Author Topic: Project debug help  (Read 824 times)

0 Members and 1 Guest are viewing this topic.

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Project debug help
« on: July 11, 2024, 07:55:11 am »
Hi All

Long time since I've done any projects and feel like a novice as I am clueless how to move forward with this.

I've built a GPS disciplined frequency reference from this design:

https://ve2zaz.net/GPS_Std_Modernized/GPS_Std.htm

The schematics are here:

https://ve2zaz.net/GPS_Std_Modernized/Downloads/VE2ZAZ_10MHz_GPS_Reference_2024_V1_Schematic.pdf

Mine (like the authors) has a ebay sourced second hand Morion MV89 double oven and the software is running on the "Black Pill" STM32F401CC module.

The project is working per-se, however, I'm seeing my frequency counter jump by 30-50Hz occasionaly and I dont like the look of the output. I have decoupling caps on the OCXO pins for +12V and the tuning voltage pin.

There are occasional odd shapes to the output signal:



generally the output signal is noisy:



and an FFT of the spectrum shows loads of harmonic content:



If I probe the output of the OCXO directly its noisy too:



and with 20MHz bandwith limit:



I built an external linear 12V PSU using a trusted LM723 + pass darlington pair design which feeds the OCXO, that also gets regulated down to 9V to feed the board itself.

I dont know:

  • How to get the 'scope to trigger so I can determine the period between my odd output cycles
  • If my OCXO from Ebay is a load of Dingos Kidneys
  • If I have a decoupling cap/PSU issue
  • Somthing else is amiss

I'm really hoping someone will be willing to help me debug this, as I'm stuck as to what my next move needs to be.

Thanks

Mark
G0MGX


« Last Edit: July 11, 2024, 07:57:55 am by g0mgx »
 

Offline MathWizard

  • Super Contributor
  • ***
  • Posts: 1530
  • Country: ca
Re: Project debug help
« Reply #1 on: July 11, 2024, 09:39:09 pm »
So what kind of "receiver"/antenna do they use for these to pick up GPS satellites?? Do they send out a 10MHz signal ?

What's with the 2 Schottky diodes between the GND pin of the 1117 regulator and actual GND ?
 

Offline Kim Christensen

  • Super Contributor
  • ***
  • Posts: 1551
  • Country: ca
Re: Project debug help
« Reply #2 on: July 11, 2024, 10:22:55 pm »
The project is working per-se, however, I'm seeing my frequency counter jump by 30-50Hz occasionaly and I dont like the look of the output.

What does the DC voltage at pin 7 of U5B look like? Does it jump around when your frequency counter jumps?
Put a scope on it, DC coupled with a slow sweep (1-10 second) and wait for the counter to jump. See if there is any correlation.
 

Offline Benta

  • Super Contributor
  • ***
  • Posts: 6063
  • Country: de
Re: Project debug help
« Reply #3 on: July 11, 2024, 10:40:11 pm »
My feeling is that your oscilloscope probe is not very good.
What are you using?
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #4 on: July 12, 2024, 05:11:46 am »
My feeling is that your oscilloscope probe is not very good.
What are you using?

I have Rigol probes.
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #5 on: July 12, 2024, 05:13:50 am »
So what kind of "receiver"/antenna do they use for these to pick up GPS satellites?? Do they send out a 10MHz signal ?

What's with the 2 Schottky diodes between the GND pin of the 1117 regulator and actual GND ?


I have a roof mounted GPS antenna (I think its designed for marine applications). The cap that links the on-board GPS antenna of the module is removed so it only takes signal from the SMA connector. GPS operates around 1200 - 1500 MHz I think - its certainly above 1GHz - there's no 10MHz involved.

The diodes are raising the voltage of the analog 5V rail to 5.3V.

Mark
« Last Edit: July 12, 2024, 05:21:29 am by g0mgx »
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #6 on: July 12, 2024, 05:18:20 am »
The project is working per-se, however, I'm seeing my frequency counter jump by 30-50Hz occasionaly and I dont like the look of the output.

What does the DC voltage at pin 7 of U5B look like? Does it jump around when your frequency counter jumps?
Put a scope on it, DC coupled with a slow sweep (1-10 second) and wait for the counter to jump. See if there is any correlation.

It seems solid and unrelated. I can (from a putty terminal) control the DAC value manually and it swings as expected when I alter the value from top to bottom.

What I found yesterday was that removing the processor board completely removes the noise. I **think** there is noise on the +5V digital power line, so I added a separate 5V regulator to my PSU board and powered only the processor board from that (removing the connections to the on board +5D line) - but it hasnt made any difference. I also tried adding a 1u cap accross C10/C2 thinking it might be noise in the comms to the DAC but that didnt do anything either.

Thanks for the help!

Mark
 

Offline dietert1

  • Super Contributor
  • ***
  • Posts: 2271
  • Country: br
    • CADT Homepage
Re: Project debug help
« Reply #7 on: July 12, 2024, 06:03:40 am »
What is the corresponding DAC voltage change of the observed frequency variations?
There could be a ground loop causing inaccuracies - like with analog audio. One needs to check carefully the ground circuit. Also that 15K resistor in the DAC output filter can "receive" noise, unless the circuit is inside a grounded metal enclosure.
Also your ciruit diagram doesn't show the recommended DAC reference bypass capacitor.
Also the schottky Gnd connection in +5VA supply is fishy, as any air movement will introduce voltage change by temperature of the diodes. And certainly R1 should deliver its load current to the diodes.

Regards, Dieter
« Last Edit: July 12, 2024, 06:26:13 am by dietert1 »
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #8 on: July 12, 2024, 07:33:23 am »
What is the corresponding DAC voltage change of the observed frequency variations?
There could be a ground loop causing inaccuracies - like with analog audio. One needs to check carefully the ground circuit. Also that 15K resistor in the DAC output filter can "receive" noise, unless the circuit is inside a grounded metal enclosure.
Also your ciruit diagram doesn't show the recommended DAC reference bypass capacitor.
Also the schottky Gnd connection in +5VA supply is fishy, as any air movement will introduce voltage change by temperature of the diodes. And certainly R1 should deliver its load current to the diodes.

Regards, Dieter

Hi Dieter

I measure 3.22mV at 0 DAC value and 4.9928V at 65535 - thats at the VTune pin of the OCXO. The VTune pin looks quite clean (I dont see the noise thats in the output) although I now have wires hanging out of the box so not ideal for PSU ripple measurements.

The DAC is 16 bit and produces a reference voltage on pin 10 which is taken to a test point TP6 on the board - there is no bypass capacitor on this pin. The DAC is powered from the +5VA line and has 100n and 10n as bypass devices. The output of the DAC goes through a rail to rail opamp to produce the voltage swing needed for the OXCO device, in my case its 0 to 5V tuning for a 12V device.

The software stabalises at a DAC value of 46424 for the 10MHz output which measures 3.5352V at VTune.

I need to look at ground loops - not sure how to investigate? Should I have a single ground connection to the board or multiple connections - I currently only have 1 from the PSU(s) to the board?

I did wonder about the LCD - could the interface or device itself be causing noise?

I dont know about the other things you mention - I assume the authors build works, so the design is ok. The device is in a metal grounded case, but the 9V PSU is inside as well as the OXCO so its quite warm in there.

Mark

 

Offline dietert1

  • Super Contributor
  • ***
  • Posts: 2271
  • Country: br
    • CADT Homepage
Re: Project debug help
« Reply #9 on: July 12, 2024, 08:33:57 am »
I was asking for the slope of oscillator tuning in Hz/V, not for the DAC voltage range. I mean you want to know what size of deviation you are looking for. The tuning voltage deviation may be 1 mV, 10 mV or more.
The capacitor on the reference voltage is shown in the DAC data sheet. It is missing in your circuit (design error).
Anyway, most engineers use a PWM type DAC for this application, for better differential linearity. Equally small frequency adjustment steps can make a difference. A MCU with 20 MHz can provide a near perfect 24 bit resolution in one second.
Those schottky diodes are another design error. I would replace them by a small resistor, like 39 Ohms. And route the current from R1 through that resistor, too. This way you can tune the output +5VA to a stable 5.3 V.
You possibly have Gnd from antenna, supply (PE), counter and scope/computer. Sometimes one can see ground currents with the scope, measuring some mV between different Gnd points.

Regards, Dieter
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #10 on: July 12, 2024, 03:43:24 pm »
I was asking for the slope of oscillator tuning in Hz/V, not for the DAC voltage range. I mean you want to know what size of deviation you are looking for. The tuning voltage deviation may be 1 mV, 10 mV or more.
The capacitor on the reference voltage is shown in the DAC data sheet. It is missing in your circuit (design error).
Anyway, most engineers use a PWM type DAC for this application, for better differential linearity. Equally small frequency adjustment steps can make a difference. A MCU with 20 MHz can provide a near perfect 24 bit resolution in one second.
Those schottky diodes are another design error. I would replace them by a small resistor, like 39 Ohms. And route the current from R1 through that resistor, too. This way you can tune the output +5VA to a stable 5.3 V.
You possibly have Gnd from antenna, supply (PE), counter and scope/computer. Sometimes one can see ground currents with the scope, measuring some mV between different Gnd points.

Regards, Dieter

Assuming I have my maths correct:

0.00322V = 9999995 Hz
4.9928V = 10000002 Hz

So we have 1.4 ish volts per Hz

Its clearly not a straight line as I am at 10Mhz with 3.5367V

Feasible?
 

Offline dietert1

  • Super Contributor
  • ***
  • Posts: 2271
  • Country: br
    • CADT Homepage
Re: Project debug help
« Reply #11 on: July 12, 2024, 04:55:23 pm »
So, if its tuning range is 7 Hz, how can it be off or jump by 30 to 50 Hz? As the tuning voltage appears more or less clean and stable, GPS sync is probably working.
Then i'd guess the jumps are a measurement problem. Maybe the counter time base is much worse, i mean some vintage solution.

Regards, Dieter
« Last Edit: July 12, 2024, 05:36:59 pm by dietert1 »
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #12 on: July 12, 2024, 05:17:33 pm »
So, if the tuning range is 7 Hz, how can it be off or jump by 30 to 50 Hz? As the tuning voltage appears more or less clean and stable, GPS sync is probably working.
Then i'd guess the jumps are a measurement error. Maybe the counter time base is much worse, i mean some vintage solution.

Regards, Dieter

The frequency counter is a XL Microwave model; it counts my Siglent Sig Gen perfectly at 10MHz, this project output jumps up by 30-50 Hz and then back again almost randomly.

My assumoption was that the "noise" in my signal which doesnt look clean is causing the error in the counting.

Mark

 

Offline Kim Christensen

  • Super Contributor
  • ***
  • Posts: 1551
  • Country: ca
Re: Project debug help
« Reply #13 on: July 12, 2024, 07:20:41 pm »
Quote
If I probe the output of the OCXO directly its noisy too:

I assume the scope is set to 5mV per division here. That means the signal out of the OCXO is only 30mV pk-pk. This seems much too low, since the DS90C402 has a ±100 mV threshold sensitivity.
Do you have R13 installed? Maybe you should try removing it.

Also: Does the  "\$\Omega\$ 5mV" on the scope screen mean the input is set to 50 ohms? It shouldn't be for that in-circuit measurement.
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #14 on: July 13, 2024, 05:21:18 am »
Quote
If I probe the output of the OCXO directly its noisy too:

I assume the scope is set to 5mV per division here. That means the signal out of the OCXO is only 30mV pk-pk. This seems much too low, since the DS90C402 has a ±100 mV threshold sensitivity.
Do you have R13 installed? Maybe you should try removing it.

Also: Does the  "\$\Omega\$ 5mV" on the scope screen mean the input is set to 50 ohms? It shouldn't be for that in-circuit measurement.

Let me try that again, with my new shiny 'scope, this is the OCXO output connection to the board:



This is the occasional noise that I'm concerned about:



Here with the output conencted via a 50R BNC coax cable to ch2:



I've done a short video to show the "noise" better:

https://www.qsl.net/g/g0mgx//BlogFiles/EEVBlogImages/20240713_062941.mp4

Yes, R13 is installed, my OXCO is definately a 50R output device.

Mark
« Last Edit: July 13, 2024, 05:33:36 am by g0mgx »
 

Offline Kim Christensen

  • Super Contributor
  • ***
  • Posts: 1551
  • Country: ca
Re: Project debug help
« Reply #15 on: July 13, 2024, 04:03:49 pm »
The video is quite illuminating.
Are the noise bursts in sync with the one second pulses from the GPS module? The noise bursts seem to peak at one second intervals. Probably from the microprocessor waking up and running code with each pulse from the GPS.
You need to figure out whether it is conducted noise (via pwr or gnd loops) or radiated noise. I'd try having everything connected together and running, but powering the OCXO from a battery supply to see if the noise goes away. If I still saw the noise, then I'd also disconnect and feed the OCXO's Vtune line with a temporary resistive divider and bypass cap. All with short leads and from the battery. If I still saw noise, I'd suspect radiated noise and start looking at shielding/layout issues.
 

Offline g0mgxTopic starter

  • Regular Contributor
  • *
  • Posts: 79
  • Country: gb
Re: Project debug help
« Reply #16 on: July 14, 2024, 03:45:18 am »
The video is quite illuminating.
Are the noise bursts in sync with the one second pulses from the GPS module? The noise bursts seem to peak at one second intervals. Probably from the microprocessor waking up and running code with each pulse from the GPS.

I think that is exactly whats happening.

I will do as you suggest and report back.

Mark
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf