[time-nuts] LVPECL logic for dummies (that would be Moi)
bruce.griffiths at xtra.co.nz
Sun Jan 23 19:14:05 UTC 2011
Michael Baker wrote:
> It has been shown that that ECL outputs are open emitters. Without
> pull-down resistors, the outputs are turned off and therefore, there is
> no output voltage. Even if the output has an internal pull-down
> it may still not be possible to measure the true output signal either,
> unless the measurement device is impedance-matched to the ECL
> output structure. The reason for this problem is that the internal
> connection between the output ECL device pin and the output connector
> is most likely a "long line," and neither the scope probe nor the high
> impedance scope input represents an impedance match to the ECL
> output structure.
The output impedance of an standard ECL emitter follower output is 10
ohms or less depending on the emitter current.
Using a 10 ohm load would result in an emitter current that exceeds the
design limits and significantly reduce the life of the part.
ECL outputs are intended to drive transmission lines terminated in their
characteristic impedance albeit to a terminating voltage other than ground.
Sometimes (rarely) the transmission lines are source terminated by
adding series resistors at the emitters to match the output impedance to
that of the transmission line being driven.
A pulldown resistor (or current source) is then required from the
emitter to the negative supply (for NECL) or to ground (for PECL).
> If one was to connect the ECL output directly to a 50 Ohm oscilloscope
> input, there would no output either, because the output emitter will be
> turned off by the ground-referenced 50 Ohm input, even if the output
> a 200 Ohm pull-down resistor. However, AC coupling an ECL output with
> an internal 200 Ohm pull-down resistor to a 50 Ohm input instrument is
> So much for not being able to measure an ECL signal, now we shall
> show how it can be measured using an ECL Terminator.
Its very easy to measure correctly terminated ECL outputs with a low
> ECL/PECL output circuits are designed to drive 50 Ohm loads
> terminated into a terminating voltage V[TT]= V[CC]-2 V.
> For ECL, V[CC] = 0 V, and V[TT] = -2 V. For PECL, V[TT] = +3 V.
> If the input of a measurement instrument is made to look just like a
> 50 Ohm/V[TT] termination, then all should be well. In fact, that is
> what an ECL or PECL Terminator is.
> An ECL Terminator is basically a biased 50 Ohm microwave attenuator.
> The input has an equivalent 50 Ohm/-2 V termination, and the output is
> suitable for driving a ground referenced 50 Ohm input instrument.
> the input of a PECL Terminator has an equivalent 50 Ohm/3 V
> In order to protect sensitive instruments, however, a properly
> ECL/PECL terminator should have a near ground level output
The usual way of ensuring this for test purposes during development is
to shift the ECL supplies so that the -2V termination voltage is shifted
For 5V ECL this requires Vcc = +2V and Vee = -3.2V.
> For measuring a differential ECL output either an instrument with a
> differential input and the proper termination or a differential to
> single-ended converter is required.
> Caution! Do not connect the output of a PECL device to an ECL
> or to a ground-referenced 50 Ohm input instrument. This could spell
> disaster for the PECL device or the instrument Although connecting an
> ECL output to a PECL Terminator may not destroy the ECL device, it
> could cause gradual degradation of the output emitter follower, due to
> possible excessive reverse bias voltage developed across the base
> to emitter junction.
> It is also shown that the collectors of the ECL output emitter
> followers are
> connected to V[CC]. When V[CC] is ground, shorting the emitter to
> merely turns off the emitter follower and no damage will occur.
> This is not the case when V[CC] is = +5 V. The transistor output
> is limited only by b times its base current, which is supplied by R
> R connected to V[CC]. In most cases, the device is destroyed
> In fact, connecting a PECL output device to a ground-referenced 50 Ohm
> load often destroys the device instantly as well.
> Now-- back to the breadboard to see if I can get this ornery LVPECL
> oscillator to show me some output... (next time, I am going to make
> such chips I use are CMOS !!)
CMOS is not a panacea, the faster CMOS families are prone to generating
lots of Vcc and ground bounce as well as supply noise.
Used properly ECL produces relatively little supply or ground noise
compared to CMOS of equivalent performance.
The close in phase noise of ECL is superior to that of CMOS.
Low close in phase noise can be important when measuring ADEV for Tau >
1s or so.
> Mike Baker
> Micanopy, FL USA
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts