GPS rollover date bug found

[noah4x4]

I reported this possibility in an earlier post, but it deserves its own thread to properly alert people.This 'bug' has now been reported to be affecting pre 2012 manufactured Celestron CPC 9.25 models. The rollover date was April 6th 2019.

The problem is that  the internal GPS units have run out of dates and have reverted to 1999. The only solution is turn off internal GPS  and use the hand controller for manual data input.

Maybe others can add to this thread as and when problems with other makes of older mounts are identified. However, Celestron telescopes manufactured after 2014 should be fine.

[Demonperformer]

50 minutes ago, noah4x4 said:

The only solution is turn off internal GPS  and use the hand controller for manual data input.

which defeats the whole object of paying extra for the gps system, surely.

Hopefully (?!) Celestron will be working on a solution ... although, as it was predicted, one might have hoped they would have produced a solution BEFORE it happened [sorry ... stupid suggestion ... where did i put my meds? ...]

[Cosmic Geoff]

If the problem can't be solved by a firmware update, Celestron deserve a big kick up the backside.

I own a CPC800 with GPS...

[noah4x4]

I think we have to keep this issue in perspective. I only posted this information to assist those that have not turned on their GPS since Sunday and today find it reporting "1999". Most scope owners will NOT be affected.

I gather this issue was a bit like the Millenium 'bug' and nobody had much idea of what actual equipment was likely to be affected until the actual 'rollover date' of 6th April. The possibility didn't even come to light within TeamCelestron until very  recently (March 2019) and my understanding is that Celestron has since been frantically trying to get information from its past suppliers of electronic components.

So far, I believe it affects only CPC scopes AT LEAST five years old as the current supplier of GPS electronics to Celestron  had resolved this when it took over supply in 2014. So far, I have only heard of adverse reports from users with (say) 2006 vintage scopes being affected, so it might possibly affect only scopes over ten + years old. This extent of the issue is unclear until more users come forward to report this problem which is unlikely to be unique to Celestron astronomy equipment.

Celestron was also unable to properly test for this until the rollover date because they could no longer get information from some previous component suppliers (presumably no longer manufacturing) and I suspect they don't have many legacy old scopes in their workshop to test with. 

I should have said "...only immediate solution is use the HC...". as I  believe Celestron is seeking a firmware update workaround now that older CPC mounts have been found vulnerable. The 'bug' did only strike two days ago and I guess only a handful of SGL members that visit TeamCelestron had any  prior idea of its existence too. So, it would perhaps be harsh to be too critical of Celestron providing that it is fixed promptly.

Edited April 8, 2019 by noah4x4

[Cosmic Geoff]

I take your points. Let's wait and see.  In the worst case, the offending GPS module can't be updated by a download and would have to be desoldered from the mount and another soldered in its place.   But I have an early model TomTom One car GPS and it still knew where it was when I turned it on just now.

I don't know the age of my CPC800 but it has Nexstar+ and XLT coatings so hopefully is not affected.

My smart TV (in common with many others) is no longer so smart because the software updates have not kept pace with the changes to iplayer and other online services.

[symmetal]

This notice from the US Naval Observatory explains the cause of the problem (bottom of notice) with an extract below. The week counter was only 10 bits so rolls over every 1024 weeks. So it can affect many older GPS receivers and not just Celestron.

Quote

• GPS Time as defined in the legacy GPS navigation message (ICD-200), uses 10 bits to count GPS Week Numbers. This representation can only cover a finite period of 1024 weeks (19.7 year epoch).

• GPS Time started on Jan 6, 1980

• The first GPS Time Epoch ended on Aug 21/22 1999.

• GPS Time is presently in its second Epoch which will end on April 6, 2019

• It’s up to the user receiver to resolve this week number ambiguity

• Newer receivers fully compliant with GPS ICD should handle this event OK

• In the Future the Modernized GPS Navigation (CNAV and MNAV) message has a 13-bit week number, which for all practical purposes solves this ambiguity

Alan

Edited April 8, 2019 by symmetal

[JamesF]

I am slightly bemused that this sort of problem wasn't picked up earlier given all the fuss about the millennium issue.  Not by Celestron, but by the manufacturers of the GPS hardware.  Or perhaps they decided they didn't care because they didn't expect anyone would still be using it at the time the problem would exhibit itself.

James

[Yawning Angel]

The Celestron Firmware guru on the TeamCelestron forum (Celestron beta test) has this to say:

Everything dies eventually. 

The exact date and time is not yet clear for the reasons that Michael explained.

I just heard back from the manufacturer, the modules we sell today (and have sold since at least as far back as 2012) will not fail next month. I am still pressing to find out what their expected fail date is. I believe it will be at least 2023, and probably 2032 before we start to see failures.

Moreover, one source I read said that only the time would be affected. That the module would still be able to calculate a proper position. If that is correct, then Celestron can correct the rollover error on the fly. 

Why doesn't the thing just get the CORRECT data from the Satellites?

Because the GPS constellation satellites don't give you the correct data. Or, rather, they don't give you complete data. Instead they tell you, very very accurately, how much time has elapsed since the dawn of the last GPS epoch. They do not report what epoch it is, just the time since the beginning of that epoch. Imagine that, for whatever reason, the satellite only provided the last digit of the current year. In that case, a satellite message might contain something like this: 16:54:01.037 March 25, 9. The 9 could stand for 2019. Or possibly 2009. Or even 2299 for all your receiver knows. So the maker of the module says something like this: "we will assume times from the satellite are from January 1st, 2010 to December 31st 2019. This creates a 10 year epoch from for that particular module.

But that is not the only 10 year epoch that the manufacture can choose. He can modify his firmware such that the modules he sells in the year 2015 to use an epoch that starts in January of 2015, and extends to December 2024. Again, it is a 10 year epoch but now we are talking about a 10 year epoch that begins shortly before the sale of the device.

The GPS epoch is 19.7 years. Why 19.7 years instead of 197 years or 19000 years? It was probably a limitation of the technology when the products were designed rather than an extreme lack of foresight. Why 19.7 years instead of 19.6 or 19.8? that is a computer numbers thing, I won't bore you with the details.

What epoch does your module use? It depends on the make and model of your module of course. And, if the manufacture changes his firmware frequently, it might very well depend on exactly when your module was manufactured.

I will reiterate that the modules we are currently selling, and have sold since 2012, will not fail in April 2019.

How should Celestron handle cases where a user has a module for 10 years or so, and then it becomes "broken" by the passage of time? Should we point out that your unit is out of warranty? Should we send you a new Accessory? That won't work if your module is installed on a larger circuit board in a mount we manufacture anymore. Fortunately, I don't have to make these decisions. But I will keep you informed as the decisions are brought to my attention.
--
derik
Firmware Engineer

[noah4x4]

Celestron aim to have a solution available in 4 to 6 weeks.

Reassuringly, the 6th April rollover date only affects GPS electronics manufactured before 2012. Later units will have future rollover dates, and will similarly be resolved.

However, it appears nobody was aware of this 'Epoch' limitation until the first rollover date was imminent. I believe it results from each week being stored as 1 byte with a limit of 1 kilobyte (hence a limit of 1024 weeks).  Nobody expected this crude basis to outlive the equipment  it was installed in. Put another way, an assumption was made by the component suppliers that in 19.7 years nobody will still be using this, but then only the astro-community hangs onto RS232/serial/Win7.....so is this any surprise? 

I wonder if Meade and other mounts are affected given they must use similar technologies? 

 

Edited April 9, 2019 by noah4x4

[michael8554]

Yes some Meade mounts too,  they have used a number of GPS suppliers over the years, so not all pre-2012 mounts are affected.

Michael 

[Gabby76]

I have a Celestron CN-16 module that now reverts to 1999.

It still inputs the proper time and location into the mount though, just the date is wrong. 

The other night I powered on the mount and waited for the GPS to link. Once done I went into menu and shut the GPS off and proceeded to alignment.

At that time it asked if the date was correct which I changed and all the other information I left alone.

[nightfisher]

This is also happening in the radio control market, some of us fly GPS equipped rc planes and a few of the early gps units have fallen over, no longer working, but most like the newer M8N units are just fine and our gps chips are stand alone at around £13 to buy so not a deal breaker, just unplug offending unit and plug in new one 

[Anne S]

On 09/04/2019 at 09:03, noah4x4 said:

 

I wonder if Meade and other mounts are affected given they must use similar technologies? 

 

Some of the earlier Ioptron mounts also have the problem. Ioptron are working on a software fix. I found out about the problem because Tomtom said my car gps needed a software update. They seem to have been on the ball!

[noah4x4]

An easy user work around for older Celestron mounts that doesn't require a firmware update has been posted in Cloudy Nights. It involves allowing GPS to insert date (which is incorrect), time, location, then using <undo> after turning off GPS then changing the year/date via the HC Menu. As year is the easiest coordinate to remember and user input accurately, you keep the benefit and convenience  of time location accuracy from GPS and its just a few extra step to update date.  Should work fine until a firmware release is made.

[michael8554]

The LX200 GPS guys are well on the way to releasing an Autostar II patch, then they'll move to the 497, then Audiostars.

All it will do is say

If YY < 19 or YY > 50, add 1024weeks to the YY data

Any working GPS's will return YY = 19 or greater for the next 19 years

so the code fix will get bypassed and the actual returned date will get used.

Michael 

[michael8554]



The patches are avaible, look for April 2019 issue

Gday All

Just a heads up, the GPS Rollover patches for the LX200s and 497 "Autostars"

are now up on the StarPatch ( many site thanks to Chris Carson )

https://stargps.ca/downloads.htm

The patch works in the same way as all others, but the new "Rollover"

option is not selected by default.

If you need that specific fix, just select the default patch options

as per usual, then select the GPS rollover option

and any others you might "like" to have.

I have tested best i can with both with my failing GPS lumps 

but have no ideas if there are any other oddballs out there , 

so all feed back on anything odd is welcome

Have fun

 

Andrew Johansen Melbourne Australia

[michael8554]

You can fix your scope's GPS problems by downloading and installing our free patch kit for the Autostar's firmware.

 

It requires a PC-to-scope connector cable and (depending upon your PC) perhaps a USB-to-serial adapter.

 

Then visit http://www.stargps.ca/starpatch.htm  to pick up the free (for this purpose) StarPatch program.

 

If you can have your PC attached to the scope and, at the same time, connected to the internet, it becomes a case of:

(a) hook it up, turn on the scope.

(b) fire up StarPatch and tell it what kind of Autostar/scope you have

(c) it will offer to fetch the newest patch kit (v21 for v4.2g in the Autostar) and firmware,  let it do so.

(d) you will be presented with a "tick list" of features in the patch ... scroll to the bottom of that and tick the 

[Fix GPS Rollover bug]                  choice.

(e) then tell it to proceed with the process ... it has a progress bar, and takes about 15 minutes.

(half of it goes very quickly, and half at "Meade speed" (paid-for copies of StarPatch do it all quickly).

(f) when it tells you it's done, you may exit StarPatch, turn off the scope and disconnect everything.

 

That's it.   The patch adds 1024 weeks to the (wrong) date it gets from the GPS receiver.

The receiver will properly work out your position.

 

have fun

--dick

p.s. "GPS Problem"?  No ... it was a known design compromise in the old GPS signal.  Some receivers (such as the Sony that Meade used in the early LX200GPS in 2002) handle it properly.   There are many work-arounds, and the Autostar's updateable firmware makes it relatively easy to address.

Our patch will also handle the problem if the Sony should mess up within the next few years.

[Jonathan Sabin]

>>>The Celestron Firmware guru on the TeamCelestron forum (Celestron beta test) has this to say:

Everything dies eventually. <<<

What a flippant response!

First off, a telescope that originally cost several thousand dollars, and has been well-taken care of, is not an appliance. And even if you think of it that way, the oven that we installed when we built our home 20 years ago is still there--and still used daily.  <knocks on wood> Frankly, I'd be ticked off if a $1000 appliance bit the dust after just six years.

I own a CPC1100 that I just acquired a couple of months ago. It was kept in near-perfect condition by the original owner, and I've had the opportunity to use it about four times now. The first three times it tracked beautifully, especially after setting it up with a 3-star alignment. This past Saturday however, the tracking seemed to be lousy...and I initially thought it was due to the fact that I set it up with a "solar system" alignment on the moon, since I aligned it prior to sunset. Because of the bad tracking, I tried after sunset four times to do a proper 3-star alignment. After four failed attempts, that's when I started scrolling through the hand-controller menus and discovered the out-of-whack date.

Later on that evening, a friend who was with me decided to try looking online to see if he could gain any insight into what was going on, and that's when we found out about the GPS bug.

Celestron needs to step up here and be a good "corporate citizen."

A response like the one posted above, even if it's not official, makes them look HORRIBLE. ("Well sure you spent six grand with us, but that was six or seven years ago, so you shouldn't expect anything more than a paperweight at this point")

I'm not even suggesting that they go out-of-pocket and absorb the entire cost of a fix.

I've seen where they've suggested that affected users can buy a new hand-controller, or a new plug-in GPS unit to fix the issue.

That's all well and good, but if they were smart, they would offer all affected parties their choice at, say, "Cost plus 10%."

If they did that, not only would they not be eating *any* costs from this debacle, they would also avoid the inevitable bad PR they'd start to accumulate if Derek the Firmware Engineer sent out any more flippant e-mails.

 

Just an idea.

[Davey-T]

Hi Jonathan and welcome to SGL.

There is a simple software patch available for the Meade with the same problem so presumably Celestron could offer something similar.

Dave

[michael8554]

Let's be clear,  neither Meade nor Celestron have found a fix, it's a dedicated team of  "amateurs " who are offering it for Meade mounts.

Michael 

[Davey-T]

2 minutes ago, michael8554 said:

Let's be clear,  neither Meade nor Celestron have found a fix, it's a dedicated team of  "amateurs " who are offering it for Meade mounts.

Michael 

Yes should have been clearer it's the Star Patch folk done the Meade, still no reason Celestron can't manage similar I'd have thought.

Dave

[Jonathan Sabin]

I had heard there was a patch available for the affected Meade equipment. (I have two 10" LX200s... though I'm somewhat relieved that they're "classics" without GPS!)

I wasn't aware that the "fix" came not from Meade, but rather a 'team of amateurs.'

Hopefully, there's an equally-capable team of Celestron-owning amateurs working on a patch!

?

[Gabby76]

I sent an email to Celestron and the response I received in reply was that I could replace my Nexstar (Version 4) handset and CN-16 module with a new Nexstar+ (Version 5) handset and module....

Using the old one when require still works other than having to change the date when aligning so 10 seconds of my time is not worth spending the extra money. 

[noah4x4]

Celestron is working on a firmware fix and I posted an interim workaround above. I think Derik's  (firmware engineer) comment needs to be put into perspective as it doesn't reflect an uncaring attitude. It was also made in a confidential forum designed for beta-testers and with an entirely different purpose to SGL. Posting that comment in SGL is a breach of TeamCelestron terms & conditions, especially as it is quoted out of context.  I don't know anybody else as passionate about customer problems or works harder to find solutions to often tricky unexpected challenges. 

It seems that NOBODY in the astronomy equipment world anticipated this problem until it had almost dawned on us. Twenty years ago, Telescope manufacturers bought GPS devices from third party suppliers either oblivious of their 1024 week lifetime or with the intent of finding a fix before it expired that then got forgotten. The firmware enginners that originally implemented this are probably long retired. It was cutting edge technology at a time when many of us still had mechanical watches.

It's absolutely true, "everything does die eventually" and merely reflects the status of that component.. Derik is a top guy who is currently managing solution finding for a huge number of challenges. Just count the number of current threads in TeamCelestron. I am confident a firmware release will appear soon. But is it any great hardship to manually input year data for a few weeks while this is fixed? Let's give the man a break, as until late-March he didn't even know there was a problem. A part has 'died' out of warranty after a lifetime of 19.7 years from its inception (albeit that the timescale for some purchasors is less). Cars regularly get recalled for unforeseen component failure. Why should telescopes be different? Thankfully, it sounds  like a no cost solution is feasible soon. 

[John Van Allen]

Hi,  I have the little Star NX01 GPS with the week rollover issue. Chris Carson, in Canada, says on his website, to send it back to Winnipeg with US15 plus US10 shipping for an update. I am in the Dominican Republic so sending things in and out is more complicated than for US folks. My system is a Losmandy G11 base with Gemini 1 Level 4 Version 1.05  It would be nice if Somebody could tell me how to do this without sending the GPS back to the factory. I'm a bit old, 73, but can figure out a fair amount of these modern things so if somebody could guide me it would be appreciated.

Best Regards

John Van Allen

Questar 3.5, Questar 7, Televue 85, Televue NP127 and AT72ED