millis() wrong result

[stan cartwright]

I tried this code and it should run at around 50Hz but usb scope says 60 odd kHz.
If I scope the 1 KHz calibrate probe pin it says 1 KHz.

#chip mega328p,16
#option Explicit
#include <millis.h>
dim elapsed as word
dir portb.1 out
elapsed=millis()
do
  if millis()-elapsed>=20 then
    elapsed=millis()
    portb.1=!portb.1
  end if
loop

[William Roth]

Looks like you have it coded incorrectly. Try this instead ( Adapted from Demo) This should set B.1 on for 10 ms then off for 10ms giving a frequency of ~50Hz (Untested)

#chip mega328p,16
#option Explicit
#include <millis.h>
Dim CurrentMs as WORD
Dim PreviousMs as WORD
Dir PortB.1 out

CurrentMs = 0
PreviousMs = 0

Do
    CurrentMs = Millis() 
    If CurrentMs - PreviousMs >= 10 then 
        PortB.1 = !PortB.1 
        PreviousMs = CurrentMs
    End if
Loop

[stan cartwright]

Thanks. I saw Chris Ropers demo
I don't see my code gave so high a frequency. and why

 if millis()-elapsed>=20 then
    elapsed=millis()

Didn't work. I shouldgiver 25Hz.

[Chris Roper]

NO - That should give 50Hz

20mS = 20E-3 = 0.020
1/0.02 = 50

if you want 25Hz then:
1/25 = 0.04 = 40E-3 or 40mS

so

if millis()-elapsed>=40 then ' will generate a 25Hz signal

[stan cartwright]

CurrentMs - PreviousMs >= 20 gives 25Hz
CurrentMs - PreviousMs >= 10 gives 50Hz

[William Roth]

Sure. The pin is on for 10 ms then off for 10ms giving a period of 20ms. 1000ms / 20 ms = 50Hz

[stan cartwright]

Don't buy a new usb scope. Get a second hand hardware scope with real knobs.
Modern ones are all software dso though. I wouldn't buy an old scope, too much component wear.
Scopes are very useful. I use a hantek 2022 with official and open source software, only 16M samples/S and not suitable for all ucontroller testing. Better than nothing I guess.

[mkstevo]

If anybody wants a USB 'scope, I could let one go for a £10 (which I will make a donation to The British Hedgehog Preservation Society) and the cost of postage. I think it cost me something like £50? It's not great, but does work. 2 Channels, good for about 10MHz. Model number is Cleqee OSC482.

I've replaced that with a Fnirsi 1014D. Claimed to be good for 100MHz but realistically 25-30MHz. It is however a fabulous little 'scope. Much better than the one I bought at work (OWON) that cost well over twice the price. Cost just under £150 from AliExpress.

I would really like an ana-digi type "real" 'scope with a CRT but a used Hameg will often fetch significantly more than the amount I paid for a newer and much lighter LCD one.

[stan cartwright]

Mkstevo Right on and all. looks tidy. How much? Postage 4u a few quid ukmail to wales,cardiff.
Typical dso in a box , nice.

[mkstevo]

Just to be clear, the picture above is NOT the one I'm selling, it is my new one.

This is the one I'm selling.

[stan cartwright]

Oh dear. I'll stick with mine until I can get a real scope, not with a crt and the faults but a lcd.... lots o money for a new one from flea bay though.

[William Roth]

I purchased a new Siglent 1202X-E about 6 months ago to replace my old TEK with a dim screen. So far I am very pleased. I use this scope almost daily with no problems. More than good enough for microcontroller stuff.

[stan cartwright]

Rigol 1202X-E looks good. 200MHz. £360 is a bit over budget for a hobby.... I don't fix stuff anymore. Think this any good? https://www.banggood.com/Hantek-DSO2C10-Digital-Oscilloscope-2CH-Digital-Storage-1GS-or-s-Sampling-Rate-100MHz-Bandwidth-Dual-Channel-Economical-Oscilloscope-p-1765866.html?utm_source=googleshopping&utm_medium=cpc_organic&gmcCountry=GB&utm_content=minha&utm_campaign=minha-gb-en-pc&currency=GBP&cur_warehouse=CN&createTmp=1&utm_source=googleshopping&utm_medium=cpc_bgs&utm_content=tanya&utm_campaign=tanya-ssc-gb-euw-all-0113&ad_id=491180619117&gclid=Cj0KCQiAosmPBhCPARIsAHOen-PzQiVJEoWhXtpqc0_Dmv-9gHuwCwkbAuSV7RpZXp-w-woDTVAhBcwaArEKEALw_wcB

[mkstevo]

The FNIRSI I bought can be bought as a "Yeapook" on Amazon for £149.99.

https://smile.amazon.co.uk/YEAPOOK-ADS1014D-Oscilloscope-Generator-Bandwidth/dp/B099S4QP3T

[stan cartwright]

Any problems with it ? There's a similar hantek screen,same price. Get what you pay for I suppose. Using a pc and a usb box is poor. £150 seems ok for a scope.

[mkstevo]

The Hantek is more expensive, and Amazon UK have the Yeapook in stock for next day delivery.

I prefer the layout of the FNIRSI/Yeapook screen. With the measurements shown on the right hand side, there is more vertical space for the traces. They can run from a USB power bank making it (semi) portable too.

[jackjames]

I had an old tektronix that reached the green grasslands of Manitou and I replaced it with this one.
It has good quality and it's honest.

https://www.amazon.it/DSO4102S-Oscilloscopio-Digital-Generator-AC100-240V/dp/B07K32RTL4

[stan cartwright]

Thanks all. I'll see what's reviewed for £150 scopes. I never got x,y mode to work with my hantek usb.

[Chris Roper]

I am getting a bit lost here Stan.
Are you saying that the code works and the USB Scope was giving incorrect results?
Or have we gone off on a tangent discussing Scopes and lost the focus?

If Millis() is not working in the new compilers I need to fix it so please advise.

[William Roth]

Stan's original code worked ok for 65 seconds, then start toggling as fast as the loop would allow. Seems that after 65 seconds the "if" condition is always true causing an a very fast toggle.

Nothing wrong with the scope.

[Anobium]

So, no issue with millis() ? Stan's code just is not right.

[stan cartwright]

The point was strange results on scope but scope gets the 1KHz probe trimmer signal correct.
I dug out some 50Hz timer 1 & 0 interrupts that worked once but gave strange results.
I tried using millis() instead. I think millis() is a long and I'm using word vars.
If I change elapsed to a long it seems to work when if millis()-elapsed>=10.
Probably scope triggers on rising edge which is every other 10ms so gives 50Hz.

[stan cartwright]

The point was strange results on scope but scope gets the 1KHz probe trimmer signal correct.
I dug out some 50Hz timer 1 & 0 interrupts that worked once but gave strange results.
I tried using millis() instead. I think millis() is a long and I'm using word vars.
If I change elapsed to a long it seems to work when if millis()-elapsed>=10.
Probably scope triggers on rising edge which is every other 10ms so gives 50Hz.
I guess if millis()-elapsed>=10 will only work as both long vars or if millis() >65535 then it won't using a word var for elapsed. Testing a new gcb release.

[Anobium]

This is Bill's method. Note the assignment of the CurrentMs variable. Why is this important? This 'If CurrentMs - PreviousMs >= 20 then' will work as expected and the variable states will be caches when the millis interrupt fires.

CurrentMs = 0
PreviousMs = 0

Do
    CurrentMs = Millis()
    If CurrentMs - PreviousMs >= 20 then
        PortB.5 = !PortB.5
        PreviousMs = CurrentMs
    End if
Loop

This may not work as expected ' if millis()-elapsed>=20 then' Why? The interrupt can fire durin g calc of millis and elapsed.... yes. Millis is four bytes and those four bytes can change messing up any calculation. This is NOT an implementation issue of millis() this is a usage issue. The way to stop this... use Bills method.

[stan cartwright]

It has CurrentMs = Millis() in the loop, CurrentMs will overflow after time.
Your point about a millis event happening during the main program is true.
Use int off, do a comparison to millis(), then int on?