Taiko Audio SGM Extreme : the Crème de la Crème

[Marcin_gps]

That is intriguing.

My assumption is that the amp I was using to drive the Kawero speakers was not powerful enough and the m12 switch made such a difference, because it significantly improved dynamics and the speakers seemed as they were getting much more power at the time.
But this is definitely not the case for Bob who is using Pilium (plenty of power)

 

[onlychild]

Your results do appear to be quite "Extreme". It would be very interested in hearing from others as well.

I'll be getting my Edge router soon. Like Bob, I also have a very busy Verizon modem. It is currently in my basement and is sending ethernet to my office on the 2nd floor through a cheap ethernet cord that I can not replace since it was put in when the house was built.

Currently, that cheap ethernet cord (maybe 50+ feet) goes into a FMC (powered by a LPS 1.2) which sends Fiber into the Extreme. I plan to have the ethernet cord feed the Edge Router which will then send fiber from its SFP port into the Extreme.

I only play locally stored files, so I am curious how much of an improvement I may see "offloading" from the heavily used Verizon router.

Will report back.

 

[Rhapsody]

I'll be getting my Edge router soon. Like Bob, I also have a very busy Verizon modem. It is currently in my basement and is sending ethernet to my office on the 2nd floor through a cheap ethernet cord that I can not replace since it was put in when the house was built.

Currently, that cheap ethernet cord (maybe 50+ feet) goes into a FMC (powered by a LPS 1.2) which sends Fiber into the Extreme. I plan to have the ethernet cord feed the Edge Router which will then send fiber from its SFP port into the Extreme.

I only play locally stored files, so I am curious how much of an improvement I may see "offloading" from the heavily used Verizon router.

Will report back.

I was just talking to someone else that is attempting to install the Edge, BUT you need a pc with a lan connection to log on to the Edge router. If you have only say a MacBook Air or ??? that doesn't have a lan output, you can't log onto the GUI of Edge. You probably know this, but.....

 

[cat6man]

I was just talking to someone else that is attempting to install the Edge, BUT you need a pc with a lan connection to log on to the Edge router. If you have only say a MacBook Air or ??? that doesn't have a lan output, you can't log onto the GUI of Edge. You probably know this, but.....

easily solved.
google 'usb to ethernet adapter'
i have a dell xps laptop running linux and never travel without it.

 

[onlychild]

I was just talking to someone else that is attempting to install the Edge, BUT you need a pc with a lan connection to log on to the Edge router. If you have only say a MacBook Air or ??? that doesn't have a lan output, you can't log onto the GUI of Edge. You probably know this, but.....

Purchased this about 5 years ago and looks like I will finally get to use it

 

[Taiko Audio]

Today I'd like to share a collection of measurements of power supplies taken from the power grid side of things. The setup is a power strip in which I've plugged in several power supplies to measure what they radiate on to the grid.

Let's start with the AC waveform:



What you see here is a flattening of the sinusoidal tops of a 50Hz waveform. This is caused by linear power supplies by the following mechanism:
We have in all its basics a transformer, a rectifier bridge and a capacitor filter bank. When the voltage stored in the capacitor bank drops below the value present on the transformer secondary, the rectifier bridge opens which allows the capacitor bank to recharge. When the voltage in the capacitor bank reaches a level similar to what is present on the transformer secondary, it closes again. Practically it will only draw power for a certain period on the top of the sinus, causing a flattening of peaks.

Now let's take a look at the spectrum analysis of a few linear power supplies in action:



Here you see harmonics becoming visible at 50Hz increments, 50Hz, 100Hz, 150Hz, 200Hz etc



An expanded scale up where you can see harmonics are quite visible up to 1KHz, expanding the view further you don't see much anymore.

A different LPS:



This LPS has a different harmonic spectrum, with harmonics every 100Hz, 50Hz, 150Hz, 250Hz etc. This one probably uses a half or full wave rectifier in stead of a bridge.

Now a 3rd LPS, this one as harmonics up to 3KHz and some resonance and/or oscillations from ~ 17 to 30KHz.

 

[Taiko Audio]

Continuing:

A SMPS:



Here you can see it switches at around 42KHz with spread spectrum applied which is quite common to flatten the peak in order to pass EMC testing. The harmonics at around 84KHz are pretty decent supressed.

A second SMPS:



This one has less well filtered harmonics, but the main switching frequency is a bit lower in magnitude.
If we expand it to a more broadband view we see some artefacts in the 500-700Khz range aswell:

 

[Taiko Audio]

Now what happens if we connect all of them to the same powerstrip?



We see an interaction going on, showing peaks, oscillations and resonances not visible in their individual spectra.

 

[Blackmorec]

Now what happens if we connect all of them to the same powerstrip?

View attachment 64038

We see an interaction going on, showing peaks, oscillations and resonances not visible in their individual spectra.

Very interesting indeed. Few questions if I may?

• What does the mains look like in Hz, kHz and MHz regions at the power strip with no power supplies connected? At the socket with no power strip connected?
• How is the power-strip built? Star earth, buss bars, anti-surge protection?
• What type of power cable was employed to connect the PSs? Always the same?
• Could this explain why certain power strips are in general to be avoided (At least in my experience)?
• Could this explain why that in my subjective observation, powering an entire system including network with a single type/brand of power supply seems to give superior results from a SQ perspective?
• Would it also explain why (again IME) adding an SMPS to an otherwise purely LPS powered system and network was so detrimental to SQ?

 

[cat6man]

Now what happens if we connect all of them to the same powerstrip?

View attachment 64038

We see an interaction going on, showing peaks, oscillations and resonances not visible in their individual spectra.

could you please explain the Y axis scale, dB relative to what and how measured, power supply terminated how?

cheers

 

[Taiko Audio]

These are all the measurements I have saved.

Powercords and/or powerstrips are not going to have a mayor impact on this type of measurement, their influence would be limited to very small amounts of inductance, capacitance and some effects of shielding, which would mainly shift frequencies in this case.

The interaction in the last graph is an interaction of a complex circuit consisting of capacitances, inductances and resistances within a wide range of frequencies (although this is still quite limited in the complete spectrum).

 

[Taiko Audio]

could you please explain the Y axis scale, dB relative to what and how measured, power supply terminated how?

cheers

I would ignore the scale, I did not correct for scope sensitivity settings in the software. EDIT: Power supplies connecting to a Netgear GS108 switch (shows virtually no difference connected or not in HF expect slightly lower charging peaks), but only connected to 1 in the final measurement, it might show some difference due to the center taps of the ethernet isolation magnetics being high passed referenced to ground, but it is below the noise floor in this case.

This actually measuring ground, or rather safety earth, relative to another with an active (battery powered) differential probe. The SMPS noise is very likely dumped there by means of input filter Y caps, I don't have medical switchers here which omit Y caps so cannot really verify that.

 

[Kris]

Now what happens if we connect all of them to the same powerstrip?

View attachment 64038

We see an interaction going on, showing peaks, oscillations and resonances not visible in their individual spectra.

Dear Emile
This is very educational. Thank you.
Did you try to repeat the graphs with Niagara or any other Balanced power solution to see its effect ?

 

[Taiko Audio]

Dear Emile
This is very educational. Thank you.
Did you try to repeat the graphs with Niagara or any other Balanced power solution to see its effect ?

Hi Kris,

No I did not, it is just a demonstration of the power grid side of things and what happens if you mix a variety of powersupplies.

A balanced powertransformer may have some attenuation at higher frequencies, an interwinding shield can help there. But you should also be aware that these measurements are on ground / safety earth, meaning it will pas straight through and the interwinding shield is not going to help as it is connected to the same safety earth ground, and the secondary center tap of the balanced transformer is as well.

The Niagara has CLC filters on AC live and Neutral, individually on each bank. I don't know the corner frequency, but it will likely attenuate some of the 40KHz + noise, it will probably do nothing for low frequency noise unless there's something special in there I'm not aware of. It does however also have a choke on each safety earth ground line which could potentially really help (again only for high frequency noise). These are all low pass filters, meaning low frequencies below the filter corner frequency is just going to pass through. But it's hard to predict how the X-caps (rated for live/neutral) and/or Y-caps (rated for L/N to safety earth ground), if fitted are going to interact with the connected power supplies.

So interestingly, low frequency noise like that generated by linear power supplies is very hard to attenuate on the AC side, SMPS noise is much easier to filter. However we do seem to be less offended by linear power supply generated noise then by SMPS generated noise.

If we generalise types of noise present we have:

white noise: the random errors are independent of each other, the spectrum of white noise is uniform
pink noise: the intensity of the noise decreases with the frequency
red noise: more of low frequency than the average
blue noise: more of high frequency than the average

We seem to be mainly interested in addressing blue noise and don't seem to mind red noise. If we want to make a controversial statement we could say we prefer to have, and may even like red noise. I will now officially distance myself from having made that statement, consider it posted for entertainment value

 

[Kris]

”But you should also be aware that these measurements are on ground / safety earth, meaning it will pas straight through and the interwinding shield is not going to help as it is connected to the same safety earth ground, and the secondary center tap of the balanced transformer is as well.”

Thats why a good grounding is important.
Lower than 2 ohm is preferable in recording studios ... and our rooms.
The more the better.
To reach that low you need a lot of backyard with many metal (cooper is best) sticks deep in the ground.
At the end of this effort you also need to be in the area that the soil is NOT very resistant.
„The Norm „ for grounding wire in my country is :
sandy or stone soil is 40 ohm
intremediate ground 20 ohm
Wetland 15 ohm
This is what we get with the house we live in.

So in short you need much better ground that most of us have at home.

If you are able to get that low grounding wire resistance ( below 2 ohm)
(and yes it is going to cost you many K of USDs Or Euros)
every component you connect to it changes very significant for good of course.
The sound is like a few levels better device in comparison to
normal grounding .

Unfortunate this low resistance creates an open door to grounding loops that might ruin all your efforts,
so be sure absolutely every single component has to grounded well.
the size of the grounding wire also plays a role here .

After 25 years of trying ”everything” In audio I got to the conclusion that Emile is right.
All we are doing In audio (analog and digital) is to modulate electricity.
So my personal main fiding is POWER AND its GROUNDING are a way more important than anything you focuse on.
Emile did it right with his Extreme. 400 kv trafo For computer !? Thausends of Euros for Caps ?!
Only top brands !
Looking at the power supply of extreme and the cost of the parts he used there
gives you a sense of what is important .

 

[microstrip]

(...) We seem to be mainly interested in addressing blue noise and don't seem to mind red noise. If we want to make a controversial statement we could say we prefer to have, and may even like red noise. I will now officially distance myself from having made that statement, consider it posted for entertainment value

Extremely controversial and a very dangerous entertainment, IMHO!

 

[Taiko Audio]

Extremely controversial and a very dangerous entertainment, IMHO!

Not as carefully formulated as usual

 

[matthias]

After 25 years of trying ”everything” In audio I got to the conclusion that Emile is right.
All we are doing In audio (analog and digital) is to modulate electricity.
So my personal main fiding is POWER AND its GROUNDING are a way more important than anything you focuse on.

+1
I got rid of all internal fuses of my audio devices and never looked back.

Matt

 

[Kris]

+1
I got rid of all internal fuses of my audio devices and never looked back.

Matt

Dear Matt
What did you use insted of fuse ?
Mundorf siver gold wire ? solgernig ?

 

[matthias]

Dear Matt
What did you use insted of fuse ?
Mundorf siver gold wire ? solgernig ?

Hi Kris,

some years ago I removed all fuses inside the cases and replaced the internal mains wiring from the mains inlet to the transformer with AWG 11 copper. The mains inlets were replaced with Neutrik Powercon 32A.
I had no single issue and get much more PRAT, slam, dynamics, the maximum undistorted volume is much higher, simply more fun

Matt