Help with a weird problem on Ice Tube Clock Moderators: adafruit_support_bill, adafruit

[JaysonL]

Hello!

I'm having a similar-yet-different issue:

Problem: When power is applied, only the bottom segment(s) of the ice tube clock light up. Sometimes the entire segment lights up for a short while before fading out, sometimes not.

After reading this thread and learning that there was some confusion regarding Q3, I removed power, installed the backup battery, and measured voltage between "the terminal nearest the edge of the board" on Q3 and ground, and got a very low figure (in the vicinity of 5 milliVolts.

Also reading that reversing Q3 wouldn't actually harm the device, I decided "what the heck, let's try reversing it"... so I desoldered Q3, flipped it around, and soldered it back in. Lo and behold, the VFD tube illuminated properly when power was applied! However, with power disconnected and backup battery installed, I read a significant voltage (I forget the exact value, but greater than 0.5V and lower than 3V) between ground and "the terminal nearest the edge of the board" on Q3, implying that Q3 was NOW installed incorrectly.

So, I once again desoldered Q3 and flipped it, which of course caused the tube to not illuminate fully.

Now HERE's the kicker: I went back to probing voltages, and discovered that the act of measuring the voltage between ground and the Q3 side of R3 caused the tube to illuminate!

I'm hoping that somebody with a bit more know-how about what's going on under the hood will be able to tell me what's going on...

--Jays

[adafruit_support_bill]

Post some photos of the front and back of the main board - as well as the solder-side of the tube-board.

[Frank_tt]

Ok, so I found a guy near by that built an ice tube clock. We took my tube and put it on his board, it worked just fine, so it is not the tube.

We then took my PLCC chip and put it into his board, again it worked. Not that.

The high voltage on his measured 20 while mine measured 14.

Help?

Atleast we know the tube is good. Should I reverse Q3 just for grins? Even though it looks ok?


EDIT**** More info

Working board Q3 Pin voltages, outside to inside 5V 4V 5V

My bad board: 3.4V with odd ripple on O-Scope, 0V , 4.8V

[jarchie]

In the YouTube video of the display failure, three segments on one digit fail in the beginning. That's odd because the display is multiplexed, so the timing of those segments failing is perfectly in sync with the multiplexing or else they would fail on all the digits simultaneously. That suggests the issue with the multiplexing control in the microcontroller or the quality of that signal. I'm wondering if the ATmega might be acting dodgy or if something else is at play. Maybe some part is drawing too much current from one of the microcontroller pins, causing the chip to be flaky? Q3? But I'm speculating.

[phild13]

This loss of digits and/or display lighting even partially then fading out might point to a possible high current condition on the pins of the processor chip. I noticed several people seem to be reporting similar issues lately, and while I obviously have no proof anything is related, I'm thinking this may be caused by a possible ground loss on part of the board.

When the display fades out and the clock is still plugged in, does the processor become noticeably warmer (hot even) on the right side Pin 1-7 pin 14-20 areas? Use caution then testing because if it is hot, it could be hot enough to burn your finger.

[phild13]

Q3 should be a FET (verify the part number) should be installed like the silk screen shows on the board. You can always jumper the outside leads of the FET which will eliminate it from the circuit for testing. Don't short the middle pin to anything. If the clock works fine after Q3 is jumpered, then it was damaged by static before install or heat during install.

[Frank_tt]

OK, just for grins I reversed Q3 from what the shape shows in my previous pics. Wouldn't you believe it, it works. Just fine, WORKS!!

Maybe there is a batch of FETs that was packaged backwards or something. Don't know....all I know is that it works.

Maybe you can't go off the shape, it is the printing that defines this part. The part goes into the board with the printing toward the flat side of the silkscreen.

Thanks everyone for the help.

[phild13]

Unplug the clock for 5 minutes then plug it back in. Does it have the correct time when plugged back in?

[Frank_tt]

Dang it Phil, I thought I had something here. It remembers what time it was when the clock was unplugged. When I plug it back in, the time flashes every 2 seconds and it starts at the time it was when I unplugged.


SO, now what??

[jarchie]

It remembers what time it was when the clock was unplugged. When I plug it back in, the time flashes every 2 seconds and it starts at the time it was when I unplugged.

That's the behavior you'd expect if Q3 is installed backwards.

SO, now what??

I'd be inclined to replace Q3 in case the one that came with the kit was just plain defective. You could use the same part that's included in the kit or try a different FET. Phil introduced me to the ZVP2110A, which I think is a better choice than the one included in the kit.

[phild13]

The only real purpose of Q3 is to turn power off/on to the VFD chip and tube filament.

The loss of time keeping when unplugged usually means Q3 is backwards and that your battery is trying to power the entire clock instead of just the processor when there is a loss of power. The battery can not supply that amount of current and will be drained if left in that condition.

When Q3 is installed properly (the way you first had it installed) the clock will keep time while on battery during a power loss. The display will flash when power comes back on to let you know normal power to the clock was lost, but the time will be correct.

It is possible that Q3 is damaged in some way and not turning on fully. That could be the reason it works when installed backwards. There is a body diode within the FET that will conduct even when the gate is off when the Drain and source (outside leads) are reversed in the circuit.

You could try turning the FET around again to the correct orientation and see if the problem comes back, but I think your solder job the first time was fine and the problem will come back.

If (Adafruit) Bill agrees with the assessment that Q3 may be bad, get a replacement ZVP3306A from Adafruit. Or You could try getting a ZVP2110A FET. I replaced my Q3 FET with a ZVP2110A FET which costs about 88 cents + postage at DigiKey. It has better specs than the one that comes with the kit which is a ZVP3306A which costs about 64 cents. So you could try that also if desired.

You could also replace the FET with a transistor and 1K resistor like in this post:
viewtopic.php?f=41&t=43908&p=219063&hilit=transistor#p219063

You can get the parts at Radio Shack for that mod.

[adafruit_support_bill]

If (Adafruit) Bill agrees with the assessment that Q3 may be bad, get a replacement ZVP3306A from Adafruit.

Yes. Please contact support@adafruit.com with a link to this thread for a replacement. If you decide to go with a ZVP2110A FET instead as suggested by PhilD, please post your results with that here as well.

[Frank_tt]

I will order the one Phil suggests and will post my results here.

Hopefully this will resolve it.

[jarchie]

It is possible that Q3 is damaged in some way and not turning on fully.

If the problem is that Q3 will not turn on fully, I would expect all segments to fade equally. To me, the problem seems more like like dodgy behavior from one of the chips, and I suspect the microcontroller given the correlation between segment failure and multiplexing. But I remembered something Frank wrote this morning... Since you have more experience, perhaps you could tell me if the following sounds plausible?

My bad board: 3.4V with odd ripple on O-Scope, 0V , 4.8V

The ATmega datasheet recommends using a 0.1uF decoupling cap near the chip to filter noise from the power supply, but the Ice Tube Clock v1.1 design omits this. Q3 supplies power to the VFD filament and the MAXIM chip, and the filament consumes a lot of power. If Q3 is outputting an odd ripple, current draw from the filament might be introducing noise into the power, thereby causing dodgy microcontroller behavior. Alternatively, perhaps the ripple in the output of Q3 introduces noise in the MAXIM chip's power, causing dodgy behavior there.

If the above explanation seems reasonable, switching to a FET that turns on fully, such as the ZVP2110A, could reliably prevent the kind of FET oscillation that's causing this problem, no?

[phild13]

John I went off the fact that when the FET was reversed the display on Frank_tt's clock worked ok without missing segments. Although that could be just the body diode conducting, I would think that would more or less rule out an issue with the processor or VFD chip. The voltages measured would indicate the FET is not fully conducting for some reason especially since the processor is pulling the gate to 0v.