he also found these high impedance leakage currents MAY also sail right through most digital isolators typically used inside USB DACs (that may promise 100% USB galvanic isolation) and even go through the Intona’s and his own ISO REGENs of the world. Shunting the high impedance leakage works he found, by grounding the PSU output if using SMPS’s.

So many (not all) of us got over the initial disappointment by using his cheap solution with the LPS-1 but also his findings about using this same cheap method to effectively block leakage currents getting into a networked streamer/endpoint/DAC, all without having to use a linear PSU…

I’m not sure anyone ever claimed perfect isolation as good as optical though? That math definitely wouldn’t add up, agreed. Most USB DAC designers will tell you, even if they have USB input isolation internally, there is still some capacitance… I haven’t seen anyone claim perfect isolation of a USB DAC input, unless they use optical isolation internally somehow? I haven’t seen that yet but it may be out there.


Another way to look at it @cxp - John S could have kept quiet and just improved things silently with his next version LPS-1.2 and not told anyone about his findings of ‘high impedance leakage’ sailing right through his LPS-1 and ISO REGEN etc etc.

But he had the class and integrity to share his findings publicly and openly, while at the same time sharing a cheap effective solution which didn’t mean having to buy linear PSU’s (and he showed measurements of his findings too).

So while I was personally frustrated INITIALLY, I came to really respect John S even more. He’s sharing some of what he learns along the way and explaining the technical mechanisms too and sharing some measurements.


Yes I agree with you 100% on that.

I am definitely excited for the EtherRegen.


Getting back to the OP’s topic - I’ve got to say I’m a fan of upsampling everything to DSD128 over usb to the DSJ.
I’m using an i7 mac mini with Roon/HQPlayer.
The initial PCM filter I’m using in HQP is the poly-sinc-xtr-2s (linear polyphase sinc filter with extreme cutoff and attenuation) along with DSD5 (rate adaptive fifth order DS modulator) to DSD128.
This is currently the best sound I’ve heard from the DSJ in my setup - though I can’t claim to have experimented exhaustively with all the available filter options (yet).
One thing I have noticed while experimenting with HQP and the Audirvana iZotope filter settings is that I tend to prefer the linear phase over minimum phase types.
If you have a machine capable of running HQP its worthwhile experimenting with the free trial version to see if its something that tickles your fancy.


I did try upsampling to DSD 128 with Roon. Didn’t use any other parts of the Roon DSP engine, like filters. I had a very slight preference for no upsampling. However I have Roon installed on a 2008 iMac that has been doing double duty as our household computer in a room separate from the dedicated music room. It works fine using JRiver to stream ripped CDs and LPs recorded at 24 / 96 to the BridgeII module. But it doesn’t have enough processing power for upsampling because get frequent dropouts. So my test of upsampling may be faulty because even when it works w/o dropouts, it may not be working optimally.

BTW I have grounding on the DC output of the Mean Well powering my LPS-1.


I’ve tried DSD128 upsampling with both Audirvana and Roon’s built in functionality but IMO they don’t achieve the superior SQ results of HQPlayer. When using Audirvana I settled on no upsampling, just sending native rates to the DSJ.
Using the HQP settings I stated above my 2.3Ghz i7 Mac Mini is only working at about 15% cpu load. There are more cpu intensive HQP filter settings that do induce stuttering on my Mac. Some HQP advocates are using crazy powerful CPU’s to do maximal filter settings and DSD upsampling. I can’t speak to the efficacy of this, but it’s moot with the DSJ anyway since it only accepts up to DSD128.