ADAFRUIT 2045- Si5351 Clock generator
Moderators: adafruit_support_bill, adafruit
Please be positive and constructive with your questions and comments.
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
ADAFRUIT 2045- Si5351 Clock generator
I have 2 questionable Adafruit 2045 Clock generators. (3 attached images). I've used a half-dozen of them over the past year- and directed orders for up 100 of them- without incident.
As installed on my PCB. (I removed the Si5351 to inactivate the module so I could piggyback a second one above it.) The second one acted exactly the same. I was able to cut/pry them off the PCB, but only one of them will ever play the violin again.
The firmware is mature. As installed on another PCB, the result is shown in the figure 'good module'. The spectrum analyzer plot is clean and at the expected frequency of 8.444 MHz.
The two questionable modules are shown in the figure 'bad modules'. The frequency was off by about 30 kHz and the spectrum energy is repeated with spurious sidebands with that same 30 kHz spacing. This came to light (it's an HF transceiver application) when I could decode any given signal every 30 kHz.
These two bad modules have something in common: the marking on the Si5351 ICs:
5351
BVRK
805
Action: I'd like to receive two replacement modules. You may also be hearing about this from other customers, and hopefully this information is useful to Adafruit.
Thanks- Dave 'smallwonderlabs'
- adafruit_support_mike
- Posts: 67446
- Joined: Thu Feb 11, 2010 2:51 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
To clarify: are you seeing the frequency shift and unwanted harmonics shown in 'bad modules.jpg' when the boards are installed, but normal operation shown in 'good module.jpg' when the boards are tested on their own?
Alternatively, is 'good module.jpg' a reference from a known-good board for comparison with the bad ones?
Something is definitely wrong in the first image, though I can't translate the image from frequency domain to time domain in my head. Could you post a scope trace of the waveform please?
Alternatively, is 'good module.jpg' a reference from a known-good board for comparison with the bad ones?
Something is definitely wrong in the first image, though I can't translate the image from frequency domain to time domain in my head. Could you post a scope trace of the waveform please?
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Hi Mike-
I have not tested the modules on their own. Given the recent issues, that needs to change. I'll test them before installing them on my PCBs. I do have spare PCBs and will build up circuitry and a 7-pin row of 'pogo-pins' to pre-test the modules.
The 'good module' image is from an older PCB with one of the Adafruit modules installed. It works exactly as expected. Neither my PCB nor the firmware has changed. // I use the Adafruit modules because they're more convenient than hand-soldering those tiny Si5351 ICs.
You're right- making the switch between time- and frequency domains can be vexing. I can get a scope trace if need be, but there's really not much to see. If we could 'look inside' the Si5351, the PLL waveform would have some amplitude modulation on it due to the sideband content. By the time the CLK signals exit the IC, though, they're in digital logic form and a constant amplitude. The only indication is that the scope triggering is somewhat 'jittery'. It's hard to see because the jitter is such a small fraction of the desired HF output. I was puzzled by the modules' behavior, and I had to take a look with the spectrum analyzer to see what was going on.
My mention of the printing on the Si5351 was intended as a caution. You probably have more of that same date code in your system. As such, you can expect frustration from other customers. I'm not really concerned about the cost of the two bad modules- it's more the time I'd invested on them.
Thanks- Dave, 'smallwonderlabs'
I have not tested the modules on their own. Given the recent issues, that needs to change. I'll test them before installing them on my PCBs. I do have spare PCBs and will build up circuitry and a 7-pin row of 'pogo-pins' to pre-test the modules.
The 'good module' image is from an older PCB with one of the Adafruit modules installed. It works exactly as expected. Neither my PCB nor the firmware has changed. // I use the Adafruit modules because they're more convenient than hand-soldering those tiny Si5351 ICs.
You're right- making the switch between time- and frequency domains can be vexing. I can get a scope trace if need be, but there's really not much to see. If we could 'look inside' the Si5351, the PLL waveform would have some amplitude modulation on it due to the sideband content. By the time the CLK signals exit the IC, though, they're in digital logic form and a constant amplitude. The only indication is that the scope triggering is somewhat 'jittery'. It's hard to see because the jitter is such a small fraction of the desired HF output. I was puzzled by the modules' behavior, and I had to take a look with the spectrum analyzer to see what was going on.
My mention of the printing on the Si5351 was intended as a caution. You probably have more of that same date code in your system. As such, you can expect frustration from other customers. I'm not really concerned about the cost of the two bad modules- it's more the time I'd invested on them.
Thanks- Dave, 'smallwonderlabs'
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Hi Mike-
I've attached a scope trace for a bad module. That snapshot itself doesn't show anything interesting....BUT... in watching the trace in real time, the frequency is bobbling back and forth. The scope trace is constantly in motion- expanding and contracting on the 'time' axis.
By contrast, a good module outputs a rock-steady waveform.
This behavior distinguishes good and bad modules without needing a spectrum analyzer.
-Dave 'smallwonderlabs'
I've attached a scope trace for a bad module. That snapshot itself doesn't show anything interesting....BUT... in watching the trace in real time, the frequency is bobbling back and forth. The scope trace is constantly in motion- expanding and contracting on the 'time' axis.
By contrast, a good module outputs a rock-steady waveform.
This behavior distinguishes good and bad modules without needing a spectrum analyzer.
-Dave 'smallwonderlabs'
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Hi Mike-
I *am* being positive and constructive about this. I'm not sure what if anything is happening at your end. I'd still like a couple replacement modules- as I mentioned in my first post of a week ago.
Thanks- Dave- 'smallwonderlabs'
I *am* being positive and constructive about this. I'm not sure what if anything is happening at your end. I'd still like a couple replacement modules- as I mentioned in my first post of a week ago.
Thanks- Dave- 'smallwonderlabs'
- adafruit_support_carter
- Posts: 29150
- Joined: Tue Nov 29, 2016 2:45 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
This should be done. We need to verify the issue is truly specific to the modules and not potentially related to usage on your custom PCB setup.I have not tested the modules on their own.
Can you demonstrate the issue with the modules used stand alone per the guide?
https://learn.adafruit.com/adafruit-si5 ... t/overview
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Carter-
Thanks for the quick reply.
The test that I referred to in my third message was done with the module removed from my PCB- that is,tested standalone.
The same firmware build was running on another board that I got the 'good module' picture from. I also confirmed that I was not commanding any frequency changes from my firmware- the I2C lines were inactive.
It's not my firmware. It's not my PCB. These Adafruit modules have always worked- until these two with identical date codes.
thanks- Dave
Thanks for the quick reply.
The test that I referred to in my third message was done with the module removed from my PCB- that is,tested standalone.
The same firmware build was running on another board that I got the 'good module' picture from. I also confirmed that I was not commanding any frequency changes from my firmware- the I2C lines were inactive.
It's not my firmware. It's not my PCB. These Adafruit modules have always worked- until these two with identical date codes.
thanks- Dave
- adafruit_support_carter
- Posts: 29150
- Joined: Tue Nov 29, 2016 2:45 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Please post a photo of this setup.The test that I referred to in my third message was done with the module removed from my PCB- that is,tested standalone.
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Carter-
I said I had already tested it stand-alone. I've now set it up again and repeated the test for your benefit (attached image). It shows the same result.
Is this simply a matter of trust or are you seriously trying to rule out something I'm doing wrong? I've already spent considerable time explaining this. I've mentioned the suspect date code several times with no response. It's a strong clue. In my time in aerospace, that issue triggered immediate corrective action. I'd be surprised that it's happening in this case.
'Positive and constructive' wearing thin- Dave
I said I had already tested it stand-alone. I've now set it up again and repeated the test for your benefit (attached image). It shows the same result.
Is this simply a matter of trust or are you seriously trying to rule out something I'm doing wrong? I've already spent considerable time explaining this. I've mentioned the suspect date code several times with no response. It's a strong clue. In my time in aerospace, that issue triggered immediate corrective action. I'd be surprised that it's happening in this case.
'Positive and constructive' wearing thin- Dave
- adafruit_support_carter
- Posts: 29150
- Joined: Tue Nov 29, 2016 2:45 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
This is not stand alone. It is still connected to your custom PCB. Do you have a breadboard and any Arduino compatible board that can be used for testing?
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Carter-
Yes- I know that, and I realized you'd point that out.
You have a decision to make. You can either send me two replacements, as I requested two weeks ago...or not. I've spend 2-1/2 to 3 hours on this thread, and the $16 (+S/H ) cost of these modules is immaterial. This discussion has worn me down, and I'm not sure that's the message you want to leave with this forum's readers.
Should you elect to authorize replacements, I can provide the necessary shipping information by PM. Based on our correspondence to date, though, I'm suspect I'm simply the owner of two defective Adafruit modules.
Thank you- Dave 'smallwonderlabs'
Yes- I know that, and I realized you'd point that out.
You have a decision to make. You can either send me two replacements, as I requested two weeks ago...or not. I've spend 2-1/2 to 3 hours on this thread, and the $16 (+S/H ) cost of these modules is immaterial. This discussion has worn me down, and I'm not sure that's the message you want to leave with this forum's readers.
Should you elect to authorize replacements, I can provide the necessary shipping information by PM. Based on our correspondence to date, though, I'm suspect I'm simply the owner of two defective Adafruit modules.
Thank you- Dave 'smallwonderlabs'
- adafruit_support_carter
- Posts: 29150
- Joined: Tue Nov 29, 2016 2:45 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Send an email to [email protected] with a link to this thread and your order number. They can send you two replacement Si5351A Clock Generator Breakouts.
- smallwonderlabs
- Posts: 9
- Joined: Mon Sep 20, 2021 2:20 pm
Re: ADAFRUIT 2045- Si5351 Clock generator
Thanks, Carter.
I was able to remove another Adafruit 2045 from a PCB I'd taken out of service. I connected that salvaged module to the custom PCB in my last photograph. It works fine- it's neither my PCB nor the firmware. Hopefully other customers won't encounter this issue.
...and thanks for the timely assistance from your Support department- replacements are on their way.
Dave- 'smallwonderlabs'
I was able to remove another Adafruit 2045 from a PCB I'd taken out of service. I connected that salvaged module to the custom PCB in my last photograph. It works fine- it's neither my PCB nor the firmware. Hopefully other customers won't encounter this issue.
...and thanks for the timely assistance from your Support department- replacements are on their way.
Dave- 'smallwonderlabs'
- jmfriedt
- Posts: 2
- Joined: Thu Oct 14, 2021 10:01 am
Re: ADAFRUIT 2045- Si5351 Clock generator
We are seeing the exact same spectra, with 30 kHz spurious. Did you end up solving the problem (either replacing the Si5351 or another part of the board)?
- jmfriedt
- Posts: 2
- Joined: Thu Oct 14, 2021 10:01 am
Re: ADAFRUIT 2045- Si5351 Clock generator
it is the spectrum spreading functionality that is enabled => disable and we get the pure sine. Set 0 to register 149 and problem solved.
Please be positive and constructive with your questions and comments.