DSO 150: Timebase

The voltage management was designed by Jyetech to reflect the values they wanted to use.
The resistors have the right values to form good voltage dividers, and as a result, converting between them and volt/div is pretty simple.

It's not so simple for the timebase.
Ideally, we'd like standard value, such as 10us/div, 50ms/div etc...
Since we use 24 pixels per div, that would mean 10us/24, 50ms/24 etc... sampling frequency

Unfortunately, we don't have the required  flexibility as far as sampling frequencies are concerned.
Here is a  table of all the available values (prescaler x sampling time)

	2	4	6	8
1,5	0,39	0,78	1,17	1,56
7,5	0,56	1,11	1,67	2,22
13,5	0,72	1,44	2,17	2,89
28,5	1,14	2,28	3,42	4,56
41,5	1,50	3,00	4,50	6,00
55,5	1,89	3,78	5,67	7,56
71,5	2,33	4,67	7,00	9,33
239,5	7,00	14,00	21,00	28,00
		Time/sample (us)

Multiple by 24, to get the time/div
Not so great in terms of user experience, 45.33 us/div is not very user friendly.

So, what we'll do is  oversample a bit the value we actually aim at, and then  decimate it.
It is a very simple decimate so that it does not slow down everything too much.

For example, if we want 100 samples at 500us/div, we'll program 400us/div, take 125 samples and decimate them regularly to keep only 100 values.

The table to deal with that is now:

Time/div (us)	ADC	Prescaler	AcqTime	Expand	Expand *4096
25	17,33	2	13,5	1,44	5909
50	45,55	2	55,5	1,10	4496
100	90,67	4	55,5	1,10	4517
500	336	4	239,5	1,49	6095
1000	672	8	239,5	1,49	6095

Where expand is the discard/expansion ratio.
If we want 100 samples, we need to acquire 100*Expand, and decimate accordingly.

We might do a real resampling later if it is fast enough

When selecting the candidates for the AcqTime/Prescaler pair, we'll take the highest Acq time to favor accuracy.


So now, we have 'fast sampling' working for both time & voltage.
Need to do 'slow sampling' & triggering, and we have something remotely usable.