Todd, When we design curent source IO cells (e.g. LVDS & PECL) the termination is usually part of the IO cell. For example LVDS has a NE and a FE terminator. This will limit the mismatch and absorb reflections. If the terminator is not on chip, then it usually is placed as close as possible to the package. The output current is usually set to the termination resistance. Thus we can control the voltage swing by matching the current to the termination. Of course the optimum situation is to control both the termination impedance and output swing. This can be quite difficult due to technology reasons. For example, LVDS requires differential termination. To design a contollable impedance not connected to the power supply can become quite complicated. Thus, we usually use a resistor with a very low temperature coefficient. For impedances connected to the power supplies, the design reduces in complexity. It is still quite difficult to match impedances over the full frequency range (>500Mhz) due to parasitics such as package and ESD. Best Regards, Bil Cohen Toshiba America Circuit Design Group Marlborough, MA "Todd Westerhoff" <twester@xxxxxxxxxxx>@freelists.org on 06/11/2001 12:54:35 PM Please respond to si-list@xxxxxxxxxxxxx Sent by: si-list-bounce@xxxxxxxxxxxxx To: <si-list@xxxxxxxxxxxxx> cc: Subject: [SI-LIST] Ideal driver characteristics Hi all, We were talking about driver characteristics a few weeks back, and Arpad questioned why someone would consider a current source to be an ideal characteristic driver. I'm wondering the same thing. Since a current source's output is independent of the voltage at its output, it seems to me that such a driver would do a pretty poor job of absorbing reflections. In fact, it seems to me that such a driver would't absorb them at all, it would reflect them the same as an open-ended line. Am I correct in this assumption? Looking at the IBIS curves for any number of models, it seems to me that the transistors are usually saturated, and therefore acting as current sources, when they first turn on. Thus, it follows (I think) that any reflections that hit the driver when it is first transitioning are likely to be reflected instead of absorbed - providing one very good reason to avoid line lengths that are 1/2 of the cycle time. It seems to me if that I'm looking for an "ideal" driver, I'm looking for one whose V/I characteristic is a staight line from the zero-current point out to the maximum current (driving + absorbing reflections) at which it will ever be operated. I'm not likely to find such a device, but that seems to me to be the perfect "linear" driver we'd like to have. What am I missing? Todd. Todd Westerhoff SI Engineer Hammerhead Networks 5 Federal Street Billerica, MA 01821 twester@xxxxxxxxxxx ph: 978-671-5084 ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list Old list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list Old list archives are viewable at: http://www.qsl.net/wb6tpu