strange behavior at T flip flop output

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
lock attach
Attachments are accessible only for community members.
user_3716231
Level 4
Level 4
25 sign-ins 10 sign-ins 5 sign-ins

I want to measure frequency and power factor,but I've gone through this issue for a week and don't understand why.when the frequency is set 60hz,I sometimes measure 20hz.the phase shift is set 30 deg,but sometimes it is measured 389.30.

I know there are some solutions in the community post,but I want to apply my method to measure values and the post is locked.

eliminate bounces at comparator output Re: bounces happened at comparator output

measure frequency Re: bounces happened at comparator output

phase shift greater than 30 Re: bounces happened at comparator output

the yellow wave is comparator output.the blue one is tff output.

the green circle is normal.the red one is abnormal.

comp tff.PNG

(please ignore disabled pages)

0 Likes
17 Replies
SampathS_11
Moderator
Moderator
Moderator
250 sign-ins 250 solutions authored 5 questions asked

Can you kindly try once more by setting the Comparator power to Low Power? Kindly let me know the behaviour you find.

Best regards,

Sampath

0 Likes

Sampath,

Sorry for the late reply.

It seems to be normal at tff output when I set comparator to ultra low power mode,but I need more days to observe because I still get the wrong value of frequency.Maybe it is because of that behavior which I just didn't capture.Any idea about it,please tell me.

Thanks your help.

comp_low_power.PNG

0 Likes
Len_CONSULTRON
Level 9
Level 9
Beta tester 500 solutions authored 1000 replies posted

Hi,

The TFF appears to be operating properly.   The issue is that you have tied the comparator output to the clock of the TFF.

At the offending spot of your scope plot, you can see that the comparator output (YELLOW) is toggling multiple times within a very short amount of time.   This is causing multiple clock transitions at your TFF clock input.   This causes potentially many toggles of the TFF output (BLUE).   Given what is happening at your comparator output, this is the expected results of your TFF.

pastedImage_0.png

You need to better filter your input to the comparator.   If you are using LINE AC from power as I suspect, it is notoriously noisy ALL points in the waveform.   This is due to many reasons including load transitions and other Conductive Emissions (CE) of upstream equipment.

Your best bet is to place an external Low-Pass Filter on the input to the comparator with a corner frequency as close to 60Hz as you can tolerate.  The down-side with any analog LPF is that the cap used introduce a time delay to the wave.  This may make using a LPF in your case (power factor calculator) not practical.

Suggestion:   Does your power factor (PF) calculation need to be made in FAST realtime?   If you can take a 'long' view, you can take multiple immediate PF measurements and throw away any that do not match the target 60Hz.  You can then average your PF measurements that match the 60Hz criteria.   This is a SW implementation without any additional external HW.

If you can take the 'long' view, this is a practical approach since the power company is reasonably reliable for their output frequency.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes
lock attach
Attachments are accessible only for community members.
user_3716231
Level 4
Level 4
25 sign-ins 10 sign-ins 5 sign-ins

1.Could someone please tell me why the value of frequency is measured 20hz without sync component?

2.The phase shift will be no more greater than 30 deg when adding sync component.why?

0 Likes

Hi,

Can I recommend an experiment to prove out your PF phase detection circuit?

  • Remove the AC stimulus from the Vadc and Iadc.
  • Insert a digital signal to Vadc at 60Hz 50% DC.
  • Insert a second digital signal to Iadc at 60Hz 50% DC.
  • Control the time delay (phase shift) of the second signal.

The digital square wave should be more friendlier to the Comparators

This test circuit should allow you to detect the frequency at 60Hz.  If you change the frequency, it should reflect this frequency in your debugging display.  This is to prove out the frequency detection part of the circuit without input noise.  It will also help to detect if there are any issues placing the SYNC component after the Comparators.

By time delaying the second Iadc input, you can control the phase shift and prove if your circuit (and SW) can properly detect the correct phase.

Once you're convinced your circuit can detect 0 to 359 degs of phase shift, then you should be able to reapply the AC input stimulus.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes

I have connected external hardware hysteresis.Should I remove them ?

0 Likes

I'm assuming you're referring to the hysteresis on the comparators since your schematic doesn't show any external components.

You can leave the hysteresis.  With a digital input it won't make a difference.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes

Len,

I insert the sqaure wave to Vadc and Iadc and the wrong value still occured when waveform is from external wave generator.

0 Likes

user_3716231

Since the wrong value is still occurring there appears to be something wrong with the logic or your detection circuit.

By using a scope on each stage of the circuit you can verify proper HW operation.

If the HW is operating to design, then there may be a SW processing issue.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes
WaMa_286156
Level 5
Level 5
First comment on blog 100 replies posted 50 replies posted

If your screen shot, which shows 80khz is the sampling rate of the logic or scope, then you may wish to up your sampling rate, or use peak detect on your scope acquisition, should that be available.  I suspect you are getting noise spikes and your comparators are sending them through.  Roughly, you will need about 10 times the frequency you wish to see correctly in order to see the basic information you want.  In this case, you need to narrow down your trigger and sample at 12 mhz to 20 mhz or better.

If you are using 80khz, then frequencies of 160khz, 320khz, 640khz, etc.  will look like slower frequencies.  Nyquist and ADC sample aliasing comes in to bite you here.  I have had a beautiful 20khz sine wave from a 2mhz input signal on a scope due to aliasing.

By using synchronization FF's, you cause the counter signal to ignore intermediate transitions due to noise, which a previous poster pointed out is being hinted at in your scope/logic shot.  That settled down your results, but probably not eliminated your problem.

To "fix" your problem, I would suggest extra power supply filtering on the analog system, or a very small capacitor on the input to the PSOC.  Try 10pf -100pf or so on the input (assuming 60hz) to filter the noise out.  Use a 100 ohm resistor in series with the input, making an R/C filter.  You might be surprised at the results.

0 Likes
lock attach
Attachments are accessible only for community members.
user_3716231
Level 4
Level 4
25 sign-ins 10 sign-ins 5 sign-ins

I have already add RC low pass filter and hardware hysteresis,but the wrong value and the behavior still occur.

0 Likes

user_3716231

I will try to reproduce your result shortly.

I've noticed a number potential issues with your latest TopDesign:

RC Filter

Your configuration for the RC filter does not work.

I've modeled your circuit and performed an AC analysis.

pastedImage_2.png

Below is the AC response from 1Hz to 10KHz.   No LF cutoff for IN1.

pastedImage_3.png

This circuit is a correct LPF for your situation.

Here is the LPF response.

pastedImage_2.png

Note:  Using the series resistor of 10K and the parallel cap of 4.7us your corner frequency (-3dB) is 3.38Hz.  This means at 60Hz you have a 20dB Loss.   This means an input voltage of 3V appears as 0.3V at the comparator.

You might want to lower the resistor value or the cap value.   For example if you keep the 10K resistor you just need a 270nF cap to get a corner frequency of 60 Hz.

Hysteresis

You placed a 150K resistor for external hysteresis.   However you have also configured the Comparator for internal hysteresis.  By "adding"  the external resistor might diminish the hysteresis voltage.   Suggestion:  Eliminate the external hysteresis and only use the internal one.

Asynchronous Comparator Output into a Synchronous Circuit

In your circuit you did not use the "Sync" option.   I believe this is the reason you are getting the following issue you circled.

This is occurring because the timing of the minimum setup and hold times to the compare inputs for the Timers is not adhered to.

To address this issue, use the "Sync" configuration  

Then add the "timer_clk" to the clock input that appears on the comparator components.

Here is a new input front-end circuit with the new RC, hysteresis resistor removal and the "Sync" option.

pastedImage_2.png

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes

user_3716231,

I've got your project loaded on my Cy8CKIT-059.

When you get

pastedImage_0.png

for TFF output, what are you using for Vadc input?   Sinewave or Digital (0V to 5V)?

I'm using a Digital input (0V to 5V) for both Vadc and Iadc.   No issue at TFF output for about an hour so far.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes

user_3716231,

You'll be glad to know that with a Sinewave source (0V to 4Vpp) I've been able to reproduce your results on the TFF output.

Sadly, it appears to be a problem with the Comparator resource.   This is even true with the "Sync" enabled.

I'm gathering more data and submitting this as a example project for Cypress to weigh in on. 

The the issue that I am seeing with the Comparator does not appear to be fixable.

However, I'm also working on a work-around for your application.

Interesting problem!

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes
lock attach
Attachments are accessible only for community members.

user_3716231,

I'm still trying to root cause why I'm getting similar results on the TFF output as you do.

It may not be the Comparator.  Right now it appears to be pointing to the pin and header connections I was using.

I have a modified version of your project.  I added two WaveDAC8s with sinewaves.  The second one is started with a 1ms delay providing about 19 deg phase delay.

I make connection between the WaveDACs and the Vadc and Iadc inputs.

With this project I have yet to get glitches in the comparator outputs and as a result the TFF output.   It has running for a few hours and no glitch triggers.

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes

len,

Have you ever try external wave generator?I appplied your project directly but using external  wave generator to generate two sine wave with 60Hz,2Vpp,2.5DC offset and there is 30 degree phase difference between them.I think 0.6hz error for frequecy is too much for me and there is actually about 2 degree error for phase difference.That's why I chose bypass mode for comparator as Evgeniy Pavlov saidRe: bounces happened at comparator output to measure phase difference correctly.But when I applied what he said approach,the wrong value and strange behaior occured in my previous post.

擷取.PNG

0 Likes

user_3716231

I no longer have access to a signal generator.  My signal generator in the file I attached earlier was using two WaveDACs on the PSoC5.  It was programmed for 4Vpp with a DC offset of 2.0V.

Hopefully my project was able to improve your results.

I've been participating on another discussion about multi-input frequency measurement.   Take a look at the link.   Maybe there is some value to be found.

Re: PSoC5LP Digital Multiplexer control without using digital pins?

Len

Len
"Engineering is an Art. The Art of Compromise."
0 Likes