($40 GPSDO working well now, my mistake)

[tkamiya]

This is getting old but I started playing with it again.

On Pakrat documentation, RS232 port has 1 as ground.  This is CD (Carrier Detect) line.  Ground is pin 5.  Oddly enough, it worked either way.  Corrected mine to the right pin.

[tkamiya]

I got it to work stably now.  No random disconnects!

Key was temperature variations.  Before, I was testing it with cover open.  It takes forever to heat up and it never actually reached equilibrium.  As temperature slowly rise, DAC value will jump and wobble and causes unlock.  (or at least status other than zero).  I found this while staring at LH display for hours. 

I closed the case.  It went up to 55 degrees C or so fairly quickly and stabilized.  At this point, lock is solid.

I'm surprised how much OVEN XCO got affected by ambient change.  My house is air-conditioned, so variation is like 2C at maximum. 

[mark03]

I just finished my GPSDO project too:



With the Extron case closed I also get ambient [inside] temperature ~ 50-55 C.  I was worried it could be much higher.  I think this is safe for everything inside.

It would be nice to know the ADEV performance of the finished unit, but this is my only GPSDO and I'm not sufficiently motivated [yet?] to purchase another standard for comparison purposes.  The system integration brings in a lot of noise sources: switching power supplies, Raspberry Pi, etc.  (Interestingly the RPi Zero W connects to my wifi access point just fine with the case sealed  )

For anyone who wants to use the Extron analog distribution amps as a GPSDO enclosure, I strongly recommend going with the taller unit which has an extra row of connectors at the bottom, which live on their own strip of PCB.  On mine there was literally nowhere to put the antenna connector in the back.  I finally gave up and made a big hole in the main PCB where one of the input BNC connectors used to be.  This necessitated some PCB rework to reconnect a couple of traces passing through that demolition zone.

[tkamiya]

That is a very nice integration of all that is necessary!  I like the idea of using Pi and having LH built in.  I have a question for you.  I'd guess you are using Linux for OS.  How do you manage proper shutdown?  Can't just yank the power, correct?

I am also a bit concerned with switching power supplies right next to the GPSDO board.  I have it in my plan to actually do a comparison between linear supply and switcher along with GPSDO. 

[mark03]

Generally it is fine to pull the plug on Linux systems unless they are in the middle of some heavy activity, and even then, proper journaling filesystems have become the norm.  RPi used to be infamous for filesystem corruption on improper shutdown due to some hardware issues, but I think those have been solved.  That being said...  I still log in manually and perform a remote shutdown if I remember; I may as well, since wifi is there.

The Extron distribution amp shipped with two switching supplies, one on the "roof" of the case supplying +15V, and another, inverting buck, on its main PCB to supply the CLC409 op amps' negative rail.  The Trueposition board has a switcher on its PCB as well.  At least I am not introducing anything completely new.  I have two Meanwell switchers, +15 and +5.  The 15 volts replaces the original switcher, and is also RC filtered and post-regulated to 12 volts for the Trueposition board.  The 5 volts is cranked to 5.5 and supplies the Raspberry Pi, plus the Extron positive rail through an RC filter and 5V LDO regulator, bypassing the on-board 7805.

No doubt there is still plenty of room for mischief and EMI issues.  The integration of GPSDO, distribution amp, and RPi sounded straightforward before I got into the details   It was a LOT more work than I imagined.

[tkamiya]

Thanks.

My unit only has the GPSDO board and a small MeanWell power supply.  (Switching type)  My plan is to either replace it with linear supply or work on filtering output.  I haven't gotten that far yet.  My case is 1U rack mount, so what I can do is really limited. 

As far as front panel goes, I intend to have a few LEDs only.  Maybe POWER, LOCK, and PPS.  I should be able to do this with a smallest Aduino.  Anything further, I'll do it remotely via RS232C connection.  Rear panel has Antenna, 10MHz out, 1 pps out, and power.  That's it.  Very simple. 

While your oven is warming up, do you see frequent UNLOCKs?  (shown as red-line on top of LH graph) and status other than zero?  I kept having that problem for long time until I closed the lid and let it warm up quickly.

I bought a bunch of Extron 3 output unit.  I intend to take it out of case and use it as modules.  These are really nice and cheap.  Performance is pretty good, too.  For general distribution in my lab, I already have two 12 channel distribution box.

[texaspyro]

With the small RPI screen, if you turn off the digital clock display (GZ) then you can see the satellite information table...

Also, touching the upper left corner of the screen will bring up the touchscreen keyboard.

[mark03]

While your oven is warming up, do you see frequent UNLOCKs?  (shown as red-line on top of LH graph) and status other than zero?  I kept having that problem for long time until I closed the lid and let it warm up quickly.

Mine was locking OK when I tested the board in the open, although it was a long time before it got happy.  A long self-survey, and then the DAC recalibration routine (or whatever it's called) may, or may not, have helped this along.  On power up with the box closed, mine takes something like 15-30 minutes before it hits "mode 0" which is the first time the red line goes away and the digital clock turns cyan.  I don't think anyone knows what all it is doing.  There are other odd behaviors, like constantly bouncing between 8 and 9 satellites on quite a regular pattern (about once a minute).

[mark03]

With the small RPI screen, if you turn off the digital clock display (GZ) then you can see the satellite information table...

Also, touching the upper left corner of the screen will bring up the touchscreen keyboard.

Thanks, I'll try this.  Actually, I was going to ask if anyone had an optimized set of graphing options for Heather+TruePosition.  What are the three plots people find most useful?  RIght now I have # satellites, temperature, and DAC voltage which is not actually DAC voltage (just a flatline).  With satellites listed elsewhere only the temperature is really compelling.  That leaves one or two curves open for anything else that might be useful to see at a glance.  It would be great to have the DAC on there but AFAIK this info is not available?

(Edit:  Maybe TFOM, and zooming out the time axis to ~ 6 or 12 hours, would be helpful.  I am always curious if and for how long it has hit TFOM=2, my record thus far.)

Unfortunately I do not have a touchscreen!  None of the "normal" LCD options would fit on my Extron front panel, because the HDMI protrusion.  I ended up with a 4.3" diagonal, 800x480 display from China (buydisplay.com) with an Adafruit parallel display adapter.  This uses every pin on the GPIO header so the serial port is over USB.  There is only one USB port on the Pi, so no touchscreen (also, there's no more room in there).  All of this could have been avoided with a larger case, but once I had the little Extron, I was committed

[KE5FX]

Brier, huh?  That's my old stomping ground (I used to live in Lake Forest Park.) 

[texaspyro]

Hmmm... you should be seeing a DAC voltage response.  Maybe you have some weird firmware that does not output the $PPSDBG message?

[Towger]

There is bug in the newer version of Lady Heather. It appears to be in the installation strings. If you start the GPSDO with and older version and switch the latest version the DAC voltage is displayed/not flatlined.

[mark03]

There is bug in the newer version of Lady Heather. It appears to be in the installation strings. If you start the GPSDO with and older version and switch the latest version the DAC voltage is displayed/not flatlined.

Hmm, first I've heard of this.  I'm running 6.14 beta with the following config:

Code: [Select]

# set com port to use (note the 'u' USB option is ignored for Windows since
# Windows treats USB com devices the same as hardware serial ports).  For Linux
# and macOS, use a number 1 greater than the operating system device id (e.g.
# -1u says to to use /dev/ttyUSB0) or use the -id= option to set the linux
# device name.  The -ip= command line option can set the TCP/IP address of
# a remote decice on the network (local or internet)
-1u

# force the baud rate to use here (if not set, a default value depending uoon
# the receiver type is used)
# -br=9600:8:N:1

# set the receiver type to use (-rx says to auto-detect)
# If you did not first set the baud rate, auto-detect tries 9600:8:N:1,
# 115200:8:N:1, 57600:8:N:1 and, 19200:7:E:1 in that order.
-rxo

# set your local time zone to use. You can use Linux standard format: CDT6CST
-tz=-8pst/pdt

# enable the digital clock display
-gz1

# enable the analog watch display
-gw1

# enable the satellite position map display
-gm1

# enable the satellite signal level display in the plot area
-gq1
-gb1

# enable the sun info display, official rise/set horizon, with sound effects
# Note: in order to make room for this display,  this will restrict the
# satellite info display to an abbreviated format.  You can use the /si=
# command line or SI keyboard command to adjust the satellite info display format
-sr=o*

# allow ESC ESC to exit the program
# -ke

# enable the singing clock (-th=4h for cuckoo clock,  -th=1b for ships bells clock)
# -th=4s

# 800x480 screen resolution
-vp

It does seem like I've seen a DAC curve in the past, although after the basic Trueposition checkout I've been so focused on the hardware that I really haven't paid attention again until now.

Edit:  Adding an explicit -gd1 to my cfg file, I do get the DAC curve.  Maybe part of the confusion is a typo in the PDF version of the manual (http://www.ke5fx.com/heather/heather.pdf) where it lists GO twice; I think one of those is supposed to be GD.

[mark03]

Brier, huh?  That's my old stomping ground (I used to live in Lake Forest Park.) 

Ha... small world!  I did wonder if I should blur that out, but I don't have any enemies with cruise missiles [yet] 

[LazyJack]

I'm resurrecting this topic, as this one seems to be with the most details and most knowledgeable people on the TruePosition board.
I have just finished my implementation of a TruePosition based GPSD, detailed in https://www.eevblog.com/forum/projects/yet-another-trueposition-gpsdo/.

However, I have some unanswered questions that came up during the build. Maybe someone can answer them.

1 - the 10MHz and 1PPS good parameters in the $STATUS. All documentation I have seem mentions it as an indicator that the 10MHz and the 1PPS output is valid and every implementation that cares about it, reports them. So does my implementation. However, I have never seen any of these to indicate that these signals are not good. My TP board always reports both as good. Under what circumstances are they not good? Is it worth bothering reporting them? I spent so much time creating the small icons, would hate to throw them away 

2 - It appears to me that the TruePosition actually saves the position after a successful fix, without any additional command. At least right after powering up, it was able to report the correct position before any acquisition. Is this true, or am I being tricked with something? I have implemented the reload of the previous position with the $SETPOS command, but it may be unnecessary.

3 - The board acquires a fix without a $SURVEY command. But I don't know if running the survey will stop further fixes? That is, once the survey has run for whatever hours and got an accurate fix after averaging etc. will it stop acquiring new fixes as those will be most likely less accurate that what was accumulated during the survey.

[cdev]

It needs to successfully complete the survey at least once in order to work properly. This can take hours or even days. I wasn't able to get it to complete with my PCTEL (Lucent) GPS-TMG-26 antenna indoors, only when I moved it to the middle of the back yard was it happy. Then it completed in just a few hours.
Once that had been done then it was able to acquire a new fix fairly quickly. I think it only needs to know the general area its in and the date to figure the solution out.

Until then it isn't happy, basically it doesn't work as it should at all.

I recently aquired another GPSDO and it acts similarly. (But is much more sensitive) I wonder if the TP would get its initial position fix faster with my (passive) log spiral antenna which has a much larger receiving area. Thats definitely something to test out soon. That would give me much more flexibility in where I put it.

[LazyJack]

Yes, that was my experience too. Survey mode is quite sensitive to having a good signal. If it has then progresses quickly, otherwise it can stop for minutes or even hours. Once it does the survey it looks to me that it actually saves it to some non volatile memory, because if I power it down and up again, and use the $GETPOS command, it will report the same position, even before tracking any satellites. So I'm guessing that there isn't really any need for the controlling software to save and load positions. If the antenna is well positioned it will acquire a fix quickly, if it is not then it doesn't really matter if I enter a location. It my speed up the time to fix, but I don't think it matters in the case of a GPSDO anyway, when you'll need some time after poweron anyway to stabilize.

[cdev]

I wonder what hangs

[CJay]

Yes, that was my experience too. Survey mode is quite sensitive to having a good signal. If it has then progresses quickly, otherwise it can stop for minutes or even hours. Once it does the survey it looks to me that it actually saves it to some non volatile memory, because if I power it down and up again, and use the $GETPOS command, it will report the same position, even before tracking any satellites. So I'm guessing that there isn't really any need for the controlling software to save and load positions. If the antenna is well positioned it will acquire a fix quickly, if it is not then it doesn't really matter if I enter a location. It my speed up the time to fix, but I don't think it matters in the case of a GPSDO anyway, when you'll need some time after poweron anyway to stabilize.

The information from survey mode is indeed stored in flash on many commercial GPSDO.

I think because the GPS antenna shouldn't have moved between power down modes and the receiver doesn't need to work out speed, direction of travel or any 'rate of change' it can hit the ground running and maintain an accurate PPS or whatever output it provided for oscillator discipline as soon as it acquires satellites because one of the major variables is already known with a high degree of precision.

The satellites provide the other high precision information required to 'solve' for the last variable, I.E. the oscillator output.

[tabajaralabs]

11 pages, what a LOT of messages! =D

But I still have a question, sorry for being so n00b about the subject:
I have an E8285A spectrum analyser, and a 53131A frequency counter. What is the best way to use them?

• Use the internal High Resolution OCXO from the E8285A to give 10MHz signal to frequency counter?
• Use my home made OCXO time base inside the frequency counter to give 10MHz signal to E8285A?
• Use a GPSDO (just got a trueposition board) to give 10MHz to both

Thanks! =)

[cdev]

Once you know the precise position you can solve the time and frequency with consistancy.

[CJay]

11 pages, what a LOT of messages! =D

But I still have a question, sorry for being so n00b about the subject:
I have an E8285A spectrum analyser, and a 53131A frequency counter. What is the best way to use them?

• Use the internal High Resolution OCXO from the E8285A to give 10MHz signal to frequency counter?
• Use my home made OCXO time base inside the frequency counter to give 10MHz signal to E8285A?
• Use a GPSDO (just got a trueposition board) to give 10MHz to both

Thanks! =)

Which one do you think is most accurate?

The problem is fairly neatly summarised by Segal's law:

"A man with a watch knows what time it is. A man with two watches is never sure."

If you have three watches, choose the two that agree with each other most closely and disregard the third, but in your situaton I'd set up the Trueposition, connect it to Lady Heather and when Heather shows it to be working correctly, use that.

[cdev]

When I connect my trueposition to my huge tindie sourced RHCP log spiral antenna I get six SVs on my TP GPSDO. No LNA just bare antenna. Will have to investigate what quality the signal is (perhaps with rtklib if that is possible) somehow later. It looks to me like the spiral is probably a pretty decent GPS antenna as navigation goes.

It's the bigger of the two that are being sold. Does anybody know why one now cannot buy it in the US?