IC9700 : Problème de verrouillage de l'oscillateur externe 10 MHz (ou pas...)

VK7MO - Avril 2019 a écrit:
Yes I do mean a 10 MHz GPS disciplined Oscillator.

Last night I started doing a test to measure the drift on 2 meters.   I input a GPS locked 10 MHz source to the IC9700 SMA reference socket and conducted the calibration. I used  another but separate GPS locked source on a Signal generator.  This is the sort of test I have done frequently with a GPS locked IC910 and it holds well within one Hz.  For this test  I used spectrum lab with a +/-10 Hz scan-width so I could see the fine variations in the drift.  After starting the test I went to bed and when I got up this morning it had drifted off scale within an hour - so more than 10 Hz drift.  This is in an air-conditioned room so very disappointing.  I will start the test again with a wider scale.  But it looks like temperature drift is much more than 10 Hz at 2 meters which would imply even more than 100 Hz at 1296.  I hope there is some simple way around this that I am missing.

Others have commented about the effects on phase noise.   I use a very high quality double oven GPS source which I expect to be much better in phase noise than a TCXO.  In any case on EME phase noise is not really an issue as I never have strong signals on the EME segment - at least not on 10 GHz in Tasmania.

73 Rex

Stephen Tompsett a écrit:Using a GPS referenced frequency reference, the automatic adjustment appears to end up several Hz off frequency on 144MHz, for comparison my IC-7610 and frequency locked transverters appear to be within 0.1Hz...
Manually adjusting the Fine value allows the frequency to be adjusted a bit closer but there did appear to be some drift.
I'm going to let the oscillator age for a while, and then do some power cycle experiments to see what the warm-up behaviour is, unfortunately there doesn't appear to be an option to keep the reference oscillator continuously powered on with the rest of the rig powered off. At least the frequency can be adjusted without dismantling the unit (as was necessary when using the internal reference in an IC-910!).

I'm a little disappointed with this implementation.

Just done some tests on my 9700 as follows.

Last night after the radio had been on rx for several hours I did a ref cal and it settled on a settings of 41/49 (coarse fine),
Checking the accuracy against my thunderbolt I found it was 33Hz high.

This morning from a cold start it measured 62 Hz high then after two hours the temp was 9Hz
but the shack was colder at 16C this morning rather than 21C last night.

Did another cal and it settled on the same 41/49 numbers but was now only 1Hz high,
so manually tweaked to 41/40 and did another cal again it reset to 41/49 but was now 9Hz high.

Clearly there's some hysteresis in the ref cal process.

Did another cal.

Check the freq again and now 2Hz high, so now went to tx on full power 1296.2MHz for two minutes,
at the end of the two minutes the tx was 8Hz low.

After about 10 mjnutes on rx the freq was back to 2Hz high.

Clearly there is some freq drift with temp - about 10Hz in this simple test or
about  0.008ppm short term over two minutes.

73 Tim Forrester

I just tested the drift of my IC-9700 key down for 2 minutes, a WSPR time period.

The rig was in RX mode for 20 minutes, and then auto calibrate run against my GPSDO.
Mode was FM set to 100 watts output fed into a Narda 150-watt 30-dB attenuator,
followed by a Mini-Circuits 20 dB attenuator producing a 0 dBm signal.

Frequency reading via an HP 5345A 11 digit counter with a 1 second gate time with a read out to 1 Hz.
Timebase for the counter is my Rubidium house standard.  The house standard and the GPSDO agree
within 1 part in 10^10th

Initial reading was 144.200000 MHz.
After 1 minute the unit read 10 Hz high.
After 2 minutes the unit read 17 Hz high.

A WSJT JT65 or JT9 50 second transmission would have had a frequency drift of 9 or 10 Hz on 2 meters.
Obviously that will be worse on 70cm or 23cm.

Rob, NC0B

I ran the same drift test at 1 watt compared to 100 watts, and it took twice as long to drift the same amount as at the higher power.  Instead of drifting 10 Hz in one minute, it drifted about 10 Hz in 2 minutes.  Instead of drifting 17 Hz in two minutes, it drifted 16 Hz in four minutes.  Maximum drift over five minutes was the same as four minutes.  The fan appears to start as soon as transmit mode is enabled, measured scientifically by holding my hand by the fan on the rear!

My IFSS data and Adam’s graph show IP+ does basically nothing, unlike with the 7300 and 7610.  It appears that the ADC chip does NOT have a dither ON/OFF function, so what IP+ is actually doing, besides raising the noise floor 1 or 2 dB is unknown at this time.

Rob, NC0B

I use my RIT whenever the station I am receiving is off more than 10 Hz.  Others obviously are not as picky as I am.
The larger issue is if there is 20 Hz drift in 2 minutes on 2 meters, that is 180 Hz drift on 23cm.
Also some digital modes are very much affected by short-term drift of a few Hz.
The main issue is the 10 MHz external reference input does NOT phase lock the internal clock.
Other Icom, Kenwood and Flex rigs allow for a 10 MHz reference (GPSDO, Rubidium or simply a stable 10 MHz oven oscillator)
to lock the internal reference on frequency.  Five Icom HF rigs have this feature, as does Flex and at least two Kenwood rigs.

Rob, NC0B

I did some tests last night with the 9700 on 2m.

Short term stability (key-down 1 minute): it started at 0Hz, deviated 20Hz and turned back to +2Hz
The next test (wait 1 minute, key-down 1 minute): it started at 1Hz, deviated 10Hz and returned to 0Hz
The drift looks like a bow on my scope.
After a few repeats it settles down to +-2Hz.

My conclusion was that the TCXO is fine, but too slow to compensate the airflow (delta in temperature).
So i stopped the airflow by just blocking the output of the blower...i know, not the professional way, but i run the TRX only at 10W.
The blower immediately kicks in after PTT.

What happened?
1 minute key-down, deviation below 1Hz, THAT is fine for me.
Now it looks like an arrow on my scope ;-)

Next step is to modify the blower, i will run it permanently and see what is happening.
Or shield the TCXO against a fresh breeze. Will see….

PS: I do not confirm this spur issue, here everything is fine with strong signal handling.

Just my 6 cents

73
Michael
DL5OCD

I made tests on 23cm with the blower blocked.
TRX warmed up abt 20min.
3 minutes key down at 3W, drift +-1Hz
On 2m and 70cm the drift is below something i can measure.

So the short term stability is better than i expected. For Long term stability i can`t tell atm, time will show.
But for that i have the auto-cal via 10Mhz. So it should be fine.

So for me (in my case) it is clear what the problem is. It is the fan and a slow TCXO, unable to compensate the rapid decrease of the temperature inside the 9700.

So my question is now how to proceed, modify the blower or shield the TCXO against a fresh breeze.
Maybe there is possibility for permanent operation of the fan via service menu.

I wait for the statement of Icom Europe now.

73
Michael DL5OCD

Thanks Rex for all your hard work.  Here are the results from my Monday spent in the Denver lab. 73, Rob, NC0B
I have an Icom R9000 in Denver, and since I brought the 9700 back to Denver, I didn’t have to wait until the weekend to do an SSB drift test with an Icom R8600.
I setup the R9000 to receive my 1296 MHz signal from the 9700, and there was no significant drift with a 5 minute transmission. (The system was warmed up for 1 hour before the test.)
At that point I noticed the fan did NOT come on when transmitting on 1296 MHz on SSB.  
The fan does come on in FM mode.
Next I did the drift test on 433 MHz on SSB.
The drift on 70cm in 5 minutes on SSB was about 70 Hz.
The fan does come on in SSB mode on 70cm.
When listening to myself on SSB, the tuning error at the beginning and end of the run is on the order of a few Hz.
My HP 5335A has the optional C input that can measure frequency to 1.3 GHz, so I repeated the tests using a carrier and the counter on 70cm and 23cm.
Here is the data on 70cm at 75 watts in half minute intervals.  Due to the start frequency error, it looks like I should have let the temperature stabilize a bit longer after the SSB run.
Since I only captured data every 30 seconds, I may have missed the maximum frequency excursion.

The total excursion is 58 Hz, which is similar to the SSB measurement of around 70 Hz.

433 and 1296 MHz data over 5 and 8 minutes.   Column 3 is a second run

Start    433.000 007 MHz       1296.000 002              432.999 997
0.5       433.000 020                1296.000 026              433.000 012
1.0       433.000 035                1296.000 055              433.000 024
1.5       433.000 050                1296.000 075              433.000 032
2.0       433.000 044                1296.000 097              433.000 042
2.5       433.000 019                1296.000 114              433.000 049  
3.0       433.000 003                1296.000 133              433.000 051
3.5       432.999 998                1296.000 141              433.000 051
4.0       432.999 994                1296.000 142              433.000 050
4.5       432.999 993                1296.000 136              433.000 050
5.0       432.999 992                1296.000 127              433.000 050
5.5                                           1296.000 117              433.000 050
6.0                                           1296.000 109              433.000 051
6.5                                           1296.000 103              433.000 051
7.0                                           1296.000 096              433.000 051
7.5                                           1296.000 092              433.000 051
8.0                                           1296.000 089              433.000 051

The fan runs on both 70 and 23cm with a carrier, but I don’t know if it changed speeds.
I did not observe the frequency increasing and then going back the other way on SSB as opposed to transmitting in FM mode.  If the retrace on SSB mode happened on 70cm, I missed it.

The second counter run on 70cm FM at 75 watts did not show a retrace of the frequency.  This difference is not understood at this time.

NC0B

Le problème est maintenant de l'histoire ancienne :
La mise à jour logicielle V1.10 permet maintenant de synchroniser l'oscillateur
interne de l' IC9700 sur un oscillateur externe de précision (10 MHz)

WB8LBZ - 09/06/2019 a écrit:I have one of those eBay GPSDO (BG7TBL type) that I bought in 2017. I have had it hooked to my Anan and just tried it on the 9700. It synced just fine. Nor bad for $148.

73, Larry

G4BBH - 09/06/2019 a écrit:I have two GPSDOs, both based on the G3RUH design using the 10kHz output from Jup‪iter T GPS boards. One locks a high quality double oven 10MHz oscillator and has a very clean output with low phase noise and jitter and is still very clean up to 10GHz harmonics. The other GPSDO uses a cheaper 10MHz oscillator which has much more phase noise and jitter.
I have used each GPSDO as 10MHz reference with my IC-9700 with tests of 5 hours each. I can see no discernible difference. It seems that the Icom reference locking system is very tolerant.
Richard (Dick)

ZL3DW - 10/06/2019 a écrit:I am using the Leo Bodnar GPSDO from Leobodnar.com or SDR kits. The SDR kits version comes in a nice carry box with the GPS antenna and a USB cable. (only needed to program the unit). Cheap unit works very well. Can be set from 450Hz-800 MHZ. +7dB output so best to use a 10dB attenuator.

NC0B - 10/06/2019 a écrit:I did some testing today as to the effect of level into the nominal -10 dBm 10 MHz input.
It will sync at -30 dBm, not that I recommend it.
I fed in 0 dBm for quite a while, and it works fine.
Also feeding in +10 dBm, and the radio didn't complain. Considering the information below from Will, that should be no surprise.
NOTE: When the signal generator switches level in steps that require the mechanical attenuators to switch, the sync faults and asks if you want to retry. Retry always works.
Back to feeding in -10 dBm.
In my case curse has stayed at 42%. "FINE", however, has been continually incrementing from 52% to now 62% over about 30 minutes.
The HP 8642A has been locked to my Rubidium, which was also checked against my Trimble GPSDO during the whole test run.
During dinner I'll run it off the GPSDO with a 20 dB in-line attenuator at -7 dBm.

Rob