AT&T glass

In a very recent video Paul mentioned there was such a thing called AT&T glass that provided galvanic isolation and much more information throughout than toslink.

I did a cursory web search and didn’t find anything. Anyone know what he was talking about?

1 Like

I found this stuff :slight_smile:

Looks like it would have been great if it had taken off - the price of the transceivers etc. could have been silly cheap now.
I still don’t know why no one is proposing standards using currently available networking transceivers etc. (apart from how difficult it is to get manufacturers to all cooperate on standards, kinda like herding cats I expect).

at - Google Search

1 Like

Thank you John! Yeah makes sense that mediocrity supplanted excellence. It’s the way of the world.

Look for “AT&T ST glass fiber optic” or similar. The problem is the price of transceivers, they cost around $40 - $60 each. A few audio companies use it: Wadia and EMM Labs come to mind.


Youch! I’m surprised they are so expensive still, do they differ from the kinda transceivers in Ethernet switches? SPFs and the like? Or am I just out of touch with IT prices…

As an aside, and I know it gets a lot of stick around here, but I still like TOSLINK.
Sure it could be better if it was updated, but it’s still pretty cool and seems to work flawlessly (here at least) up to 192 kHz with no issues (and bit perfectiness) :slight_smile:

1 Like

I like optical too. Second to I2S via HDMI.


Yeah I like the sound of that (excuse the pun) but I don’t have I2S stuff here, source or DAC.
AES is my ideal I suppose (for what I have here) but isolating earths with TOSLINK is very useful.
Plus it’s cool (did I say it was cool? anything with Lasers is always cool :slight_smile: )

1 Like

The price has come down a lot. When I first priced them they were around $80. The cable prices have always been reasonable.


Isn’t the fiber cable some use for internet the same or similar?

Thanks, still more than I thought now, and of course that multiplies up for a retail product.
Shame though :slight_smile:

Yes, AT&T ST a standard internet cable, also often used for network patch cords in internet installations, offices, etc. ST is built to be robust with locking connections and precise alignment of the ends of the glass fibers… I used it in my original DSD DAC prototypes to connect to EMM Labs (Meitner) gear. We are planning to use a more modern simpler and cheaper network optical connection in the TSS, but it still has an ECL interface to the transceiver :slight_smile:


I know Playback Designs uses and prefers the AT&T glass connection. Is it superior to i2S?
Also, it appears that the fiber optic cables are less expensive than a high-end HDMI cable, which would offset the cost of the connectors used in the unit.

ST optical can definitely be better than a wired connection (ignoring cost). It has the bandwidth required and provides galvanic isolation. It depends on the design of the DAC whether unembedding of the clock in an optical connection is better or not. EMM Labs used a separate cable for the clock. In fact, it used three optical connections, one from the master clock (which might be in the DAC or elsewhere in a large installation) and then a connection for the data to the DAC and a separate connection for the clock for that data. Since the DAC was providing the original clock it knew that the returned clock was at the correct rate and it only needed to deal with any phase offset, i.e. it didn’t need a PLL which could add jitter.

If you are using only one connection, you can think of ST as an idealized TOSLink. Better bandwidth, more reliability… With good software receiving the signal it can also be as jitter free as I2S.


I used to have a Wadia CD transport and DAC, and the connection was AT&T glass. It was retailed for $125 if my memory serves me right. Of course, Wadia had to put a healthy premium on top and that was 25 years ago.

The video Paul did was why some streamers sound better than others. I am new around here but I’m going to go out on a limb. I’m new to streaming and can only get Tidal in Canada. I can stream through my Bridge II or my Aurender N100H.

I find the source is the biggest variable and it changes all the time. By this I mean the quality of the Internet connection at the time, and the quality of the music track being streamed. Neither of these remain constant than, say the digital stream coming out of a CD transport.

Isn’t it possible you could put so much research and development into perfecting galvanic isolation only to be held hostage by what was coming into your streamer in the first place?

Tranformers and separate power supplies could create the type of isolation demanded by the most critical audiophiles.

1 Like

@Rob_W Well unless you have relatively high jitter, a decent download speed (somewhere around 10-20 Mbps should be plenty for high res. And even if it drops once in a while it should be ok with buffering (again with relatively low jitter.)

DSD4 is 11.2MHz so you’ll need at least 11.2Mbps under perfect conditions. Someone can correct me if I’m missing something.

1 Like

I have a good Internet connection so I’m speaking in generalizations. Galvanic isolation is a precision that could easily be comprised by factors beyond our control as a listener. As an example, on Tidal I will listen to one track that is 96 kHz / 24 and sounds amazing only to hear another 96 kHz / 24 track that sounds awful, likely because it was a poor remaster.

I notice on my laptop at 5 GHz to my router will let me do a download speed test of 100 Mbps which will eventually drop to 25 Mbps. This is usually the laptop having to renegotiate a speed with the router for any number of reasons. I am sure this type of negotiation is present on the Internet as the Tidal connection makes it hops towards my destination. The point being, we are dealing with variables that we don’t have control over when looking for perfection in the design of the streamer.

1 Like

No designer obsesses over galvanic isolation at the expense of other potential sources of non-ideal behavior. Perhaps a way of looking at it is that the sound quality has to pass a series of gauntlets and any one of them can bring it down, it’s only when everything along the path is engineered well that you can get the maximum sound quality from the source.

With streamers there’s a buffer that gets rid of almost all of the nastiness upstream. There’s still always a little leakage from noise on the inputs, power etc. but they are mostly dealt with with good engineering principles. Often, it’s not noise from the internet that’s the issue, so much as noise from the power grid which is often correlated similarly with the time of day.


No. It’s not how the internet works. The data being sent back and forth doesn’t vary in quality. It either arrives exactly as it should be or it doesn’t arrive at all.

1 Like