Electronic digital logic circuitry – Interface – Current driving
Reexamination Certificate
2000-10-30
2002-05-21
Tokar, Michael (Department: 2819)
Electronic digital logic circuitry
Interface
Current driving
C326S030000, C326S086000, C326S090000, C710S108000, C710S120000
Reexamination Certificate
active
06392442
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to data transmission in digital systems. More specifically, the present invention relates to a driver that compensates for skin effect losses of the interconnection media by using a lower impedance when data switches at the maximum switching rate and using a higher impedance when data switches at less than the maximum switching rate.
DESCRIPTION OF THE RELATED ART
In the art of digital signal processing, switching frequencies continue to increase. As is known in the art, the problems associated with transmitting high-frequency signals tend to be more difficult to solve when designing interconnect fabrics, which link together integrated circuits, circuit boards, and the like.
At high frequencies, such as 100 MHz and above, current is primarily carried by the outer skin of the conductor. Skin-effect resistance causes the attenuation of a conventional transmission line to increase with frequency. However, this attenuation is only present for the high-frequency components of the signal, and does not effect the low-frequency components. This phenomenon causes intersymbol interference, which degrades noise margin and reduces the maximum frequency at which the system can operate.
FIG. 1A
shows a prior art transmission circuit
10
, which includes a driver
11
having an output impedance of 20&OHgr;. Driver
11
drives high signals toward VDD and drives low signals toward VSS. In
FIG. 1A
, typical values of 1.8 V for VDD and 0.0 V for VSS are shown. The driver is coupled to a transmission line
12
. The other end of transmission line
12
is coupled to a receiver circuit, which is not shown in
FIG. 1A
, and is terminated with a 100&OHgr; resistor coupled to VDD and a 100&OHgr; resistor coupled to VSS.
FIG. 1B
shows the Thevinen equivalent circuit
13
of the prior art transmission circuit
10
of FIG.
1
A. The two terminal resistors in
FIG. 1A
can be modeled as a single resistor having a Thevinen resistance of 50
106
and coupled to a Thevinen voltage of 0.9 V. Consider that driver
11
is driving transmission line
12
high for an extended period of time. The 20&OHgr; output impedance of driver
11
forms a voltage divider with the 50&OHgr; Thevinen resistence of the termination resistors. Accordingly, the receiver circuit will be provided with a DC signal of 1.54 V. Similarly, if driver
11
is driving transmission line
12
line low for an extended period of time, the voltage divider will provide the receiver circuit with a DC signal of 0.26 V.
If transmission line
12
where lossless, the signal provided to the receiver circuit would swing between the high and low DC values. Accordingly, the signal swing would be 1.28 V. However, because of skin effect losses of transmisssion line
12
, the signal swing will be attenuated when the signal is switching at high frequencies. A nominal attenuation for a circuit such as that shown in
FIG. 1A
is 40%. Of course, the magnitude of attenuation will vary with frequency and the characteristics of transmission line
12
. Applying the nominal attenuation to the lossless signal swing results in a signal swing of 0.77 V.
FIG. 2
shows a timing diagram of a signal
15
applied to driver
11
in FIG.
1
A and the resulting waveform
16
observed at the receiver circuit. Assume that signal
15
has been low of an extended period of time, and therefore signal
16
has discharged down to the low DC value of 0.26 V, as described above. When the first pulse of a series of high-frequency pulses is transmitted at driver
11
, signal
16
will rise 0.77 V, which is the signal swing calculated above. Accordingly, the pulse will rise to 1.04 V, which is just above the receiver detection threshold of 0.9 V. Note that the “eye opening” of this first pulse in signal
16
is very small, so there is little chance that the pulse will be properly detected by the receiver circuit.
As the series of high-frequency pulses continues to be transmitted, signal
16
centers itself about the average signal value, producing progressively better eye openings for the remaining pulses in the series. After the last pulse in the series, signal
15
remains low for several cycles, and signal
16
once again discharges down to the low DC value of 0.26 V.
After several low cycles, signal
15
once again goes high at pulse
17
. Once again, the result is a pulse in signal
16
with a very small eye opening. Signal
15
then goes low for a cycle, and then goes high and remains high for several cycles. The result is that signal
16
charges up to the high DC value of 1.54 V.
At pulse
18
, signal
15
goes low for a single cycle. Since the signal swing is 0.77 V, and the signal starts at 1.54 V, signal
16
will only fall to 0.77 V, which is just below the receiver threshold of 0.9 V. Accordingly, the pulse in signal
16
produced by pulse
18
in signal
15
also has a very small eye opening.
Now consider what happens if the circuit designer attempts to compensate for the small eye opening by using a lower output impedance in driver
11
. The high-frequency signal swing of signal
16
will increase. However, the resulting voltage divider created by the termination resistors and the output impedance of the driver will cause the low DC value to become lower and the high DC value to become higher. Accordingly, while the signal swing is larger, the signal starts from a point farther from the receiver threshold, thereby negating the larger signal swing. Similarly, if the designer uses a higher output impedance in driver
11
, the starting points will move closer to the receiver threshold, but the signal swing will decrease.
This problem of skin effect losses at high frequencies was addressed by William J. Dally and John Poulton in a paper entitled “Transmitter Equalization for 4 Gb/s Signaling”, which was first presented at the Proceedings of Hot Interconnects at Stanford University in August of 1996. This paper is hereby incorporated by reference. To solve the problem, Dally et al. propose using a 4 GHz finite impulse response (FIR) filter in a current-mode transmitter. The FIR filter increases the width and height of the “eye opening”, thereby making it easier to detect the pulse. In essence, the FIR filter prevents the transmission line from discharging down to a low level or charging to a high level. Unfortunately, the FIR filter consumes a relatively large amount of logic on an integrated circuit, and increases propagation delay.
SUMMARY OF THE INVENTION
The present invention is a driver circuit that compensates for skin effect losses in a transmission line by using a lower output impedance when data switches at the maximum switching rate, and using a higher output impedance when data switches at less than the maximum switching rate. As is known in the art, skin-effect resistance causes the impedance of a transmission line to be higher for high-frequency signal components. The present invention compensates for this effect by lowering the output impedance of the driver when transmitting high-frequency components having alternating data values, and using a higher output impedance when transmitting low frequency components having consecutive data values. When transmitting low-frequency consecutive high or low data values using a higher output impedance, the resulting voltage divider formed by the output impedance of the driver at the beginning of the transmission line and the termination resistors at the end of the transmission line causes the high and low DC levels at the end of the transmission line to move closer to the detection threshold of the receiver circuit, thereby causing the next isolated low or high pulse, respectively, to start from a point closer to the threshold. Furthermore, when transmitting high-frequency alternating values at the maximum switching rate using a lower output impedance, the resulting voltage divider produces a larger signal swing in the signal received by the receiver circuit. The result is that all pulses cross the receiver threshold with an excellent “eye opening”, thereby ensuring detectio
Plettner David A.
Tan Vibol
Tokar Michael
LandOfFree
Driver circuit that compensates for skin effect losses does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Driver circuit that compensates for skin effect losses, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Driver circuit that compensates for skin effect losses will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2878174