UF896 7minutes problem

[sky59]

Hi, I and some other people have the same problem. I tried original Android (never worked well), two different OpenWrt and this Debian.
Problem is that modem disconnects from ISP after 7..8 minutes. So it works only for 7 minutes.

I suspect that it could be the SIM card is rechecked if still present in modem.
I also read somewhere that SIM_SEL and SIM_EN are exchanged. (so probably seller flashes incorrect FW)

So I believe it might be that after power on SIM is available due to all signals active. Then if it wants to reaktivate SIM it crashes?

Is it possible to exchange in source code these 2 signals? I tried to find but there is no *.dts file for UF896.

Thanx!

[kinsamanka]

Try adding a heatsink or active cooling, these devices run hot. I used silicone thermal pads on mine.

Is it possible to exchange in source code these 2 signals? I tried to find but there is no *.dts file for UF896.

The *.dts can be extracted from the *.dtb files. They can be found under the /boot directory.

[sky59]

Thank you for the fast answer!
I think it is not due to heat, the time is pretty the same independent from ambient temp.
Also stick completely works except disconneting from ISP 4G provider.
I think it is more likely that even chinese seller makes chaos programming wrong code into original device - Android.
Somebody sucesfully solved the problem with exchanged pins.

Yes, I decompiled dtb file into dts file using dtc but I do not see any pin for SIM card, but I can see at the end pins for LEDs.
Or do I use wrong program dtc and I should use something else?
Do you know what is for SIM card? Is it "sdc1" signal? And sdc2? but they have 3 different signals not EN and SEL

[kinsamanka]

According to this, there is no sim select for UF896.

Have you tried another dongle? It could be a hardware issue too.

[sky59]

It is a bit strange:
this is from forum:
https://forum.openwrt.org/t/uf896-qualcomm-msm8916-lte-router-384mib-ram-2-4gib-flash-android-openwrt/131712/127
articel 119

Change GPIO power of SIM card in file:

/ https://github.com/kirdesde/Handsomemod/blob/uf896/target/linux/msm89xx/files/arch/arm64/boot/dts/qcom/msm8916-uf896.dts /

Change SIM_EN and SIM_SEL

and yes, I did change it, but it is OpenWrt, I will try it this evening
It is weird it is not the same dtb for debian as for OpenWrt......(?) I would prefer debian :(

Anyway, I will let you know tomorrow, I hope I manage to replace new *.dtb in /boot

[kinsamanka]

If you have a working kernel from another firmware, you can easily use them here.

Just copy the /lib/modules/* files, kernel vmlinuz and the dtb. Edit /boot/extlinux/extlinux.conf and it should work.

[sky59]

thank you for help, but what you suggest probably would not solve problem, here is why:

yesterday I made 2 tests: I put modem into freezer then out on the window (temp was -3celsius) and after 7 mins disconnected
then I put it on the radiator (about 50 degrees) and after 7 mins disconnected, changed from connected->enabled, restart not possible

this behaviour is the same for all possible systems, debian, 2x openwrt (different forks) and also original Android
I am not the only person to have this problem as you could see on forum I quote above

I think you know well debian, if you could tell me which log and where to see, eventually another way of digging this problem I would try it

as I said yesterday, I tried openwrt with exchanged EN and SEL in dts and no change (I am not sure it was really compiled with exchanged signals, I have to find how to get dtb from openwrt, it is not anywhere......)

I suspect there is a problem with ModemManager and it is copied everywhere from the same origin.... I think

[sky59]

I verified that I correctly exchanged and recompiled OpenWrt:

• using binwalk I extracted from boot.img flattened device tree
• using dtc I reversed dtb into dts
• comparing two obtained dts files I found they ARE exactly the same except this:

NORMAL:
sim_sel {
label = "sim:sel";
gpios = <0x34 0x02 0x00>;
default-state = "on";
};

	sim_en {
		label = "sim:en";
		gpios = <0x34 0x01 0x00>;
		default-state = "off";
	};

EXCHANGED pins:
sim_sel {
label = "sim:sel";
gpios = <0x34 0x01 0x00>;
default-state = "off";
};

	sim_en {
		label = "sim:en";
		gpios = <0x34 0x02 0x00>;
		default-state = "on";
	};

So it means I did it correct.
But I am left in shock, though two pins were exchanged modem behaves in the same way including 7minutes working modem

Question is if I changed pins that control SIM card???
Or both pins are controlled in the same time in the same way so it dos not matter if they are exchanged?

I will try to exchange sim and sim2 if it stops working!?

[kinsamanka]

Question is if I changed pins that control SIM card???
Or both pins are controlled in the same time in the same way so it dos not matter if they are exchanged?

This shows that UF896 does not use the sim select pins.

Try another way of checking the hardware:

• remove the plastic covers
• use a hair dryer to heat up the board to around 70~80 degC
• observe how long before the sim disconnects

[sky59]

yes, you are right in this:

• your debian does not use SIM definition pins in tds file
• openwrt has got definition in dts but definitely ignores it (yesterday I even exchanged SIM1 <-> SIM2 and no effect, still working but only 7 minutes)

I already made temp test and it is not temp problem, see above

Do you know where in source code the SIM is handled? Because one guy (see above) proved already different hardware with exchanged pins

[kinsamanka]

No idea where the SIM handling is done, maybe it's done by the Qualcom firmware.

[sky59]

I have discovered something strange, this could be the problem:
I put in modem with debian SIM card (valid but not registered oficially) and let it go for some time. When I started it the mmcli was reading searching/connecting, but it could not because of not registered SIM.

Later I dicovered that mmcli reads "no sim card". I measured VCC and it is only 1.8V. The card in reader with 5V normally replies ATR.

I think there is something wrong with selecting VCC for SIM?

[kinsamanka]

I think there is something wrong with selecting VCC for SIM?

here

[sky59]

Yes, I believe this is the problem. Even before this I did not know why one old SIM card is not recognized in the modem.
I think it is 5V SIM (very old) so modem somehow did not recognize the card.

(I soldered small SMD led with SMD resistor 10kOhm on SIM on VCC and voltage is lower than 2.2V all the time, modem does not increase not even to 3.3V, not to 5V)

[sky59]

I spent today whole day measuring the staff. It is modem that deceives us. It is NOT true "Sim missing"

I measured Power, Data and Reset signals (from top to bottom) All SIMs I have really work on 1.8V and modem DOES know it is not needed to increase the voltage. Because all SIMs I tested communicate with modem.

But for some modem reads "Sim missing" - absolute nonsense. Attached are waveforms for "sim missing" but you can see it answered ATR and then communicating a long long time.......

I do not know why modem claims for some sim cards they are missing but they DO communicate!?

Attached is a long time record and first byte of ATR 0x3B after Reset released to log1.

What the heck is going on?! I am sure this is also answer why modem drops connection after 7 minutes.

[kinsamanka]

Do you have another modem to test?

Do note that majority of these boards use recycled components.

[sky59]

Unfortunalety I have only one. But I have different types that are only NDIS modems which need some computer, they are usb sticks. In these all sim cards I have are "visible"
I think this is the first step to understand why some sim are "missing"

I believe no usb openstick (UF896, UZ...., and so on) work
They probably all work for 7 minutes....

Or problem is really on my side. I use "modem.bin" from my stick which had Android. Could be that I transfer 7-min problem with this modem file. I will go back to Android I have backup and try to root it to make some analysis. Could be that this modem file is also responsible for "sim missing" problem

[kinsamanka]

You need to test with the original firmware first. If that fails, it's either a firmware or hw issue.

The sticks I own has no modem issues. The only problem that I've encountered are flash failures (due to recycled chips).

[vernonet]

Hi, I and some other people have the same problem. I tried original Android (never worked well), two different OpenWrt and this Debian. Problem is that modem disconnects from ISP after 7..8 minutes. So it works only for 7 minutes.

Thanx!

Does your modem have a battery? If the modem does not have a battery, then stable operation can only be achieved when powered from an external source, not from the computer's USB. I modified the USB cable (connected an external power supply) for FY_MF800 and now disconnects happen very rarely (once a day). I removed the battery, since its charging does not work adequately on Debian.