[time-nuts] 9390 GPS RX
ed_palmer at sasktel.net
Fri May 31 12:02:39 EDT 2013
On 5/31/2013 5:24 AM, Mark C. Stephens wrote:
> Hi Ed,
>> Well, since you've apparently given up on time-wasters like sleep, there are a few things that you can do while you wait.
> My daytime job is level 2 remote support so I seem to get by on about 6 hours sleep every couple of days.
> Dunno how that works, it probably isn't healthy. The good thing is I get to work from home. And work they make me :)
>> 1. Check your power supplies for noise, ripple, and proper value. Not just the 28V, but there has to be 5V, maybe +-12V or +-15V, etc. Don't forget that you've got capacitors in there that are over 20 years old so anything's possible.
>> 2. Check the EFC to the FRK for noise & ripple.
> I'll get back to you on the PSU and EFC measurements later on tonight if I can sneak out.
> It is 9:30pm Friday night here and I have a deal with Mrs Wife, weekdays, I can work late but weekends must be spent as family time. Fair enough too :)
>> 3. Are there any software commands that have to be or should be used to set up the disciplining? I glanced through a manual for a 9390 and it looks like there's a 'z' command to initiate disciplining.
> I have only found 3 commands that work over the serial port "!", "#" and $.
> It spits out a bunch of numbers periodically but they aren't corresponding to anything on the display.
> I assume the Serial commands are pretty primitive on this early model.
> Certainly z does not work.
>> 4. Do the status reports indicate that everything's good? Number of satellites tracked? Signal strength? EFC value? Anything else?
> When doing GPS corrections, the 9390 always sees at least 4 birds which I think is its maximum. Nothing on SS I am afraid.
> You remember I have to use a N to BNC adaptor with a BNC "tee". I have a 75R terminator in one end and a Nokia L1 antenna in the other.
> That's the only way I can lose the <ANTENNA FEED ERROR>
A 75 ohm impedance may be loading down the signal from the antenna and
loading down the DC voltage to the antenna. Can you just ignore the
Antenna Feed error or does the unit refuse to do anything as long as the
error is present? Maybe build your own terminator with a larger
resistor and maybe a small inductor in series. It's hard to do that
properly at 1.5 GHz, but for a test don't worry about 'properly', focus
on 'it works'.
> I have set it to "Auto" mode. So, when the HDOP gets low enough it picks a satellite and switches to timing mode.
> Then I start to see the EFC values. I am upstairs at the moment and unable to see the LCD.
> Perhaps a web cam would be the ideal tool in this situation ;))
> What I have done is run Time-lab remotely and I am getting 2.98E-12 for an hours samples and it was still dropping at the end of acquisition..
> And although the frequency is constantly varying, how can the phase be so good?
> Am I missing something here? Is it jumping 360 degrees or something?
> Wrap around? I am at a loss, I am afraid.
If the frequency is overshooting and undershooting consistently by the
same amount, it will average out over time to give a good Allan
Deviation. But the short term numbers will be poor. Look at Timelab's
Phase Difference (Unwrapped) plot. If you're comparing to another
GPSDO, there shouldn't be a frequency difference so the line should be
horizontal. Any slope means a frequency difference. Next, is the line
straight as a ruler or does it wobble back and forth? I'm expecting a
wobble because that matches the counter readings. A wobble would
confirm that the short term Allan Deviation will be poor. What you're
seeing might be quite normal for the 9390 - particularly one that was
designed over 20 years ago. Just something to keep in mind.
I have a recent-model commercial GPSDO that I won't name. I think it
meets all of it's specs, but from a Time-Nuts point of view the Allan
Deviation is rather poor and the phase difference plot looks absolutely
awful. But good Allan Deviation numbers weren't one of it's design
criteria. It was designed to give reliable time of day information and
it does that quite well. I suspect that we've been spoiled by the HP
and Trimble GPSDOs that have something of a 'no compromise' mentality to
> One thing I noticed is the Lock LED (decimal point) is significantly brightest and clearest LED on the panel.
> Either it has not been on for a long time or the series resistor has gone walkabouts.
> As it is muxed, it must be the former, we can give ourselves a pat on the back for that one, Ed :)
>> I'm looking at the picture of the unit on your website and I'm a bit confused. I see the FRK in the back corner, but I can't see it's cabling. I see what looks like an OCXO on the second board from the front with white cables coming
>> off of it. There's also what looks like the outline of an OCXO on the main board with white cables connected where the output and EFC would be. What's connected to what?
> The FRK cabling comes out on the side facing the main board at the bottom. The EFC and 10Mhz go to the outline on the main board you so observantly noticed.
> The "card" has a 16.618 (I think from memory) OCXO that goes to the main board on the 2 white SMB cables closest to the front.
> The next 2 SMB cables go down to the monster GPS RX mounted upside down under the main board platform.
> This seems to be the reference for the GPS and the logic board is doing something with the clock to steer the GPS in some way.
Quite a mess, isn't it? Thanks for the tour.
> I must put a 4 channel scope on and have a goosy gander. The only 4 Channel scope here is a rather ancient HP 180 series with a monster screen.
> I think I have one of those early horrible 4 channel sampling HP scopes here as well. Anyway, I digress as usual..
I don't think there's any need to look at everything at the same time
but I guess it won't hurt.
>> By the way, in the back right corner of the main board next to the EPROM, I can see a big Dallas Semi chip. Are you aware that there's a lithium battery in there that's probably stone dead? I don't know what the effect of that
>> would be, but it's something to check.
> Upon Power up All settings are lost, error 17 - Battery Failed.
> The battery is a as dead as a dodo I am afraid.
> I have my eye out for the same type, I have found a few but the last 2 Digits of the Dallas part number differ even though they look the same.
> As it has a sealed Lithium Battery (Who's smart idea was that then, I wonder?) I may have to buy a new one, perish the thought ;)
Make sure you don't buy one with an old data code. The battery might be
dead on it too. I think I've seen datasheets that guarantee the battery
for 10 years which is a reasonable life for most equipment.
> I powered the 9390 off in the afternoon for 30 minutes and when powered back on, I left it with default settings.
> The Rubidium locked within 5 minutes this time. Must have been still a bit warm.
Yes, but you might also find that now that the FRK has had some recent
runtime, it will lock in quicker after a cold start.
> After about 20 minutes it had even found its location and was pulling down the satellite data.
> So, Although inconvenient, I can live with that.
> However if it holds self-calibration details that is a different story.
> You have me pondering again, as Usual, Ed :)
I often find that when I'm trying to figure out a piece of equipment
with poor documentation I usually end up going in 12 directions at the
same time until I can answer some questions and decide which directions
are dead ends.
> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On Behalf Of Ed Palmer
> Sent: Friday, 31 May 2013 3:58 PM
> To: time-nuts at febo.com
> Subject: Re: [time-nuts] 9390 GPS RX
> On 5/30/2013 9:44 PM, Mark C. Stephens wrote:
>> Oh No, it stayed locked, I sat there until 4:30am watching it..
>> But it's still drifting +/- 2 Hz 8 hours later which annoys me so much!
>> I will take a page from Ed's book and exercise patience, I guess.
>> Not much else I can do without disturbing its natural course.
> 1. Check your power supplies for noise, ripple, and proper value. Not just the 28V, but there has to be 5V, maybe +-12V or +-15V, etc. Don't forget that you've got capacitors in there that are over 20 years old so anything's possible.
> 2. Check the EFC to the FRK for noise & ripple.
> I'm looking at the picture of the unit on your website and I'm a bit confused. I see the FRK in the back corner, but I can't see it's cabling. I see what looks like an OCXO on the second board from the front with white cables coming off of it. There's also what looks like the outline of an OCXO on the main board with white cables connected where the output and EFC would be. What's connected to what?
> By the way, in the back right corner of the main board next to the EPROM, I can see a big Dallas Semi chip. Are you aware that there's a lithium battery in there that's probably stone dead? I don't know what the effect of that would be, but it's something to check.
>> -----Original Message-----
>> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On Behalf Of J. L. Trantham
>> Sent: Friday, 31 May 2013 12:32 PM
>> To: 'Discussion of precise time and frequency measurement'
>> Subject: Re: [time-nuts] 9390 GPS RX
>> So, probably means that it is 'flashing', on and off. :^).
>> So, it's been sitting there for a couple of hours before the Lock LED came on.
>> Interestingly enough, I woke up (its 3am here) and I thought, Hmm wonder if its locked.
>> I walked down stairs to the workshop and looked at the front panel just as the Lock LED came on!
More information about the time-nuts