OK, I have limited TC to max 1024 (also because I couldn't believe that it could be reached, but you reached it anyway : -)
MY QO100 SETUP
-
-
PA3FYM Hi Remco, Lars says that he had problems with OCXOs when using temperature coefficient. I think to modify your routine to set a stable every x sec to check (about 15 to 30). Also to lower more aggressively that increasing. Alternatively idea is to check for bigger dema than 16 and apply greater TC decreases. Any idea for a number to check? Also I think that exist a min,max TC limit that can be used to your routine. TNX again !!!
-
OK, I have limited TC to max 1024 (also because I couldn't believe that it could be reached, but you reached it anyway : -)
It costs me more to find a better OCXO that yours
-
min TC = 16, max TC = 1024
Yeah, that could also be an idea, to let the TC decrease as a function of the DEMA and/or a more 'sadistic' approach, i.e. 'punish' harder than rewarding. They are now equivalent (>>4) , perhaps the punish part must be twice the reward (from >>4 to >>3)DEMA value was chosen arbitrarily , although somewhat based on monitoring its value and my ns-offsets with my contraption.
Room for improvement, so if you want to experiment with it, I would greatly appreciate that. After all, it's your airconditioner ;; -)
-
PA3FYM Hi Remco, it seems that the code is working. It slowes down from 1024 to a value that have TC <16 without loosing lock . I observed the frequency relative to Cesium and no importand jums was seen. I send you at email the modified code. I check every 20 sec (so a change is occured immidiately after the banner you added) and I check for dema >= 64 and then I divide by 2 as when it loose clock. I try without the 64 , with >>3 at >16 but it loosed lock.
-
Hi, updated block diagram.
-
SV1BDS OK, the 'punishment' has to be real hard in conjunction with your airconditioner. When the airconditioner keeps running, or is off, the system behaves nominal? I ask this because now TC is increased very quickly in the beginning, which may result in loosing lock before the OCXO is 'settled' (at least, I experienced that, that's why I checked mod TC.)
Thanks for the addition, I'll try to work that out in some general conditions, perhaps an extra menu-item to deal with discontinuities like your airconditioner ; -)
-
PA3FYM I have seen that with stable temperature the OCXO needs about half an hour to stabilize. When I have the aircondition off the GPSDO can reach high TC. When it is on it can not reach more than 200. My explanation is that it has flaps that are variyng the temperature and the OCXO can not reach high TC values. When aircondition is off it reach easy values as 1024. I believe that you will not have problems with 'faster' TC increase as it monitors the DEMA and if this become >16 it slows down the TC. I have never seen NoLock at startup period (after first lock...) even when I start the OCXO cool. I have initially try to 'panish' with >>3 only but did not work. Then I see that DEMA takes high values and when I see this I slow down TC before it loose lock. Sometimes the DEMA increases to as high as 30 and then the TC is dropped without loosing lock. With this code I have faster converge without loosing lock. At least I have not yet a case.
-
Understood, perfect. Note that perhaps using DEMA (double exponential moving average) may not be 'scientifically correct' (that algorithm has advantages and disadvantages) but it at least gives some additional discipline : -)
-
-
Hi, a 4 way 10 MHz (not only) splitter was finnaly build in a recycled box. The attenuation is 8 dB and isollation is 12,5 & 23 dB between output ports.
-
Hi, a down converter from 739 MHz to 28 MHz in a surplus box. It is a 739 BPF, a SBB5089 MMIC amplifier (worked better that so called SPF5189 from eBay) a SBL1 and a circuit for 50 ohm termination at 28 MHz. With H polarization the CW beacon is clearly heard !!!
-
Hi, up converter from 28 MHz and IC735 to drive the roof upconverter at 430 MHZ with MAR8,SBL1,BPF 430 MHz, MAV11.
-
Observation using the same LNB:
With my normal RX setup (LNB + 25 MHz GPS ref, IF = 739 MHz - 307 MHz = 432 MHz) I determine around 30 dB SNR of the CW-beacon.
Injecting 25.78... MHz, to produce an IF of 432 MHz directly, the SNR of the beacon decreases here with ca. 5 dB (around 25 dB SNR now)
Anyone else did this experiment with his setup (using the same LNB) and what were the results?
Note: I know the 432 MHz IF is way out of spec and different LNB's behave different
-
Hi, with SDR-Console , Amstrad F101 , 60 cm offset disk, I have 34 dB SNR at 739 (as measured by the latest version - not the difference of levels). SV8QG with a larger disk and unmodified Octagon LNB has 40 dB SNR. I will try to inject frequency to archive 430 MHz and also 950 MHz to see the differences. PA3FYM do you have wobling?
-
-
-
Yes, 25.78 is bigger than 25 ;; -) But I think you don't mean that. In other words, what is your question?
-
-
PA3FYM The available frequency generator has relative high phase noise when used for near 25 MHz creation for reference for LNB and the resulted received signal has very low SNR. So when I will have a very low noise genarator for near 25 MHz I will do the experiment. One solution I thought is to divide the signal using a MB506 2.4 GHz divider or a similar solution that I have available.