|
I have several pieces of test equipment that have an input for a precision 10 MHz reference. Specifically an HP freq counter.
Would a Progrock and QLG1 GPS set 10 MHz provide a lab grade reference for a counter?
A poor mans version of quality reference?
Bob? WA1EDJ Monroe, GA
|
I just read through the ProgRock manual and it looks like using the OCXO oscillator with the ProgRock and the GPS will get you close, but it doesn't look like it will be real "Lab Grade" but will be close enough for most ham radio stuff. ?It can be programmed to correct the output every 1 pps from the GPS module, but that could give you a continuously varying (by 1 or more Hz) output frequency from the ProgRock. ?If you can live with that variation, I'd think it would be okay, but if you were doing an ARRL FMT (frequency measuring contest), it probably wouldn't be accurate enough. ?That's just my take on what I skimmed quickly from the ProgRock operation manual. ?Hans might have something different to say about this if/when he gets time from his busy schedule to chime in.
Jim - W0EB
------ Original Message ------
Sent: 11/5/2017 8:18:06 AM
Subject: [QRPLabs] Progrock + QLG1 GPS as 10 MHz Reference?
I have several pieces of test equipment that have an input for a precision 10 MHz reference. Specifically an HP freq counter.
Would a Progrock and QLG1 GPS set 10 MHz provide a lab grade reference for a counter?
A poor mans version of quality reference?
Bob? WA1EDJ Monroe, GA
|
???????? YES Bob - when it is all functional.????
John - VK6JY
????????? I will check it against my two rubidium standards
?? ?????? when all is ready. I'll add some buffering.
On 5/11/2017 10:18 PM, Bob wrote:
I have several pieces of test equipment that have
an input for a precision 10 MHz reference.
Specifically an HP freq counter.
Would a Progrock and QLG1 GPS set 10 MHz provide a lab
grade reference for a counter?
A poor mans version of quality reference?
Bob? WA1EDJ
Monroe, GA
|
Would a Progrock and QLG1 GPS set 10 MHz provide a lab grade reference for a counter? A poor mans version of quality reference?
Bob, I suspect you could do better with a good modern OCXO. (Mine stays within about 0.1Hz at 10MHz over a year checked with GPS.) But I await John's tests with interest. 73 Alan G4ZFQ
|
Ok, Thanks guys.? So if I understand it, using the OCXO/Si5351A with the Progrock instead of the standard non-OCXO 5351A version would be an improvement?? That makes sense.? All of the GPS discipline comes from the Atmel controller bottom board on the Progrock anyway I guess, and the OCXO board plugs into the Progrock normally.
Just trying to understand this clearly before I spend $$!
I'm not expecting Rubidium quality for <$100 but this approach sounds very good.? I would like to hear John, VK6JY's results.
TNX....Bob, WA1EDJ
On Sun, Nov 5, 2017 at 9:56 AM, Alan G4ZFQ <alan4alan@...> wrote:
Would a Progrock and QLG1 GPS set 10 MHz provide a lab grade reference for a counter?? A poor mans version of quality reference?
Bob,
I suspect you could do better with a good modern OCXO. (Mine stays within about 0.1Hz at 10MHz over a year checked with GPS.)
But I await John's tests with interest.
73 Alan G4ZFQ
|
Hi Bob and all,
I’m following this thread closely because I have built a 10Mhz Reference from a Progrock and a UBox M-6 module. It works, but is not as accurate as thought it would be. I think it could be better if the firmware was changed to utilize the GPS calibration better, but I don’t know for sure. It looks cool though.
toggle quoted message
Show quoted text
On Nov 5, 2017, at 8:10 AM, Bob < bobc784@...> wrote: Ok, Thanks guys.? So if I understand it, using the OCXO/Si5351A with the Progrock instead of the standard non-OCXO 5351A version would be an improvement?? That makes sense.? All of the GPS discipline comes from the Atmel controller bottom board on the Progrock anyway I guess, and the OCXO board plugs into the Progrock normally.
Just trying to understand this clearly before I spend $$!
I'm not expecting Rubidium quality for <$100 but this approach sounds very good.? I would like to hear John, VK6JY's results.
TNX....Bob, WA1EDJ
|
Hi Bob,
Like some of the others posting here, I'm disappointed in the ProgRock and QLG1 GPS as a 10 MHz reference. I'm using the recommended Fox TCXO which is just excellent. And for cheap. But the GPS discipline was problematic.
With GPS discipline the frequency accuracy was poor but without it I was able to "dial it in" against a rubidium standard and it "stays put" pretty well over a range of room temperatures. I'll be using the GPS for something else.
73,
Joe, K7JD
|
So if I understand it, using the OCXO/Si5351A with the Progrock instead of the standard non-OCXO 5351A version would be an improvement? Bob, I meant a commercial 10MHz OCXO, the QRP Labs one might be able to be setup as good but I am thinking of probably a around $30 with the advantage of no GPS. (If you think that is an advantage.) And, as Joe says a modern TCXO like the FOX can be fairly close to that. (Although I've never liked the idea of the frequency jumps of the FOX even if it's never noticed.) 73 Alan G4ZFQ I suspect you could do better with a good modern OCXO. (Mine stays
within about 0.1Hz at 10MHz over a year checked with GPS.)
But I await John's tests with interest.
73 Alan G4ZFQ
|
Ahh, Ok, Thanks!? ?Can you guys recommend some commercial ones?? ?PN's, source.? I'm just learning the details of this high accuracy freq control.
Just had a good weekend at the hamfest so have some $$ available.?
TNX....Bob WA1EDJ
On Sun, Nov 5, 2017 at 2:30 PM, Alan G4ZFQ <alan4alan@...> wrote:
So if I understand it, using the OCXO/Si5351A with the Progrock instead of the standard non-OCXO 5351A version would be an improvement?
Bob,
I meant a commercial 10MHz OCXO, the QRP Labs one might be able to be setup as good? but I am thinking of probably a around $30 with the advantage of no GPS. (If you think that is an advantage.)
And, as Joe says a modern TCXO like the FOX can be fairly close to that. (Although I've never liked the idea of the frequency jumps of the FOX even if it's never noticed.)
73 Alan G4ZFQ
? ? I suspect you could do better with a good modern OCXO. (Mine stays
? ? within about 0.1Hz at 10MHz over a year checked with GPS.)
? ? But I await John's tests with interest.
? ? 73 Alan G4ZFQ
|
Hi Bob, all
I see several people have answered this already. This is my take on it all...
What type of precision are you looking for when you say "lab grade"?
The ProgRock when it is properly set up and working properly, should be within +/- 0.5Hz of the desired 10,000,000 Hz. That is, 0.05ppm (50 parts per billion). The frequency "steps" as the algorithm hunts around making corrections, are NOT several Hz, they will be a fraction of a Hz. Theoretically the steps will be sized 0 to 0.3Hz at this output frequency. They depend on the arithmetic with the particular reference clock measurement that is made by the 1pps.?
In order to get this precision, the "GPS correction threshold" register 03 must be set to value 0. This is described in section 6.7 of the manual.?
Joe, Joel, I would be very interested in some numeric data on your units - what kind of accuracy you see and why. I am very interested to know also whether you set the "GPS correction threshold" register to zero. The default value is 5Hz which means that the measured reference frequency must change from the last time by at least 5Hz before a correction is applied.?
There ARE firmware improvements which are possible, that can improve the accuracy. I can work on those if it is necessary. However I still wonder, of the people who find the accuracy is not as good as they expect - have they set "GPS correction threshold"?to zero??
73 Hans G0UPL
On Sun, Nov 5, 2017 at 5:18 PM, Bob <bobc784@...> wrote: I have several pieces of test equipment that have an input for a precision 10 MHz reference. Specifically an HP freq counter.
Would a Progrock and QLG1 GPS set 10 MHz provide a lab grade reference for a counter?
A poor mans version of quality reference?
Bob? WA1EDJ Monroe, GA
|
Bob, Hans & All .............. de John?
VK6JY
Back to your design basics Hans, would the ProgRock be
inherently more accurate in parts per million if operating at
say 50MHz or 100 MHz and subsequently? divided down to 10 MHz???
(ie Less ripple?)? GPS and control optimum.
On 6/11/2017 3:32 PM, Hans Summers
wrote:
Hi Bob, all
I see several people have answered this already. This is my
take on it all...
What type of precision are you looking for when you say
"lab grade"?
The ProgRock when it is properly set up and working
properly, should be within +/- 0.5Hz of the desired 10,000,000
Hz. That is, 0.05ppm (50 parts per billion). The frequency
"steps" as the algorithm hunts around making corrections, are
NOT several Hz, they will be a fraction of a Hz. Theoretically
the steps will be sized 0 to 0.3Hz at this output frequency.
They depend on the arithmetic with the particular reference
clock measurement that is made by the 1pps.?
In order to get this precision, the "GPS correction
threshold" register 03 must be set to value 0. This is
described in section 6.7 of the manual.?
Joe, Joel, I would be very interested in some numeric data
on your units - what kind of accuracy you see and why. I am
very interested to know also whether you set the "GPS
correction threshold" register to zero. The default value is
5Hz which means that the measured reference frequency must
change from the last time by at least 5Hz before a correction
is applied.?
There ARE firmware improvements which are possible, that
can improve the accuracy. I can work on those if it is
necessary. However I still wonder, of the people who find the
accuracy is not as good as they expect - have they set "GPS
correction threshold"?to zero??
73 Hans G0UPL
|
Hi John
No. At 50MHz or 100MHz the parts-per-million is still the same, when divided down you end up with the same thing.?
The limitation is the even-integer mode of the Si5351A which is used in the ProgRock (and other QRP Labs kits) as recommended by SiLabs for lowest jitter (best performance). The reference frequency to VCO frequency PLL multiplication ratio consists of an integer part and a fractional part, whose numerator and denominator are 20-bit integers. The 20-bit resolution is the limitation here.?
There ARE things which can be done in the firmware, if I work on it, to improve that! The current choice of numerator and denominator uses by default a denominator of 1048575. Only the numerator is adjusted. There are algorithms which I could write which would optimise both numerator and denominator to reduce the "frequency step" size, which would allow it to get closer to 10,000,000Hz (say - if as in this case, 10.0000000MHz is the target). One could also abandon the even integer division approach and adjust the MultiSynth divisors, this can result in a very fine adjustments. So there are ways to improve it. If necessary...
73 Hans G0UPL
On Mon, Nov 6, 2017 at 11:17 AM, JohnY <VK6JY@...> wrote:
Bob, Hans & All .............. de John?
VK6JY
Back to your design basics Hans, would the ProgRock be
inherently more accurate in parts per million if operating at
say 50MHz or 100 MHz and subsequently? divided down to 10 MHz???
(ie Less ripple?)? GPS and control optimum.
On 6/11/2017 3:32 PM, Hans Summers
wrote:
Hi Bob, all
I see several people have answered this already. This is my
take on it all...
What type of precision are you looking for when you say
"lab grade"?
The ProgRock when it is properly set up and working
properly, should be within +/- 0.5Hz of the desired 10,000,000
Hz. That is, 0.05ppm (50 parts per billion). The frequency
"steps" as the algorithm hunts around making corrections, are
NOT several Hz, they will be a fraction of a Hz. Theoretically
the steps will be sized 0 to 0.3Hz at this output frequency.
They depend on the arithmetic with the particular reference
clock measurement that is made by the 1pps.?
In order to get this precision, the "GPS correction
threshold" register 03 must be set to value 0. This is
described in section 6.7 of the manual.?
Joe, Joel, I would be very interested in some numeric data
on your units - what kind of accuracy you see and why. I am
very interested to know also whether you set the "GPS
correction threshold" register to zero. The default value is
5Hz which means that the measured reference frequency must
change from the last time by at least 5Hz before a correction
is applied.?
There ARE firmware improvements which are possible, that
can improve the accuracy. I can work on those if it is
necessary. However I still wonder, of the people who find the
accuracy is not as good as they expect - have they set "GPS
correction threshold"?to zero??
73 Hans G0UPL
|
Bob, Hans, I just checked my ProgRock with a PPS, it was about 2Hz out at 9.996000MHz (The Russian RWM standard.) BUT that is not with a QLG1. Ah! just as well I waited for Hans, that was with register 03 at 1. In the manual register 03 0 is not recommended, the extra accuracy not mentioned, maybe it needs an edit? Now with register 03 at 0 it is still about 2Hz low, but it's an old Rockwell not a QLG1. The same with a PPS from a uBlox NEO-7. The correction is working, looking with a SDR I see it stepping until it settles. Not propagation, it's the same, 2Hz low with a GPS-derived standard. So is it the PPS? recommend some commercial ones?? ?PN's, > source. The ones I have used were from Ebay suppliers, not new. The 10MHz one on my counter is an Isotemp, after a short warmup it settles no more than 0.1Hz from a GPS derived signal. This has hardly varied over many years. Similar to this Others, on different frequencies, CTS, Vectron have similar performance. These are all trimmed exactly with a multiturn pot to compensate for any ageing. 73 Alan G4ZFQ
|
Hi Hans, a ProgRock version, where the frequency has the highest presision when uset with the Olg1 gps would really make me happy. I have checket the ProgRock and the U3s, and both are not god for using as timebase for a frequency counter. If it just give 1 excakt frequency or perhaps 2 frequency out, that would be nice. I would need 10 MHz and 1 MHz . I hope that would be posible. By the Way, i have 2 U3s running, 1 with a 5 meter vertikal ant. oz8wsp and one with a 10 meter vertikal ant. oz8pz, both with just one bs170 as output transistor . Fantastic to see how fare away they are heard. Very Nice kit. 73 stig Oz8pz - Oz8wsp? Sendt fra min iPad
toggle quoted message
Show quoted text
Hi John
No. At 50MHz or 100MHz the parts-per-million is still the same, when divided down you end up with the same thing.?
The limitation is the even-integer mode of the Si5351A which is used in the ProgRock (and other QRP Labs kits) as recommended by SiLabs for lowest jitter (best performance). The reference frequency to VCO frequency PLL multiplication ratio consists of an integer part and a fractional part, whose numerator and denominator are 20-bit integers. The 20-bit resolution is the limitation here.?
There ARE things which can be done in the firmware, if I work on it, to improve that! The current choice of numerator and denominator uses by default a denominator of 1048575. Only the numerator is adjusted. There are algorithms which I could write which would optimise both numerator and denominator to reduce the "frequency step" size, which would allow it to get closer to 10,000,000Hz (say - if as in this case, 10.0000000MHz is the target). One could also abandon the even integer division approach and adjust the MultiSynth divisors, this can result in a very fine adjustments. So there are ways to improve it. If necessary...
73 Hans G0UPL
|
Hans, I just checked my ProgRock with a PPS, it was about 2Hz out at 9.996000MHz (The Russian RWM standard.) BUT that is not with a QLG1.
The one thing I forgot is that I'm using a non-standard reference crystal. Not 27MHz but a 26MHz TCXO. So that might well invalidate my tests? I'l look for a synt with the standard crystal (it I have one) maybe try again. 73 Alan G4ZFQ
|
Hans,
Yes, sorry.
With a FOX 27MHz TCXO it does indeed settle closer, I'm not looking close enough but better than +/- 0.5Hz at 10MHz.
73 Alan G4ZFQ
toggle quoted message
Show quoted text
The one thing I forgot is that I'm using a non-standard reference crystal. Not 27MHz but a 26MHz TCXO. So that might well invalidate my tests?
I'l look for a synt with the standard crystal (it I have one) maybe try again.
|
On Sun, Nov 5, 2017 at 11:32 pm, Hans Summers wrote:
Joe, Joel, I would be very interested in some numeric data on your units -
what kind of accuracy you see and why. I am very interested to know also
whether you set the "GPS correction threshold" register to zero.
Hans, I no longer have the (borrowed) rubidium standard but my notes indicate a value of about 27000155 in Register 2 with GPS discipline. Removing the GPS allowed me to set this register to 27000021 which was nearly perfect frequency accuracy as well as I could measure.
I did experiment with different values in the GPS correction threshold register (briefly including 0) but the results were always inferior. Even if I started with Register 2 "dead on", it would degrade when GPS discipline was used. It just wouldn't get close enough regardless of my settings and I eventually abandoned GPS discipline. Sorry I didn't take better data while I had the setup.
Please pardon this off-topic question Hans, but would you comment on the current capacity of the rechargeable battery used on the QLG1 GPS please? I'd like to trickle charge it with a resistor during normal operation. Thanks,
73,
Joe, K7JD
|
Hi Hans,
I’m not sure what register 03 is set to, so I’m going to program it for 0. Can I key in 030 enter or do I have to put in all the other 0s like 0300000000 then enter?
toggle quoted message
Show quoted text
On Nov 6, 2017, at 6:57 AM, K7JD, Joe Dubner via Groups.Io < jdubner@...> wrote: On Sun, Nov 5, 2017 at 11:32 pm, Hans Summers wrote:
Joe, Joel, I would be very interested in some numeric data on your units -
what kind of accuracy you see and why. I am very interested to know also
whether you set the "GPS correction threshold" register to zero.
Hans, I no longer have the (borrowed) rubidium standard but my notes indicate a value of about 27000155 in Register 2 with GPS discipline. Removing the GPS allowed me to set this register to 27000021 which was nearly perfect frequency accuracy as well as I could measure.
I did experiment with different values in the GPS correction threshold register (briefly including 0) but the results were always inferior. Even if I started with Register 2 "dead on", it would degrade when GPS discipline was used. It just wouldn't get close enough regardless of my settings and I eventually abandoned GPS discipline. Sorry I didn't take better data while I had the setup.
Please pardon this off-topic question Hans, but would you comment on the current capacity of the rechargeable battery used on the QLG1 GPS please? I'd like to trickle charge it with a resistor during normal operation. Thanks,
73,
Joe, K7JD
|
Hi Joel
Key in 030 <Enter>
That is fine. No need for all the other zeroes (though that will work too).
73 Hans G0UPL? ?
73 Hans G0UPL?
On Nov 6, 2017 6:15 PM, "Joel Caulkins" < caulktel@...> wrote: Hi Hans,
I’m not sure what register 03 is set to, so I’m going to program it for 0. Can I key in 030 enter or do I have to put in all the other 0s like 0300000000 then enter? On Nov 6, 2017, at 6:57 AM, K7JD, Joe Dubner via Groups.Io < jdubner@...> wrote: On Sun, Nov 5, 2017 at 11:32 pm, Hans Summers wrote:
Joe, Joel, I would be very interested in some numeric data on your units -
what kind of accuracy you see and why. I am very interested to know also
whether you set the "GPS correction threshold" register to zero.
Hans, I no longer have the (borrowed) rubidium standard but my notes indicate a value of about 27000155 in Register 2 with GPS discipline. Removing the GPS allowed me to set this register to 27000021 which was nearly perfect frequency accuracy as well as I could measure.
I did experiment with different values in the GPS correction threshold register (briefly including 0) but the results were always inferior. Even if I started with Register 2 "dead on", it would degrade when GPS discipline was used. It just wouldn't get close enough regardless of my settings and I eventually abandoned GPS discipline. Sorry I didn't take better data while I had the setup.
Please pardon this off-topic question Hans, but would you comment on the current capacity of the rechargeable battery used on the QLG1 GPS please? I'd like to trickle charge it with a resistor during normal operation. Thanks,
73,
Joe, K7JD
|
Hans,
Okay, I’m locked on 10 satellites with the UBlox M-6 module feeding the 1PPS output to the ProgRock, register 03 is set to zero and I’m reading 9.999989.3 on my HP 5384A counter. I’m not sure whether this is super accurate or not. The problem is, it’s this counter I want to use the 10Mhz Reference for. Not sure how I can use it and measure it at the same time. I guess I need a standard to check my counter against so I can test the output of the ProgRock accurately.?
toggle quoted message
Show quoted text
On Nov 6, 2017, at 7:21 AM, Hans Summers < hans.summers@...> wrote: Hi Joel
Key in 030 <Enter>
That is fine. No need for all the other zeroes (though that will work too).
73 Hans G0UPL? ?
73 Hans G0UPL?
|
Jim Sheldon
Joel Caulkins/N6ALT