Memory Flag: delay question

[J V]

Hello,

I think I've been misinterpreting the delay-option in a Memory Flag... According to the description, it is "Delay input, by x cycles, to the output". Suppose I have a memory flag A with delay 0 and a flag B with delay 1, both connected to some analogue value V (edit: V is an analog output).

I thought that at any point, the value A reflects the current value of V, and the value B reflects the value in the previous cycle, but it appears that if V does not change, neither does B. As such, the description "by x cycles" is NOT machine cycles (as I thought) but rather "value changes of V"...

I tried to explain it in a table, suppose V varies as shown in the second row.

cycle	1	2	3	4	5	6	7	8	9
V	5	5	6	7	8	8	8	9	9
A exepected	5	5	6	7	8	8	8	9	9
B exepected	-	5	5	6	7	8	8	8	9
A observed	5	5	6	7	8	8	8	9	9
B observed	-	5	5	6	7	7	7	8	8

In bold I highlighted the values that were surprising to me.
I hadn't used those delays much, but started to use them and was struggling...

Do I now have the correct interpretation of how the delay works?


Thanks!


EDIT: I'm unable to reproduce this behaviour, the observed output is the expected output.

Jörg

[Christian Fenzl]

You cannot connect two analogue values to one analogue input. The design of your test is wrong.

[J V]

There is one analogue value V, V is an analog output... and then one memory flag A with delay=0, and one memory flag B with delay =1, both taking V as an input.

The lines "expected" is what I thought it would give, the lines "observed" are what it gives...

[romildo]

I cannot confirm your observations.
In my attempts it behaves as expected in the list

Cofig V11.1.9.14

[J V]

So it should be like my expected behaviour? Thanks!

I'll have to check again, but I had the impression it behaved differently for values coming from KNX...

[Christian Fenzl]

Could you show your Config of this test.
I seem not to understand what your logic looks like.

[J V]

Config is quite complicated... it just seems that delay 1 is only updated it the analog value to which it is connected changes, not with the machine cycles. I need to check more as it may be some side effect.

[J V]

I cannot reprocude it anymore... now it behaves as one should expect, on all the values I tested yesterday... Weird...

But at least the confirmation is that "cycles" in the delay documentation referes the cpu cycles.