Ice Clock - hour digit blank only when it's an 8 Moderators: adafruit_support_bill, adafruit

[mqueue]

The right digit of the hour digits is blank only when it's 8am or 8pm (or 1800 when in 24-hour mode)

Has anyone else had this issue?

Everything else works great. Initially, some digits were flaky, but it turned out to be the 9VDC 1000mA adapter that I bought from adafruit. I swapped it out with an old 9VDC 300mA adapter, and then the display worked properly.

[adafruit_support_bill]

That is an unusual one. I haven't seen that problem before. If you can post some clear photos of the front & back of the main board and the solder-side of the tube board we can chack that over.

Initially, some digits were flaky, but it turned out to be the 9VDC 1000mA adapter that I bought from adafruit.

What was the problem with the 9v adapter? Was it not putting out 9v?

[jarchie]

The right digit of the hour digits is blank only when it's 8am or 8pm (or 1800 when in 24-hour mode) Has anyone else had this issue?

I haven't seen this exact issue, but flaky digits are often caused by a flaky Q3. As bill suggested, it's worth posting pictures of the board. But if you're convinced the problem isn't with the soldering, it might be worth replacing Q3 with a ZVP2110A from Digi-Key or Mouser.

Everything else works great. Initially, some digits were flaky, but it turned out to be the 9VDC 1000mA adapter that I bought from adafruit. I swapped it out with an old 9VDC 300mA adapter, and then the display worked properly.

Were you able to determine that the adapter was faulty? Or was it just that replacing the adapter seemed to fix the problem? If it's the latter I'd wager that Q3 is the problem.

[mqueue]

I will disassemble the clock and get some photos posted as soon as I can.
I measured 9V from the original adapter, so I'm not sure why the erratic digits occur when I use it. I'll try it again just to be sure.

[jarchie]

I will disassemble the clock and get some photos posted as soon as I can.
I measured 9V from the original adapter, so I'm not sure why the erratic digits occur when I use it. I'll try it again just to be sure.

In that case, it seems like both adapters are working and that switching adapters causes the flaky digit problem to become far less severe (e.g., only affecting the one "8").

I don't think anyone's figured out for certain why a faulty Q3 causes the flaky digit problem, but I've speculated that a faulty Q3 causes power ripple which in turn causes the ATMEGA chip or MAX chip to misbehave. If my speculation is correct, it makes sense that changing part of the power supply (the adapter) could also effect the power ripple and hence the severity of the problem. My guess is that replacing Q3 will fix the problem.

[phild13]

I would be interested in what turns out to be the solution and if a Q3 replacement is the solution.

I would speculate that since it is winter this means much more static than in warmer more humid times of the year when it is less of an issue. This might account for what seems to be a larger than normal reports of flaky Q3 FET's.

FET's hate static and can be easily damaged by static charges while handling. Simply removing the FET from its package, picking it up, or placing it on the work surface can cause damage to the FET. You don't have to zap the FET with an arc (like what sometimes happens when reaching for a doorknob) to damage it. Your body can easily have a large enough charge to damage the FET and other electronic parts such as the VFD chip and processor.

[jarchie]

I would speculate that since it is winter this means much more static than in warmer more humid times of the year when it is less of an issue. This might account for what seems to be a larger than normal reports of flaky Q3 FET's.

That makes sense. The kit-provided Q3 is only partially on during normal tube operation, so any change in power level will effect how much current Q3 provides to the VFD filament and MAX chip. That would make the kit-provided Q3 especially prone to power noise and oscillation issues, but begs the question of why only some clocks have the flaky segment problem. I suspected that an occasionally marginal or faulty Q3 was causing the problem, but minor ESD damage seems like a more plausible explanation. I suspect that replacing Q3 with a transistor that turns on fully (like a ZVP2110A) would make the design more robust against the flaky segment problem.

I will disassemble the clock and get some photos posted as soon as I can.

In my opinion, that's the best way to proceed--just to rule out any obvious soldering or component placement problems. I've seen people with much more experience than myself make assembly errors and miss them when examining their own boards. For whatever reason, some issues are only obvious to a second set of eyes.

If there are no assembly issues, Adafruit support will probably offer to send you a replacement Q3 for free.

But my unofficial advice is to buy a different part to replace Q3 from Digi-Key or Mouser: a ZVP2110A. The advantage of the ZVP2110A is that it will turn on fully. In addition to making the clock more robust to flaky segments (my speculation), the ZVP2110A will provide more voltage to the VFD filament and prevent two other problems that occasionally effect Ice Tube Clocks: First, the dim digit problem is where the initial and/or final digit appears significantly dimmer than the others and is caused by the low filament voltage in the Adafruit design. Second, low filament voltage also accelerates cathode poisoning, which can shorten tube life and make the 3rd and 6th digits appear dimmer than the others. A clock with both issues is pictured in the initial post of this thread.

Another solution with the same benefits is to replace Q3 with a PN2907A and resistor. Although a bit less elegant that replacing Q3 with a single transistor, a PN2907A and resistor need not be ordered online. Your neighborhood Radio Shack should carry them.

But I'm getting ahead of myself... I still recommend posting photos of the front and back of your boards before doing anything else. The problem might be obvious to a second or third set of eyes and, with luck, will have an easy fix. :-)

[mqueue]

It works!
I took the clock apart so that I could take pictures of the board.
I connected the original power adapter (1000mA) to verify if the display was still flaky with it.
The display looked good though! So, I changed the time to 8 o'clock, and the 8 digit worked!
I hooked up the power adapter I was previously using (300mA), and the 8 digit no longer worked.

So, I'm not sure why the display was acting flaky with the original adapter, but that's no longer an issue.
And, the 8 not working was caused by the 300mA adapter I was using -- not sure if that was due to being under powered or having noisy power or what...

But, all is well now! Thank you!

I may try changing out that Q3 some day because the 1st digit flickers slightly and the 6th digit is dimmer than the rest.

All the best to all of you.

[jarchie]

But, all is well now! Thank you!

You're very welcome, although I suspect the problem might be intermittent and periodically resurface.

I may try changing out that Q3 some day because the 1st digit flickers slightly and the 6th digit is dimmer than the rest.

Just to satisfy my own curiosity, I would be willing to mail you a ZVP2110A to see if that fixes the slight flickering and evens the digits. I'd just need a mailing address; visit the URL in my footer to get my email. And the only catch is that I would ask you to post your results back to this thread (e.g. if the ZVP2110A fixes your problem or not).

[mqueue]

jarchie, you were right about the problem being intermittent.
One day after the "8" started working, it stopped working again.

Today, 26 Feb, I installed the component that you sent (ZVP2110A). The display looks much better!. The right-most digit is still just a tad dim, but the left digit is as strong looking as the rest. And, none of the digits are flickering any longer.
I will wait a few days to see if the "8" stays stable and report back.
For now, thank you! It looks very promising. 8)

IMG_0926.jpg (61.13 KiB) Viewed 844 times

[jarchie]

And, none of the digits are flickering any longer. I will wait a few days to see if the "8" stays stable and report back.

I agree; waiting a few days (at least) before pronouncing the problem fixed is smart.

The right-most digit is still just a tad dim, but the left digit is as strong looking as the rest.

The rightmost digit appearing dimmer is probably an artifact of your particular IV-18 tube and the way the clock drives the tube. Once you're absolutely sure the "8" problem is fixed, there are a couple things you could try to brighten the last digit:

A dim final digit is often caused by low filament current, so replacing R3 with a smaller resistor or a jumper might help. Some people find it easier to leave R3 in place and solder a second resistor or jumper below the board in parallel to R3. Although increasing current should help with that final digit, increasing current too much might introduce a brightness gradient across the display. It may take some trial and error to find a happy medium.

Another approach is to install an alternative firmware. My xmas firmware allows the user to adjust the brightness of each individual digit to compensate for uneven displays. To install xmas yourself, you'd need an ATmega328p and a programmer, such as the Adafruit USBtinyISP. Alternatively, I could mail you a preprogrammed ATmega328p. The ATmega328p is a drop-in replacement for the ATmega168v that comes with the kit, so you would pull the original 168v and insert the 328p.

For now, thank you! It looks very promising. 8)

You're very welcome!

[phild13]

Replacing R3 with an 11 ohm resistor or soldering a 22 ohm resistor on the bottom in parallel (which gives 11 ohms total resistance) with R3 will be a good compromise of current and help even out the digits.

[mqueue]

Day 2: All digits are healthy and bright -- including the rightmost one! I'm not sure why, but the rightmost digit seems to have fixed itself. Maybe it needed to warm up a bit? It was only a tad dimmer than the others.

[jarchie]

I'm not sure why, but the rightmost digit seems to have fixed itself. Maybe it needed to warm up a bit? It was only a tad dimmer than the others.

I believe you are correct. The dim digit issue is temperature dependent, so I suspect the clock literally needed to warm up a bit.

The IV-18 tube filament is intended to be run under 5 volts of center-tapped alternating current, but the Adafruit design drives the filament with just under half that voltage and direct current. As a result, the filament sometimes doesn't get hot enough at the ends where the metal attachments can pull heat away to the outside wires. And when the filament doesn't get hot enough, it won't emit enough electrons to fully light the nearby segments. When the tube is cold, the problem is worse.

You've already replaced Q3 with a ZVP2110A--a transistor that turns on fully. The voltage across your VFD filament should be around 3.1 volts which eliminates the dim digit problem on the majority of IV-18 tubes. Phil's suggestion of reducing the resistance of R3 to 11 ohms should increase the filament voltage to around 3.6 volts. (That's just an estimate... There's significant variation among IV-18 tubes, so the exact filament voltage depends on the particular tube used.)

I've probably used a couple dozen IV-18 tubes for various purposes--mostly Ice Tube Clocks and my hardware revisions. Intuitively, I think that driving the filament with 3.6 volts direct current, as Phil suggests, should work well for most--if not all--tubes. On some tubes, there might be a very slight brightness gradient at low brightness, but I doubt most people would notice. But most importantly, the dim digit problem should not be an issue. And 3.6 volts direct current should eliminate any potential cathode poisoning issues as well.

So here's my opinion: Assuming the "8" problem is indeed fixed, I think your clock is fine as-is, but if the occasionally dim final digit bothers you, try Phil's suggestion. (Besides, Phil has much more electronics experience than I do, so you should trust his opinion more than mine anyway.)

[mqueue]

Day 4: All digits healthy and bright. I'm calling it cured. :-) Cheers.