SOtM dX-USB-HD Clock replacement.

[Harpy]

Replaced the clock in the SOtM dx-USB-HD from a Sunny Crystal to a Vanguard TCXO and just wanted to post results. A 24.576000 MHz value Vanguard will not work for the USB Clock (?) clock. Easy upgrade, just pop the top lid and lift the oscillator out of it's socket and put the new one in with the shinny label facing the board edge or the indent of the socket. I just did a quick check of voltages and matched the values, there may be much more to it. It plays fine though, and the software says it has stable clock. Cheap 10 minute project if you need to fool with something in your stereo.

 

[sandyk]

According to a 6Moons review," the sCLK-2224 has two clock circuits for the audio- and video-related sampling frequencies - a 22.5792MHz clock for 44.1, 88.2 and 176.4kHz, a 24.576MHz clock for 48, 96 and 192KHz. When one clock circuit activates the other automatically shuts down completely.... "

I use a Vanguard 0.1PPM 24.576 mHz TCXO in my highly modified MF X-DAC V3 and find that it takes several minutes until performance suddenly jumps up a notch.

 

Alex.

[Harpy]

Sandy, any idea why only the Sunny clock will work in the 24.576 socket? I know you shouldn't play with what you don't understand, but I learn by doing. I ordered both clocks.

[sandyk]

Sandy, any idea why only the Sunny clock will work in the 24.576 socket? I know you shouldn't play with what you don't understand, but I learn by doing. I ordered both clocks.

 

I can't explain that, unless the 24.576 MHz TCXO is faulty. I would expect that the oscillators at both locations use similar power supplies and have similar load conditions. Are the pins on the 24.576mHz TCXO still straight after removal from the socket ?

[Harpy]

Alex, thanks for the info, it really helped. You may have touched on something when you mentioned one clock was turned off while the other the other was in use. I believe both oscillators to be good as they both play on the 22.5792 MHz socket and neither work in the 24.576 MHz socket. I'll play again tonight, though it may be a fool's errand.

[sandyk]

Alex, thanks for the info, it really helped. You may have touched on something when you mentioned one clock was turned off while the other the other was in use. I believe both oscillators to be good as they both play on the 22.5792 MHz socket and neither work in the 24.576 MHz socket. I'll play again tonight, though it may be a fool's errand.

 

Harpy

Why not send the supplier an email and mention this ?

 

Kind Regards

Alex

[Harpy]

Mainly because if I can't integrate them into a circuit I don't think the burden should fall on them unless there is a defect in the part. I wouldn't want Email SOtM asking for support on modifying their product either, though I suppose you could argue you want to freshen up aging clocks. I was hoping someone had tried something similar.

[Harpy]

I messed with it tonight and discovered as long as there is a Sunny Oscillator is in either socket you can choose which Vanguard clock you want to use. With both Vanguards it has a good clock and the spy tool shows an output, it just doesn't play. It looks like there are different configurations of pin 1 on the oscillators, but who knows. Going to keep the 22.579 MHz Vanguard clock and maybe try some other clock configurations.

[4est]

I doubt the pin outs are different. All of them I have used are the same, even when the case style is different. Some use the last pin as an able/disable pin, and your circuit may require that.

[sandyk]

I doubt the pin outs are different. All of them I have used are the same, even when the case style is different. Some use the last pin as an able/disable pin, and your circuit may require that.

 

Hi Forrest

The data sheet for this oscillator from ebay shows pin 1 as N.C.

There is no guarantee that this is correct though. It may be helpful to see the data sheet for the original oscillator ?

Given correct data, it may be possible to fit a small toggle switch to change between TCXOs ?

 

Kind regards

Alex

[4est]

I am talking about the originals. Somehow they need to be turned on and off or something in order to switch. That is unless they let both sit and run, which may add considerable interference. This really is a question for SOtM, Jesus @ Sonore or wherever it was purchased. It may aslo be in the specs of the originals.

 

The toggle would be built in. For the record, there are adapters for surface mount to the DIP socket. There are more clocks available in surface mount for obvious reasons. If it were me, I'd be thinking about using one of those to hold a Crystek or eq. and a new regulator from the main PS (5v?)>the 3.3v required. Now that I think of it, what are the voltages for the crystals? I have seen 5 and 3.3v.

Hi Forrest

The data sheet for this oscillator from ebay shows pin 1 as N.C.

There is no guarantee that this is correct though. It may be helpful to see the data sheet for the original oscillator ?

Given correct data, it may be possible to fit a small toggle switch to change between TCXOs ?

 

Kind regards

Alex

[sandyk]

I am talking about the originals. Somehow they need to be turned on and off or something in order to switch. That is unless they let both sit and run, which may add considerable interference. This really is a question for SOtM, Jesus @ Sonore or wherever it was purchased. It may aslo be in the specs of the originals.

 

The toggle would be built in. For the record, there are adapters for surface mount to the DIP socket. There are more clocks available in surface mount for obvious reasons. If it were me, I'd be thinking about using one of those to hold a Crystek or eq. and a new regulator from the main PS (5v?)>the 3.3v required. Now that I think of it, what are the voltages for the crystals? I have seen 5 and 3.3v.

 

Hi Forrest

Harpy rejected the suggestion of contacting the supplier, however I think that this a legitimate suggestion, as the designer has made provision for different oscillator modules, when it would have been more economical to simply have soldered them in directly.

 

Regards

Alex

[Harpy]

Alex, I just felt bad asking. I do see it as a valid suggestion since, like you said, they were put in sockets to be replaceable when they could have just soldered them in. According to the data sheet the Sunnys loose or gain 5ppm a year (auditable to anyone I don't know). If my replied seemed a little terse it was out of frustration.

I am going to try and find a better data sheet for the Vanguards. Pin 1 on the Sunnys have Tri-State Voltage.

No Connection Enable Output

VIH : 2.0VDC Enable Output

VIL : 0.8VDC Disable Output: High Impededance.

The pin outs are the same.

[sandyk]

Harpy.

It would be interesting to see what they have to say. You may need to find a way to disable the output of the applicable Vanguard , but this may involve surgery to the supply rails via a toggle switch perhaps?

Personally, if I had to choose , I would use the 24.576mHz Vanguard for 48 and 96kHz and the Sunny for RB CD.

It has been my experience that the 24/96 and 24/192 has more to gain from a precision oscillator.It's not the long term drift that matters so much, but how close to the marked frequency the oscillator is. Xtal ovens are an even better way to control the actual output frequency, and used to be used extensively in Communications receivers.

Regards

Alex

[mcduman]

harpy

did you notice in improvement in sq after the oscillator upgrade?

[Harpy]

mcduman, I only had experience with the 22.5792 MHz chip playing at 44.1 where I sat down and listened for a bit. I also changed to 2012 r2 and had flashed the SOtM firmware to the latest version. From what little I heard it was positive. I am waiting for a connector to go fully battery. There is a jumper to change from USB to Chassis ground and a mystery one. The USB power regulator also just lifts out. The proper way to do this (depending on budget) would be using the SOtM clock upgrade where the circuit is optimized for a better clock. Sounding good or not the unit is still out of spec which may not be an issue, but it could too. Nicely built unit though becoming a SOtM fan boy. Like their mini server.

 

I am also having trouble sourcing 4 pin oscillators. Any help?

[Harpy]

It would be interesting to see what they have to say.

 

Will do.

[Harpy]

Had time to poke around with the multi-meter and found the switch that enables or disables the clocks. It uses pin 1 of the oscillator to enable or disable the clocks. I fear using one Sunny and one Vanguard that both clocks are enabled which is not how the unit is designed. I am pretty sure you need a oscillators that has the Tri-State option on pin one. I haven't been able to find much in a 4 pin or an adaptor for 4 pin to surface mount where there is a greater selection. Scraping this idea all together till I can find a suitable alternative.

 

 

q101 appears the be the switch.

[mcduman]

harpy,

sorry to hear that you are putting the project on hold. was curious to hear your comments with the vanguard 22.5792 MHz part playing 44.1 khz. i have the stock dx-usb-hd with the battery supply. as much as i like the sound, the unit is wanting in air and top-end sparkle. the clock upgrade could perhaps cure that on the cheap...

[JohnSwenson]

Many crystal oscillators have an output enable which is controlled by pin 1. The most common configuration is pin one is high to enable the output. BUT some manufacturers also have pin one LOW for enabled, although it is pretty rare.

 

A common connection is to tie the outputs of the two oscillators together and feed pin one with opposite polarity signals, that way only one is enabled at a time.

 

One way to test this is with your original oscillators, put the 22.5792 in its original socket, leave the other socket empty. Play a 44.1 file. You should be able to hear music. Play a 48 file and you should not hear anything.

 

Now take it out and play a 44.1 file and measure the voltage on pin one of the two sockets. The most likely scenario is that the socket with the 22.5792 is high and other is low, but it might be opposite. If they are opposite you have low enabled oscillators. If they are both low or both high it's not using the enable pin to select the frequency, most likely a mux is used to select which one to send out to the rest of the circuit.

 

A word on TCXOs. A TCXO is a temperature compensated crystal oscillator, it is used where you need to control the long term change in frequency with change in temperature.

 

This is NOT the case for digital audio. The absolute long term average of the frequency makes very little difference to digital audio. What DOES matter is the short term variations, otherwise known as jitter or phase noise. The problem with a TCXO is that it adds a lot of jitter in order to improve the long term average. The TCXO consists of a regular crystal oscillator, with a temperature sensor, some anlog circuitry to get the change in voltage into particular curve, which drives a varacter diode across the crystal. This circuit tweaks the frequency as the temperature changes. Unfortunately those parts (sensor, curve fitter and varactor) all generate noise, that noise rides directly on top of the signal which is varying the frequency. So the long term average improves, but the short term gets considerably worse.

 

The result is that you are making something which doesn't matter better, while making what DOES matter worse. Not really a good thing.

 

This does not mean all TCXOs sound worse than regular oscillators, a really good TCXO will sound better than a not so good regular XO, BUT a XO of the same quality as the TCXO will always sounf better.

 

Given this I personally would never use a TCXO for digital audio, spend the money on the lowest jitter non TCXO you can get, it will sound better than a TCXO unless you spend a LOT more on the TCXO.

 

In this particular case where non SMD parts are being used, it might be the case that you can get TCXOs in the right frequencies that have lower jitter than XOs with the same case. But I pretty sure that with careful looking you can find low jitter XOs with the right frequencies in those cases.

 

John S.

[sandyk]

In this particular case where non SMD parts are being used, it might be the case that you can get TCXOs in the right frequencies that have lower jitter than XOs with the same case. But I pretty sure that with careful looking you can find low jitter XOs with the right frequencies in those cases.

 

John S.

 

Hi John

In a highly modified MF X-DAC V3 I get markedly improved results using the close tolerance Vanguard 24.576 TCXOs as discussed here. The most precise one in the series is even better for high res. It's a weird effect to hear it sounding almost identical to the original XO ,or even the previous lower tolerance TCXO, then you suddenly become aware minutes later that the SQ has stepped up quite a notch and remains there. This is when using a very low noise, low drift PSU with it. A member of another forum living in the U.S.A. noted the same effect with this unit, and I have been able to compare 2 versions of the Vanguard TCXO in my X-DAC V3 against the original XO ,then the 2nd best Vanguard TCXO in a friend's similarly modified X-DAC V3.

In the mad stampede to SMD by manufacturers, there seems to be very little available still in the extremely close tolerance through hole types of XO , expect perhaps to special order.

 

Kind Regards

Alex

[Harpy]

John, thanks for the info. Unfortunately the unit does not function without both chips in. I did order Chinese TCXOs with VC/NC on pin 1, but again it has a crappy data sheet. There are so many good candidates with better data sheets, but the big sellers aren't stocking the 4 pin dip. Like a Crystec I found matched perfectly except it was a surface mount and I was hoping for a drop in solution. A 4 pin adaptor to surface mount would be ideal, but I can't find one.

 

Here is the listing description 1pcs TCXO 24.576MHz Ultra precision Crystal Oscillators DIP 0.5ppm(NT). Hopefully VC is all that they need to do the switching and I didn't miss anything else.

 

mcduman, It may take a week or two to get these and I can report back. I thought the clock change had potential for better sound on the cheap. I don't have the gear to measure the clocks. I can only tell you if they work and my perceptions. Still looking for other clocks too.

 

Alex, Very nice - stupid question can I solder dip pins on a surface mount oscillator and stick them in the sockets.

 

Thanks for the help and ideas. Think I would get flamed else ware.

[sandyk]

Harpy

Modern surface mount Oscillators are much smaller than the size of the Vanguard and what you have already.

You would need some kind of adaptor PCB.

In my case, the Vanguard was twice the size of the original DIP version and I had to extend a couple of it's leads and relocate a ceramic capacitor or 2 under the PCB to make room for it. I soldered it in. The high stability TCXOs and XOs are only as good as the power supply feeding them. That's the function of the small additional PCB in the photo. It's fairly similar to the P.F.M. "Flea"

Regards

Alex

[Harpy]

OK Thanks, I don't think there is to much I can do for the power supply. I removed the power regulator and flipped it upside down so you can see. About the only thing I would feel comfortable soldering are the 220uf 10v. It will be on battery, but I am going have to rely on the standard circuit.

 

[mcduman]

anyone looked at these?

XO