-
Ok I see that level shift now - my initial thoughts were blinkered to the transient disturbances.
I can see that the control voltage level change could result from a change in the U5 regulated rail, and/or the VDD rail, and/or the gnd points.
The driving change is increased U7 supply current on the VDD supply and its return. That could be due to the VDD trace resistance from U3 to U7 causing a voltage change on U4 (if the PWM value was sensitive to device supply voltage). That influence could be checked by cutting the VDD trace near U7 and rerouting U7 directly to U3, then there is just the load regulation of U3 at play.
It seems difficult to know where to start carving up the gnd plane without some indications of where there may be ground current influences. I'd suggest that static voltage changes in the gnd plane could be assessed by retrofitting direct additional wire links, eg. from U3 to U7, and U3 to J7/4, and U3 to U4/14, such that the gauge of an added link diverts or suppresses gnd related voltage drops that show a net benefit in the frequency ppm deviation with 50 ohm load steps.
If available from the pcb maker, a simple way to suppress static voltage drops in the gnd plane is to increase the copper thickness spec for that plane.
-
Increasing the copper thickness sounds like a pretty drastic method to me. I'd first try to increase decoupling, adding capacitors close to the critical circuits, maybe adding a choke.
-
Further investigation of the reason the errors were halved when the averaging period doubled. A bit of stupidity (typical). The previous algorithm was designed to steer the 1pps so it arrived in the interval 0 to 25ns. As described earlier, this allowed the system to settle anywhere in that 25ns window so tended to wander in that window. My simplistic solution was to add 1 to all +ve deviations so 0-25ns became 1, 25-50ns became 2 etc, with -ve deviations -25to0ns is -1, -50to-25ns is -2 etc. This did force the system to try to settle on the transition from -ve to +ve. What I didn't realise was this gave too much weight to the 0to25ns and the -25to0ns intervals, resulting in corrections which were too large. This can be seen on the graphs, the voltage becomes a sawtooth overshooting every time.
When the measurement period was doubled, the overshoot was the same. But the algorithm now had twice as long to fix it, so the correction was halved.
The perfect solution to this is to measure the arrival time exactly rather than to the nearest 25ns. But this adds more hardware, against the principal of this build. Also it may test the power of an 8 bit processor to deal with the larger numbers involved.
So the algorithm is changed yet again. The solution was to add weight to deviations greater than ±25ns. The actual solution was double the number and add 1. So 0-25ns is still 1, 25-50ns becomes 3, 50-75ns becomes 5 etc, with -ve deviations -25to0ns is still -1, -50to-25ns is -3 etc. Calculations of corrections uses close to 32 bit precision, and shifts (divide by multiples of 2) the final result. It was a simple matter to add 1 to this to remove the effect of doubling the raw deviations.
So does it work. Yes it does. The GPSDO was off for a different reason before changing the software, so it has been running less than a day. Even so, sawtooth is gone and many corrections are less than 0.01 ppb. That's very satisfying. It seems temperature is now the biggest disturbing factor. Always something to think about.
I'll post another graph when I've got more data.
-
Here's the results of further testing with the u-blox LEA-M8T. Each bar on the lower graph is the adjustment after 1024 seconds of averaging so there's about 2 days of data. As can be seen there's a definite up and down wander to this which I originally put down to some error in the disciplining algorithm. Between X=105 to 140 I tried two modifications to the algorithm. Both seemed to make the situation worse, and as I couldn't see a flaw in the algorithm, the original was reinstated around 140.
Then I looked at the pattern displaced by 84 on the X axis, and it seems to repeat. 84 bars is about a day. Lightbulb moment, this is variations in the 1PPS. It makes sense, when the M8T was surveyed in, it came up with a standard deviation of position of 10 metres. The antenna is in a poor location and it is likely sometimes the calculations are done on a reflected signal. ±10 meters equates to ±30ns in arrival time.
Best way to prove it is put the antenna in a better position. Maybe in a few day's time.
Even so, happy with this. Some room for improvement but isn't there always.
-
I quickly started up your GPSDO earlier today to test and adjust the high stability OCXO option 001 on a HP3325B, which was 0.1Hz high for 10MHz. It certainly makes the adjustment process easy, especially as I only needed to reach the 0.01 ppb level on the GPSDO for this application.
My other 'budget' based adjustment on the HP3325B was to null the Analog Phase Interpolation 3kHz spur using a PC soundcard and REW spectrum analyser software, rather than use a full-blown spectrum analyser from that vintage. -
My introduction to GPSDOs was not a good one.
Oct 2021
I had a HP 10811A crystal oscillator that needed a cal check using a higher authority. The DIY nature and low cost of a GPSDO seemed the way to go so I started building the @AndrewBCN project. https://www.eevblog.com/forum/projects/yet-another-diy-gpsdo-yes-another-one/250/
I am not a programmer so after several failed attempts at programming the STM32 chip using the Arduino IDE, I gave up as I had more pressing things to do.
Jan 2024
Recently I revisited @AndrewBCN project and reread all the posts and downloaded the latest IDE, STM and firmware. This time I got lots of IDE compile errors, so gave up again!
In Reply #143 of the @AndrewBCN thread I saw a short description of a GPSDO by @MIS42N. This one used a preprogrammed PIC to do the disciplining. Fantastic! I chased up all the parts and PCB and here it is. My 'Budget GPSDO' is still in the build phase but it does work and it works well.
Initial tests show my FA-2 counter to be low by about 2Hz. Later I will test the HP 10811A crystal oscillator. I also logged the Vco (OCXO control voltage) for a couple of hours and am still trying to understand the control mechanism.
@MIS42N, I made some changes to the build:
1) I used linear voltage regulators throughout (I prefer them).
2) I did not use the Vco buffer FETs and connected the PWM directly. This does mean a longer wait for stability when connecting to the 10MHz 50R output.
3) I used an Isotemp 143-141 5v OCXO. The two used OSC5A2B02 OCXOs I bought on eBay were faulty.
enut11
EDIT: The note on the Vco plot about hot-plugging the GPS module is a leftover from an old test and should be ignored.
-
@MIS42N, I made some changes to the build:
1) I used linear voltage regulators throughout (I prefer them).
2) I did not use the Vco buffer FETs and connected the PWM directly. This does mean a longer wait for stability when connecting to the 10MHz 50R output.
3) I used an Isotemp 143-141 5v OCXO. The two used OSC5A2B02 OCXOs I bought on eBay were faulty.
enut11
1) - no problem. The SMP was there to lower dissipation in an enclosure, and so it could be used by the Amateur Radio people who like a 13.xV supply. Any supply that works is OK.
2) - connecting the PWM directly leaves you at the mercy of supply variations to the uP and the output buffer. Also, if you connect up to the serial output it will report a -ve control voltage slope and an incorrect Vco. That's an artifact of the reporting, the software was written to handle both +ve (normal) or -ve (inverted) slope. It determines the slope during the calibration run. The FETs act as an inverter and the reporting software assumes it is there. Neither of those are show stoppers, just "nice to have".
3) - Also the software is written to handle OCXO of different sensitivities, also determined in the calibration. It's good to hear this works in the real world.
The control mechanism is written up in Software.pdf in the GitHub repository. It's a bit weird as there's no accurate timer of the 1PPS, just to the nearest 25ns. It relies on the sawtooth jitter of low end GPS modules to randomize arrival times which (perhaps counterintuitively) improve accuracy. So collecting lots of data and averaging, unlike a normal PID control.
I still can't understand GitHub and might shift to SourceForge which seems more user friendly.
GitHub repository https://github.com/ajcashin/budget-gpsdo -
Thanks @MIS42N for a fantastic GPSDO project that just works and is great for non-programmers.
I bought some of the hard-to-get parts from Silicon Chip Magazine and the rest from Jaycar Electronics Australia.
https://www.siliconchip.com.au/Issue/2023/May
https://www.jaycar.com.au/
You might have noticed that I mounted the OCXO on stand-offs. After the failure of two used units from eBay, I needed a way to easily change the OCXO.
To finish the project I am building an analog power supply and I also have a nice case to mount it in. You might say it was made by HP for this project
In case you were wondering, the box at present is just an empty shell. I know not what happened to the original internals.
enut11
-
All the bits mounted inside the HP box. I am currently monitoring the OCXO Vco from a cold start to see how it tracks by the time the 1PPS signal stabilises.
The 'GPS Lock' LED appears to be off but is in fact flashing once every second. The designer, @MIS42N, tells us that this is when the best performance is available.
Apart from measuring the Vco, how else can I check the circuit for performance? I am a novice wrt precision frequency/time references and my main interest at this stage is a way to check my frequency counters and 10MHz distribution box.
enut11
-
As I was an early adopter (so to speak), and MIS42N kindly gave me two of his Rev1 pcbs, I built a second of these GPSDO's to 'confirm' performance of the first built unit. To satisfy any further doubt, I luckily had access to a HP5065A Rubidium standard that gave a reference that was independent from GPS. As I recall, I didn't have tools (or incentive) to push the comparisons below 0.01Hz, or do statistics or further design tweaking that MIS42N has since done. Since then I have happily tweaked a HP 10811-60159 10MHz OCXO, as this budget GPSDO was easy and convenient to turn on and quickly stabilise.
-
The 'GPS Lock' LED appears to be off but is in fact flashing once every second. The designer, @MIS42N, tells us that this is when the best performance is available.
The LED has a variety of flash patterns. They are documented in GPSDOnotesV1.pdf in the repository. A design aim was to minimise the hardware to get a functioning system, the LED flashes try to convey useful information at all stages (e.g satellites in view before the GPS module reports a lock).
A single flash (50ms) once a second indicates the software thinks the frequency is within 1ppb (i.e. 10MHz±0.01Hz). This was a design aim and considered 'good enough' for most hobbyist uses as it is better than most standalone frequency references. The GPSDO frequency may be more accurate but that doesn't change the LED. The actual estimate is reported via the 9600 baud user interface. The output frequency is locked long term to the GPS 1pps, guaranteed (10M * period in seconds) cycles ±1.25 cycles if the LED is single flashing. So can be used to compare a rubidium standard (which is very stable but can have a small frequency error) using a phase comparison over a long period (e.g an hour). -
As I was an early adopter (so to speak), and MIS42N kindly gave me two of his Rev1 pcbs, I built a second of these GPSDO's to 'confirm' performance of the first built unit. To satisfy any further doubt, I luckily had access to a HP5065A Rubidium standard that gave a reference that was independent from GPS. As I recall, I didn't have tools (or incentive) to push the comparisons below 0.01Hz, or do statistics or further design tweaking that MIS42N has since done. Since then I have happily tweaked a HP 10811-60159 10MHz OCXO, as this budget GPSDO was easy and convenient to turn on and quickly stabilise.
@trobbins, I too have a HP 10811 reference and, compared to this GPSDO, it was running approx 2Hz low. This may have been just an ageing phenomenon as the HP specs say something like '10 to the -11' which is very good indeed. Anyway, I adjusted my 10811 and it is now better than '0.02Hz' as read on my FA-2 counter. See Reply #83.
enut11 -
This is a plot of the Vco (OCXO trim voltage controlled by the PIC) over a period of 2 hours from a cold start. This was an open box test.
The sudden Vco variations near the start are part of the software OCXO voltage/frequency sensitivity calibration.
The 1PPS GPS Lock signal happened around 30 min. From then on the plot is very flat suggesting only minor frequency control changes were required.
-
A close up of the last 20 min shows the Vco varies by about 300uV PP short term. Larger cyclic variations are also visible. Sample rate for this test was 4 sec. Ambient temp was 30C.
-
My GPSDO locked up a couple of times and had to be rebooted. Discussions with @MIS42N ultimately showed that I inadvertently created the problem.
The HP box that I used to house the GPSDO had some front panel LEDs that I re-used. These 'last century' LEDs were very inefficient, requiring up to 10mA drive. This current was upsetting the PIC and ultimately the Vco stability. While the exact nature of the problem was not fully understood, it appears to have been solved when I installed a very efficient modern PIC LED.
I also skimped and left off the 2 FETs that buffer the Vco and would recommend that they be used as per the circuit.
In any case, the PIC based GPSDO is working well now and holding 10MHz to within 0.01Hz or better.
A great project. Thanks @MIS42N
enut11 -
enut11, did you use a 10k series resistor for your LED, or lower that value for better visibility ?
I had surreptitiously located various smt caps directly from various device pads to local gnd plane as a pro-active way of hopefully avoiding such disturbances, given the 10MHz operation and the nuanced manner of deriving the OCXO control voltage. -
Hi @trobbins. I had previously lowered the 10K resistor to 270R (ouch) to get sufficient visibility from those very old LEDs. Anyway, I found a green LED that needed only 0.1mA (x100 less current) and it was still bright! Installed an additional series 18K for a total of 28K so minimal loading now on the PIC.
Tell me more about your ground plane and how you use it to improve circuit performance.
enut11 -
'Improving' circuit performance may be a step too far, but the photo shows where I located local bypass caps on the PIC, 740AC04, J1, J2, OCXO, and power input (I used 5V plugpack and an on board LC for buffering). The pcb is V0.0.
Ciao, Tim
-
A close up of the last 20 min shows the Vco varies by about 300uV PP short term. Larger cyclic variations are also visible. Sample rate for this test was 4 sec. Ambient temp was 30C.
I was curious about the short term Vco variations shown in Reply #88. @MIS42N had indicated that they were too large and that something was affecting the PIC supply.
The only other significant load on the PIC supply was the GPS receiver so I powered it with its own 5v regulator directly from the 12v input. This made a noticeable improvement to Vco stability.
The first chart is a 9hr plot of Vco. The large sudden variations are under software control as the program uses the Rx 1PPS to lock in the 10MHz.
The second chart is zoomed in to show the smaller variations. These are now around 10uV. Compare this to the 300uV in Reply #88.
The Budget GPSDO is now boxed and ready to use. Very happy with this DIY 10MHz reference. I have used it to calibrate a Racal Dana frequency meter, a 10MHz Oscilloquartz reference and a HP 10811 OCXO.
Is it finished? Not yet. I want to add a 10MHz output to the front panel and maybe even a buffered 1PPS. Also, I think a 240AC line filter will help clean up the power supplies even more.
enut11
-
With the metal cover on, the Budget GPSDO seems to have stabilised to the third decimal place. It has been running for about 8 hours in a room with an ambient temperature of 26-29C. Counter settings were 1s gate and HP 10811A external reference. Not sure if the GPSDO was checking the 10811 or vice versa
-
Í´d say I´m good.
Half of the sky shaded by building, in non-heated room, no enclosure...
>h Time 204314 UTC. Ctrl 1.760273 -0.049 ppb
Time 204730 UTC. Ctrl 1.760203 -0.070 ppb
Time 205146 UTC. Ctrl 1.760194 -0.008 ppb
Time 205602 UTC. Ctrl 1.760196 0.002 ppb
Time 210018 UTC. Ctrl 1.760378 0.182 ppb
Time 210434 UTC. Ctrl 1.760338 -0.040 ppb
Time 210850 UTC. Ctrl 1.760435 0.097 ppb
Time 211306 UTC. Ctrl 1.760323 -0.112 ppb
Time 211722 UTC. Ctrl 1.760218 -0.105 ppb
Time 212138 UTC. Ctrl 1.760233 0.014 ppb
Time 212554 UTC. Ctrl 1.760255 0.022 ppb
Time 213010 UTC. Ctrl 1.760339 0.085 ppb
Time 213426 UTC. Ctrl 1.760386 0.047 ppb
Time 213842 UTC. Ctrl 1.760370 -0.016 ppb
Time 214258 UTC. Ctrl 1.760319 -0.051 ppb
Time 214714 UTC. Ctrl 1.760295 -0.023 ppb
Time 215130 UTC. Ctrl 1.760288 -0.008 ppb
Time 215546 UTC. Ctrl 1.760330 0.042 ppb
Time 220002 UTC. Ctrl 1.760366 0.036 ppb
Time 220418 UTC. Ctrl 1.760416 0.049 ppb
Time 220834 UTC. Ctrl 1.760378 -0.038 ppb
Time 221250 UTC. Ctrl 1.760322 -0.056 ppb
Time 221706 UTC. Ctrl 1.760216 -0.106 ppb
Time 222122 UTC. Ctrl 1.760214 -0.002 ppb
-
Í´d say I´m good.
Excellent.
I (and maybe others) are interested to know where you got your info and parts from. Version 1 or Version 2? And is it a part or full build. [enut11 did a partial build with his own variations]. -
Excellent.
I (and maybe others) are interested to know where you got your info and parts from. Version 1 or Version 2? And is it a part or full build. [enut11 did a partial build with his own variations].
Huh... Parts source: Aliexpress, PCB from JLPCB, I used design published here, modified power supply part and added LC filter to the output, repoured copper pour...
-
Is it finished? Not yet. I want to add a 10MHz output to the front panel and maybe even a buffered 1PPS. Also, I think a 240AC line filter will help clean up the power supplies even more.
Try this trick i used when I was chasing microvolts of variations on 5V line in another GPSDO - use 12v power supply and add 7808 in front of 5V regulators.
For 1pps use PICDIV from leapsecond.com for generating 1pps from 10MHz, in that way you´ll still have 1pps if gps looses signal, and if something goes wrong with 1pps out, you´ll replace 1€ PIC, not gps. -
Is it finished? Not yet. I want to add a 10MHz output to the front panel and maybe even a buffered 1PPS. Also, I think a 240AC line filter will help clean up the power supplies even more.
Try this trick i used when I was chasing microvolts of variations on 5V line in another GPSDO - use 12v power supply and add 7808 in front of 5V regulators.
For 1pps use PICDIV from leapsecond.com for generating 1pps from 10MHz, in that way you´ll still have 1pps if gps looses signal, and if something goes wrong with 1pps out, you´ll replace 1€ PIC, not gps.
Hi @ddosegov. Yes, I already use a 7808 in front of the 7805 supply to the PIC. That is part of the original design by @MIS42N.
Tell me more about PICDIV.
Thanks
enut11