Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
2001-02-05
2004-08-17
Baderman, Scott (Department: 2114)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C714S030000, C714S719000, C365S222000
Reexamination Certificate
active
06779136
ABSTRACT:
BACKGROUND OF THE INVENTION
Field of the Invention
The present invention generally relates to information memories that require a refresh device in order to refresh at certain time intervals the information that is held in the memory cells. The invention specifically relates to a method for testing such a refresh device of an information memory that is designed to refresh the information stored in a multiplicity of cells of the memory as a state of the respective cell, in each case before a guaranteed minimum retention time has elapsed. The refresh device includes a refresh selector for selecting memory cells to be refreshed; a sensing device for sensing the state of each cell selected by the selector; a restorer for setting each selected cell into a fresh state, which, in a refresh operating mode of the restorer that effects the refreshing, represents the information derived from the sensed state. An important, but not exclusive, application of the invention is DRAMs (Dynamic Random Access Memories), i.e. dynamic memories with direct access, in particular semiconductor memories of this generic type.
The extent to which a cell state that has been imprinted by the writing of an information item remains stable depends on the nature of the memory cells used in an information memory. If the cells include bistable electrical circuits (flip-flops), then the information that has been written is preserved as long as the power supply is not interrupted. However, certain memory cells of a different type are configured in such a way that in the course of time they lose the information that has been written, and therefore have to be “refreshed” from time to time.
This applies for example to memory cells in which the actual memory element is an electrical capacitance (capacitor) with different possible charge states, an information item that has been written being represented by the level of the charge. On account of inevitable leakage of the capacitor, the introduced charge volatilizes in the course of time to such an extent that an information item represented by introduced charge no longer can be unambiguously identified. The cell state can then be interpreted incorrectly during reading. If the cells are operated as binary memories, by a distinction being made only between the two cell states “charged” (high or H level) and “discharged” (low or L level), for the representation of the binary values “1” and “0”, then after a certain time the charge of a cell which has had “1” written to it may have decayed to such an extent that a “0” is read at this cell. Quite similar problems arise to an even more pronounced degree if the cells are each used to store more than two discrete information values, by a number of information values being assigned to specific intermediate levels of the charge.
In addition to the capacitive memory cells mentioned above, other kinds of memory cells may also require refreshing. In general, the invention applies to all types of memory cells in which at least one of the information-describing states is volatile. In this case, these states may be of an arbitrary physical or chemical nature.
In principle, a refresh includes the following: the cell state is sensed in good time before it might have volatilized so far that the information represented by it could no longer be unambiguously identified; after, the information identified by the sensed cell state is written afresh to the relevant cell.
The period for which an information item can be retained in a cell, i.e. the “retention time” for which the information item that has been written remains unambiguously identifiable in the cell, is dictated by construction and can differ greatly from cell to cell within the same memory module. In commercially available DRAMs effecting capacitive storage, the guaranteed minimum retention time of a “1” (that is to say of the information described by the H level) is usually a few milliseconds, whereas the actual retention time of the “1” may randomly be much longer in some cells, in many cases even up to a few seconds. When choosing the time intervals for the refresh, however, it is necessary, just for organizational reasons, to comply with the guaranteed minimum retention time, i.e. the intervals between the refreshes must not be longer than this period of time.
An information memory whose cells are in need of the refresh requires and uses, as is known, a refresh device having the following constituents: a refresh selector for selecting memory cells to be refreshed; a sensor for sensing the state of each cell selected by the refresh selector; a restorer for setting each selected cell into a fresh state. The refresh device constructed in this way is normally operated automatically in such a way that the selector selects all the memory cells in accordance with a sequential program set by the user, that the sensor senses the state of each selected cell, and that the restorer sets the relevant cell afresh into that state which corresponds to the information derived from the sensed state. The aforementioned sequential program of the refresh selector must be configured by the user such that no cell remains unrefreshed for longer than the guaranteed minimum retention time of the memory.
In the course of the design analysis and in the production test, a check must be made to determine whether the refresh device can carry out the desired refresh reliably and at all of the cells which are to be selected. A possible malfunction may be that the refresh selector does not correctly follow the set program. This can happen in particular when, in the selector, a cyclically operated refresh counter is used for the cyclically repeated selection of the addresses of the cells or cell groups to be refreshed and the overflow function of said counter does not work correctly or the counter stutters in another way. Another malfunction may occur when a cell that is selected for sensing is not reached by the restorer.
A method for testing the refresh device is insufficient if it only includes the following: writing a known information item tending toward volatility to the entire cell array, then completing a refresh cycle over all the cells shortly before the minimum retention time has elapsed, and subsequently verifying whether all the cells still contain the information that was written. This method is insufficient because those cells whose actual retention time is distinctly longer than the minimum retention time may, at the instant of verification, have retained their information even if they were passed over in the refresh cycle.
In order to yield a really meaningful test result, it is customary, therefore, to write the information and then to carry out many successive refresh cycles, at intervals of in each case not longer than the minimum retention time, but in total for a duration which is longer than the maximum retention time to be expected only in this way is it possible, using the subsequently sensed information content of the cells, to ascertain whether and which cells were regularly passed over during the refresh cycles. However, this method requires long test times and is ruled out, therefore, in particular when the tests, in the case of a relatively large memory, ought to be carried out only in sections on small memory blocks.
As an alternative, instead of checking the refresh result itself, the mode of operation of the refresh counter might just be checked. However, this does not allow identification of many defect states, such as, for instance, the lack of actually being able to reach a selected cell. A defect for example in the wiring between the refresh counter and the address decoder would remain unnoticed, as would a defect in the multiplexer which is usually provided for selection between normal address and refresh address. Moreover, a counter check requires the detection of the respectively existing count status (instantaneous count) in order to be able to ascertain any defects in the operation of the counter. In many refresh counters, displays indicating the counter reading or a
Richter Detlev
Spirkl Wolfgang
Baderman Scott
Greenberg Laurence A.
Infineon - Technologies AG
Lohn Joshua
Mayback Gregory L.
LandOfFree
Method for testing the refresh device of an information memory does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for testing the refresh device of an information memory, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for testing the refresh device of an information memory will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3314275