Why do I need to hold the state of certain pins/nodes?

Non-boundary scan devices such as DDR or flash memories need to hold one or more nodes (chip select/enable) to fixed states during boundary scan testing, so that the device is disabled during boundary scan testing and its output pins (i.e. data) are not activated, as otherwise, it will cause unstable tests or damage to the circuitry. Boundary scan “hold states” are node/device pins on a board that are held to a logic of 1 or 0 by a boundary scan driver on the node, when the boundary scan device driver/output pin is in EXTEST.

An example is shown in figure 1. The boundary scan device (U1) is connected to another boundary scan device (U2) and to a non-boundary scan DDR device (U3). The DDR device data lines are connected to the U2 data lines and the DDR (U3) should be disabled during boundary scan testing so that it will not conflict with the U2 data lines at any time. This means that the U1 chip select nodes should be “held high” during boundary scan testing to avoid data bus conflict or worse, damaged devices.

         Note: Chip Select should be hold high “1” to disable U3 DDR device

Figure 1. Boundary scan device connected to a non-boundary scan DDR device.