PCIe USB 3.0 host card

[praphan]

This SoTM card looks interesting, seems like it may not be available yet but could be a useful upgrade.
 
http://www.sotm-audio.com/products7.php

Nick.

I am currently running NOS1 USB through SOtM tx-USB with satisfactory result. Folks on CA forum have mentioned about tx-USBexp for quite a while but no one knows the release date. I used to order stuff directly from Korea. So I recently wrote to them asking if they use NEC/Renesas chipset and about the release date. So far no response.

Hope they don't use Etron Tech chipset since its drivers (all versions) are not compatible with NOS1 USB.

Regards,
Praphan

[boleary]

Using a Startech card as suggested by Nick I have interesting results. Using either a standard USB 2 cable or a USB 3 with adapter, the NOS1 DAC driver is recognized (it appears in Device Manager), but when I try to play a song the OS crashes. However, I can use the standard Hiface plugged into the Startech card, then into a 24/192 Dac, my Denon 4806, and all works well.

Might my 32 bit system be an issue here?

[Nick]

Using a Startech card as suggested by Nick I have interesting results. Using either a standard USB 2 cable or a USB 3 with adapter, the NOS1 DAC driver is recognized (it appears in Device Manager), but when I try to play a song the OS crashes. However, I can use the standard Hiface plugged into the Startech card, then into a 24/192 Dac, my Denon 4806, and all works well.

Might my 32 bit system be an issue here?

Brian hi,

Which version of the Renesas driver software are you using ?

Version 2.0.34.0 is the version known to work ok with the NOS1 USB. As far as I know most folks are using 64bit OS with this version of the driver, but I would start by checking this first. I think there may be a link or a mention as to where the driver can be downloaded earlier in this topic if you don't have it.

Best Nick.

[juanpmar]

Hi Brian,
Here you can download the driver:
http://drivers.softpedia.com/get/Other-DRIVERS-TOOLS/MSI/MSI-990FXA-GD65-Renesas-USB-3-0-Driver-20340.shtml

Regards,
Juan

[boleary]

Thanks Guys, I'll give it a try when i get home this evening.

[praphan]

Folks on CA forum have mentioned about tx-USBexp for quite a while but no one knows the release date. I used to order stuff directly from Korea. So I recently wrote to them asking if they use NEC/Renesas chipset and about the release date. So far no response.

Hope they don't use Etron Tech chipset since its drivers (all versions) are not compatible with NOS1 USB.

Regards,
Praphan

Hello,

I had a reply from SOtM. They will release tx-USBexp in late June.

Not a Renesas chipset but Texas Instrument 7320 instead. http://www.ti.com/lit/ds/symlink/tusb7320.pdf

From the data manual, any chance to predict the compatibility of its driver with NOS1 USB?  Can be too expensive for a wild experiment. I expect that it will carry a price tag of USD 300+

Best regards,
Praphan

[Nick]

Hi,

I was looking for better ways to establish a good signal ground from my PC into rest of my system. In my case the PC connects to a NOS1 USB.

The signal ground for the whole of my system is established from the connection to the ground of the USB port on the PC (and via the PC's ATx supply back to the mains earth) this follows Peters design for the NOS1 USB (Peter, please straighten me out if I am going wrong  ). Working the ground back towards the mains earth, as I use a PCIe USB 3 card (as a lot of us are using now) the ground is referenced from about three of the edge connection pins between the PCIe card and the motherboard PCIe bus. From the bus connector it goes over the motherboard back to the motherboard 24 pin ATx connector and via one of the earth leads back the the ATx ground plain in the supply.

It could be better....It got me thinking most PCIe USB 3 cards have that molex or SATA power connector (to provide enought current for the USB 3.0 in line power standard). These aux power connectors are bad for sound if connected as they are no use to put clean power into the USB chipset BUT the aux power socket is connected to the PCIe USB cards ground plain with a much better connection than the ground pins of the PCIe bus AND you can choose how the ground is routed from here back to the ATx supply.

To see if the ground connection could be improved, I took a molex to SATA power wiring link and cut the 12v and 5v power wires (yellow and red) leaving just the black ground wires intact. This was pluged into the PCIe USB card SATA power connector and from here the aim was to connect back in the cleanest way possible to the ATx supply. In my case I have a modular type ATx supply so I was able to run a dedicated molex distribution lead with no other periferal devices connected to it from the ATx supply directly to the ground only connection to the PCIe USB card.

I think this has helped sound quality in my system but a lot has changed recently so it's hard to be specific about the exact changes. It may be worth giving this a try particularly if you have a system who's signal ground if referenced to the USB ground.

It would be good to know if this works for others.

Regards,

Nick

Ps I was determined to find a use for that aux power connection 

[boleary]

Hey Nick, any chance you could post a photo of your latest tweek? I think I need to generally "see" what you mean, and particularly what the "ATx supply" is. No technical knowledge here.....

Oh, and thanks to both you and Juan, my Startech card now works with the recomended driver. Very much appreciate. The sound is great.

[PeterSt]

The signal ground for the whole of my system is established from the connection to the ground of the USB port on the PC (and via the PC's ATx supply back to the mains earth) this follows Peters design for the NOS1 USB (Peter, please straighten me out if I am going wrong  ).

The contrary Nick; The whole of your description sounds to my heart, although I can't judge whether you really accomplished what you want. I think you hope you did, but how to be sure ? You'd almost need to cut the ground PCIe fingers which go into the bus slot. Or ?

When indeed you think you can't be sure yourself, what I would do is trying to put power to the PCIe card first, before the whole MoBo (officially) is powered up. With that you will (hopefully) have created a "current loop" which should stay when the MoBo is powered up (but read : when the OS is started).
Then I would still like to check it, by means of looking at the currents flowing, although then I see a problem with some references. This almost comes down to looking at the current in the ground lead (somewhere) of the PCIe towards the MoBo without the Molex connection, which is to be compared with the Molex connection being active. In that latter case the current through the PCIe (fingers, so to speak) should drop obviously.
Where to stuff in those multimeter leads is something else. At the Molex side this is easy of course. Ehm, take out the PCIe card, measure the consumption of the whole PC in Wattage at the wall outlet, write down the difference with-PCIe and next calculate the Amperes the PCIe should be taken and try to find that back in the Molex connection later ?

[Nick]

The contrary Nick; The whole of your description sounds to my heart, although I can't judge whether you really accomplished what you want. I think you hope you did, but how to be sure ? You'd almost need to cut the ground PCIe fingers which go into the bus slot. Or ?

Peter,

At the moment I have the two ground paths (not ideal I know, one via the molex and the other via the mother board) as I have not yet cut the PCIe edge connector pin ground traces. I wanted to order another PCIe board first to make comparison of any effect easier.

Results so far really rely on the return resistance of the molex ground route [hopefully] being lower than via the Mobo and therefore establishing the current return via this route first.

More tests coming up soon with cut tracks.

When indeed you think you can't be sure yourself, what I would do is trying to put power to the PCIe card first, before the whole MoBo (officially) is powered up. With that you will (hopefully) have created a "current loop" which should stay when the MoBo is powered up (but read : when the OS is started).
Then I would still like to check it, by means of looking at the currents flowing, although then I see a problem with some references. This almost comes down to looking at the current in the ground lead (somewhere) of the PCIe towards the MoBo without the Molex connection, which is to be compared with the Molex connection being active. In that latter case the current through the PCIe (fingers, so to speak) should drop obviously.
Where to stuff in those multimeter leads is something else. At the Molex side this is easy of course. Ehm, take out the PCIe card, measure the consumption of the whole PC in Wattage at the wall outlet, write down the difference with-PCIe and next calculate the Amperes the PCIe should be taken and try to find that back in the Molex connection later ?

Defiantly agree with your logic here.

As an observation where I have generated alternate grounds for other components such as HDDs (when applying linear supplies) I have found that Bass SQ in particular often suffers unless the ground to the component (HDD etc) is still referenced back directly to the ATx grounds.


Best Nick.

[Nick]

More tests coming up soon with cut tracks.

So I cut the ground pin tracks on the PCIe card so that the ground was provided only via the Molex connector back to the ATx PC Power Supply. This gave an interesting if not successful effect.

Bass had the best delineation and tone differentiation I have heard from my system, like listening to a melody played out in the mid range but delivered with Bass frequency notes, nice. Bass weight dropped overall however.

The real problem and the reason the experiment is not successful overall is that the sound is thin and forward eg voices lost the sense weight and tone of a real person singing and the depth and tonal of instruments is to a large extend lost.

I have seen this presentation before when tweaking USB PCI cards. If I had to guess I think the improved analogue ground did significantly improve bass, stereo image and sound noise floor a lot, but the digital transmission to the card suffers from the ground changes resulting in the characteristic thinness.

I would not recommend anyone try cutting the ground traces, but I will listen to a [new] USB 3 board with and without the molex earth arrangements to see if this gives any sort of SQ edge post findings.

Anyway a good excuse to use Mobo USB 3 [not too bad actually] whilst the new PCIe USB 3 card is delivered.

Nick

[Telstar]

Great choice for the pcie controller. I was going to get the much more expensive Buffalo card.

I see that nobody tried to power completely the card from an external (linear/battery) PSU.

The ground separation was an interesting idea but i think is better to cut all power lines and power externally, but i dunno if:
1- the card is still recognized, depends on pcie pinout which i dunno.
2- the power will be enough. the older version of this controller or a different one may be required to achieve this.

[Tore]

I want to try USB3 and have found some cards with 4 USB3 ports.

They have this chip: NEC uPD720201

Older cards have this chip: NEC uPD720200

Do you think it will be problems with the new chip?

Is it OK for the NOS1 driver?

Tore

[PeterSt]

Tore, I hope someone else can help out, because I am really not experienced on the USB3 cards ...

Peter

[Tore]

This is the 4 port USB 3.0 PCIe card:
http://vr-zone.com/articles/buffalo-launches-4-port-renesas-based-usb-3.0-card/13741.html


Tore