News:

;) This forum is the property of Proton software developers

Main Menu

Do Logic analysers have glitches

Started by RGV250, Apr 09, 2023, 03:06 PM

Previous topic - Next topic

RGV250

Hì,
I am trying to analyse some signals and not getting exactly what I was expecting. I am getting what look like genuine pulses but I cannot see what the y would do. They seem to coincide with a rising or falling edge on another pin.
The smallest one is 1ns which I think is impossible. The device has a 6mhz xtat so even with a 1:1 instruction the dest it could be is 167ns if my maths is correct.

Bob

TimB


What logic analyzer are you using, what software, what sample rate?

 

david

I think it's unlikely to be the logic analyser.
Also, it's not the clocking rate that determines the speed of the pulses - this could be a slow clock with very fast rise/fall times coupling via a few pF into a pin.   As an example 2pF into a 1k resistance will produce some very narrow pulses regardless of clock rate.

Cheers,
David

trastikata

I got similar glitches on my DSO with a faulty probe.

Couldn't find the exact reason for that but I've determined that the BNC connector had to do something with it.

Anyway, does the signal appear if another probe at another port is being used for the same signal?

Jean Morin

Hi TimB

I got Saleae Logic 8

The best for me
73 de VE2TI / VE2CSN

RGV250

Hi Tim,
It is a Picoscope 2205MSO and Picoscope 7 software, the sample rate is 20ms/s. If I sample slower I miss pulses.
I am trying to find out how to control a colour matrix display with a PIC as the manufacturer does not have any information on the data format required.
These little pulses do not make any sense, (D6 A0) is for the row selection so should only change after the 8 clock pulses at the bottom (D2 Clock). Also the ones on (D3 Red & D4 Green) do not look right, Green looks OK apart from the one on the left but Red doesn't. Unfortunalely I can only run the test program which goes through all 3 colours and scrolls the message.
trace 1.jpg

Bob

david

Is it fair to say that most of your glitches relate to the clock rising edge with a few relating to the falling edge?  That may suggest capacitive coupling - possibly between logic probes. Try adding about 470R on the clock source and connect the logic probe to the end of that. (effectively a low pass filter)
Any sign of the glitches if you use the two analog inputs on the 'scope?  Try a different ground reference point for the logic ground in case you're getting ground bounce on the point you're on.  Add more decoupling on the logic circuit - clock edges are where most current is drawn because gate and track capacitance has to be charged and discharged.   Sorry - fresh out of ideas but good luck tracking down the gremlins. 

Cheers,
David


Stephen Moss

#7
I was going to say the same as david, that it could be capacitve coupling along the ribbon cable, but ususally you would expect to see them on adjacent cores but in this case it seems to permiate further that I would expec and it is not consistant, i.e. the glithched on D10 and D11 are differnt so maybe they are not random errors/capacitive pickup but a result of something in your code/PIC hardware resetting.
I have just obatined a PicoScope MSO for work, but cannot take a look to see it is may be a common issue with them as I am currently on holiday.

First thing I would do is disconnect the effected signals from the MSO, if there is no change then I would still suspect either capacivitve coupleing in the Ribbon Cable taking the signals to the PicoScope or between PCB tracks inside the PicoScope, or possibly an source internal to the PicoScope is likely to be the problem.
However, if disconnecting those signals causes the glithes to dissapear then it would seem logical that either capacitive coupleing on your PCB or something in your code/the PIC is likely to be the source of the issue.

Finally, are you using the locic analysers internal clock to take samples or an external clock?
If the former then it could be that the appeaance of the glithes are an effect of a lack of sysnchronisation between the analyser and signal clocks, in which case using say the CLKout pin of the PIC as the Clock source for the logic analyer may result in them not being captured or thier location/size altering as the better synchronisation results in more accurate signal capture.   

RGV250

Hi,
I am using the sort of ribbon cable supplied, I have a cheap Salea clone somewhere, I might try to find that and see if it is the same. I am not sure if it is fast enough though.
I am going to give Picotech a call, strange they have all the other crap like facebook and twitter but no email.

It is not the PIC, I have not got that far yet, it is just an internal test on the display. If it only did one colour it might help but I have traced all but one of the signals so this is more to confirm things.

Finally, are you using the locic analysers internal clock to take samples or an external clock?
If the former then it could be that the appeaance of the glithes are an effect of a lack of sysnchronisation between the analyser and signal clocks, in which case using say the CLKout pin of the PIC as the Clock source for the logic analyer may result in them not being captured or thier location/size altering as the better synchronisation results in more accurate signal capture.

I will have to look that up, at the moment I am using the reset pulse as the trigger.

Bob

RGV250

Interesting and confused.
I asked on the Picoscope forum and it is resolved, I am not sure why though.
The first question was if the GND connection was good, I thought it was but put all 4 of the connections to GND. This made it a bit better, the second was the threshold, I changed it from 5v to 4v which I thought would throw up more glitches but I am now gertting exactly what I thought I should be getting.

Bob

david

Good to hear you got it sorted.  Sharing a single earth return was perhaps pushing it a bit as the shared inductance for the current spikes on leading and trailing edges would provide a common coupling path. 
Not sure what your Vdd was (5V?) but setting the trigger point close to either Vdd or ground may pickup "bounces" off the rails but this is usually only seen with high speed edges.

David

RGV250

Hi David,
Just out of curiosity I removed the other GND connections leaving the one and it is still good. I am going to put them back in case I am pushing my luck.

The VDD was 5v which is why I set it so, I definitely will not forget that next time. I am glad it is sorted as I am not sure if it is possible to alter on the cheaper analyser I have. It would not have been any use though as it only samples at 12ms/s which is where I was missing pulses as well.

Bob

david

Hmmm.   If the Vdd is 5V I would be setting the trigger point to about half that unless you're specifically looking for ringing in the supply line.

David

RGV250

Hi,
It shows I know next to nothing about this, I thought that the closer the threshold to VDD was it would filter out small glitches. 4v seems to work fine but I will take that on board when I switch to 3.3v assuming the logic IC's are OK with that.
I have worked out what all the signals do now apart from one, it is a litte annoying that I cannot find it with a meter either.
At least I have all the timing details so it is now on to the programming.

Bob

david

4V should be fine but the closer you get to the rail the more likelihood you could get triggering on ringing with fast logic. At some frequency decoupling capacitors go through self resonance and start looking inductive.
You're a brave man to take on a project like that with such limited data.  I hope it all goes well for you.

Cheers,
David

RGV250

Hi David,
QuoteYou're a brave man to take on a project like that with such limited data.  I hope it all goes well for you.

As Les said on the LOL shield page of his site.
QuoteEven now, after years of working within this industry, LEDs have a fascination to me. And an LED matrix is something that I can't wait to try out.

Well this thing has 2560 of them (and in 3 colours) so even more fascinating.

I am a hobbyist and now retired so I have spare? time and it seemed like too good a challenge to let go.
My sister asked why bother as I have no idea what I will use it for but I replied why do people do crosswords and Soduku, it is fun. And as I think John Drew pointed out, it keeps the brain cell active. And I bet ChatGPT could not work it out :)

I am also learning things on the way, it is even more fascinating that I have worked out there are only 11 signal/data lines and one does not appear to be needed.

Bob

david

That's the spirit!   I suspect there's a bunch of us in the same category (retired tinkerers) who would applaud your enthusiasm taking on this project. Good luck unravelling its mysteries and perhaps we might see some pictures at a later date?

Cheers,
David