Correct Termination

[AmongstTheLiving]

I've tried sorting through all the forums about terminating wordclock and can't seem to find the correct answer for my simple setup. I'm running Pro Tools HD2 with an Apogee 16x A/D w/XHD card. I want to use that as the master clock and slave my Rosetta 800 (Also XHD card). So the Rosetta doesn't have internal termination, so I need to terminate this. I've read a few different ways of doing this and was just wondering what was correct for my particular setup. Also, where the hell do you find a 75ohm BNC terminator? Are there different levels of quality amongst them?

[Mastering101]

Redco Audio

really good company...click on connectors then fish around

[Doc Mixwell]

Originally Posted by AmongstTheLiving

I've tried sorting through all the forums about terminating wordclock and can't seem to find the correct answer for my simple setup. I'm running Pro Tools HD2 with an Apogee 16x A/D w/XHD card. I want to use that as the master clock and slave my Rosetta 800 (Also XHD card). So the Rosetta doesn't have internal termination, so I need to terminate this. I've read a few different ways of doing this and was just wondering what was correct for my particular setup. Also, where the hell do you find a 75ohm BNC terminator? Are there different levels of quality amongst them?

As long as you've got a 75 Ohm terminator on a T connector [WC INPUT] than you'll be more than fine.

[Barish]

Below is a diagram that I had posted in my forum a while ago. Sorry it is in Turkish, but you'll get the idea.

Top one is the wrong way of doing it and the bottom one is the correct way (Red and Green text should tell you: YANLIS = Wrong, DOGRU = Correct).




Buy a BNC T connector for every unit to be slaved, and chain them like you see it in the second diagram. And when you reach to the last slave in the chain, put a 75 Ohm BNC terminator into the redundant end of the last T connector like a stopcock (so that the wordclock signals don't fall down, haha).


This is a BNC T connector:




And this is a 75R (75 Ohm) BNC terminator:




If you can't find a terminator in the store then just make one. Buy a 75 Ohm resistor from Radishack along with a blank BNC connector, and put one end of the resistor into the pin in the middle and jam the other end between the body and the screw-in outer sleeve where usually the braided ground sleeve of the cable would be attached. And then use it as a terminator. That's how I did it.

Mine even has two 150 ohm resistors connected in parallel because I didn't have 75R resistor handy at that time and I had to put something together so I used two 150Rs in parallel, which makes 75R in total. It works, because that's what a terminator is. A BNC connector shorted inside via a resistor. Nothing so fancy about it.

M.

[peeder]

Thanks Barish...do you feel that the T-connector daisy chain is actually superior to using multiple outs from a central clock source? I suspect that it is, but I'd like someone with more expertise to confirm.

If it is, then people who buy external clocking devices because they have more BNC outputs have just yet another reason why they have wasted their money obtaining inferior results.

[Barish]

That question opens a can of worms. I wouldn't want to make a comment on the effects of both practices over the sonic quality, as it is pretty subjective. However, I could make a comment on the technical aspects of it as far as my knowledge goes.

The principle is told that, the signal level at the end of the chain must be 2Vp-p or over so that the latter units don't lose sync or become unstable (although most of the new designs are pretty immune to signal losses; but you never know). The signal is usually around 4Vp-p at the source output, and the longer the chain and more the number of equipment, the more will be the voltage drop when you get to the end of the chain.

If the chain is too long and crowded, then you may need to separate the slaves into parallel groups so that you avoid excessive drop along one single branch, which is where the necessity for a multi-output clock source arises. It allows you to create multiple branches all in sync with each other, while being electronically independent from each other, thereby keeping the signal level healthy across each branch.

But we are talking about tens of yards and devices in the chain before having a serious voltage drop in wordclock signal here. It's hard to say how many devices how far apart before the signal starts suffering. An oscilloscope is the ultimate decider in each particular case. But in most project studios, the potential equipment lists are way below the threshold, I can say that.

If you don't have such a large operation in your set up, then buying a dedicated clock generator just for connecting 5 slaves to 5 different output sockets is pretty extravagant IMO. Totally unnecessary.

Some may argue that, as the cables get longer, the interference and the distortion in the wordclock signal increases so, by using a multi-output source, you can use shorter cables for each unit and avoid all those "hi-end unfriendly" conditions etc, but personally speaking, I don't buy any of that. I'm from "if it ain't broke, then don't bloody fix it" school of technology. I would only invest into a device like that if I had a problem with syncing in our system, and I currently don't have any.

But then again, to each their own. It's their money, if it makes them feel better to splash it onto a dedicated unit and then put it on the rack with pride, then so be it. We are all gearslutz.



M.

[AmongstTheLiving]

Thanks a ton. Seeing it in a whole chain actually makes way more sense than looking at it with 2 pieces of gear. You want to run everything parallel.

Question: If you have a device set as a slave with incoming wordclock, does wordclock still come out of the slaves output?

[Doc Mixwell]

The converters I tested will still pass clock signal set to external, as with our RADAR V, which is currently clocked by the Grimm Master clock via short run AES cable right now [soon to be installed with BNC to all chassis]

The RADAR is the master clock of our Apogee System System, as well as our IZ ADA's MADI interface PCIe Card [RME HDSPe MADI]. RADAR WC clock feeds both using a T connector; The first AD16x [which is daisy chained to another DA/AD/DA16x and terminated [Last DA16x] and connected with our Symphony System] and the MADI card for the IZ ADA.

So upon switching the RADAR to External Clock, the rest of the system is using the Grimm as well. This may be delaying the clock signal to the other converters, which is why the Grimm should be connected to all of these devices via BNC.

But these two systems, are apart from each other in there conversions mostly, so right now its a non issue, but in the future [Monday's Service the Studio Day!] I will absolutely use BNC cables to each chassis......Its a better idea, because its designed for timing precision..........you never want to delay the clock signal through I/O.

[Loki]

Originally Posted by AmongstTheLiving

Thanks a ton. Seeing it in a whole chain actually makes way more sense than looking at it with 2 pieces of gear. You want to run everything parallel.

Question: If you have a device set as a slave with incoming wordclock, does wordclock still come out of the slaves output?

Most devices yes. (But you don't want to do that because technically, there is some delay between the clocked incoming signal and the (possibly reclocked) outcoming wordclock signal on the same device.)

[Loki]

Originally Posted by AmongstTheLiving

Also, where the hell do you find a 75ohm BNC terminator? Are there different levels of quality amongst them?

Radio Shack is a good source, also, in LA, Pacific Radio. Most electronic parts/hobbiest radio stores will carry them. Just make sure you get 75ohm and the right size barrel connectors (there are 50ohm and others in different diameters for other a/v purposes.)

[Barish]

Originally Posted by AmongstTheLiving

Thanks a ton. Seeing it in a whole chain actually makes way more sense than looking at it with 2 pieces of gear. You want to run everything parallel.

Question: If you have a device set as a slave with incoming wordclock, does wordclock still come out of the slaves output?

Yes but No. You don't want to use that.

You want to use only one OUT in the system, and that is the Wordclock Master's OUT.

And use IN for everything else in the system that's set to be Slave.

If it's a Master, then there can't be an Input to tell it what to do.

If it's a Slave, it's output is incompetent. It can not tell anyone what to do, hence it can't output anything, or even if it does, nobody wants to listen to it anyway.

You don't want to create sub-masters out of slaves.

Hence, all the slave devices are connected to the system via their IN sockets, and their OUT sockets are not used.


Follow the bottom part of the diagram, OUT-IN-IN-IN-TERMINATOR scenario and you'll be fine.


As to where to buy the connectors, I am not American but I've seen Radioshack everywhere I went during my visits to US -some 14-15 states- so I'm thinking there must be one in your area as well, you can buy whatever you need from your local shop.


M.

[Barish]

Originally Posted by Roc Mixwell

RADAR WC clock feeds both using a T connector; The first AD16x [which is daisy chained to another DA/AD/DA16x and terminated [Last DA16x] and connected with our Symphony System] and the MADI card for the IZ ADA.

If you are not having any issues with your system then I wouldn't attempt to fix it, but if I understand what you mean there correctly, if I were you I wouldn't use a T connector at the source OUT socket to branch out two lines from one node feeding leftbound and rightbound. I don't think that it is a good practice. It is prone to anomalies due to both branches inducing different properties from their sides on the centre node, should the branches start reaching out farther and getting crowded. Because you are gonna end up terminating both ends to avoid reflection, this time you'll be terminating the entire line twice, bringing the overall termination resistance down to 37.5R rather than 75R across the network. No good.

I'd go straight out as one single branch and put all the devices on it. But that's me anyway, if you say it's not broken, then... or may be I got you wrong. You are using the T adaptor on the slave units and not the master OUT, but that's how I understand it from your wording.

M.

[foldback]

Markertek has these parts. America's Broadcast | Pro-Audio | New Media Supply House

BNC T = part # B-2BF ($3.89ea)

75 ohm terminator = part # B-75TM (forgot price)

I was just researching these parts for one of my new mixing rooms and had the numbers handy.

Apogee has some tech notes about termination on their website and it includes thoughts about Motu converters and others.

In the latest Ensemble manual, Apogee shows the way to use T-connectors to daisy chain multiple units and then they go on to detail a better way which illustrates separate cables from discrete outputs on a BigBen. We are currently playing with all of these devices and will report back if we can hear a difference.

[Barish]

The current Ensemble and BigBen manuals provided on their website do not provide any diagrams as such as far as I can see.

Only BigBen manual v1.0 (June 2003) says that:

Word Clock Termination

As the frequency of a word clock signal increases, termination requirements become more critical to precise transmission. In order to ensure the integrity of the clock waveform, each connection should be terminated by a 75 ohm load impedance. Over-termination (a load impedance lower than 75 ohms) will attenuate the signal excessively, while under-termination ((a load impedance greater than 75 ohms) introduces overshoot and other waveform distortions. Both conditions compromise clock accuracy, and are indicated by Big Ben’s termination sensing LEDs.

Connecting Clock Signals

From the preceding information , these four Rules of Clock can be established:

1) Ensure that all A-to-D and D-to-A converters are clocked by the most precise source possible, to ensure the highest quality for the entire digital audio system.

2) Each digital device should be connected directly to the Master Clock Source with the shortest possible cable, in a “star” configuration.

3) Each Word Clock line should be terminated with a 75 ohm load, as indicated by Big Ben’s Termination sensing LEDs. When the word clock input of connected devices is un-terminated, it’s acceptable to chain a few devices with a BNC “T” connector on the word clock input.

4) Don’t delay the clock signal to any device by chaining a signal through input/output clock circuitry. Every device should receive the digital “downbeat” simultaneously.

Which I would agree, except for the recommendation of a "star" configuration, which contradicts with their earlier warning of "over-termination", as I have pointed out in my previous post. I wonder how they would explain that.

It must be a wording error in their documentation or something.

M.

[Doc Mixwell]

Originally Posted by Barish

I'd go straight out as one single branch and put all the devices on it. But that's me anyway, if you say it's not broken, then... or may be I got you wrong. You are using the T adaptor on the slave units and not the master OUT, but that's how I understand it from your wording.

M.

My wording may be off,

The Master Clock is our RADAR, so the WC OUT of that machine has a T connector with both ends feeding our other digital devices. For testing the Grimm Master Clock, I simply used a AES cable to the RADAR's 2-Channel AES input, and designated the RADAR to see external clock from the AES port, so in theory the resulting WC T connection is following along, but it may be delayed, which is why the Grimm should feed all devices separately.

[Barish]

Originally Posted by Roc Mixwell

My wording may be off,

The Master Clock is our RADAR, so the WC OUT of that machine has a T connector with both ends feeding our other digital devices. For testing the Grimm Master Clock, I simply used a AES cable to the RADAR's 2-Channel AES input, and designated the RADAR to see external clock from the AES port, so in theory the resulting WC T connection is following along, but it may be delayed, which is why the Grimm should feed all devices separately.

Right. I got you right then.

If you don't take my opinion as a patronizing move, when Radar is the master clock provider, I'd remove the T from the Radar's WC OUT and take only one cable out from there and take that single line around all the other digital devices' WC INs via T connectors as the only WC carrier line, and terminate it at the end.
And when slaving Radar to the Grimm master, I'd put a T to the WC IN of Radar, remove the cable coming out of the Radar's WC OUT and move it onto one side of the T, and feed the other side of the T from the Grimm with a straight cable, and then set Radar to see the External WC at the BNC IN as the master.

That would be much healthier in my opinion.

Ask Bruno what he says if in doubt. That's what my knowledge and experience suggests to me.

I see their manual says that CC1 has a switch that enables the WC OUTs to operate at 30R low-imp levels, but I'm not so sure that every equipment out in the market today can adjust itself to that. Unnecessarily risky IMO.

M.

[Doc Mixwell]

Originally Posted by Barish

Right. I got you right then.

If you don't take my opinion as a patronizing move, when Radar is the master clock provider, I'd remove the T from the Radar's WC OUT and take only one cable out from there and take that single line around all the other digital devices' WC INs via T connectors as the only WC carrier line, and terminate it at the end.
And when slaving Radar to the Grimm master, I'd put a T to the WC IN of Radar, remove the cable coming out of the Radar's WC OUT and move it onto one side of the T, and feed the other side of the T from the Grimm with a straight cable, and then set Radar to see the External WC at the BNC IN as the master.

That would be much healthier in my opinion.

Ask Bruno what he says if in doubt. That's what my knowledge and experience suggests to me.

I see their manual says that CC1 has a switch that enables the WC OUTs to operate at 30R low-imp levels, but I'm not so sure that every equipment out in the market today can adjust itself to that. Unnecessarily risky IMO.

M.

So you're suggesting [if daisy chaining] using the T connector with the next device in the chain so it then splits to the third device in the line? I've tried this.....I suppose I'll try it again.

Yea the Grimm does some pretty cool tricks with termination, so I'll look into the best way to run these devices.

So far the Grimm has made a signifigant audible "change" to our RADAR. Its one of the best sounding clocks we've tested with the RADAR.

[Mastering101]

I prefer one master clock system" Big Ben" At least my ears like it. Everything digital gets the Big Ben.My Logic tells me daisy chain is not the way to go. Maybe 2 units at the most if i had too... But i don't want to argue about it

[Barish]

Originally Posted by Roc Mixwell

So you're suggesting [if daisy chaining] using the T connector with the next device in the chain so it then splits to the third device in the line? I've tried this.....I suppose I'll try it again.

I think when we say "daisychaining", it means an OUT-IN-OUT-IN scenario (correct me if I am wrong, English is not my first language). We certainly want to avoid that in Wordclock syncing -and even in any other syncing to my knowledge. A slave must always be a slave. We don't want to let one slave act as the next slave's master.

I used to work for a post-production studio in Istanbul as a newcomer video editing/systems engineer in the mid 90s and everytime I tried OUT-IN-OUT-IN style daisychaining somewhere in the system, there would be terrible syncing problems. Some of the devices would lose VITC and we'd end up seeing the black burst contents across the screen and all sorts of stuff. We had to use the OUT-IN-IN-IN-TERMINATION method for the healthiest results. I believe it would also be the best for digital audio syncing.

M.

[Doc Mixwell]

Originally Posted by Barish

I think when we say "daisychaining", it means an OUT-IN-OUT-IN scenario (correct me if I am wrong, English is not my first language). We certainly want to avoid that in Wordclock syncing -and even in any other syncing to my knowledge. A slave must always be a slave. We don't want to let one slave act as the next slave's master.

I used to work for a post-production studio in Istanbul as a newcomer video editing/systems engineer in the mid 90s and everytime I tried OUT-IN-OUT-IN style daisychaining somewhere in the system, there would be terrible syncing problems. Some of the devices would lose VITC and we'd end up seeing the black burst contents across the screen and all sorts of stuff. We had to use the OUT-IN-IN-IN-TERMINATION method for the healthiest results. I believe it would also be the best for digital audio syncing.

M.

Nah....I meant using T connectors in a daisy chained manor.

[Matt Syson]

Hi
Jumping back to using a 'Big Ben' in a 'star' configuration.
This is correct as I believe the subtlety of the setup possibly got missed.
The relevent bit is that all clock outputs from BigBen are buffered and exactly in time with each other. This way you would run an individual cable to each destination, each of which is terminated 75 Ohms. Ideally each cable would be the same length but workclock is not as critical as component video.
The termination resistor damps out any possible 'reflections' (echoes) which would otherwise go back up the cable and interfere with the signal waveform in exactly the same way as having a microphone picking up sound wave reflections from a hard surface, which as we all know causes peaks and cancellation depending on frequency and exact placement.
As a general note 50 Ohm and 75 Ohm BNC connectors are interchangeable, certainly mechanically and up to at least VHF frequencies. If you are going UHF (towards microwave) then there may be some issue but we are not going up that far.
Matt S

[peeder]

If a cable transmits signal at 70% the speed of light, how much timing difference are we talking about exactly?

When you say "short cable runs" are you talking about under 100km or under 1000km?

[Barish]

If you consider that timing fluctuations (jitter) around 10 picoseconds (which is 10*10^-12 of a second, equating to a cycle in a 100GHz signal) can cause an audible difference to the result during AD/DA conversion -unless the manufacturers are bullshitting us in their released product spec sheets-, you can calculate how the differences in lengths of cable can effect the output of an entire system.

To give an example, one of the reasons why the chip structures and operation voltages are getting smaller and smaller in computer technologies as the operation speeds get higher is because it gets to a point where even a tiny tiny fraction of a millimeter between units/components can make or break the system, for the signal may arrive too late or too early to a gate from the previous one. It can be so critical. That's why what we use as the latest technology today is usually what manufacturers had thought about and designed a few years ago behind closed doors. It takes years of prototyping and testing phase to bring a product on the shelves, and even then we have so many problems with them. Heck, it even takes 3 to 6 months to finalise the PCB design of a pishy conventional power supply, if you want it to operate at respectable specs in todays standards, let alone designing pretty complex structures like AD/DA converters and such.

Not as unimportant as you may think.

And they come down here and we bash them for not sounding as good as we expected them to, only for producing a bunch of below-average songs that we can't even stand hearing anymore ourselves and will probably never sell

M.

[Barish]

Originally Posted by Roc Mixwell

Nah....I meant using T connectors in a daisy chained manor.

Gotcha

Thanks for the info also, Matt.

M.

[PMoshay]

The Apogee site had a good tutorial on clocking........ email them and ask for the pdf they have on clocking with Apogee products....... it will clear it all up for you direct from the people who know best.

i.e. - some other gear does not show up as properly terminated (like a Motu Traveller for example), Apogee will give you a list of these pieces so you wont go crazy trying to figure it out.

[peeder]

Originally Posted by PMoshay

The Apogee site had a good tutorial on clocking........ email them and ask for the pdf they have on clocking with Apogee products....... it will clear it all up for you direct from the people who know best.

Apogee knows best how to make money off sync products and cables based on misconceptions that they have cultivated.

I would rather get my information from a less biased source lacking a long history of profiteering.

[Doc Mixwell]

Originally Posted by peeder

Apogee knows best how to make money off sync products and cables based on misconceptions that they have cultivated.

I would rather get my information from a less biased source lacking a long history of profiteering.

You think they'll steer you wrong?? Who knows, there might be a vast conspiracy to over-or-under-terminate your converters...?

Hmmmmmm.....

If a company makes money, its because they build a product that people purchase. In terms of profiteering; by your logic, Digi-Design should be Darth Vader and the dark side has already conquered the galaxy.

Why even support a product that you build?? Is that even good practice??
{I'm being sarcastic}

The Big Ben Termination paper deals with an Apogee product, but speaks to all termination issues, as the above poster stated. Take a look, it cant hurt to read it so you know that a Rosetta 800 is UN-Terminated and requires a T connector with 75 Ohm termination resistor at the end when clocking the unit from another source.

[polf]

Originally Posted by Barish

Below is a diagram that I had posted in my forum a while ago. Sorry it is in Turkish, but you'll get the idea.

Top one is the wrong way of doing it and the bottom one is the correct way (Red and Green text should tell you).



M.

also, i just read (thread below)that if you do it this way every device has to un-terminated except the last device....

Lavry Blue & Digi 192's

[peeder]

Originally Posted by polf

also, i just read (thread below)that if you do it this way every device has to un-terminated except the last device....

Lavry Blue & Digi 192's

Actually no devices can be internally terminated...you have the one external terminator and that is it.

[Audio Hombre]

Originally Posted by Roc Mixwell

Why even support a product that you build?? Is that even good practice??
{I'm being sarcastic}

i agree with roc here. pretty ludicrous implication,peedy. it's in their interest to provide their customers with the best advise they can.