Boot loop and network issue

[Stephen Seer Snively]

Greetings,

I recently received my Hermes, filter board, I/O board, and case from Makersfab and have been trying to get it working. The assembly wasn't hard and the I/O board fit pretty well in the case, better than I expected from the photos on this forum. 

For the most part, it appears that my Hermes will power on, attempt to get an IP address, and then be 'alive' for a few moments, perhaps 4 seconds. And then all the lights will turn off and it looks like it starts the boot sequence again. Here are two short videos of the process, with and without a short on on the PTT

https://photos.app.goo.gl/az66KqQ63Bb1caSm6 

Could I be shorting something inside the case? I was surprised to find the block to connect the I/O board was longer than the pin block connecting the two other boards, leading to the possibility of putting them in the wrong pins. I don't think that happened as the back plate lines up well, but I want to be transparent that this isn't my strong suit, but I am a system administrator and understand networking very well. 

I was, one time, able to see the radio on the 169 address in SparkSDR, but after I connected to it and saw a waterfall for a moment, I got to 'configure radio' and found the IP address setting, but the waterfall stopped a few moments before I clicked 'set' and the radio appears to have rebooted again. 

I'm at a lost as to what to do next. My laptop in my living room, where I'd like to operative from, is two switches away from the Hermes and doesn't see the arp entry for the Hermes at all, but maybe that's just because it's been alive so shortly. 

[Mike Lewis]

Did you solder in the small 3 pin header to the main board, skipping the holes as specified?  This makes the overall length of the main connector longer matching the IO board.

 

Have you tried a fully wired connection?  Sounds like you may be wireless, that can be iffy on some networks.  If the radio and your laptop are on a home LAN then the radio should pickup a valid IP address and not use a 169  address.

 

You can also use the hermeslite.py script to see if it discovers your HL2, same process as SparkSDR, Quisk, and others.

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/e3486e47-1d49-4348-91a5-911926a7fc13n%40googlegroups.com.

[Stephen Seer Snively]

I've not soldered anything. Am I missing a step? Is there a link to the instructions for how to do this that I missed, and would not doing it show what I'm describing? 

Perhaps I will remove the I/O board and use the 'jumper block' for a moment to see if I can get the radio up and working first. 

[Mike Lewis]

Always best to start with the basics, then add on. 

 

All the info is on this github page.  Note the 3 pin header added to the main board.  You should have received a header strip, cut it down to 3 pins and place as specified, there will be a few holes empty as seen in the picture at jimahlstrom/HL2IOBoard (github.com).

 

For the IOBoard, there are instructions on how to load firmware, starting with the test firmware.  After a good test result you then load the basic example. You should first know what you want the IOBoard to do for you though, it is basic functionality and exists to customize outputs requiring you edit code, compile, and upload and test yourself. The HL2 does not require it to operate for most folks.

 

[Stephen Seer Snively]

Thank you so much! I feel like such a fool as I've read that months ago, but it didn't register as a step to complete before putting in the I/O board. Thanks for pointing that out, and you are correct, I should take baby steps first. 

While I understand that the I/O board isn't needed for normal operation, I'm a 'computer nerd' and the software side of this project really interests me. It's like being inside the radio, except I understand how to read wireshark streams better than multimeter readings. 

My network is fully wired, but it is thru a Dell switch and a unmanaged switch and a docking station via USB-C to my laptop. 

UPDATE: I've removed the I/O board, but it still doesn't work. I think I might have broken the antenna port, as the connector now spins. Yes, I can see the square part of the connector spin, and I'm not sure what to resolder. I used the coax jumpers I was using for my Kenwood with an SMA adaptor and it looks like the tension was too much for it. Did I just ruin my new radio before I even got it working? 

I know a local club member with a microscope and good electrical skills, and I might be able to schedule some time with him if I know exactly where it could be fixed. Again, I'm not skilled in this level hardware might need words to tell someone else who might help. 

[Mike Lewis]

If the center pin is intact and lines up to look like the other SMA, you could get away with just soldering the flange to the old pins/pads.  These are common SMA edge connectors and replacing it would be easy and preferred. 

 

A thin flexible jumper is advised to adapt to a have cable to prevent such issues.  6 or 12 inches of RG316 for example.  Be sure to support the cable, you still cannot pull on it too much.

 

Regardless of the connector, assuming no other hardware issues, the HL2 should get a proper DHCP address and communicate with your SDR app, receive something, if not just noise until you fix your antenna connector or use the Alt RX jack.

 

As a temporary workaround for receive you can get the basic HL2 talking, add in the IOBoard, then use the SDR app settings to Rx on the Alt RX port.  How you do that varies by SDR app and/or custom firmware.  Many posts have been made about that (configure and use the Alt RX port).

 

[Stephen Seer Snively]

I just ordered a cable like that.

Even with nothing attached to the antenna port, I'm still getting the 'boot loop' behaviour. The very first cycle after power on seems to keep link and activity lights for 3-4 seconds, and then all lights turn off and the cycle restarts. The next three boots or so will have it reboot a fraction of a second after the 'activity' light first turns on, and then every boot cycle after that won't light up the Ethernet ports at all, just the D3 and D4 LEDs. 

Shorting the PTT gives a similar response. 

Any ideas?

[Mike Lewis]

Don’t know what the booting issue is but there are 2 kinds of ethernet chips due to the original going out of stock. 

 

Maybe look at this thread and similar one in the past for some clues and links to various procedures to help narrow things down.

Eathernet stopped working need to get hl2 repaired (google.com)

Troubleshoot Network · softerhardware/Hermes-Lite2 Wiki (github.com)

 

[Steve Haynal]

Hi Scanner Brightly,

For the reboot issue you describe, I first suspect a power supply or over current issue. What power supply are you using? Please report the voltage and current draw when the HL2 is connected.

Is there an intermittent connection at the power supply barrel jack? Try wiggling to the power supply connector to see if you can trigger the reboot.

There is a F1 fuse which will trip for over current conditions and then reset. It is possible that it is defective but first we should verify that there isn't really an over current situation.

I don't suspect an ethernet issue as you say all lights go off for a period of time. One of the four LEDs will stay lit the entire time if IP is renegotiated. How long is the time period when all LEDs are off while it reboots?

73,

Steve

kf7o

[Stephen Seer Snively]

Interesting angle. I have a Jetstream JTPS32MAB switching power supply with Andersen poles in the front, which I have used for my Kenwood TS-140s, TS-570, and TV-71a (all currently disconnected) in the past. I did try to wiggle the barrel connector, but it didn't seem to do anything. I don't have a power monitor besides the analog front plate of the power supply, which shows very little draw. With that SMA cable I ordered above, I'll also be getting an in-line power monitor and will be able to tell you more. 

Here is another 30 second video of the boot loop LEDS and a look at my power supply. https://photos.app.goo.gl/2tavXPiagG9BxQdg9 . I could also do a slightly longer, tripod version of a 'cool boot' for 2 minutes or so, so you can see this entire thing (see below) play out if you like.

But I did notice something very interesting: I hadn't touched it since yesterday, and I booted it for the first time just now. At first, it seemed to fully work, getting to the ADC LEDS solid and the D2 and D3 lights blinking together, with both link and activity lights on the Ethernet jack. I was impressed! I watched it seem okay for about 10 seconds. Just as I was thinking of turning the local computer on, all lights went out for half a second, and the boot sequence started again. This time, it got to the same place but only for a second and then rebooted. Next time, it just lit up the "activity" light on the Ethernet before it rebooted. Each reboot was quicker up to a point. 

My only build problem was when I first drilled the whole, it was about 1mm off, so I had to re-drill it and it slides a little bit. Since I will be removing it from the case to re-solder the SMA connector, I'll try it without the heatsink (and without transmitting) to see if that changes anything. Could that heat shim being about 1mm off (frontwards) cause a short? 

Seer

[Steve Haynal]

Hi Seer,

I wouldn't expect a short near the hole to cause a problem like this as the PA uses a different internal power supply, but it is a good idea to check when removed from the enclosure. 

Do you see the same behavior if no ethernet cable or antenna is connected? (Only connection being power.) I expect clk to blink, hw to remain solid and no activity for IP and SPD, The hw LED going off briefly and the rhythm of clk being disrupted is not normal.

This still feels like an internal power regulator is resetting. With the top off, can you measure 1.2V, 2.5V and 3.3V as seen in the picture below and see if there is any brief disruption that is in sync with the hw LED going off? A trigger on a sampling scope would be ideal but a DVM may pick it up. Does anything feel hot to the touch?

https://github.com/softerhardware/Hermes-Lite2/wiki/Hardware

Also, your left clk LED is blinking at the unexpected slow rate for a new HL2. Can you please verify if you have a ksz9031 or ksz9021 installed, and share a picture of R34 similar to this thread?
https://groups.google.com/g/hermes-lite/c/FMcByH2uupU/m/zoOSVYs5BAAJ

https://groups.google.com/g/hermes-lite/c/FMcByH2uupU/m/ms-pHbQ5BAAJ

73,

Steve

kf7o

[Steve Haynal]

Also, please apply slight downward finger pressure on the ksz9021 (ethernet IC near jack) towards the circuit board and power up the HL2. Does this cause any change in behavior?

[Stephen Seer Snively]

First, you are correct that if I plug in only power, the ACL 100 LED stays lit and the RUN LED will blink at a steady rate forever. I've tried it for a few hourse and it never fails.

If I plug in the Ethernet (with or without downward pressure, doesn't seem to change the outcome) it will appear to light up and work for about 10 seconds, and then all lights will turn off for 0.2 seconds, and it will do the 'loop', but never stay with all lights like that again, even for a moment. 

Here is an okay photo of the whole board, I'm not sure exactly what you want to see, I can get a better lit photo of the part you'd like to see if you just let me know where in the above photo to take. 

My cheap multimeter doesn't seem to refresh quick enough, but I believe I might be going to someone's house tomorrow who has much better equipment and skill than I do. And the SMA connector will be fixed. 

I'm getting about 

* 600 between the 2v5v and GND 

* 800 between 3v3 and GND. 

* 1603 between Vop and GND, 

* 20.36M om between Vsupply and GND. 

* 10.70K om between Vpa and GND

*  152k om between Vbias and GND

* 18M om between 1v2 and both the 3v3 and 2v5 rail.

* 1.8k between the 2v5 and 3v3 tail. 

So that all seems "above 500", but visual inspection shows that the L3 look more open, with copper showing thru. What is that about, I never noticed it before. The chips at U4-U6 and such look perfectly done. A photo with flash, for the chip. 

Any ideas? 

[Rick Langford (N8SDR)]

Is it possible that the ethernet pins have been bent and crossed so that when you plug in a cable there shorting?

Also, what about the cable itself? can or did you try just leaving the end on the Hermes plugged in and see if it stays on

Also are you perhaps using a POE switch or a POE active port perhaps when you're plugging in the Hermes?

[Stephen Seer Snively]

I have tried several cables before, but just now swapped my known good-at-1 gigabit PC Ethernet with the radio one. The radio one also connected at gigabit speeds on the PC and has no errors. I've tried store bought and CAT5e cable I've crimped myself. I don't use POE at home, but I also tried removing my Dell switch from my network to connect only thru an unmanaged switch to the NetGear router, which, again, I can see in the router logs as this:

[DHCP IP: (10.0.0.48)] to MAC address D8:BB:C1:D6:C9:CD, Saturday, Jan 13,2024 11:23:08

That's the IP address I've set for it to give that MAC address, which is what my Hermes broadcasts. I haven't done any WireShark yet, but I'm not sure it's a network issue, with the full LED blink out it does.

Here is today's 13 second video of it boot looping with the cable I was using for my PC: https://photos.app.goo.gl/AFYLdqX2BFhR1MsC8 

This video starts after the 'good 10 second boot' on the first time around. You can see the ADC 100 light blink off for a moment at 3 second and again at the 11 second mark.

Thanks for keeping interest in helping me. I appreciate it. Sorry for my typos above, I recently had surgery on my left hand and just got my stitches removed, and it's still a little strange to type. It's not ham radio without random medical knowledge of strangers, right? Yet another great reason to move my ham radio experience to Thetis: leveraging PC accessibility tools! I can't wait to start! 

Seer Snively - KM6WMO

[Stephen Seer Snively]

I see the MAC address  00:1C:C0:A2:13:DD in the router logs 8 times, from the 11th to the 13th. That might be how many times I've done the 'cold boot' and it's gotten it's 'good 10 seconds of link light' the first boot around. Looks like it's grabbing 10.0.0.8, but SparkSDR still doesn't see it, but I'll try to force it. 

Router log:

[DHCP IP: (10.0.0.8)] to MAC address 00:1C:C0:A2:13:DD, Saturday, Jan 13,2024 13:44:58

[Steve Haynal]

Hi Stephen,

The picture clearly shows a KSZ9031 is installed. This explains the slower blink of the left LED. If Makerfabs installed the KSZ9031 but with a R34 value for the KSZ9021, it could explain what you are seeing. The resistor R34 sets the transmit output level for the ethernet phy. Please take a look at R34 and report its value or take a clear picture. You can measure the value in circuit with a simple Ohm meter provided no power is applied to the HL2. 

R34 is on the bottom side of the PCB:

Here is a close up picture of the proper value for KSZ9021. Your value markings must be different from this to be correct for the KSZ9031.

73,

Steve

kf7o

[Stephen Seer Snively]

Here's a photo of the bottom of my board, showing the same "688" or "889" marking as your photo. Does that mean I have the wrong  R34? What do I do now?  :

[Steve Haynal]

Hi Stephen,

Yes, it does look like you have the wrong value of R34 for the KSZ9031. When Makerfabs tests and programs units, they use a Raspberry Pi with short ethernet cable and no ethernet switch. That setup does not seem to be sensitive to this issue. I will let Makerfabs know that this problem is not caught by tests.

The markings on these resistors can be confusing. Please verify the value of the resistor with an Ohm meter. I expect you will measure 4.99K when it should be 12.1K.

Do you feel comfortable changing this resistor? I can send you a new one of the proper value. If not, I can do the repair for you since you are located in the US. Your shipping costs would be covered from the repair pool. Let me know if you prefer this and I will provide shipping instructions in a private e-mail.

73,

Steve

kf7o

[Stephen Seer Snively]

You are correct, my meter reads it as 5k. I do not feel able to change the resistor, I am sorry. 

Thank you so much for your time and effort. 

[Steve Haynal]

Hi Stephen,

Thanks for confirming. I sent you a private e-mail regarding repair.

73,

Steve

kf7o

[Robert Herendeen]

Only one 12.1K resistor is used on the HL2 build. I suspect they loaded the wrong reel and every unit passed testing with the wrong value. 

Is there any way to see how many units were run in that build run? 

[Steve Haynal]

4.99K is correct for the KSZ9021 which is usually used. I suspect it is more likely they built a few boards with the KSZ9031 but did not change to the 12.1K which is required. I have informed Makefabs of this potential problem. 

73,

Steve

kf7o