Check your sample rates. I'm sure it's an undersampling problem...
it's not a sampling problem...
when i lay down in bed, blood flows easy into the brain, and the answer pops up easy.
The Conclusion:
Firmware v1.08 the Boxcar Average Filter "Hi-Resolution Mode" was controlled only by the HW Frequency Counter, controlled by the Trigger.
does not matter how i moved the Horizontal/div knob, the Boxcar filter does not move, because it's following the Frequency counter.
I understand that English is not your first language, so it's very hard for me to follow your logical argument, but I still thinkn that it's a under-sampling issue -
From what I can tell your scope is a 350MHz model with 5Gsps or something, but only 25kpts/channel.
This means that in the slow time-base settings your sampling rate is memory limited.
your write in teh first post:
"200ms/div Waveform disappears at 12.2KHz & 48.8KHz & 97.7KHz & 194.99KHz."
with 10 divisions per screen, those 25kpts are divided over 1sec of time, at 200ms/div
That means that the effective sampling rate is 25ksps, and you would expect a null in your spectrum at 25kHz, and multiples thereof.
Now the boxcar filter probably operates on the sample memory, and if it takes an average of two samples, I would expect it to have a null at 12.5kHz (and odd multiples of that frequency)...
Now if the boxcar filter is a average of two points, it will
If it were a memory problem, Normal Sampling Mode would also be affected, and any other sampling mode, Peak & Normal Average.
"Problem" was that: when i upgraded the firmware,
found that, when looking a frequency 200x times higher than the visible on the screen per time/div,
makes the Square wave disappear, but only in Hi-Resolution mode.
That didn't happened with older firmware,
because the Hi-Resolution Boxcar algorithm, was following the Frequency counter of the waveform, not the screen time/div,
if frequency moves up, the filter moves up, and i can never reach the other side of the filter.
but because Scope has an Alternate Trigger On/Off "dual Trigger/one for each channel in a dual channel scope"*
that feature was changed to allow Hi-Resolution Mode in every channel independently,
that would be impossible with 1 hardware Frequency counter per Oscilloscope,
the ideal method would be to have a hardware Frequency counter per channel, and each Boxcar algorithm following each frequency counter in every channel.
*don´t know if the 4 channel version has 4 triggers.
Relation of sample rate, memory & time is: Capture Interval = 1/Sample Rate X Memory.
&
Capture Interval/10 = Time Per Division.
Every Scope has a Sweet Spot, where refresh rate, sample rate, wfm/s are the best.
Because of Memory Depth Limitations the Sample Rate Changes,
Because of Speed Limitations the wfm/s Changes.
but the Boxcar algorithm is just a very strong brick wall low pass filter, at a very high frequency following the maximum visible screen frequency,
Variable, not fixed like BW Limit,
it´s like a super smart Bandwidth Limit, similar to a "Rocktron HUSH Noise Reduction." that follows an envelope follower,
cleaning up the signal of sometimes unnecessary hi-frequency noise but keeping the super fast rise times & all the advantages of a high speed scope when working at low speed.
As far as I know is Not Possible to Maintain max-settings constant in every setting, at least in this price range.
Here is a table for this scope: Settings vs. Memory Depth.-
Sample Rate vs. Time/div.: 25 SPS: 100s/div <--- From Here Starts Roll. <--- 25Kpts can record 1000sec. or 16.6 minutes at this time/div.
50 SPS: 50s/div <--- Records >8min.
100 SPS: 20s/div
250 SPS: 10s/div
500 SPS: 5s/div.
1 KSPS: 2s/div.
2.5KSPS: 1second/div.
5 KSPS: 500ms *** Here Starts the Fun.
10KSPS: 200ms/div.
25KSPS: 100ms/div. <--- Roll Mode Ends Here. <--- Records 1sec. Of data at this time/div.
50KSPS: 50ms/div.
100KSPS: 20ms/div.
250KSPS: 10ms/div.
500KSPS: 5ms/div.
1 MSPS: 2ms/div.
2.5MSPS: 1ms/div. ---> Records 10ms. data.
5 MSPS: 500us/div. ---> Records 5ms. data.
10MSPS: 200us/div. ---> Records 2.5ms. data.
25MSPS: 100us/div. ---> Records 1ms. data.
50 MSPS: 50us/div. ---> Records 500us. data.
100 MSPS: 20us/div. ---> Records 250us. data.
250 MSPS: 10us/div. ---> Records 100us. data.
500 MSPS: 5us/div. ---> Records 50us. data.
1 GSPS: 2us/div. ---> Records 25us. data.
2.5GSPS: 1us/div. ---> Records 10us. data.
5 GSPS: 500ns/div.
5 GSPS: 200ns/div.
5 GSPS: 100ns/div.
5 GSPS: 50ns/div. *** Here Starts the Rising Edge Fun.!
5 GSPS: 20ns/div. <-----Here is Real 5GSa/s, no ET, Measuring Jitter in all its Glory!
******************** Here is the Limit of RealTime.
25Kpts at 5GSa/s 20ns/div allows to record 5us. of data.
Basically if you hit [Stop], and go back in <Horizontal Time/div>, the 25Kpts allows to capture a full screen at that resolution, and half screen at 1 time/div back, or 1/5 screen <<2 time/div back.
...
if [Stop] at
1ms/div, 25Kpts captures full screen, when moving the <<Horizontal to
2ms, shows ½ screen of data, at 5ms shows 1/5 data., 10ms shows 1/10 data, 20ms/div shows 1/20 screen data.
So when dealing with 25kpts memory [Run/Stop] it´s not a Magic button where you push and look everything, must be done meticulously,.
******************** The Equivalent Time Starts Here.
ET 10 GSPS Equivalent Time, or 5 GSPS [sin(x)/x]: 10ns/div.
ET 20 GSPS or 5 GSPS: 5ns/div.
ET 50 GSPS or 5 GSPS: 2ns/div.
ET 100 GSPS or 5 GSPS: 1ns/div. ***Here is how it would look in a $1/4 Million dollar scope in RealTime.
The Scope has 2 modes, Equivalent Time &/or [sin(x)/x]
ET looks digital without VPO, with VPO-Active works the same way, but with average sampling, not real sampling, a bit slow.
[sin(x)/x] looks Amazing, like Real Time VPO.
Equivalent Time was designed to increase the time divisions "less than <20ns/div" and increase resolution.
5GSa/s has 200ps per sampling point,
This scope cannot go lower than 1ns.1ns = 1000ps, has 5 sampling points at 1ns, but ET & [sin(x)/x] looks like if it had more SamplingPoints.
10sp at 2ns,
25sp at 5ns
50sp at 10ns …
100sp at 20ns RT.
5 sampling points are needed minimum to measure a Sine wave, like a Triangle.!
.
. . .
.