Re: ProgRock + TCXO + GPS Stability/Accuracy

Phil Crockford <pcc@...>

I have done investigation on the accuracy of the GPS control using the Synth module in both the PRO-ROCK and the VFO Sig-gen and I also confirm a frequency difference between an off air standard and the GPS locked frequency derived by the two products.

If a synth module is used in the PRO-Rock it produces a frequency which deviates from the programed frequency by around 2 Hz and does not seem to attempt to correct any further in order to align with the off air standard. However if the same synth module is used in the VFO SIG-Gen and corrected by the same GPS used for the PRO-Rock the Sig-Gen produces an accuracy of better than .5Hz from the programmed frequency and continues to correct itself from then on.

Until now I have resolved to apply a correction to the programmed frequency when using in the Pro-Rock and was content not to communicate my findings to the group. I have a number of Synth modules , OCXO and standard and find the same with all modules when compared in this manner.

The initial acquisition process in the Pro-Rock seems to work well but once this is achieved it seems reluctant to control the frequency any further.

This test have been done with the correction set to 001 and also set to 000 but the pro-rock seems to fail to control the frequency to within less than 2Hz.

Both the Sig-Gen and the Pro-Rock have been run at the same time and controlled by the same GPS using two Synth modules known to exhibit the 2Hz error. The Sig-Gen produces a frequency that is correct to within less than .5Hz but he Pro-Rock Controls to within 2Hz.


This email is by way of information and is not intended a criticism of any  product.





From: [] On Behalf Of Hans Summers
Sent: 28 May 2020 11:14
Subject: Re: [QRPLabs] ProgRock + TCXO + GPS Stability/Accuracy


Hello Giuseppe


Setting register 3 to zero affects how often the frequency is updated. It does NOT affect how often the EEPROM is updated, using the Reference Clock value held in RAM. 


The calibrated Reference Clock value held in RAM is only written to EEPROM if the difference between the existing EEPROM value and the RAM value exceeds 10Hz. Therefore in practice the EEPROM is updated rarely. 


73 Hans G0UPL


On Thu, May 28, 2020 at 10:27 AM Giuseppe Marullo <giuseppe@...> wrote:

Hi Hans,

this is what is written in the manual:

"If you set the register to have a value of 0, then the microcontroller
will adjust the Si5351A output frequencies once per second, every
second, depending on whether it thinks the 27MHz frequency has drifted
upwards or downwards. This continuous change in output frequency may not
be desirable. It is more practical to have small jumps in output
frequencies, less often – i.e. only when the reference oscillator has
changed by more than a threshold.
The QRP Labs website has some notes on how to obtain excell"

How often I could expect the eeprom to be written if I set the register
3 to 0?
Could you recommend 24/7 operation?


Giuseppe Marullo

On 5/28/2020 8:45 AM, Hans Summers wrote:
> Hi Giuseppe
> FYI the ProgRock does not write to EEPROM once per second. It only
> writes to EEPROM once in a while when the error (stored ref freq value
> to actual value held in RAM) exceeds a threshold. So writing to EEPROM
> is quite rare and does not wear out the EEPROM (rated 100,000 cycles
> in the datasheet; I saw some practical experiments on an ATmega328 to
> try and destroy the EEPROM and it actually lasted nearer to 1 million
> in practice).
> I have a beta ProgRock version which allows update of the frequency
> via the serial port, without writing the registers to EEPROM - so that
> would be useful in an application where the constructor wishes to do
> things like repeated frequency sweeps, and not wear out the EEPROM.
> 73 Hans G0UPL
> On Thu, May 28, 2020 at 5:39 AM Giuseppe Marullo <giuseppe@...
> <mailto:giuseppe@...>> wrote:
>     Hi Sid,
>     I know that some people got their "quirks" with the accuracy of this,
>     you were not the only one looking back the ML.
>     I have received a new MCU, to rule out eeprom wear, since I kept it
>     powered several weeks and if fhe eeprom is written once a second,
>     well I
>     expect it not to work anymore, theoretically it will die after
>     27hours.
>     I was past 10x this period, easily.
>     My problem is not that it is not spot on on 10MHz. The problem is
>     that
>     it stays about 2.4Hz below 10MHz of my Trimble. It is almost
>     constant,
>     like either my Trimble is not 10MHz and it is stable, let's say it
>     is @
>     10.0000024MHz, or ProgRock decides that 9.9999976MHJz is the right
>     frequency. I am evaluating getting another source for my lab, if I
>     could
>     acquire at a decent price. Unfortunately there is no visible feedback
>     from the ProgRock about what is really doing, if it consider itself
>     within range or what.
>      >The most common "solution" recommended is to replace the xtal
>     with a
>     TCXO which is no solution that simply masks the problem. That's the
>     discipline...use a clock source that does not drift!!
>     Hehe, mind to give me a DigiKey part number with less than 12
>     digits in
>     the price column? Everything drifts within a hobbyist budget, that
>     the
>     point in the disciplined part of their name.
>     I am already using QRPLabs TCXO BTW.
>     Waiting to perform the surgery on the ProgRock, then I will
>     analyze the
>     PSU too(didn't check it yet). Not much left to look at though.
>     I am building some GPIB interfaces to connect the frequency
>     counter and
>     log data programmatically, it will take 2-3weeks tough, waiting for
>     AR488 pcb to arrive.
>     Giuseppe Marullo
>     IW2JWW - JN45RQ

Join { to automatically receive all group messages.