Wordclock amplifier

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

gnd

Well-known member
Joined
Jan 24, 2006
Messages
285
Hi.
Is there an easy way to amplify wordclock signal?

Usually that signal is TTL 5V square wave of 44kHz for 44kHz sampling rate, minimum 0V, max 5V.

But some units output less, just arround 2V, and that is not enough for other units. So in need to bring it from 2V to 5V.

Would simple opamp work? Video amplifier? Any suggestions for a simple solution?

thnx
 
The best way would be to run the clock into a schmidt trigger chip and have it drive several buffers in parallel run off a 5 volt TTL supply. You really don't want an analog opamp for this kind of work where you would have to bias it to be positive going square wave.
Also don't forget that you're talking 75 ohms here, and most clock distribution amps use a ferrite transformer to talk to the outside world on every output.
 
[quote author="Jim Zuehsow"]The best way would be to run the clock into a schmidt trigger chip and have it drive several buffers in parallel run off a 5 volt TTL supply. You really don't want an analog opamp for this kind of work where you would have to bias it to be positive going square wave.
Also don't forget that you're talking 75 ohms here, and most clock distribution amps use a ferrite transformer to talk to the outside world on every output.[/quote]

So it has to drive 5Volts into 75ohms? Usually I connect wordclock without 75ohm terminator, because that 75ohms drops voltage, and there are problems. If it is without terminator, then load impedance will be much higher, is it?

And regarding biasing op-amp, with gain 2, would it be done by applying negative 1.25V DC to negative pin? This would give 2.5V offset on input.

Would this work, without terminator and 1.25V input offset?

gnd
...
 
TI makes a clock distribution IC that will lock onto an incoming clock and output something like 8 outputs. I remember it could drive pretty good too.
 
I know there is a super complicated (ultra professional) solution to this. But how about a real simple solution?

Why not just take single TL071, feed WC to non-inverting input, set gain at 2 with 10k resistors, and thats it. That would give no error, no phase shift, no jitter, just amplified voltage. It is just a analog signal after all, simple 44kHz square wave on scope, almost audio range, nothing special at all.

I did simulation with single TL071, level is ampified, DC offset too, so at output there is square wave with ofset and everything, just double level. Everything looks just fine to me...

Why would this not work? :?:

gnd
 
Drive 5V peak into 75 ohms with a TL071?

Good luck!

PS: The output impedance needs to be built out to 75 ohms to avoid reflections in the transmission line. With a 6dB loss from the matched termination, your amplifier would need a gain of 4, not 2.

PPS: Simulators lie.
 
Oh yeah the IC i'm thinking of has a PLL with jitter reduction circuit and 8 outputs but I can't seem to find the copy of the datasheet in my pile. I'll try to find it and get back to you. It was a BB part though..

However there are lots of solutions from a lot of manufacturers as you mentioned but you have to watch out, most clock distribution is for much higher freqs(MHZ) and only certain ICs are stable at KHz freqs.

Gnd:

go ahead and try it! there are no rules when it comes to learning. Be forewarned though that simulations LIE. most opamp models are rather idealistic and refined, almost perfect. This of course is great for proof of concept but not good at all for real world usage, thus the R&D cycle of design, testing, redesign, testing, refining, testing, etc.

I'll digress as I'm sure you are already aware of this. Just be wary of overshoot, ringing, protection circuits, and great decoupling otherwise your circuit will happily introduce jitter, noise, ringing, etc. Also be wary of your cabling and proper termination of the signals both incoming and outgoing or else you'll get reflections and harmonics. And this is just the peripheral stuff!

When you get to choosing your opamp, all of this will either work for you or against you. Some opamps like driving loads, some hate it. Some like seeing capacitence on their outputs(designed for it..) and some fall down into oscillating piles of useless silicon when faced with it. Some opamps will overshoot while another will round the edges of a wave when placed in the same exact circuit.

Above all, I would suggest using an ADSL opamp. plenty fast, designed to drive cables, designed to work in demanding conditions with low voltage rails.

Good luck!

EDIT: it seems while i was taking my time writing this, the (ir)regulars came and stole my thunder.. oh well. I second what they are saying.
 
[quote author="Larrchild"]Dave Notes:
The best way (in terms of being easiest/cheapest) is to buy a used video pulse distribution amplifier off eBay. Very Happy
Get a less expensive one without a clamp circuit, however,
But they work great![/quote]

Larr, what is clamp circuit?
 
Absolutely. this is why I would suggest a nice ADSL opamp, these things have gone through leaps-and-bounds regarding performance in slew and recovery with as little overshoot as can be had. Anything ADSL should work good for video and vice-versa.

I think the BB part might be obsolete now unfortunately. I'll keep looking.
 
OK, I have done something, which is of course never to be done. Better sit down, this will knock you off... :grin:

I measured input resistance of different WC inputs, and it measures from 56k to 100k ohm. Schematics of Beh ADA8000, for example, shows 100k resistor over WC input, and it measures same 100k.

So I thought, whatever.... :idea: I connected W/C from LynxOne card, which has low W/C output, over BNC->cinch adapter, over 2meters of cheapest homemade cinch audio cable, into tape-in input on cheapest UB802 Behringer mixer, then output tape-out from mixer to cinch/BNC adapter, then to BNC cable and to W/C input on another unit. Sync works!!!!

Can you imagine? Can thic connection be any worse? Out of mixer it is not even TTL output, +- but symetrical square, 5V p-p, 10V amplitude. On scope square is not so sqare anymore, but a bit rounded. Still much less rouned than for example SPDIF signal.

So, what is going on here? Where is all the W/C alchemy? All magic is gone. :grin:

Any comments?

gnd
 
[quote author="clintrubber"]Functionality & performance are quite different things I'm afraid. The former isn't the hardest one of the two...[/quote]

Yes, but how could performance suffer from such a thing? On scope it is a square wave, just a bit rounded. But it is not perfectly square even on other W/C sources. Actually, W/C signal is basically the same out of Behr mixer, as is on LynxOune output, just amplified and slightly more rounded.

How can jitter be increased by simply amplifying signal in linear way? I'd say that much more jitter would be introduced by too low W/C signal, when unit is just on the limit of loosing sync, and triggering off the rounded edges of W/C signal. But with strong signal, and no additional noise in 5V range, how can there be more jitter?

...
 
just a bit rounded

this is a lot of the answer for all the questions. Most equipment locks onto a rising or falling edge of a signal. if the signal is slightly rounded and moves slightly in amplitude and time alignment(jitter) then the input of the slave clock circuit will jitter as well. You want the squarest square you can give and you want it to stay perfectly still.
 
[Oh good, another Word Clock thread. Excuse me while I get on this here soap box]

gnd, have you seen the word clock discussion we had recently ?

[quote author="gnd"]I measured input resistance of different WC inputs, and it measures from 56k to 100k ohm.[/quote]
So ? One way to design a proper WC input is to have it AC-coupled and then terminated to 75R, with maybe a 100k resistor to ground to keep the input capacitor from floating.

[quote author="gnd"][quote author="clintrubber"]Functionality & performance are quite different things I'm afraid. The former isn't the hardest one of the two...[/quote]
Yes, but how could performance suffer from such a thing? On scope it is a square wave, just a bit rounded. But it is not perfectly square even on other W/C sources. Actually, W/C signal is basically the same out of Behr mixer, as is on LynxOune output, just amplified and slightly more rounded.[/quote]
It's edge jitter that matters, not whether it 'looks' square on a scope. If you can tell 1ns of jitter from a scope display of a 48kHz wave, you have better eyes and/or a better scope than I. And nanosecond performance does matter (see previous thread).

JDB.
 
[quote author="gnd"]So I thought, whatever.... :idea: I connected W/C from LynxOne card, which has low W/C output, over BNC->cinch adapter, over 2meters of cheapest homemade cinch audio cable, into tape-in input on cheapest UB802 Behringer mixer, then output tape-out from mixer to cinch/BNC adapter, then to BNC cable and to W/C input on another unit. Sync works!!!![/quote]
Here's a simple analogy.

Take your antenna/cable connection out of the back of your TV, and connect it to a coax->banana plug adapter. Connect another coax->banana plug adapter to the back of your TV, and wire the two adapters together with a pair of test cables a few feet long. Perfectly linear, right ? Now turn on your TV and tune to a few channels. For bonus points, do the same with a PAL/NTSC monitor but switch off the cable termination on the monitor. Odds are that your TV's VSYNC PLL will manage to lock (just like the WC sync on your converters), but image quality will be degraded. So, here too, Sync works!!!! but performance suffers.

For jitter performance, picking the right rise/fall time of the signal is a major trade-off. Too slow and your transition point gets more noise sensitive; too high and your interconnect needs to deal with hundreds of MHz without group delay. Take HCMOS, for example: with rise times around 10ns, you'd want to be linear without group delay up to about 30MHz for best jitter performance. That's not going to happen with an audio mixer in the system, Behringer or otherwise.

JDB.
 
[quote author="mediatechnology"]I have the formula somewhere that relates rms jitter to S/N degradation. I think it was for a 16 bit system and the effect was profound. Perhaps we should Google it.[/quote]
I posted a derivation a few months ago.

Now, in all fairness, with a good PLL and oscillator the sample clock jittter will be less than the word clock jitter. The amount of jitter attenuation depends on the PLL filter, see the TI SRC4392 datasheet for an example of a rather good filter. Note that this filtering is easier to do in the digital domain than in analog; most analog PLLs aren't quite that good. And then there is VariSpeed to deal with; if your PLL supports it, its filter has to be wider than you'd want for optimal jitter suppression.

Having said that, there are no brick wall filters. Any increase in WC jitter will result in an increase in sample clock jitter, which will impact audio quality.

JDB.
[ta-ta-talking to myself here, as M. Headroom would say]
 
Back
Top