ProgRock OCXO and GPS locked frequency accuracy.


Mark Horn
 

Hi, 

I've recently purchased a ProgRock which I use with one of the OCXO's and the GPS receiver.   The documentation states the accuracy should be at least 0.1ppm and when I queried this with Hans he started it should be better than 0.05ppm however I'm not getting anything like that the best I can get is 0.3ppm when GPS locked.  Interestingly I lost GPS this evening, the antenna isn't in a very good position, and the accuracy improved to 0.08ppm which is just odd.  

I've attached a PDF with some pictures of the measurement from the VNWA for both another Leo Bodnar GPSDO and the ProgRock as well as pictures of the waveform from the ProgRock and some SA measurements as well. 

So does anyone have any idea why the ProgRock seems to be so far off frequency when GPS locked? 

73 Mark
M0WGF


Hans Summers
 

Hi Mark

What is the value you have in the GPS Threshold correction register, register 03?

If you don't know, that means you haven't changed it, which means it is the default 5Hz. For best accuracy, change it to zero.

73 Hans G0UPL 

73 Hans G0UPL 


On Fri, May 1, 2020, 22:14 Mark Horn <mhorn71@...> wrote:
Hi, 

I've recently purchased a ProgRock which I use with one of the OCXO's and the GPS receiver.   The documentation states the accuracy should be at least 0.1ppm and when I queried this with Hans he started it should be better than 0.05ppm however I'm not getting anything like that the best I can get is 0.3ppm when GPS locked.  Interestingly I lost GPS this evening, the antenna isn't in a very good position, and the accuracy improved to 0.08ppm which is just odd.  

I've attached a PDF with some pictures of the measurement from the VNWA for both another Leo Bodnar GPSDO and the ProgRock as well as pictures of the waveform from the ProgRock and some SA measurements as well. 

So does anyone have any idea why the ProgRock seems to be so far off frequency when GPS locked? 

73 Mark
M0WGF


Mark Horn
 

Hi Hans, 

Thank you for the reply. 

You're correct the I was using the default threshold of 5Hz I'll change it to zero and see how it goes. 

Many thanks and 73's

Mark
M0WGF


Mark Horn
 

Hi Hans,

Sadly changing the GPS threshold to zero didn't make a difference.

Interestingly before I changed the threshold when I had lost GPS lock I measured the frequency at 999999.25Hz and once the unit had locked to GPS again it now reads 10000004Hz so it further off when locked.

I left it for a couple of hours to see if it would correct however it was still off by 4Hz.

So I don't know where to go from there?

73 Mark
M0WGF


Giuseppe Marullo
 

I had the same problem, but didn't want to investigate much (until now!).
Apparently it was consistently below 2.7Hz@10MHz, even with the register to zero.

I only have another GPSDO so it is not easy to jump to any conclusion easily.

I keep the QRPClock indoor, while the Trimble had the proper missile head GPS antenna outside, on a balcony (south) with clear view of the sky.

Hans, how do you test its accuracy? How many watches do you have? LOL

Giuseppe
IW2JWW - JN45RQ

On 5/2/2020 1:27 AM, Mark Horn wrote:

Hi Hans,

Sadly changing the GPS threshold to zero didn't make a difference.

Interestingly before I changed the threshold when I had lost GPS lock I measured the frequency at 999999.25Hz and once the unit had locked to GPS again it now reads 10000004Hz so it further off when locked.

I left it for a couple of hours to see if it would correct however it was still off by 4Hz.

So I don't know where to go from there?

73 Mark
M0WGF


geoff M0ORE
 

Hi Mark, it would appear that the GPS is doing its job of maintaining the frequency, i.e. the stability. Have you changed the value set in register 02 from the default 27.004.000. I think you have to set this first to get your desired output then let the GPS maintain that frequency. The GPS only maintains the value in register 02. A change of 00.000.001 in register 02 can have a large change in the output frequency.

Maybe I have got this wrong but it worked for me.

On 02/05/2020 00:27, Mark Horn wrote:
Hi Hans,

Sadly changing the GPS threshold to zero didn't make a difference.

Interestingly before I changed the threshold when I had lost GPS lock I measured the frequency at 999999.25Hz and once the unit had locked to GPS again it now reads 10000004Hz so it further off when locked.

I left it for a couple of hours to see if it would correct however it was still off by 4Hz.

So I don't know where to go from there?

73 Mark
M0WGF


Giuseppe Marullo
 

Hi Mark, it would appear that the GPS is doing its job of maintaining
the frequency, i.e. the stability. Have you changed the value set in register 02 from the default 27.004.000. I think you have to set this first to get Geoff,

many thanks for the insight...however this is not how I hope it to works.
I mean the whole point of a GPSDO is that, despite its performance may vary(and it will fail at certain stage in its life to compensate), it should know how much it is off the desired frequency and act to reduce the error. If what you say it is true, it is not able to "close" the control loop using the final frequency(let's say 10MHz), just satisfied to keep the same frequency (27MHz something) the local oscillator is at.
Aging the XTAL for example, would make the frequency shift again without a correction.

AFAIK the final frequency should be directly related to the 27MHz, so it should basically work anyway but at this point I am puzzled.

Should not require anything special to self correct, but I hope I am missing something.

Giuseppe

IW2JWW - JN45RQ

PS: this is the "equipment" I was using, I need to pull out everything from the storage boxes I am not using it at the moment. I found the draft of the email I was going to send out at the time:

----------------------------

I have problems to get a properly aligned 10MHz out of a ovenized ProgRock using GPS as well.
Apparently comparing its signal to a Trimble GPSDO it is consistently 2.4Hz below a "perfect" 10MHz.

I have put the ProgRock inside a QRPLabs Clock enclosure with clock already connected to GPS (just PPS signal is used from the ProgRock).


Reference Equipment:
- Outdoor GPSDO timing antenna (symmetricom 26db) on a fiberglass mast some meters above ground, about 1m apart from the building good but not complete sky view
   using COAX CABLE H155 some meters (less than 10)
- Trimble GPSDO using 10MHz reference out
- 2m RG58 cable
- RACAL Dana 1998 Frequency counter using said reference signal
- Linear PSU (DAIWA PS-43) for both


Trimble is checked against itself using LadyHeater V5.0, reported precision usually is within 100ppt:



On 5/2/2020 9:34 AM, geoff M0ORE via groups.io wrote:

Hi Mark, it would appear that the GPS is doing its job of maintaining the frequency, i.e. the stability. Have you changed the value set in register 02 from the default 27.004.000. I think you have to set this first to get your desired output then let the GPS maintain that frequency. The GPS only maintains the value in register 02. A change of 00.000.001 in register 02 can have a large change in the output frequency.

Maybe I have got this wrong but it worked for me.

On 02/05/2020 00:27, Mark Horn wrote:
Hi Hans,

Sadly changing the GPS threshold to zero didn't make a difference.

Interestingly before I changed the threshold when I had lost GPS lock I measured the frequency at 999999.25Hz and once the unit had locked to GPS again it now reads 10000004Hz so it further off when locked.

I left it for a couple of hours to see if it would correct however it was still off by 4Hz.

So I don't know where to go from there?

73 Mark
M0WGF


Mark Horn
 

Hi Giuseppe & Geoff, 

Thank you both for your replies. 

Geoff as Giuseppe has mentioned setting the LO of the Si5351a shouldn't make a difference as the whole point of the GPS 1PPS counter is to calculate the correct offset that needs to be applied to the 5351 LO to give the correct output.    Interestingly when the GPS isn't connected the output is never off more than 2Hz while when the GPS is connected it generally off by 4Hz.   It's almost like the ProgRock is calculating the offset in the wrong direction?

My next step today if I can get the time is to

1. Double check the setting on the ProgRock using an Arduino to query the serial bus.
2. Build the 5351 that came with the kit and see if that makes any difference.
3. Sniff the IC2 bus and try and see what data the ProgRock is sending to the 5351.
4. If possible see if Lady Heather will read the NMEA data from the QLG1 

Bar those four things unless the ProgRock has some kind of debugging that can be set I'm not sure what else I can do. 

73 Mark


Hans Summers
 

Hi Mark

2. Build the 5351 that came with the kit and see if that makes any difference.

This is either new news or news I missed in earlier posts. So what 5351 are you using currently?

73 Hans G0UPL 


Mark Horn
 

Hi Hans, 

I'm currently using one of your OCXO's. 

Mark


geoff M0ORE
 

This is most likely one for Hans to comment on.

How does the GPS 1 sec pulse know what to calibrate to? As the manual says, register 02 is set to 4kHz high as that is where most crystals end up. If the crystal is 4.5kHz high, then what tells the GPS pulse that it is different.

Just to confuse things some more, I am using a 25.0000MHz crystal with my own oven and register 02 is now 25,008,020 which gives me very close to the required 10MHz output.

Looking at the manual page 8, example 1, the register 02 is 27,002,894.

Section 4.3 on pages 12 / 13 describes how to calculate the value to be used in register 02 to get the correct output. This is how I started my calibration sequence.

I do strongly suggest using the serial interface to set the registers and then, when it is correct, take a hard copy and file with the manual.

Geoff

On 02/05/2020 14:36, Mark Horn wrote:
Hi Giuseppe & Geoff, 

Thank you both for your replies. 

Geoff as Giuseppe has mentioned setting the LO of the Si5351a shouldn't make a difference as the whole point of the GPS 1PPS counter is to calculate the correct offset that needs to be applied to the 5351 LO to give the correct output.    Interestingly when the GPS isn't connected the output is never off more than 2Hz while when the GPS is connected it generally off by 4Hz.   It's almost like the ProgRock is calculating the offset in the wrong direction?

My next step today if I can get the time is to

1. Double check the setting on the ProgRock using an Arduino to query the serial bus.
2. Build the 5351 that came with the kit and see if that makes any difference.
3. Sniff the IC2 bus and try and see what data the ProgRock is sending to the 5351.
4. If possible see if Lady Heather will read the NMEA data from the QLG1 

Bar those four things unless the ProgRock has some kind of debugging that can be set I'm not sure what else I can do. 

73 Mark


Mark Horn
 

Hans,

I've replaced the OCXO with the supplied 5351 and it's still off by 4Hz after 30 minutes of running.

Interestingly it'll occasional flip between 10000004Hz and 9999996Hz so it corrects by 8Hz not by four.

Geoff not 100% sure how Hans is doing the calculation but the general gist of it is :

CLK2 is set by default to say 2.5MHz the AT chip monitors that frequency say over a 40 second period which would equal 100MHz
The 1PPS second pulse triggers a counter on an interrupt when the counter reaches 40 e.g 40Hz
Then if the internal frequency counter measures say 99.98MHz you can can divider that by 40 which would be 2.4995MHz and from that you can apply an offset to the 5351 LO to bring it on frequency.

I'm not saying that's how Hans does it but it's a way of doing it.

If you read section 4.3 of the manual it also states with the GPS 1PPS connected the unit will self calibrate in less than a minute and set the 5351 LO if you have to set this by hand then the 1PPS calibration isn't working.

73 Mark


geoff M0ORE
 

Mark,I think my interpenetration of 4.3 is different. As I understand it, the GPS will maintain the calibration that is set in register 02.

Why else does Hans describe the method of calculating of offset?

On 02/05/2020 15:46, Mark Horn wrote:
Hans,

I've replaced the OCXO with the supplied 5351 and it's still off by 4Hz after 30 minutes of running.

Interestingly it'll occasional flip between 10000004Hz and 9999996Hz so it corrects by 8Hz not by four.

Geoff not 100% sure how Hans is doing the calculation but the general gist of it is :

CLK2 is set by default to say 2.5MHz the AT chip monitors that frequency say over a 40 second period which would equal 100MHz
The 1PPS second pulse triggers a counter on an interrupt when the counter reaches 40 e.g 40Hz
Then if the internal frequency counter measures say 99.98MHz you can can divider that by 40 which would be 2.4995MHz and from that you can apply an offset to the 5351 LO to bring it on frequency.

I'm not saying that's how Hans does it but it's a way of doing it.

If you read section 4.3 of the manual it also states with the GPS 1PPS connected the unit will self calibrate in less than a minute and set the 5351 LO if you have to set this by hand then the 1PPS calibration isn't working.

73 Mark


Mark Horn
 

Hi Geoff,

I agree it a little unclear however paragraph 5 in section 4.3 says:

'If you are NOT using a GPS, you can manually calibrate the ProgRock by entering the correct value into register 02'

To me that implies if you are using a GPS you don't need to apply a calibration manually.

Well have to wait and see what Hans says. 

73 Mark


ajparent1/KB1GMX <kb1gmx@...>
 

Geoff,

Unless one uses really expensive crystals their marked frequency and actual
frequency are nominally good to two maybe three decimal places some better.

Most applications use circuits that can be trimmed to frequency, the 5351
doesn't  so we use an offset.  Calibration s obtained when the desired set
frequency occurs during the 1PPS time  provided by the GPS precision
time base.  NOTE the GPS is not a transmitter, nor does it measure the 
users frequency.  It uses the signals from the Sats that are of known very
high accuracy to stabilize its own time base.  The 1PPS is derived from
that and is of very high accuracy for frequency and more importantly time.


Allison
-------------------------------
Please reply on list so we can share.
No direct email, it goes to bit bucket due address harvesting in groups.IO


Hans Summers
 

Hi Mark, Geoff

The manual section describes a way of calibrating without GPS.

If you have a GPS then you should not need to do anything. Just connect the 1pps and everything should take care of itself. The ProgRock will measure and compensate its own error and write the correct value in 02. No need for any manual correction. 

The discipline method is a kind of Huff and Puff see http://hanssummers.com/huffpuff and http://qrp-labs.com/ultimate3/u3info/u3hp.html (old U3 article). The Clk2 output is set to a divided down version of the 27MHz reference (low enough for the AVR to measure, which means less than 40% of the system clock); the 1pps is a frequency count gate. A correction step is applied to the 27MHz reference value, depending on whether the frequency measurement was more, or less, than expected. The correction step size is dynamically adjusted so as to ensure rapid initial calibration in under 1 minute and precise long term discipline. The finest correction step is 0.01Hz on the 27MHz reference.

It is certainly strange that you are seeing the larger than expected difference, I don't have an explanation for it right now and will have to investigate further.

73 Hans G0UPL 


On Sat, May 2, 2020, 18:06 Mark Horn <mhorn71@...> wrote:
Hi Geoff,

I agree it a little unclear however paragraph 5 in section 4.3 says:

'If you are NOT using a GPS, you can manually calibrate the ProgRock by entering the correct value into register 02'

To me that implies if you are using a GPS you don't need to apply a calibration manually.

Well have to wait and see what Hans says. 

73 Mark


geoff M0ORE
 

Allison, I fully concur with your comments about crystals and correction methods. The issue is, does the register 02 need to be changed to give the desired output or does the GPS do it automatically.

My ProgRock is working OK and frequency is stable but with register 02 changed to reflect the crystal error.

On 02/05/2020 16:19, ajparent1/KB1GMX wrote:
Geoff,

Unless one uses really expensive crystals their marked frequency and actual
frequency are nominally good to two maybe three decimal places some better.

Most applications use circuits that can be trimmed to frequency, the 5351
doesn't  so we use an offset.  Calibration s obtained when the desired set
frequency occurs during the 1PPS time  provided by the GPS precision
time base.  NOTE the GPS is not a transmitter, nor does it measure the 
users frequency.  It uses the signals from the Sats that are of known very
high accuracy to stabilize its own time base.  The 1PPS is derived from
that and is of very high accuracy for frequency and more importantly time.


Allison
-------------------------------
Please reply on list so we can share.
No direct email, it goes to bit bucket due address harvesting in groups.IO


Alan G4ZFQ
 

Interestingly it'll occasional flip between 10000004Hz and 9999996Hz so it corrects by 8Hz not by four.
Mark,

I'd say this was typical of a counter that is not triggering precisely.
Therefore the exact frequency will be in question.
Unless you have the U3 configuration all wrong it will not jump like that.

73 Alan G4ZFQ


Mark Horn
 

Hi Hans,

I finally found my Ardunio and interrogated the ProgRock over serial other than the fact reg 3 hadn't got set to zero there's some funny chars at the start.  I don't know if that's relevant or not?

L'j1:123,456,794
2:27,004,333
3:0
4:25,000,000
5:10,000,000
6:0
7:0
8:0
9:0
10:0
11:0
12:0
13:0
14:0
15:0
16:0
17:0
18:0
19:0
20:0
21:0
22:0
23:0
24:0
25:0
26:0
27:0
28:1,678,772,748


geoff M0ORE
 

That looks exactly like mine including the 01 register contents.

On 02/05/2020 17:34, Mark Horn wrote:
Hi Hans,

I finally found my Ardunio and interrogated the ProgRock over serial other than the fact reg 3 hadn't got set to zero there's some funny chars at the start.  I don't know if that's relevant or not?

L'j1:123,456,794
2:27,004,333
3:0
4:25,000,000
5:10,000,000
6:0
7:0
8:0
9:0
10:0
11:0
12:0
13:0
14:0
15:0
16:0
17:0
18:0
19:0
20:0
21:0
22:0
23:0
24:0
25:0
26:0
27:0
28:1,678,772,748