Cheap Chinese optical SPDIF switches

[rah]

Hi,

I'm considering buying a cheap Chinese optical SPDIF switch like one of these:

https://www.aliexpress.com/item/4000105840424.html

https://www.aliexpress.com/item/4000116739399.html

Ordinarily I steer clear of dirt cheap stuff but that's in the analogue domain. In the digital domain, would a switch like these have any impact on the data? And if so, would any impact ever be audible? Any advice would be appreciated.

Thanks,

rah

 

[Killingbeans]

I wouldn't worry too much about it affecting the audio. My main concern would be reliability and build quality.

 

[Blumlein 88]

I would be suspicious of it having very, very high levels of jitter.

Here is a post where I measured the Toslink output of a device which switched HDMI and also put out Toslink digital audio. So not what you are looking at exactly. The Toslink ouput had so much jitter it raised the noise floor and added many spurious tones. In the graphs in this linked post you see the cheap switcher in red and a Vlink Toslink output in green.

This doesn't mean the one you are looking at would perform this way, but it shows what some very inexpensive converters can mess up.

https://www.audiosciencereview.com/...mer-ultrahd-hdmi-3x1-switcher.1560/post-61147

 

[RayDunzl]

I would be suspicious of it having very, very high levels of jitter.

Is it fair to compare a cheap HDMI switcher with an S/PDIF repeater?

All my coax/optical goes through one - AudioAuthority 1177A - bought used off eBay years ago.

The measurements I've taken don't seem to show defects related to the repeater/switcher.

 

[Blumlein 88]

Is it fair to compare a cheap HDMI switcher with an S/PDIF repeater?

All my coax/optical goes through one - AudioAuthority 1177A - bought used off eBay years ago.

The measurements I've taken don't seem to show defects related to the repeater/switcher.

Maybe, maybe not. I said suspicious. With all such cheap, cheap items I'd be suspicious.

 

[rah]

In the graphs in this linked post you see the cheap switcher in red and a Vlink Toslink output in green.
...
https://www.audiosciencereview.com/...mer-ultrahd-hdmi-3x1-switcher.1560/post-61147

In this test, you've compared the HDMI output of a laptop converted to Toslink with the USB output, presumably of the same laptop, again converted to Toslink. In the frequency analysis you see more noise with the HDMI output converted to Toslink. Why have you attributed that noise to jitter in the Toslink output of the HDMI switch? There could be jitter in the HDMI output of the laptop but even so, I don't understand why the noise would be attributed to this jitter?

 

[Blumlein 88]

In this test, you've compared the HDMI output of a laptop converted to Toslink with the USB output, presumably of the same laptop, again converted to Toslink. In the frequency analysis you see more noise with the HDMI output converted to Toslink. Why have you attributed that noise to jitter in the Toslink output of the HDMI switch? There could be jitter in the HDMI output of the laptop but even so, I don't understand why the noise would be attributed to this jitter?

I also measured this device which included a DAC, and the DAC has the same jitter and noise floor. So I surmise the clocking in the device has caused this on HDMI, Toslink and analog outputs. Also the Toslink output from USB is clean. I've connected other devices fed from the same HDMI output of the same laptop and they are also clean. So it isn't coming from the laptop HDMI, isn't intrinsic to Toslink, and appears to only result when using this particular device.

I'm not saying you should expect the device under consideration to act this way. It was simply an example that with a very low cost device involving Toslink clocking might be poorly done. You might measure the results fed through that Toslink switcher and find no problems. I wouldn't assume that is the result with a device costing so little.

 

[ernestcarl]

Hi,

I'm considering buying a cheap Chinese optical SPDIF switch like one of these:

https://www.aliexpress.com/item/4000105840424.html

https://www.aliexpress.com/item/4000116739399.html

Ordinarily I steer clear of dirt cheap stuff but that's in the analogue domain. In the digital domain, would a switch like these have any impact on the data? And if so, would any impact ever be audible? Any advice would be appreciated.

Thanks,

rah

I’ve got a similar looking clone of the first link. Runs 24/7 for about three years now. Primarily used with a Panasonic BD player. Thing is, there are a dozen or so clones of the same switcher. Which exactly is the original one is anyone’s guess. Mine was probably a few more dollars, but definitely not over $30.

I suppose it’s cheap enough that I ain’t worried if I had to replace it every couple years. But like I said, it’s running 24/7 for about three years now so It seems reliable enough. I have no means to test jitter though.

 

[mansr]

I would be suspicious of it having very, very high levels of jitter.

Why? It's probably just a bunch of Toslink receivers feeding into a mux connected to a transmitter. The amount of added jitter should be minimal.

 

[Blumlein 88]

Why? It's probably just a bunch of Toslink receivers feeding into a mux connected to a transmitter. The amount of added jitter should be minimal.

You could be right. But unless someone measures one we don't know. And stuff that cheap often has corners cut. It could be perfectly fine at that price.

Maybe a good candidate to have drop shipped to Amir for testing. A few dollars for extra shipping to have it measured and know what you are getting. A service to anyone having need of such a switcher in the future.

Here is one from the same brand name that made the device I measured.
https://www.amazon.com/ViewHD-TOSLINK-Digital-Optical-Switcher/dp/B00G188Z7A

 

[rah]

the DAC has the same jitter and noise floor

Here you're distinguishing between two different things: jitter and noise floor. In the write-up of your tests, you listed two "Jtest"s, are these tests of jitter? If so, how does that test work?

As I understand SPDIF, the clock timing is encoded into the data signal and the receiver recovers the clock signal using a PLL or similar. Presumably this clock recovery will not be susceptible to changes in timing between single samples and will average out any jitter?

I've connected other devices fed from the same HDMI output of the same laptop and they are also clean.

Ah, I see!

 

[rah]

jitter and noise floor

Hello?

 

[Blumlein 88]

Hello?

@rah
Okay, jitter and noise floor.

I use a Jtest, quarter sample rate tone. Sidebands of periodic jitter show up with that. Strictly speaking a J test is also supposed to have a low level square wave of another frequency added. Jitter is a timing variation. As well as being periodic frequency specific sidebands around the main tone, random jitter can raise a wideband noise floor.

The switcher in question when fed to DACs (I've actually tested it with a few) creates nearly identical sidebands of a higher than usual level, and it raises the noise floor for a few thousand hz each side of the 11.025 khz or 12 khz tone. Those same sidebands and raised noise floor appear in the analog output of the DAC built into that switcher. I take it the clock for the digital portion of that device is being effected by noise getting into and modulating the clock.

So what are the questions about that?

 

[Blumlein 88]

Here is an example of adding some gaussian jitter to signal. Red is a result of added jitter.
https://www.tnt-audio.com/clinica/jitter2_e.html

 

[Blumlein 88]

And here is the switcher I measured in Red, vs the Vlink device with a much lower jitter clock. In places the jitter raises the noise floor more than 30 db. You also see some spikes of periodic jitter. The highest of which are less than -70 db from the signal.

 

[mansr]

And here is the switcher I measured in Red, vs the Vlink device with a much lower jitter clock. In places the jitter raises the noise floor more than 30 db. You also see some spikes of periodic jitter. The highest of which are less than -70 db from the signal.

View attachment 40608

That looks like a poor recovery of the audio clock from the HDMI data stream. This requires extra care when the audio sample rate, video frame rate, and pixel clock don't have simple relationships.

The device the OP asked about is a simple S/PDIF switch. There is no reason for it to do anything with clocks. All it does is connect the chosen inputs to the outputs. The simplest way to do this is with an analogue mux such as http://www.ti.com/product/SN74LV4052A, though it's probably using something even cheaper from a Chinese supplier.

 

[rah]

I use a Jtest

Could you possibly explain how you do that? I tried searching the web for "how to do a Jtest" but found nothing useful.

 

[rah]

here is the switcher I measured in Red, vs the Vlink device

What exactly does that graph show? Is each plot an FFT of a sample from an ADC fed by a DAC fed by the device under test?

Also, what exactly do the graphs for your Jtest results show?

Thanks.

 

[Blumlein 88]

Could you possibly explain how you do that? I tried searching the web for "how to do a Jtest" but found nothing useful.

@rah
J test so called because it was developed by the late Julian Dunn. Let us use a 48 khz system for an example. A Jtest consists of a 12 khz square wave and a low level square wave at 250 hz. In short this would stimulate possible jitter in the SPDIF interface the maximum amount. If you aren't using SPDIF, then the 12 khz alone is enough. Jitter can be periodic, or noisy. Jitter is any variation in the clocking rate. Instead of each bit being clocked out exactly on time it may oscillate around too fast and then too slow, or wander somewhat randomly around too fast and too slow.

You play a DAC and record the analog output with an ADC. If there were no jitter, the clock was perfectly timed, you'd see a single FFT spike at 12 khz and the low level analog noise floor otherwise. But let us say something is modulating the clock to be slightly early and slightly late at 1000 hz. What you would see in the FFT is the spike at 12 khz, and a shorter spike at 11 khz and 13 khz. These are the sidebands to the main tone caused by a jittering clock. Or if the clock is modulated by power supply noise at 60 hz, you would see these sidebands spread out at 60 hz intervals above and below the 12 khz tone. Or you can have wideband jitter from noise modulating the clock. You will get something like the raised noise floor around the 12 khz tone in the graphs up thread. It leads to noise floor modulation. You'll often see just the base of the 12 khz spike raised and widened a few hundred hertz. Or done badly a wide highly raised band like in the cheap video switcher. In the green FFT of the Vlink you see only a very small widening of the 12 khz base, and a small number of low in level spikes as sidebands. Most of the noise floor is un-effected at all.

Now a simple optical switcher the Op asked about may not have poor jitter injected into the signal (though without testing you can't be certain).

If I've messed up the jitter explanation partly mansr can probably straighten it out.

More in depth is the paper by Julian Dunn.
http://www.audiophilleo.com/zh_hk/docs/Dunn-AP-tn23.pdf

 

[Blumlein 88]

That looks like a poor recovery of the audio clock from the HDMI data stream. This requires extra care when the audio sample rate, video frame rate, and pixel clock don't have simple relationships.

The device the OP asked about is a simple S/PDIF switch. There is no reason for it to do anything with clocks. All it does is connect the chosen inputs to the outputs. The simplest way to do this is with an analogue mux such as http://www.ti.com/product/SN74LV4052A, though it's probably using something even cheaper from a Chinese supplier.

I agree it is poor recovery of the clock, but as that gets passed on then anything downstream has its clock polluted by this device. And doesn't necessarily mean the optical switcher asked about by the OP will act this way.