Talk:First-order hold

Switch of terms pfoh vs foh

The term first order hold seems to be switched with predictive first order hold. In the literature predictive first order hold is used when the output is a interpolation of the two nearest input (ref http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=203948). While zero order hold has the slope given by the last two samples (ref http://www.mathworks.se/help/toolbox/simulink/slref/firstorderhold.html). --Tibnor (talk) 17:44, 5 June 2012 (UTC)

First of all, a ZOH has a slope of zero and the value is equal to that of the most recent sample in the past. The FOH is simply doing linear interpolation, but if it is causal it cannot interpolated to the immediately following sample if it does not know what its value is. The delayed FOH is the same thing, but there is enough delay (1 sample) so that it knows both ends of the linear segment used for interpolation. The predictive POH extrapolates rather than interpolates. That's the main difference.

Can you get this IEEE document? The title to the document is predictive FOH, so I am curious as to in what sense it is predicting. The first FOH is not realizable anyway, but is that what the IEEE author is talking about? 70.109.182.9 (talk) 19:14, 5 June 2012 (UTC)

6 years later: I have the previously discussed terminology problem again. Most literature seems to use the term FOH for what this article calls "Predictive first-order hold". (see wikibooks) I have not found any literature for the (acausal) "Basic first-order hold". Most likely because an acausal filter is a quite useless construct to begin with. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=203948 introduces the term "Predictive first-order hold" for a completely different concept. One delay element it canceled from a prior system block (its presence is a requirement for this method). Then the paper uses a "Delayed first-order hold". The cancellation compensates the extra intoduced delay and the result is what this article would call a "Basic first-order hold". It is a neat method that makes sense in some modeling concepts. I actually do not like that the paper calls it "Predictive"; there is no precidtion. But published is published.

The "Predictive first-order hold" in this lemma uses the past two points to "predict" the slope of the following interval. This is to me a resonable use of the word. But we use the terms differently as some of the (older) literature. Does someone have good Literatue (established books) for a solid citation? --Jahobr (talk) 09:28, 9 April 2018 (UTC)