Segmented Memory mode:
We previously talked that v1.20 hack enabled 2 MPoints of Single acquisition memory and referred to Keysight’s explanation how it is used in different acquisition modes. Now, in reality things are a bit more complicated and may be confusing to an extent. This specifically applies to Segmented memory mode. Surprisingly, in fact FERSA 1.10 hack already made 2M Points of Segmented memory available, but it could only be used by the scope in Segmented mode (vs 1M points in Single acquisition mode). This new v1.20 hack has liberated 2M points for Single acquisition mode while Segmented memory stayed at 2M points. It is probably fair to say that most of the users will use Single acquisition mode much more often than Segmented one, so doubling the Single acquisition memory in v1.20 hack was of a great value to them.
To add to the confusion, the v1.20 hack actually also increased utilization of the Segmented memory by the scope. Stay with me to understand why. Let’s for the moment refer back to the following part of the above screenshot about the maximum number of data points:
- Whether segmented memory (available on DSOX1000-Series models) is on. Acquisition memory is divided by the number of segments
I found the second part of it not being entirely accurate. Calculating the number of points per segment is not as simple as dividing total memory by the number of segments. The scope seems to have a sophisticated algorithm for memory partitioning in Segmented mode and the number of points not always adds to 2M points. I am guessing this may be because some of the memory is used by the scope to perform the task for calculations, indexing of segments and such. It may vary as much as 50% depending on the number of segments selected by the user and is especially pronounced at low segments count, approaching more or less even and full memory utilization at higher segments count.
If that was not enough of a confusion and you want more, here it is. In Segmented memory mode there are actually two of them: Linear and Circular. Up until this point we were talking about Linear segmented mode. When you select Linear segmented mode, the scope will wait until all of segments are filled with data and then automatically stop the acquisition and display the captured data. In Circular mode the scope will continuously wrap around after all of segments are filled and it is up to the user when to hit the Stop button (while running in Circular the scope will display a rough segment counter but will not display the signal being acquired).
But this is not the only difference between Linear and Circular modes. They also seem to have different memory allocation algorithms. Number of memory points per segment in Circular mode always is less than that of in Linear mode under all other conditions being equal. My guess is the same as in the Linear mode this may be because of additional overhead required by the scope to perform the task, plus some more for additional calculation or alignment of segments, or to maximize the speed, do not know. The difference in number of samples per segment for a given Time/div between Circular and Linear modes can be as big as 50 %, that is – in Circular mode you may get twice as less number of points per segment then in Linear mode (depending on the number of segments). I think this is something that may need to be understood and kept in mind.
Now to the good news: the v1.20 mem hack provides an improvement in memory utilization in both Linear and Circular segmented modes, as much as 30% at some lower segments count. The chart below represents memory utilization in Linear mod for v1.20 with no mem hack and the same v1.20 with one.
Home
Help
Search
Login
Register
