Integral, time, complex formula issues.

Forum for discussing PicoScope version 6 (non-automotive version)
Post Reply
lab!fyi
Newbie
Posts: 0
Joined: Mon Dec 12, 2016 6:58 pm

Integral, time, complex formula issues.

Post by lab!fyi »

Hello!

Scope fails integral(derivative(A)) math test. Started investigating, in the process did find out:

- derivative goes off the scale at the start of capture
- derivative autoranging does not work
- tried to solve with time skew, discovered that:
-- A[0.00000005] works
-- A[50n] does not give error but also no output
- time skew cannot be used in complex formulas

Tried on 2 different scopes, official and beta software.
Attachments
PicoScope_derivative_and_time_bugs.png

Martyn
Site Admin
Site Admin
Posts: 4491
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Integral, time, complex formula issues.

Post by Martyn »

One thing to note is that you only have 100 true data samples across the screen, the rest of the sample points are created from interpolation.

What is the purpose of these calculations ? What are you trying to determine from the data ?

With the equations you are creating you are performing maths, on maths, on maths, so a better understanding of the required results would help.
Martyn
Technical Support Manager

lab!fyi
Newbie
Posts: 0
Joined: Mon Dec 12, 2016 6:58 pm

Re: Integral, time, complex formula issues.

Post by lab!fyi »

Martyn wrote:What is the purpose of these calculations ? What are you trying to determine from the data ?
Purpose is functional testing of the scope. Reversing the operation is very good for that. Before processing real data and drawing (sometimes very important) conclusions these kinds of tests are essential. I have stuck with Pico because do not know any other scope that can support page wide formulas in real time without time-consuming use of external software.

Specific test was requested from me by scientist to see if Pico performs better than scope from other company (that he is currently using) with differentials/integrals. It would, if not described bugs.

Picture of similar bug attached. Problem is not the sample points count. Probably error lies in incorrect internal boundary condition handling at the start of the trace.

Bug 1:
integral(A[0.0001]) works (integral on time advanced channel)
integral(A[-0.0001]) does not work (integral on time delayed channel)

Bug 2:
Time skew does not work in complex formula,
will display "Invalid formula".
Example: A-B phase calculation:
acos(integral(A[0.0001]*B)/(sqrt(integral(A[0.0001]*A[0.0001]))*sqrt(integral(B*B))))/pi*180

Suggestion: Channel time skew compensation is basic feature of DSO. It must work with full complexity math. However it is cumbersome to have it only in math channels. Would be very helpful to also adjust global time skew for channel in channel menu (where probe X is set etc) and in math independently. VHF freq range is problematic with multiple channels and BNC cables w/o time skew compensation (non-equal length cables may be a problem etc).
Attachments
PicoScope_integral_bug.png

lab!fyi
Newbie
Posts: 0
Joined: Mon Dec 12, 2016 6:58 pm

Re: Integral, time, complex formula issues.

Post by lab!fyi »

Hi again, just discovered that last stuff I posted was already gotten a ticket #TS00079098 by e-mail discussion. Had not noticed reply e-mail somehow... I was asked several questions in e-mail and just sent reply with explanations why is it so I want time-skew math feature with at least 1 picosecond resolution :)

Martyn
Site Admin
Site Admin
Posts: 4491
Joined: Fri Jun 10, 2011 8:15 am
Location: St. Neots

Re: Integral, time, complex formula issues.

Post by Martyn »

I have taken a look at the email, and the other two you have sent, and will follow up through the ticket system. Our offices are closed until the 3rd of January, and I don't have access to any scopes to test with.
Martyn
Technical Support Manager

lab!fyi
Newbie
Posts: 0
Joined: Mon Dec 12, 2016 6:58 pm

Re: Integral, time, complex formula issues.

Post by lab!fyi »

Hi,

magenta trace - derivative(A)
lime trace - (A[0.000000002]-A)/0.000000002
Latter is 1 sample point time advanced (2ns) due to nature of function.

Cant integrate none of them. Only way to make ref wfm and integrate then.
Result perhaps sheds some light why it is not working in real time.

purple trace - integral({derivative(A)})
green trace - integral({(A[0.000000002]-A)/0.000000002})

None of them lands on where it should, purple on blue, green on (blue - 2ns).
Attachments
PS6_derivative_integral.psdata
(12.48 KiB) Downloaded 506 times
PS6_derivative_integral.jpg

Post Reply