Building log for "EEZ BB3 Enclosure & Bare Boards"

[prasimix]

"Do not power on the MCU module without the display connected or with a loosely connected flexible display cable. Doing so may damage the power supply of the display backlight located on the MCU module." Oh shi~...

Applicable only to assembled MCU modules where ZD1 with too high voltage (30 V) is mounted. In the last BOM list it is lowered to 24 V and it should be fine. If accidentally a failure occurs, there is no damage to the display. All that can happen is that D2 breaks.

[prasimix]

Phew, seems to be working after all . At least I was able to upload firmware. I'd like to share my experience so other people won't do same mistakes.

Not the best soldering, but good-enough for Australia . A few things I messed up. One is my zener flew away when I tried to grab it with twezers. I replaced ZD1 with two zeners (16V+7.5V). I think I damaged my tweezers by using them to open resistors on tape.

Then I couldn't solder the usb port, it seems it's meant to be soldered with hotair or reflow. I removed rear tab of the connector to get access to pins.
I ran out of cr2032 batteries, so I put cr2025, hope it will work.
The boot0 button seems damaged by flux: it's stiff, but seems to be working.

In some cases it might worth soldering ICs first, and then decoupling caps as they get in the way. But I mostly used hand soldering, users using hotair probably find it much easier to solder components. I put plastic parts at the very last because: 1) I was worried to damage it with hotair and flux 2) it's easier to work with flat board without components sticking from it. If I was assembling it again, I'd use solder paste and hotair gun for most components, rather than soldering iron.

Oh, and double-check the BOM .

Now time to assemble the enclosure...

Congratulations! The choice of method obviously depends on one’s soldering skill. I still manage to solder almost everything by hand. On this module I only used hot air for the LDO thermal pin. For USB you need a sharp tip.

There is one other thing that I did not mention in the building instructions and it is worth the effort is to further strengthen the SD card holder. Without that, it could easily happen that the metal shield pops out of the slot the first time you carelessly insert the card. Therefore, the metal shield should be soldered at three points as in the picture.



Attention! While this seems like a trivial task, it’s not for dilettantes like AssemTec Europe (aka Bolek & Lolek Enterprise). Namely, they managed to destroy over 30 holders that I had to replace. What's the problem? They applied too much solder that the solder spilled on the other side so the SD card could no longer be inserted!

[exe]

There is one other thing that I did not mention in the building instructions and it is worth the effort is to further strengthen the SD card holder. Without that, it could easily happen that the metal shield pops out of the slot the first time you carelessly insert the card. Therefore, the metal shield should be soldered at three points as in the picture.

Very interesting, the sd card holder I bought have only one tab on metal shield. I have this one: https://www.molex.com/molex/products/part-detail/memory_card_socket/5033981892 (there is 3d view on the right). Anyway, I'll just be careful, that's the least problem I have .

I did the first power up and... nothing happened except the fan worked for 10s or so. Is it possible to see any debug log via, say, usb? I wonder if it's "just" a display issue, or the board doesn't boot at all. I think this kit is a bit beyond my skills but I don't give up. I just need more spare parts . I suspect the mcu can be partially destroyed due to previously incorrect orientation. I'll try to see if there is any signal on display pins with my oscilloscope and 10x probe.

PS I forgot to buy front push button, but I can 3d-print it.

[prasimix]

Very interesting, the sd card holder I bought have only one tab on metal shield. I have this one: https://www.molex.com/molex/products/part-detail/memory_card_socket/5033981892 (there is 3d view on the right). Anyway, I'll just be careful, that's the least problem I have .

You have the right holder. It has only one tab, but it is possible to solder it on all three points.

I did the first power up and... nothing happened except the fan worked for 10s or so. Is it possible to see any debug log via, say, usb? I wonder if it's "just" a display issue, or the board doesn't boot at all. I think this kit is a bit beyond my skills but I don't give up. I just need more spare parts . I suspect the mcu can be partially destroyed due to previously incorrect orientation. I'll try to see if there is any signal on display pins with my oscilloscope and 10x probe.

PS I forgot to buy front push button, but I can 3d-print it.

I'm confused now. I thought you changed the MCU because it would be a real miracle if it survived the wrong assembly. How you manage to download firmware if it cannot power up? Did you disconnect USB cable before power up? On the power up even if some errors are detected, with functional MCU you should hear the sound (OK, that presume that audio path is functional).

[exe]

I'm confused now. I thought you changed the MCU because it would be a real miracle if it survived the wrong assembly. How you manage to download firmware if it cannot power up? Did you disconnect USB cable before power up? On the power up even if some errors are detected, with functional MCU you should hear the sound (OK, that presume that audio path is functional).

No, I just re-soldered the mcu. DFU works with it, but no sound, and tft brightness output is just 4.8V (I don't see pwm signal controlling brightness). Ordering new mcu...

[prasimix]

Yes, I think that is the right move.

[Kean]

Attention! While this seems like a trivial task, it’s not for dilettantes like AssemTec Europe (aka Bolek & Lolek Enterprise). Namely, they managed to destroy over 30 holders that I had to replace. What's the problem? They applied too much solder that the solder spilled on the other side so the SD card could no longer be inserted!

Yes, one of mine seems to be suffering from this.  Getting the microSD card in was not easy, and getting it out again... 
I should have a suitable replacement part in stock.

[billrule]

....wait....so you are implying we should read those wordy things BEFORE we ...O.... #$%$@!!

[Kean]

....wait....so you are implying we should read those wordy things BEFORE we ...O.... #$%$@!!

OK, so what did you do?

[exe]

I replaced the mcu with a new one, flashed firmware with "sudo dfu-util -a 0 -D ~/Downloads/bb3-v1.4.dfu", removed usb cable, power-cycled the board, and... nothing . I checked that 3.3V is present on decoupling caps, I saw a ~25MHz sine across Y1 oscillator (amplitude ~1V or thereabout), I checked that NRST pin is at 3.3V (measured on C8 capacitor footprint, no C8 is installed). No signal is present on Q2 (buzzer driver). I tried to start the board with and without screen, with and without backplate. I tried to replace the IDC cable too (just in case), no luck.

I suspect that there could be a soldering issue with RAM or something (for some reason solder doesn't wet pads well) and mcu stucks in initialization. I'll try to build a custom firmware that just beeps. And I'll try to resolder pins and I'll use more aggressive flux.

[exe]

Oh boy, I added tons of flux and solder on SRAM chip, thoroughly resoldered it and...  it seems to be working! I wonder if the previous mcu worked too, but I'm not gonna recover it from the trash bin and resolder .

[exe]

Okay, the "easy" part is done, now it's time for my lovely defective dcp405 modules. Denis, how do you test them?

I quickly checked them. Two of dcp405 seem to be half-working. At least they are detected and output something, but I see some weirdness in their behavior (such as voltage doesn't go to zero, or OVP triggers when trying to change voltage). One module doesn't recognize . I fixed one soldering joint on it, but it still doesn't recognize. I tried to initialize the module, but somehow I destroyed the sd card slot. No visible damage, it's just the card doesn't fit in. When I forced it in, it still not detected. Adding a new one to my next order...

[prasimix]

Okay, the "easy" part is done, now it's time for my lovely defective dcp405 modules. Denis, how do you test them?

I quickly checked them. Two of dcp405 seem to be half-working. At least they are detected and output something, but I see some weirdness in their behavior (such as voltage doesn't go to zero, or OVP triggers when trying to change voltage). One module doesn't recognize . I fixed one soldering joint on it, but it still doesn't recognize. I tried to initialize the module, but somehow I destroyed the sd card slot. No visible damage, it's just the card doesn't fit in. When I forced it in, it still not detected. Adding a new one to my next order...

Did you try to initialize it with MicroPython script as described in assembly instructions?

If OVP trips that is clear sign that CV control loop is not functional, i.e. cannot maintain set voltage.

Most of the junkyard boards are ready for complete resoldering or moving all parts section by section to the new PCB.

Also most of them should have properly functioning bias power supply (IC2, IC5, IC6) but still that is the first thing to check.

[exe]

Can I use an IDC cable to connect module outside the enclosure? Say, a 30cm cable, do you think it will work?

PS found why my sd card doesn't work. When I soldered front right corner the solder flowed inside and blocked the right side of cardhole .

[prasimix]

We have to check that, currently there is an active discussion about that on #prototyping channel at Discord server

[Kleinstein]

There is a good chance it can work with some 30 cm of cable. It is not ideal, but chances are not so bad. The critical signal would be the SPI clock and if there is significant ringing with the added line. It would help if the driver pin at the µC side is not set to highest speed, even if the means lowering the SPI speed a little.

[exe]

A quick update on my build:

1) soldered fan pin in reverse way. The interesting thing is the fan still worked, but readback didn't work and MCU thought the fan is faulty and was spinning it at full speed.
2) fixed the sdcard slot. Turned out card detect pins on the back were not soldered (I didn't notice them). The card holder got some cosmetic damage because I tried too hard to insert a card with a solder blob inside, but seems to be fully working. The blob inside was removed with a thin solder wick.
3) fixed all the dirty hacks I applied (put the proper optocoupler, replaced zener, replaced boot0 button, put proper FB1 ferite bead, installed ethernet socket, etc).

It seems that active discussions are happening on discord, so I think I should go there, and leave this thread only to report results.

PS I'll order a long ribbon cable to see if it just works.

[exe]

Quick update on my build. I managed to fix two out of three dcp405 modules. One had soldering issue on shunt sensing opamp, another one had output enable resistor badly soldered (see the pic, although it was the other side of resistor, I wasn't able to take a good picture of the problem). In many cases visual inspection didn't help much, I had to probe with a dmm to detect failures. Soldering is actually quite bad, even though at first glance it doesn't appear so. I suspect they have problem with soldering paste, the alloy feels brittle or not completely melted or something. The board also produced cracking sounds when I dried the board with hotair (the temperature was well below 100C, I could keep my hand). I hope that weren't solder joints cracking...

My last dcp405 is a bit more difficult. I resoldered a few ICs with hot air, many joints resoldered with my leaded solder just in case, also found one disconnected capacitor (see the pic). The CC mode still misbehaves: it outputs 21uA when set to zero on low current range, and oscillating ~0.2mA on hight current range. Sometimes it outputs 22mA on that range when set to zero amps. I'll continue my investigation.

Meanwhile I found that common ground coupling didn't work. It was T2N7002AK,LM from toshiba which failed open between source and drain). My very-very first dead mosfet! I replaced it, and replacement died the same way... ESD problem? EMF from relay? I checked zenner across relay, it appeared to be working. I replaced the mosfet with 2N7002-7-F from diodes inc., and it worked. Weird, isn’t it?

[exe]

Small update, there is indeed a lot of oscillation on the last dcp405. Looking at broad noise spectrum I suspect pre-regulator is not functioning well.

PS I also checked my meanwell 48V supplies to be sure the noise doesn't come from them.

[exe]

Resoldered C13 capacitor (the one on the previous picture), and it seems the module works as expected, or there is an intermittent failure which disappeared.

So, now I have three modules, they work the same way. So, either they all fixed now, or they are all broken in a way I don't know yet. I've done stress test, all are capable of outputing 0.2V-40V, 50mA and 5A. Minimum current is ~22uA in 50mA range, I wonder if other users have the same.

[Kean]

So, now I have three modules, they work the same way. So, either they all fixed now, or they are all broken in a way I don't know yet. I've done stress test, all are capable of outputing 0.2V-40V, 50mA and 5A.

Sounds like good progress.  I've repaired 2 of my 3 so far.  Did you run the downprogramming test suggested by prasimix?
https://www.eevblog.com/forum/projects/eez-h25005-a-possible-successor-of-eez-h24005-programmable-power-supply/msg3303354/#msg3303354

Minimum current is ~22uA in 50mA range, I wonder if other users have the same.

Although the low range current display doesn't work in the 1.4 FW, you can still set low currents. I took some low current measurements (as suggested by a friend) via a 10k resistor and 20V compliance voltage.  Set at 100uA was reading about 97uA (3% low).  Set at 20uA was reading about 15uA.  Anything lower (including 0) delivered 10uA.  My low range calibration could possibly be a bit better, but I'm pretty happy.  Measured with my Keysight 34465A.

[Kleinstein]

From the pre-regulator there can be some current for the control: there are about some 50 µA flowing through R7 and R10. Depending on the voltage some of this current bypassed the load, but not all. So the minimum current is expected to depend on the output voltage. With low voltage is may not go much below 50 µA, with more voltage ( > 0.5 V) the 50   µA should flow trough R27, R32 so that the load current could approach zero.

[exe]

With 10K resistor I too have ~10uA . Indeed, as Kleinstein says, the min output current depends on output voltage. With 51k resistor I got ~3.43uA output current at ~0.18V drop. Cool!

One weird (or not) thing. I got really weird oscillations on output, something I can't explain. I connected my trusty (not anymore?) analog discovery 2 to see if downprogrammer works. The good news it does . But what I also noticed is horrible ripple on output that didn't depend on power supply settings. The way I "fixed" it was... attaching ground wire to the module. I don't get how this works. What's worse, when I want to zoom in on ad2 and used 500mV sensitivity, I got channel overloaded (see last pic). Floating my laptop didn't help.

Omg, I got a ground loop that made my ad2 crazy. And the loop was due to... externally mains-powered sound card. Still, I don't get, how attaching protective earth connection to the power module helped. Shouldn't power modules be isolated?

PS attached a shot with DP on and off, jumping between 0V and 5V, current set to 1A, 10ms each step. With small current it was going into CC mode due to output cap .

PPS I ran out of broken dcp405 modules, bought another one for repair.

[prasimix]

Still, I don't get, how attaching protective earth connection to the power module helped. Shouldn't power modules be isolated?

Power modules are isolated. PE could help because both Vout+ and Vout- have 10n to PE.

[motozak]

I am very interested with the EEZ DIB BB3 PSU, I want to build one but there is a problem with the MCU (STM32F769IIT6), this is very difficult to find in my area
can I replace it with another type like STM32F767IIT6 or STM32F767IGT6?