DIYHiFi.org

This is just my opinion... but I'd use sine wave clock distribution. Then I'd use an enhanced cheap-ass CMOS sine-square wave conversion at each receiver. The pico gate type chips cost less than 10 cents. Good bypassing costs more than that. You can get cheap transformers from MiniCircuits and M/A Com. With a sine wave, you only have to worry about reflections of the fundamental, not all of those harmonics. Easier on the transformers, easier on the EMI, easier on the power system at the oscillator, and so on. If you want to supply 3 Vp-p at the inverter input, with a 4:1 transformer at the receive end you need about 7 dBm (assuming 50 Ohms) at the transmit end. Even if you used separate transmit amplifiers for every single destination, that's still pretty small cuz you only need 1 transistor amplifiers for each. 2N3904s are pretty cheap, too.

However, I'm still presuming (uh-oh) that you are talking about separate boards connected with ribbon cable. That might be way off base.

But, I'm harassing the designer. I'm not even a big bucks cash customer. That's wrong and not fair and I won't do it any more.

You will recall that Guido Tent was firmly against this when I proposed it. Back in the early days of the forum.

Way back then...........my oscillator operated on +12 V, so I had the ability to make nice sine-square converters, and the common-base drivers that Wenzel advises.

But.............here is the other problem:

The regulator is noisier at 12 V than it is at 5 V. I lowered the phase noise by operating the oscillator at a lower voltage. I would have to do some research on making common-base stages that will drive 50 ohms. Not saying that I can't do it. Only that it will take some work to come up with a suitable circuit.

So.............all of you guys who may want one.............speak up now!

Keep in mind that I may have to use a cheap 1:4 transformer, if the output level of the oscillator is too low. (Broadband toroid transformers are easy to make with a ferrite core.) Of course, that makes the oscillator a much smaller PCB. So, speak up. Does the idea of a sine "clock" appeal to you?

Jocko

Yes, but you must be honest and state that I also hate comparators.

Jim Hagerman uses one, and his circuit measures OK. Still, I have intuitive feeling that they are evil

I could be wrong. I am a bitter old fart, and in a crappy mood.

Jocko

Why not, I'm in for two as always (actually for me and ozone_stink).
Are the 33.8xxx fundamental crystals good enough to justify your effort? (I guess it is easier to make a good one on 11.2xxx...) If the design needs too much fidling to make it work at our frequency then just scrape us from the list.

tks

I am not doing 33.whatever because all I have in-house are Philips machines. All 11.2896 MHz.

You can build a tripler!

Jocko

Folks -- before I go searching the datasheets, does anyone offhand have a good equivalent to the CCS FET specified by Jocko?

It's one of those only-at-Mouser products, and if this thing has any more Texas content I'll have to barbeque it.

Now we have mystery FET too? [joke]

BF245A works OK as in Pedja's CCS circuit for the TDA1541 offset nulling.
J309 also works as I used it in my TDA1543 regulator circuit. I had lots of J309s, wonder why?

You don't have to use that CCS. Any JFET will work. Just pick one with 1 mA or so of current.

No, it is not TI. I live down the street from them, so I have more reason to hate them than you do.

I don't know how they get people to work there. Strange bunch of folks.

Jocko

...one of those Phred's LNDxxx thingies is really good for your circuit.
oopps ...maybe you use it



sidiy,
don't you think is better to use 11.2896 MHz "before" the DAC?
Only think about the harmonics of a 33.xx clock. 3th is 100MHz!!! It's easy for such a frequency to couple via parasitics or even through the air, and reach the DAC. I'd avoid this situation.

I quote myself:


Here's the passive tripler tuned for our frequencies.
Sine in (left), sine out (right).
Input has to be well isolated from the "clean" 11.xx output to the reclocker.
Output has to be squared and divided by two to obtain 16.xx MHz

_________________
Ciao
Roberto

Asynchronous eclocking at exactly 11.2896 MHz gave erratic results to me

Elso, I'm not asynchronous reclocking anything!!
I know that exist Elso's clocks and asynchronous circuits, but it's Jocko Homo section ...so ...no crappy asynch thingies here ....and no crappy comparators!!!!

I thought Sidiy's PMD100-to-AD1862 is synchronous! Isn't it?

_________________
Ciao
Roberto

Crappy things? In Jocko's section?

Jocko,

Just my opionion but as we all know jitter use to be called word clock jitter. But now I think you can see jitter on a bit wise level also. Allot of designers are now reclocking bit clock and data as well as word clock. I am not a real fan of reclocking clocks like bit clock. I would rather divide down or if there is a hint of delay somewhere use a fifo of some sort and clock in with bit clock and clock out to the dac & fifo using something cleaner.

As for cables you may want to check out what DigiKey offers. They have some solderable 0.1" cables that I use for remote control boards and stuff that are available at fixed lengths that would be consistently the same. There are also ribbon cables with connectors both shielded and not available terminated at fixed lengths. I would suggest in either case to run ground in between each signal.

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio

Ah, but I am only going to use old '1862 that I have. Unlike BB DACs, you don't need 2 extra bit clock pulses after everything else stops. Makes my job easier.

And leaves space for ECL-style sine-square converters. Something that I am better suited for.

Jocko