differential clock

Status
Not open for further replies.
I have started to build the next reclocker using EL52, EL11, blue LEDs.
Next Upgrade will be E1651 comparator, only userfriendly chip I could find.

For the LVPECL it is better to make a layout for a printed circuit bord, I have no motivation to hand wire that tiny comparator package. Someday...
 
I would not take it for granted, as we know there are different generations of ECL, what did he use ? SE or diff. ? Translation to the DAC ?
Also NECL logic levels are referenced to ground and not to a regulated voltage. Concerning the isolation, I still think about injecting the divided clock into the DAC to avoid jitter feed through the flipflop.

The E1651 seems to work PECL also.
 
Last edited:
You mentioned something about LED in post #167 ??

Patrick

LEDs have parasitic capacitance of 120 pF or so, I measured with C meter.
If you drive an LED with a source impedance Zp = 7 ohm for ECL output and add a load resistor Zs = 300 ohm > Zp on the other side of the LED, there will be overshoot on LED/Zs, undershoot on LED/Zp and a slow rise and fall across the LED.

I did simulation with parallel cap and it showed these effects.
Also measurement on real translator using bypass cap.

So I would not expect any bandwith problem.

Even more if you add a cap, you will gain voltage swing.

Upper trace ECL output, lower trace DAC input with 100n cap.

ecl_led.jpg
 
But what if one does what I suggested :

Terminate ECL with say 100R before LED. Load LED out pin (cathode in case of PECL) to -5V with a J508 or equivalent. Then there should be no current fluctuation across the LED, and thus no charging discharging of capacitance, hence no RC effects.

At least this is the theory.


Patrick
 
I will stick to LED + voltage divider V+ to gnd on the DAC input.
That is proved to work for me and allows to adjust for symmetrical swing around the DAC switching point.
The CRDs have high tolerances which together with the LED tolerances shift the voltage swing away from the switching point. You would have to adjust with the termination resistor or select LEDs or CRDs.
 
Last edited:
Bernhard,

It's been a while.
2 long weekends means some progress on PCB layout and circuit details (and not just the diff clock).

A question came to mind. Do you have a photo of the sinusoidal output of your balanced Colpitts ?
I wonder what the amplitude would be, and whether they exceed 4V single ended.
The MAX9601 is specified at 5V differential maximum, and the ADCMP567 at 7V.
But the latter is really very fiddly to solder.

🙁

Any hints would be a great help.
Many thanks in advance.


Patrick
 
Do you have a photo of the sinusoidal output of your balanced Colpitts ?

No, my HF sampling plugin does not allow very good resolution at > 10 MHz.
But I checked with the spectrum analyzer and it is clean.

I wonder what the amplitude would be, and whether they exceed 4V single ended.

The amplitude is 600 mV pp per output which means 600 mV diff.
This is also determined by the 1k resistors on the output.

Anyway with 5V supply it will get hard to exceed 4V without clipping.


I did not continue to work on the LED reclocker, was fiddling with my speakers all the time...
 
Last edited:
Status
Not open for further replies.