USB Clock Upgrade My Experiences (so far)

[Nick]

Peter hi,

Just to add to Pauls points.

Most cards will use the 3.2v pcie bus voltage by default. In this situation this bus voltage is used to power the nec chip and via a dc to dc converter the the 3.2v is stepped up to 5 volts to power the aux usb supply (5volts carried via the usb cable)

With the molex I think only the 5 volt supply is used. I know it will be used to supply the 5v aux supply (the enable pin on the dc to dc converter above is used to switch off the 5v generated from the 3.2v pcie bus). I am not sure if the molex 5v supply is used to power the nec chip, I suspect not.

One thought about the card that will not work. Is the clock live signal connected to the right xtal pcb pad. There will be two traces from where the old xtal is mounted to the nec chip. The correct connection usually has a series resistor on its way to the nec chip. If in doubt it should be ok to just try to solder to each pad in turn (this is what I did on the nos pcb to work out which xtal pad to inject the signal. If the nec chip has no clock signal the pc will not see the device.

Apologies if you already know some of the above. Hope it might help and best of luck.

Regards,

Nick.

Ps I lost another post yesterday.
Im reasonably sure now that sound quality is dependant on both absolute clock speed AND relative clock  speed. There are some other effects in play as well which ill come back on later but tuning the usb link clocks appears to have more variables that need to be understood. The good news is that this is about getting the last 20 to 30% of performance on top of the really significant improvement that just putting the clocks in provides. 
 

[Scroobius]

Maybe Paul Pang is using regulators to derive 3.3v from the 5v rail hence the chunky heat sinks - dc-dc converters could be expected to significantly more noisy.

P

[acg]

Guys, those bits of aluminium on the Pang card are not for heat, they are for rfi.  He glues those pieces on to components that may be impacted by or radiate the nasties.

Anthony

[Scroobius]

Ha ha the replacement power supply for my Dexa arrived today. What a relief to get two Dexa's back into service. Oh my the sound quality is just stunning with 2 Dexa's. The fantastic bass is back and just as fantastic everything else.

Anyway now that I have two Dexa's again I have been playing with the "heavenly trimmer". And this time the effect on sound quality is unmistakable. From the as delivered setting a quarter turn one way makes the sound softer. A quarter turn in the other direction again seems to make the sound softer but also a refined sound. After just a short time it is not possible to say which sounds correct or even the best - it does not sound clear cut.

Paul

[Scroobius]

Anthony - thanks for the clarification the one that I did get the back of my finger on was quiet warm hence the erroneous assumption.

P

[Nick]

Maybe Paul Pang is using regulators to derive 3.3v from the 5v rail hence the chunky heat sinks - dc-dc converters could be expected to significantly more noisy.

P

Paul hi,

That seems like a better way to do it I guess on Paul Pang's card.

Sorry I realised that I said that I was adding to your point but actually my comment above were about the molex on "vanilla" USB cards, sorry if I did not make that clear. As you know I have only seen your Paul Pang card for a limited amount of time so I'm not really able to say anything about the supply of the card.

Best,

Nick.

[acg]

From what I can tell the modded Pang card uses the external 5v supply to run only the upgraded TXCO.  The remainder of the card runs as normal from the pcie bus.

[Nick]

Ha ha the replacement power supply for my Dexa arrived today. What a relief to get two Dexa's back into service. Oh my the sound quality is just stunning with 2 Dexa's. The fantastic bass is back and just as fantastic everything else.

Anyway now that I have two Dexa's again I have been playing with the "heavenly trimmer". And this time the effect on sound quality is unmistakable. From the as delivered setting a quarter turn one way makes the sound softer. A quarter turn in the other direction again seems to make the sound softer but also a refined sound. After just a short time it is not possible to say which sounds correct or even the best - it does not sound clear cut.

Paul

Paul,

Fantastic !   normal service resumed.

Interesting to here that the trimmers do have an effect now. It took a while to work out what they were doing because most settings seem to have qualities that are appealing in some way, so I agree that it is not so clear cut.

The sound I'm homing in on has a nice balance of dynamic, tone sound stage and emotion. It seems to be just moving from the slight sharp setting to towards the smooth setting on the trimmer, just so that the edge disappears.

It might be worth having a chat about this, as mentioned above I think absolute speed and relative speed of the clocks seems to have an effect so it would be good to bounce a few ideas about and check and compare results.

Speak soon,

Nick.

[Scroobius]

I can confirm my PC does not recognise the pcie card when there is no clock in it (when the crystal is removed) or when the Dexa signal is not switched on.

I have another problem and that is when I reboot my PC with the pcie card clocked by the Dexa (and the Dexa switched on) the Bios does not fire up correctly - I always have to reset the Bios just as the reboot starts and then select default Bios values - then and only then will the PC boot. The output of the Dexa clock to the pcie card is completely isolated as it comes out of an isolating pulse transformer. I have replaced the BIOS battery on the mobo earthed the Dexa card to the PC but none of this makes any difference. Also I powered the pcie card from a molex supply from the ATX but again made no difference.

If I remove the card I can get the PC to boot properly.

Any ideas Gents would be appreciated.

Cheers

Paul

[Scroobius]

Hey Nick - I did exactly the same and just moved slightly away from the point where the sound had a slight edge.

I was expecting that maybe the ps would need some running in (and maybe it does) but the sound is really smooth and just effortlessly glides through difficult pieces. String quartets are sounding wonderful.

P

[CoenP]

The latest nec chip runs on a very low voltage (like 1.3V or something). Like a pc motherboard this voltage is made from probably the 3.3v by a switching regulator but I did not measure its source. The switching regultor is on the pp card on the right of the nec chip.

Pci-e has no native 5V connection, so that has to be made on the board or by an external supply. In case of the pp card this is the only function of the molex. The voltage is filtered by an rf pi-filter with a "big" rf choke on the top part of the board.

The card designer has two voltages available on the pci bus to make the 5V voltage from: 3.3 and 12V. I guess the 12V will be preferred for the 5V USB power if no external supply is available, at least that makes the most sense to me fom an efficiency and power availability perspective.

Anyway you for shure don't need the extra molex for the 12v and in case of a molex, likely the 12v is not used at all. Too bad you can't prevent it from entering the card since its also on the bus .

Regards, Coen

[CoenP]

I can confirm my PC does not recognise the pcie card when there is no clock in it (when the crystal is removed) or when the Dexa signal is not switched on.

I have another problem and that is when I reboot my PC with the pcie card clocked by the Dexa (and the Dexa switched on) the Bios does not fire up correctly - I always have to reset the Bios just as the reboot starts and then select default Bios values - then and only then will the PC boot. The output of the Dexa clock to the pcie card is completely isolated as it comes out of an isolating pulse transformer. I have replaced the BIOS battery on the mobo earthed the Dexa card to the PC but none of this makes any difference. Also I powered the pcie card from a molex supply from the ATX but again made no difference.

If I remove the card I can get the PC to boot properly.

Any ideas Gents would be appreciated.

Cheers

Paul

Paul,

You are probably violating the startup requirements with a running clock on the card before power-up. This could be either a bios or a pcie card startup problem. My guess is the latter.

I would suggest you use a card that works for others. Nick?

Regards, Coen

[Scroobius]

Hi Coen - This is my second pcie card both have exhibited the same problem. What I will do is try Nick's he was expressing an interest in coming down here soonish. Nick does not have this problem.

P

[Scroobius]

Just a point about the "heavenly trimmer" the point where the sound has a slightly hard edge to it is where the Dexa cards trimmers were factory set. I would have to assume that the Dexa's are set very close to 24Mhz in testing simply because they can be!! So maybe the slightly hard edge comes when the two Dexa's are very close in frequency.

Nick is that what you found?

Wish I could measure it - Nick presumably you could see the two clocks on your funky scope?

Paul

[Nick]

Just a point about the "heavenly trimmer" the point where the sound has a slightly hard edge to it is where the Dexa cards trimmers were factory set. I would have to assume that the Dexa's are set very close to 24Mhz in testing simply because they can be!! So maybe the slightly hard edge comes when the two Dexa's are very close in frequency.

Nick is that what you found?

Wish I could measure it - Nick presumably you could see the two clocks on your funky scope?

Paul

Paul,

Bingo my thought too !

 This is one of the reasons I have been saying that I think clock speed faster than 24mz may be needed. My scope will not measure to 8 digits but something that will is on the way. Im really wanting to understand the variables,  I think the understanding could really help tune the usb link .

Cheers,

Nick.