On Saturday, August 13, 2022 we will be working on the Adafruit Customer Support Forums. If you visit over the weekend and things are not complete yet, please check back later, https://adafruit.com/forumupdates
0

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

Please be positive and constructive with your questions and comments.

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Tue Jan 21, 2014 1:04 pm

PhilD13 wrote: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.

I agree that the FET is not fully conducting, but that was pretty much what I measured on my Adafruit clock. I believe this is normal behavior and is sufficient to drive the filament on the majority of tubes, although occasionally a tube will exhibit the dim digit problem.

Since all segments use the same filament, the problem should be similar across all segments if low filament power is the problem. Nearby segments should fail similarly because they share a filament, but that isn't the behavior described. The problem must involve the grids and segments, which are controlled by the MAX chip and the microcontroller.

I do not mean to suggest that microcontroller or VFD driver chip are bad. As you noted, reversing Q3 did solve the problem. But I was trying to provide an explanation of how an oscillating, partially-on Q3 could be affecting the behavior of one of the chips. Reversing the diode would cause conduction through the body diode, which should reduce or eliminate the oscillation that I suspect is generating too much power noise, and causing dodgy chip behavior.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by phild13 on Tue Jan 21, 2014 1:55 pm

Ok, I understand your point now.

phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Tue Jan 21, 2014 1:57 pm

I found another thread with display trouble that seems related. The problem was solved by swapping out Q3 with a more reliable switch. This case is similar in that only one segment showed, indicating the problem was on the anode (segment) side of the tube:

redwire wrote:I had odd troubles after assembling my Adafruit ice tube clock display.
The symptom is only one segment lights! Occasionally, the display flickers and shows the time or fades in/out. Soldering/parts placement is fine. Oddly, putting my DVM from GND to JP2-2 got the display to light up steady...
I took some measurements, and found the P-channel mosfet Q3 (ZVP3306A) is lazy, it is only partially on. Verified pinout- the rounded-side matches the pcb silkscreen. Source=4.72V Gate=0V Drain=3.06V (*too low); Vin=9V, Vcc=4.72V, Vboost=14V, Vfilament=2.0V (*low)

Fix: Swapped in a venerable 2N3906+1k base resistor and the display lights now (RBase= 1k gave 4.56V out; 10k gave only 3.76V out, B=200)

Looked at datasheet and Q3 (mosfet) is a dog, as finding TO-92 P-ch mosfets with low-threshold is very difficult.
http://www.diodes.com/datasheets/ZVP3306A.pdf
FJN4301R, PDCT123T, DTA114 digital transistors might work, but pinout is different.

Just letting people know, as the symptoms were pretty strange and that mosfet is not best suited for this.


For me, this was encouraging news for Frank and anyone else who decided to try replacing Q3 with a ZVP2110A.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by phild13 on Tue Jan 21, 2014 3:10 pm

That is one reason why I suggest (someone else originally suggested it) the ZVP2110A as a replacement for Q3, and also gave the link to the option of changing the FET to a transistor/resistor modded to fit in the same space as the Q3 FET occupied which will also work just fine as a replacement. viewtopic.php?f=41&t=43908&p=219063&hilit=transistor#p219063 According to Russell 27, the parts are available from most Radio Shacks so that would be a quick fix for those that have a local Radio Shack and don't want to have the hassle and/or wait time of a new or different part.
Radio shack should have a 2N3906 in store and I think should work in place of the PN2907A though I did not check the spec sheets.
http://www.radioshack.com/product/index ... Id=2062585
Pack of 1/8 watt 1K resistors
http://www.radioshack.com/product/index ... Id=2994579

phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Tue Jan 21, 2014 4:49 pm

PhilD13 wrote:That is one reason why I suggest (someone else originally suggested it) the ZVP2110A as a replacement for Q3

I think I found the first ZVP2110A post by HolstCentre. It describes "non working display segments" caused by "some spikes coming from the microprocessor, corrupting the timing sequence of the display driver chip." A clever arrangement of capacitors smoothed out the spikes and solved the problem.

Unfortunately, HolstCentre was still left with the dim digit problem, which was solved by replacing Q3 and reducing the resistance of R3. (Because the Adafruit-provided Q3 does not turn on fully, voltage across the filament is insufficient to fully heat the filament on some tubes. The result is usually a final digit that is significantly dimmer than the others. This is the "dim digit" problem that I mentioned.) I'll wager that if HolstCentre had replaced Q3 with a ZVP2110A first, that substitution would have solved both problems at the same time.

The electrical noise from Q3 could be caused by parasitic oscillation, and part-to-part variation in the FETs (some would be more prone to oscillation) and microcontrollers (some would be more susceptible to electrical noise) could explain why most people don't experience this problem. On the other hand, using a ZVP2110A for Q3 would ensure that Q3 turns on fully. Small oscillations at the gate would have no effect on the output, as Q3 would remain in complete saturation. A constant output from Q3 would in turn prevent prevent the induction effects that cause parasitic oscillation in the first place. And the reduction in electrical noise would ensure reliable microcontroller operation.

Still just speculation, but I think it's a pretty good guess at this point.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by jaysonl on Wed Jan 22, 2014 10:24 am

Hi,adafruit_support_bill.

Sorry for the delay in replying, here are pics as requested.

I noticed some discussion in this thread about potentially faulty FETs. Would it be possible that this is the case? How would one test the FET?

--Jays
Attachments
DSC00932_cropped.jpg
DSC00932_cropped.jpg (915.14 KiB) Viewed 731 times
DSC00931_cropped.jpg
DSC00931_cropped.jpg (910.81 KiB) Viewed 731 times
DSC00929_cropped.jpg
DSC00929_cropped.jpg (996.56 KiB) Viewed 731 times

jaysonl
 
Posts: 5
Joined: Sat Jan 18, 2014 3:39 pm

Re: Help with a weird problem on Ice Tube Clock

by phild13 on Wed Jan 22, 2014 1:24 pm

The easiest way to test is to jumper the two outside leads on the fet. This will remove it from the circuit. If the clock display works good after jumping the fet, then the fet is likely bad for some reason. Don't short the middle lead to anything.

Another way is to carefully measure the voltage between each outside lead of the fet and the ground tab of the regulator while the clock is powered and the tube is inserted. I have 4.622 on the inside lead and 3.936 on the outside (nearest the board edge) lead, so your voltages should be near those. I am using a ZVP2110A instead of the ZVP3306A, so your voltages may be a bit different so post them if you like and someone will let you know if they are to low.

I don't see anything wrong with the soldering.

phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Help with a weird problem on Ice Tube Clock

by jaysonl on Wed Jan 22, 2014 4:48 pm

Hi, PhilD13!

Thanks! I jumpered the outer leads of Q3 as per your instructions, and yes, the tube illuminates fine.

Looks like I need a new ZVP3306A?

I noticed mention of replacing this component with a ZVP2110A instead... but my local will-call electronics retailer (Jameco) doesn't seem to stock that item (but they DO stock ZVP3306As)... are there any MOSFETs that they DO stock that would be better suited for the job? I only ask because if I place my order within the next couple of hours, I can go pick it up by EOB and be done with it.

Thanks again,
--Jays

jaysonl
 
Posts: 5
Joined: Sat Jan 18, 2014 3:39 pm

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Wed Jan 22, 2014 5:11 pm

PhilD13 wrote:I have 4.622 on the inside lead and 3.936 on the outside (nearest the board edge) lead, so your voltages should be near those. I am using a ZVP2110A instead of the ZVP3306A, so your voltages may be a bit different so post them if you like and someone will let you know if they are to low.

I believe the normal Q3 voltages for the Adafruit part (ZVP3306A) are ~3.4v (outside), ~0v (center), ~4.7v (inside).

If my speculation is correct about parasitic FET oscillation causing display failures, you might get normal Q3 measurements, and Q3 could still be problematic. So I suggest doing both of Phil's tests. If the voltages on Q3 look good, but adding the jumper allows the display to function, Q3 could still be bad. Please keep in mind my speculation is just a theory supported by some circumstantial evidence; I could be wrong.

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

The quote above is from a previous post comparing a working board to a faulty board. I suspect that these are the approximate normal voltages for an older kit which used a PN2907A for Q3. The PN2907A was replaced by the ZVP3306A because, without a resistor at the base, the PN2907A pulled too much current from the microcontroller, occasionally causing faulty operation. The Q3 center pin is likely forced to 4v due to excessive current draw from the microcontroller.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Wed Jan 22, 2014 5:18 pm

JaysonL wrote:I noticed mention of replacing this component with a ZVP2110A instead... but my local will-call electronics retailer (Jameco) doesn't seem to stock that item (but they DO stock ZVP3306As)... are there any MOSFETs that they DO stock that would be better suited for the job?

The other solution known to work is the PN2907A & resistor that Phil mentioned. I'm not aware of other suitable MOSFETs, but maybe Phil is.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by Frank_tt on Wed Jan 22, 2014 9:15 pm

OK all, I replaced Q3 with ZVP2110A as recommended and it atleast fixed the fading segments. The clock stops when power is pulled but it could be that the battery has run down some. I will get a new one and see if that fixes that or do I have yet another problem?

It does remember the time when the power was pulled but doesn't advance it without external power.
Frank_tt
 
Posts: 13
Joined: Mon Dec 30, 2013 11:16 pm

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Wed Jan 22, 2014 10:18 pm

Frank_tt wrote:The clock stops when power is pulled but it could be that the battery has run down some.

The reason the clock didn't keep time when Q3 was reversed was that the body diode in Q3 was conducting even when the clock was asleep. So when the clock was unplugged, the battery was outputting as much power as it possibly could through the VFD filament, and those conditions will kill the battery in no time. Replacing the battery should fix the problem.

EDIT: I suppose it is possible that there is another problem preventing sleep from working properly, but I think it's much more likely that your battery is simply dead.

Frank_tt wrote:It does remember the time when the power was pulled but doesn't advance it without external power.

This is normal behavior with a dead battery. The power caps store enough energy for the clock to detect the power loss and save the current time to EEPROM. When power is restored, the last known time is restored from EEPROM.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by jaysonl on Thu Jan 23, 2014 1:55 am

Hey, all.

I just wanted to report back in... I decided to go the PN2907A + resistor route; it seems to be a resounding success! Good thing, too... the pads on Q3 were starting to lift.

The tube lights up evenly and without flicker, even at the lowest brightness setting. I can't say for sure whether the battery backup is functioning as intended, as I currently lack the desire to either run down the lithium coin cell prematurely (finding replacements can be irksome) or open the enclosure back up (since I went through the trouble of cleaning off the fingerprint smudges on the inner surface), but I presume it is.

The transistor set me back a whopping five cents, and I already had the resistor on hand, so this easily qualifies as cheapest. repair. job. evar.

It appears that many folks are having issues with the stock Q3 FET, not to mention the "last set of digits are dimmer than the rest at low brightness settings" bit in the Ice Tube Clock FAQ... have the folks at Adafruit put any consideration into rev bumping the kit?

Anyways, back to admiring the cyan glow of the Soviet tube at the heart of my new timepiece! Thanks again for the helping hand.

--Jays

jaysonl
 
Posts: 5
Joined: Sat Jan 18, 2014 3:39 pm

Re: Help with a weird problem on Ice Tube Clock

by jarchie on Thu Jan 23, 2014 2:29 am

JaysonL wrote:I just wanted to report back in... I decided to go the PN2907A + resistor route; it seems to be a resounding success!

Awesome! I've used both the PN2907A/resistor substitution and the ZVP2110A substitution. Both work extremely well.

JaysonL wrote:It appears that many folks are having issues with the stock Q3 FET, not to mention the "last set of digits are dimmer than the rest at low brightness settings" bit in the Ice Tube Clock FAQ... have the folks at Adafruit put any consideration into rev bumping the kit?

I've been hoping that Adafruit would revise their kit ever since establishing the cause of the dim digit problem almost a year ago. Maybe the recent bout of Q3 issues will prompt an official solution? <fingers crossed> In your case, the PN2907/resistor substitution will provide sufficient voltage to the filament to prevent the dim digit problem. It will also make your tube less prone to cathode poisoning.
--John <www.jarchie.com/email>

jarchie
 
Posts: 608
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Help with a weird problem on Ice Tube Clock

by jaysonl on Thu Jan 23, 2014 7:39 pm

jarchie:

Excellent to hear about the whole "preventing cathode poisoning" thing... I want this tube to have a long, happy life.

I noticed that your profile info says you're from Santa Cruz... greetings from south San Jose! I feel like I owe you a beer for pointing me towards this mod... do let me know if you ever want to meet up at 99 Bottles for a pint!

--Jays

jaysonl
 
Posts: 5
Joined: Sat Jan 18, 2014 3:39 pm

Please be positive and constructive with your questions and comments.