Skip to main content ↑ ↓
Topic: Unreliable Output (Read 3091 times) previous topic - next topic

Unreliable Output

I have my OLS hooked up to my STM32F407 dev board which is outputting three waveforms.

1kHz    +/- 0.0001kHz
5Mhz    +/- 0.03MHz
42Mhz  +/- 0.1MHz

I have put my scope into persistent mode and these signals are VERY clean. The ranges are shown above. I have a problem with how much jitter the OLS client shows on the channels and all of the captures show an erratic wave train.

1) Jitter. This problem is always present on any signal that has <100khz signal present. It shoes the proper overall state but with hundreds of bumps between on and off.

2) Erratic. This is easy to see from the non uniformity when zoomed out. For example at 200MHz capture, 1 channel group, max depth, no RLE if I measure one cycle I can get anything from 40MHz to 50MHz and their are fewer short pulses than longer so I assume it averages out to my 42MHz.

I borrowed my friends saleae logic analyzer and it shows the 5MHz and 1kHz signals beautifully and as clean as the scope. It can not pick up the 42MHz signal reliably as its maximum accuracy is ~5MHz signals. That's the whole reason I bought the OLS...

The device metadata shows:
Device type: Open Logic Sniffer v1.01
Firmware: 3.07
Protocol: 2
Ancillary: -

Re: Unreliable Output

Reply #1 –
42MHz is pretty much on the edge of what you can capture with OBLS. The 200MHz capture I personally almost never use as it really depends on the version of firmware/bitstream you use 'cause the FPGA is running at 100MHz so the 200MHz sample is made by sampling on both rising and falling edge of the clock. You know that you can't count on duty cycle out of the oscillator to be 50% but it actually floats (if you want to have both edges align perfectly you need to double the frequency and then to half it). That's one side of the problem, the other side of the problem is the probes. The probes that come with OBLS are far from being able to properly probe a 42MHz signal. In order to probe signal that fast you need to use proper probe. Also if you look at the schematic you will notice that there's no termination on the OBLS, the signal goes straight from the SIL connector into the buffer and then from buffer into the FPGA. You need to add some termination resistors if you wanna be able to probe signal that fast. Also, I'm not sure if you can probe 50MHz signal trough that buffer, you might want to use the wing connector and probe using those unbuffered pins (note it's 3v max !!). I was recently (about year ago) analyzing some 45MHz singnal and I made a probe out of some old differential scsi cable where I connected differential pairs as probes (one wire from the pair was grounded) and with some 680R resistors at the wing connector between each probe and ground.. It looked super ugly but worked like a charm :)

Re: Unreliable Output

Reply #2 –
Most of my signals I deal with are around or under 10MHz. I was running a 24MHz one as a test as it controls the ethernet MII for that board.

I might look into making a differential probe wing for the board if I can get all the other problems with this board sorted out. It has not been a reliable tool for anything so far and I spend more time getting it to work right than getting the actual project working.

Any idea why the lower speed signals all have jitter and are misaligned?

Re: Unreliable Output

Reply #3 –
Hi uMinded,

I'm sorry about the problems with the board, this is a highly reliable mature project and your problems are very unusual. Probably the easiest thing is to contact Seeed Studio for a replacement, please reference this thread.
Got a question? Please ask in the forum for the fastest answers.

Re: Unreliable Output

Reply #4 –
[quote author="uMinded"]Any idea why the lower speed signals all have jitter and are misaligned?[/quote]

That is a weird problem. Do they have jitter on 100MHz too?

One question - did you upgrade the system or you are using the firmware it came with?
I am not sure but I think 3.0.7 is the latest version so it should be latest but haven't checked for a while...
I assume you are using http://www.lxtreme.nl/ols/ 0.9.6sp1

Can you get some snapshot of the problem up? Maybe it's just how client is aliasing it when you zoom out

Re: Unreliable Output

Reply #5 –
I will put up a picture when I get home tonight.

The noise appears on any signal lower than 1MHz. That seems to be the limit but the jitter is on everything I measure, every pin and capture speed.

I believe I am running the latest FW, I updated in the fall I believe. The OLS client is the latest version from the website.