mikser

There's some analog processing performed on the incoming signal, and the battery voltage could make a difference there. But in the digital part voltage cannot make any difference: it either works or not, there's no middle ground.

I don't just look at the difference between the starting and ending locations, but with a good (basically self made) map I can tell when it's starting to drift during the hike. Others have posted much better screenshots, but here it is. Waypoint 002 was taken when the actual position was at the cross in the center. After marking the waypoint it drifted towards the actual position while I stood still for some minutes. Then I power cycled it and the new track segment (different color) continues from the center cross, and doesn't drift anymore in this part of the route. It is interesting to note, that before the drift there was bad reception, as indicated by the jumping around in the tracklog. It has been observed earlier that bad reception triggers the drifting behavior, but drifting persists even after the reception becomes good.

On today's walk I've seen a significant (300m/1000ft) drift, as well as smaller ones with the new unit. As usual power cycling restored the accuracy. So it seems those were right who said that this is purely a software issue. Now back to the good ol' 2.30 firmware...

It may be a combination of hardware/software: my theory is that the chipsets in the bad batch didn't tolerate some of the firmware changes. As Garmin tested (uhm, maybe tried out) the 2.60 firmware only with the newer units which weren't effected, they didn't notice the problem. And as you said, their bug tracking system is so incredibly broken, that they simly cannot identify/fix this issue based on customer feedback. Etrex H and Colorado manufacturing probably overlapped quite a bit, so the bad batch might have made it into early Colorado units as well. In both cases there are some people who are not seeing this strange drifting issue, even when they are explicity looking for it.

I still haven't had opportunity to verify the behavior of the new unit, but FWIW here are the serial numbers: old: 16D037... new: 16D141... I don't see anything in the test mode about chipset revision (bravo version? doesn't sound like it). Everyone trying test mode (ENTER+PowerOn) should be aware of the following: only do it with no sattelites in view, or some (non-permanent) miscalibration could result: http://gpsinformation.net/main/diagnost.txt

Some places on this route have consistently triggered the drifting behavior on the old unit with the 2.60 fw, but it could have been just a lucky constellation. I'll report when I can confirm this theory with 100% confidence.

Rubber band came off my unit, so I took it back to the retailer, and to my surprise got a brand new one (and a restarted warranty period, which seems to apply in such cases here). Versions are 2.60/2.60 factory installed. So I took it for a walk, and no drifts. So as I think many of us suspected, this looks like a combination of software and hardware problems: i.e. the inaccuracy only happens on certain chipset revisions with certain software versions. For me a happy ending, ...but I still feel Garmin could have handled this a lot better.

Latest response: "I will forward you concern and see what I can do to research this a bit more." (RRLover: thanks. And long live the railways!)

RRLover, your kind words are appreciated when there seems to be no hope. I sent a reply, explaining the downgrade process, and gently asking for the information in the original message to be forwarded to engineering. I'm now waiting for the evil magic to break.

Here's the reply I got from a "Product Support Specialist": Dear X Y, Thank you for contacting Garmin International. This isn't some I have heard of as of yet. From what I read from the forums is that a lot of this may be error of GPS. The unit will have a drift. If you leave a unit alone in the same place 5 minutes it will drop a track log that is jumbled. For old Garmin versions: www.Gpsinformation.net I couldn't find a way to down grade the Gps SW version. With Best Regards, ... One has to wonder what clue level their non-specialist support staff has... Should I give this up as a lost cause, or should I explain kindly to Garmin support staff, how the GPS SW version can be downgraded on their products?

I sent the follwing email to several garmin addresses I could find on their site. I'll post any responses. Dear Garmin, There's a widely reported and well documented problem with accuracy of some units with the high sensitivity MediaTek chipset (Vista HCx, Colorado 400t, etc). These are just some of the forum topics that cover the observed problems: http://forums.Groundspeak.com/GC/index.php...186593&st=0 http://forums.Groundspeak.com/GC/index.php...184158&st=0 Garmin product support has constantly denied the existence of this issue, even though on the forum several people have reported that they have talked to Garmin representatives, sometimes multiple times. The problem has been narrowed down to the 2.60 chipset firmware for these units. On the forum someone posted downgrade instructions that restores the factory default 2.30 chipset firmware, and it has been verified to fix the accuracy problems. Several factors have been eliminated as the possible cause of the problem: - bad satellite reception - custom maps - WAAS being turned on or off - backlight on/off - low battery power People are very unsatisfied with the response from Garmin. We have paid a lot of money for these otherwise excellent units, and may rightly expect them to perform their most basic function of determining position as accurately as possible. Please respond to this email or to the forum regarding the status of your investigation to this problem. Thanks in advance.

Did a hike with the downgraded chipset firmware (2.60/2.30) and did not experience the position drift problem, and everything (except the speed and the odometer errors) seemed to work fine. On the same route I usually got several such drifts with the 2.60 chipset fw. So I'm happy (sort of)...

Thanks astronomenov! I can confirm, that with this method it is possible to downgrade the chipset firmware (ephemeris wasn't lost for me). I'll test it over the weekend to see if the inaccuracy is fixed. Thanks again.

Same here on the HCx. In fact my old Geko201 shows better speed readings basically all the time, than the HCx. These MTK chipsets (used in all new high sensitivity Garmin units) seem to be incredibly broken with low speed navigation. They just didn't test it properly in that mode. Probably drove around with it all the time, and hadn't tried walking.

Maybe Garmin is aware of the issue, only not admitting it. Either way, they are handling it in a pretty sh*tty way. What harm would it be to admit that they are investigating? What harm could it cause to distribute the previous chipset firmware to affected people?

Heh. The only way this will get their attention, if it's mentioned in a high profile review of one of the units. ...or if we manage to directly contact an intelligent enough Garmin engineer.

That's my feeling as well. There's a valley on a usual trail I take where the reception isn't very good, and even before the firmware upgrade I got some variation (~15m/45ft) on the tracklogs. But after leaving the valley, the tracks quickly became very tight again. After the firmware upgrade going through this place almost always triggers this "drift mode", the inaccuracy persist for a long time, and sometimes even after all satelites come back at full strength. I'm not patient enough, and usually power cycle it before it can return to normal by itself. WAAS: I have it normally turned off, and that's how originally I've seen this issue. Then I've tried with WAAS turned on as well, but it happens just the same. Apparently it doesn't have anything to do with WAAS.

I agree completely. The fact that power cycling it restores the expected accuracy points to a software/firmware issue. What I've also noticed is that this drift only happens when moving slowly (walking), when moving with any substantial speed the accuracy always stays very good, even with poor reception (just like before the firmware upgrade). So it seems to me, that this may have to do something with the earlier "low speed odometer" issue, that was fixed with this firmware. Apparently that fix may have broken the position calculation for low speed movement.

Hear, hear! The problem is that the H series has a new chipset (MediaTek) and Garmin probably doesn't control the firmware, or if it does, it has very little expertise in it. In hindsight I should have bought a 60CSx, which is a bit more expensive, but much more mature unit. Oh, and I wish there was a little more competition in the handheld GPS market.

Definitely. I had this unit for about half a year before upgrading to 2.60, and never noticed anything like this. I use it mainly for mapping (OpenStreetMap and turistautak.hu), and so these inaccurate tracks made the unit basically unusable for me. Yes, it's weird, that only a small percentage of HCx users seems to experience this problem. Just guessing, that some older ones have a different revision of the chipset, that makes them susceptible to this bug. If Garmin doesn't release a fix soon, I'll have to send it back for repair/replacement under warranty.

Checked with and without WAAS, and it's the same. It does seem to have "good days" when it seems to work perfectly. My current "record" for this bug is 50m (150ft) shown accuracy and about 150-200m (500-600ft) actual inaccuracy, with good satellite reception. Turning unit off/on restored the accuracy to 4m (12ft) and to the correct position on the map at a road crossing. N2CUA, the Legend HCx has exactly the same firware as the Vista HCx, but I wouldn't worry about the factory installed software. I'd advise against upgrading to the latest "GPS software" though, until Garmin resolves this issue. The main software doesn't play a part in this bug, so that can safely be upgraded.

I have WAAS always turned off, as I've never seen any improvement with it on the HCx. Maybe the act of turning it off flipped something in the chipset firmware. I'll try turning it on and then off to see if that helps. Thanks for the info. BTW, I got this reply for my latest query to product support: "There is no known fault with the firmware and no way of downgrading, so the only option is a repair." At least they are trying to be helpful. But still, replacing a unit because of a buggy software update is a rather stupid thing to do IMHO. Especially since this seems to affect quite a number of people (I've got reports of this on a Hungarian forum as well).

Yes, it's unfortunate that it's impossible to downgrade the chipset firmware. But finding more cases increases the chance, that Garmin will do something about this quickly. For example they could rename the 2.50 chipset firmware to 2.70 and allow us to "upgrade" to it in WebUpdater. Or just create an exe with the old chipset fw for the affected people. Have you contacted their support yet? I think it may speed things up if more people would complain about this bug.

Ahh, yes! That is very much how it looks like. For example the red track does the same turns as the blue, but it's wandering further and further away from the real position. That's exactly the same as what I've been observing. I see that you both have the same firmware versions. So it must be something that goes wrong during the firmware upgrade in some cases, and not in others. Garmin support told me to reset the unit and leave it for 30minutes in clear view of the sky. The reset did erase all the user settings, but didn't seem to have any effect on the GPS receiver, the fix was instantaneous after the reset just as before. So the problem may be that the GPS chipset itself needs to be reset, but which apparently the normal reset doesn't perform. Thanks for the info, that's very useful!

No. 2.60/2.60 here and I'm getting highly accurate fixes. You don't have "lock on road" turned on do you? Which map are you using? I'm also getting highly accurate fixes, after turning the unit on. But after some time (half an hour, hour) of walking in not so open terrain, the accuracy just drifts away, and is not always regained when the reception becomes good. In these cases turning the unit off and then immediately turning it on improves the accuracy by 10x, so there is clearly a bug somewhere... "lock on road" is turned off. The map is a free tourist map of Hungary made by volunteers from gps tracks (turistautak.hu), and it is quite accurate in most places. But I also know these routes and places well and have many tracklogs from before and after the firmware upgrade. Anyway, thanks for the response.