Ice tube clock battery backup problem

[russell 27]

Recently put together Ice tube clock, had same problems with battery backup as others. For the record, don't know it all, but I've been an electronics man for a long time. At first all good with clock. First time I pulled power, battery back up did not work. Looked at board and schematic. The suggested mounting for the ZVP3306A P channel MOSFET is NOT CORRECT. DRAIN and Source are reversed. This MOSFET uses a body diode. With Drain connected to +5 rail, MOSFET diode is forward biased and will conduct current to VFD if MOSFET is on or off. Result of this; at power loss CV1220 Battery is quickly drained by VFD and cannot supply current to Atmega to keep correct time. So I removed ZVP3306A, which is not easy on this board, and replaced with new ZVP3306A. At first this seemed to correct the problem. small power loss time span, clock kept correct time. Left unplugged for several hours, power up, again, time lost, I had installed a new CV1220 Battery. This time I noticed display had lost brightness( same brightness setting), and overall radiance between digits was not even as had been before. I did not originally have any problems with display digits varying in brightness as others did. I jumpered drain and source, display regained original brightness, and digits were even. I can try and check the MOSFET again. A few things. The original circuit used a PNP transistor, and this transistor cut out the MAXIM and the vfd display, according to the schematic. The current board seems to only cut out the VFD. Also the ZVP3306A can only source 160 Ma, seems like enough but..... Bottom line is older kits seem to have used a 2N2907 PNP transistor, and there were no problems with the kit then. Is this MOSFET not right for the job ? If I had built this using the open source I would have expected bugs to be worked out. I bought the kit form for ease of construction, didn't expect all this. Especially wrong installation of MOSFET. Any input would be appreciated. If there is another place to work this out directly please advise. There is a problem here.

Russell

[russell 27]

If anyone with an older kit can tell me what was used instead of the ZVP3306A_P Channel CMOS I would appreciate it.

Thanks

Russell

[russell 27]

I believe I have answered my own question. All of this was covered some time ago. Not only was the old kit design a little lacking with the battery backup circuit, but the new MOSFET fix didn't cut it either, not to mention mounting was reversed, which is really a problem. Nothing personal here, I LOVE ELECTRONICS. I bought this kit as ready to go, nothing suggested otherwise. I'm not talking programming here, circuit wise. OPEN SOURCE would have been the way to go. BANNED.

[adafruit_support_rick]

Thanks for the feedback. I'll make sure to pass this along to the engineering team.

[sparkejf]

I've just completed the Ice tube clock. Everything works perfectly except the battery backup. When I remove the clock from the main power source, it goes blank. I removed and resoldered the battery holder. Added more solder to the center tab. Made sure that the schottkey diode was facing the correct direction. Replaced the battery with a new one. Nothing. I am a novice when it comes to electronics. I guess I'm just not going to have a battery back up.

[adafruit_support_rick]

This battery backup problem is a new issue - we're still trying to figure out what's going on

[phild13]

@Russell
Your findings are exactly what I found when comparing the schematic and the pin out of the FET Q3. That the instructions to turn the ZVP3306A around are wrong and will connect the drain to +5v, causing the fets body diode to conduct draining the battery when on battery power.
You can see my post here on this issue along with instructions on how to reverse the fet so the source is connected to +5volts:
http://www.adafruit.com/forums/viewtopi ... 41&t=43209

The 2N2907A (in a TO92 package P2N2907A) was the original transistor used. If I can find the source again I will post it for reference.
P2N2907A Datasheet:
http://media.digikey.com/pdf/Data%20She ... 20Rev4.pdf

The ZVP2110A FET is a better fit for the clock at 230 mA and 8 ohm on resistance vs. 160 mA and 14 ohm on resistance for the ZVP3306A ) and it works good. I would suggest changing the fet in the parts to this one.
ZVP2110A datasheet
http://www.diodes.com/datasheets/ZVP2110A.pdf

And of course the current fet datasheet for comparison.
ZVP3306A
http://www.diodes.com/datasheets/ZVP3306A.pdf

Changing the 22 ohm resister to a 11 ohm resister or a straight jumper (easiest and won't hurt anything) in place of the 22 ohm resister and reportedly helps some clocks that have issues.
I think using a 22 ohm resister and the ZVP3306A (14 ohm on) is to much of a voltage drop on the bias for some tubes.

The current schematic on GitHub shows the FET Q3 as connected to the VCC of the MAXX vfd chip and to the 22 ohm resister. Q3 is switching +5 volt power on or off to the MAX vfd chip and also the tubes bias pin which is what the other end of R3 (the 22 ohm resister) is connected to. Gate drive for Q3 comes directly from pin 5 of the ATmega168. A low signal on the pin turns the fet on, and a high signal turns it off. I checked my recently ordered kit after I received it against the schematic and it looks to be the same so I don't think there have been any revisions to the board.

Couple other things I found
The fet is very static sensitive which will damage it causing it to not work correctly, so don't touch the leds with your fingers when handling it and it is best to not lay it on the work bench unless it is wood. I usually just drop it in an antistatic bag for safe keeping. A note should be put in the instructions about properly handling static sensitive devices such as fets, vfd chips, processors.

An Overcurrent will also damage it and I found damage in such a way as it will not fully turn on or off.

Excessive heat will damage the fet. It is much easier to damage it during desoldering than when soldering it. If the solder pads delaminate from the board during desoldering, too much heat was used for to long and the fet is likely bad. I recommend clipping the leds flush with the top of the board and desoldering the pieces left. There is usually enough led left to solder it back in the right way.

[phild13]

I've just completed the Ice tube clock. Everything works perfectly except the battery backup. When I remove the clock from the main power source, it goes blank.......

It is supposed to do that. Upon power loss the processer records the current time at power loss and goes on battery. The processer will keep up with the correct time so when power comes back on the clock can display the correct time. The display will not remain on when on battery as Q3 should turn off the display circuitry. The clock should keep the correct time while on battery for at least a week. When plugged back in the display will flash with the correct time of day. If I remember correctly, pressing a button on back should stop the flashing. If the clock does not keep the correct time then there may be an issue with the Q3 component or the associated circuitry, which is what is being reported recently in the forums.

[phild13]

Attached is a screenshot of the schematic showing the 2N2907A transistor instead of a fet. I don't think it or the P2N2907A is available any more but you should be able to find one like it.

EDIT: Here is the link to the version 1.0 schematics that I made a screen shot of. http://learn.adafruit.com/ice-tube-clock-kit/downloads

For those following the thread, I would recommend the 1.0 version only for reference unless you really know what your looking at and doing.

The current version 1.1 schematics are located at https://github.com/adafruit/Ice-Tube-Clock and are what are currently being used for the kits.

Also, here is a thread detailing some findings (there are other discussions also) on the use of the ZVP3306 fet, and the possibility of the fet possibly causing a low voltage condition that may cause issues. http://www.adafruit.com/forums/viewtopi ... 45#p190790

[russell 27]

Thanks for your reply fellas. I did some testing, still working on it. There is nothing wrong with using a MOSFET, actually it's ideal, this one seems to be lacking. For the moment, I removed the MOSFET and jumpered the DRAIN and SOURCE pads together with a jumper loop. This gets rid of any display problems. I believe this MOSFET is either not turning on fully or can't supply full current. The tube really needs close to 5 volts on its filament anode to illuminate segments fully and evenly. Try the jumper and see, You can carefully do this with the MOSFET installed. Of course the battery backup will not work with the jumper. At the moment I think the best option, sans a better MOSFET, is using a PNP (2N2907, 2N3906) with a base resistor of about 1K. Changing the 22R resistor may also help, but I don't know if it's needed. You posted the link about redwire's findings, I saw that too, totally agree. The hardest part is finding a nice place for the resistor to fit. I thought about using an 1/8 watt Ik resistor, and soldering it to the base lead in a U shaped bend, to almost make a single part, that would fit the available pads. The transistor would fit right in, Emitter to source pad, base via 1k to gate pad, collector to drain pad. I don't know what was originally done, a transistor there without resistor would also cause problems. If it works out I'll let you know.

Russell

[russell 27]

PhilD13

Did you use the ZVP2110A, Cut it ? I've got a gigantor IRF9520 that has worked splendidly in other projects, fits probably not an option, interesting to see what it does.

[phild13]

I used the zvp2110a and it works fine. I also used an 11 ohm resistor for R3 along with the fet change and the combo works good.

[sparkejf]

Thanks PhilD13. I didn't understand that the battery backup didn't keep the clock running. Unfortunately, when I plug the wall power back in, it does not display the correct time. I'm assuming that it is a problem with the Q3 component. I'll wait and see if someone smarter than me can figure it out. Thanks for replying to my post.

[phild13]

Thanks PhilD13. I didn't understand that the battery backup didn't keep the clock running. Unfortunately, when I plug the wall power back in, it does not display the correct time. I'm assuming that it is a problem with the Q3 component. I'll wait and see if someone smarter than me can figure it out. Thanks for replying to my post.

You can see my post on how to check and verify if Q3 is doing what it is supposed to and if not, and a method that can be tried to fix the situation. Besides the information in the first post in this thread we are currently in ( http://www.adafruit.com/forums/posting. ... 0#pr217743 ), my post also contains some information on the reason the clock won't keep time while on battery so you can understand a bit better the "why" of the issue.
http://www.adafruit.com/forums/viewtopi ... 41&t=43209