Author |
Topic |
|
ullix
Germany
1164 Posts |
Posted - 07/31/2024 : 04:29:18
|
I got a brand-new GMC-800 counter (firmware: 1.09) and opened it, see picture. (You find a high-resolution picture on the GeigerLog site: https://sourceforge.net/p/geigerlog/discussion/general/thread/76cdbfd23b/ )
To my surprise I found the surfaces of all major chips "sanded off"! I can imagine two reasons for doing that:
- To make it harder for your competitors to copycat your device. - To hide that you are using counterfeit chips.
Which one is it here?
The tube is a J321.
I believe from some details that GQ is now using an STM32 cpu, and no longer the classic, Intel 8051 based, ones. It does give hope that the clock drift is much reduced now over the old design, as the STM32 have a real-time clock built-in. |
Edited by - ullix on 07/31/2024 04:33:07
|
|
Reply #1
Kaninbjerget
Denmark
18 Posts |
Posted - 07/31/2024 : 10:23:30
|
Yeah. I've noticed that too and yeah it annoys me as well (what's to hide?). Even those IC's that are quite obvious like the USB to serial bridge (identical to latest revisions 300 series). Funny thing is that they're not sanded. They're lasered and with a bit of care and electronics knowledge you can actually still read what chips they are. And beside soldering the missing 4-pin header there's probably more than one way to find out which chip it is. No surprize they'll need a more powerfull IC to drive higher resolution color display. It even has a voice for the volume. Why? The only thing I've found with voice. I like short distinct beeps - not long running power on/off chimes and for volume voice doesn't represent the beeps that well which is the most important thing.
I agree with the real time clock. In the 300 series it is absolutely horrible(!) But one thing I don't understand. At least the latest 300-series revisions has DS1302 realtime clock IC - why does it drift so much then? I mean even a really off spec and badly loaded crystal shouldn't drift more than a handful of seconds each day and using the correct crystal and loading capacitances drift should easily be within 2 s/day (even IKEA's non RC clocks are much better than that). Are they constantly reading and writing back the time? I've build numerous Chineese clock kits with various RTC's including DS1302 - they keep "perfect" time. |
|
|
Reply #2
EmfDev
2238 Posts |
Posted - 07/31/2024 : 10:52:33
|
The current versions for the gmc-300/320/s units have improved ds1302 clock.
As for the voice, you can hold the back button to have it read out the dose rate. We are planning to add an automatic voiced reading every minute/10min/custom set by the user. |
|
|
Reply #3
Kaninbjerget
Denmark
18 Posts |
Posted - 07/31/2024 : 23:32:29
|
quote: Originally posted by EmfDev As for the voice, you can hold the back button to have it read out the dose rate. We are planning to add an automatic voiced reading every minute/10min/custom set by the user.
Wow thanks! Wasn't aware of that. Perhaps I didn't read the instructions enough. For most parts it's simple to use I'll try it when I get home
Not sure how much I'll use it though but a really nice gimmick. If anything is missing I think it's the timed count. Not needed to set own duration. Just the default ones from the 300 series would quite suffice. It is very useful to be able to rate different objects against each other as instant readouts by nature are a little random. Perhaps also the navigation could need a little tweak. I think it's faster on the 300 series.
quote:
The current versions for the gmc-300/320/s units have improved ds1302 clock.
I wonder whether I have the lastest hardware revisions? I have a 300S with mini USB and a 320S with USB-C. They are both equally and really horrible timekeepers. But reminds me that it is probably time to ask if there's a newer firmware than 1.13b (as I remember them have). They was updated a year ago as the 300S seems to have an issue and the 320S while at it anyway (can't remember if it got dynamic estimation enabled or it already got it). Updating firmware didn't fix it back then. I can set time manually but it's unable to sync with computer. It says synchronized successfully (no error) but it doesn't happen. The 320S have no issues. I wonder if it is fixable by firmware. I also has the impression that I got an open package from Amazon and wondering if it had been better to just send it back. The 320S was no doubt fresh, I really hate ordering especially electronics on Amazon for exactly those reasons - you risk getting a package that has been returned even though you want a new one and don't mind paying a little extra for it - but living in EU it's by far the easiest and most economic way to get quality geigers as other ways involves very high shipping costs and taxing fees (not the taxes themselves but the fees to have it taxed). |
|
|
Reply #4
ullix
Germany
1164 Posts |
Posted - 08/01/2024 : 01:30:59
|
quote: ... soldering the missing 4-pin header ...
Any idea where to solder this?
|
|
|
Reply #5
Kaninbjerget
Denmark
18 Posts |
Posted - 08/01/2024 : 02:26:45
|
quote: Originally posted by ullix Any idea where to solder this?
Look above the µC. There's 4 vertical through holes and the top one marked with a circle. That is the positive - it goes to another point indicated that's the stable 3,3v internal supply. The second seems to be negative and then two for communication. Such connection points are quite common in electronics. Device can be assembled and before it ships to customer µC (and/or EPROM) can be programmed directly usually using pogo pins so no pins needed to be soldered. When it's empty is doesn't understand serial communication. Essentially a low level access to its programming. |
|
|
Reply #6
ullix
Germany
1164 Posts |
Posted - 08/01/2024 : 03:05:04
|
Clock Drift
When you do long time recording (weeks, months, ...) the clock drift of your counter can become a problem in particular when it is as much as 20 sec per day as almost universally seen with GMC counters. So, now that the GMC-800 has a new chip, how is it faring with respect to clock drift?
I took overnite measurements with a GQ GMC-300E+ counter, a new GQ GMC-800 counter, and a FNIRSI GC-01 counter, all shown in the pic.
The GC-01 is a low-cost counter - less than half the price of the cheapest GQ counter - and with its RadPro Open-source firmware is fully supported by GeigerLog. If you want to try: follow the instructions in the GeigerLog 1.5 manual, and download the latest pre-release version of GeigerLog (currently pre05) from https://sourceforge.net/p/geigerlog/discussion/devel2/
The FNIRSI GC-01 is a look-alike to the GQ GMC-800, however, a look on the inside reveals a very different design!
If you use a GMC-800 the same GeigerLog pre-release is also recommended, as it has a variety of improvements for the GMC-800 as well.
The GMC-300 has the often observed, almost obscene, 20 sec per day clock drift. The GMC-800 has a 4 fold improved drift of "only" 5.3 sec per day. I hope this is not an outlier; so far it is measured on a single device only!
However, the by far cheapest kid on the block is outperforming both of the GQ counters with the lowest clock drift of 3.5 sec per day!
This is interesting, because the GC-01 uses an STM32 chip - I know for sure because nothing inside is lasered away! - and I presume that the GMC-800 is also using an STM32 chip.
Then with a chip from the same design family, why is the GQ GMC-800 performing twice as bad as the low-cost counter?
Furthermore, now that more CPU power and memory is available, why has GQ removed the clock-correction introduced only recently in its firmware? With proper code, it could have mitigated the drift (almost) completely.
Well, GQ has opted for spending the effort and money on a loudspeaker, a cable, a connector, code, memory, some cicuit board real estate to have a female voice say "six-teen-cee-pee-emmm" after you press a button for some 1 sec.
Irrelevant for a quality device, but alas ...
|
|
|
Reply #7
Kaninbjerget
Denmark
18 Posts |
Posted - 08/01/2024 : 09:56:38
|
quote: Originally posted by EmfDev As for the voice, you can hold the back button to have it read out the dose rate.
Tried it. Sure a funny gimmick! But it seems like there's a bug/odity with numbers in the 30s.
10.25 uS/h reads out as ten point twenty-five uS/h 10.35 uS/h reads out as ten point three five uS/h Any of them is probably equally right unless some ISO standard or similar says otherwise - but the inconsistency is odd.
2586 CPM reads out as two-thousand five-hundred eighty-six C P M (can we get counts per minute?) 2534 CPM reads out as two-thousand five-hundred three four C P M (that is for sure odd)
I don't have any real strong sources so while I probably could get past 3000 CPM with right setup mix of beta and gamma much higher than that will be a challenge.
Firmware Re1.09 |
|
|
Reply #8
PLSoftware
Italy
9 Posts |
Posted - 08/04/2024 : 07:13:12
|
I bought it a few days ago with firmware 1.09 and I can say that it is fast and very precise, it's a shame it doesn't have the "timed count" functions |
Edited by - PLSoftware on 08/04/2024 12:47:27 |
|
|
|
Topic |
|