0

Ice Cube Clock - Low HV
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Re: Ice Cube Clock - Low HV

by phild13 on Wed May 14, 2014 11:23 am

Glad you have it working!

I based the suggestion of base resistor on the tests and results done by Russell which you can read here ICE TUBE Q3 TEST: MOSFET or PNP TRANSISTOR

I can confirm a 1K base resistor presents no issues for the clock.

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

Re: Ice Cube Clock - Low HV

by Barry914 on Wed May 14, 2014 11:35 am

I guess I should have read that thread too. But I never thought using a 1K base resistor would be a problem, just thought it was driving the base a bit harder than it needed to saturate the CE. But I stand corrected.

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by jarchie on Wed May 14, 2014 5:30 pm

Barry914 wrote:...just thought it was driving the base a bit harder than it needed to saturate the CE. But I stand corrected.

I'm not so sure you're wrong. From Russell's results, I suspect a 10k would be alright. Using a 10k is certainly an improvement over the Adafruit-provided ZVP3306A, and the ZVP3306A works fine for most people.

But the ~1k solution provides more voltage/current to the VFD filament, has been tried by a handful of users, and has always worked well. To my knowledge, no one has tried a 10k in an actual Adafruit clock. Although a 10k would probably work fine, I suggest going with the solution that is known to work well.

Barry914 wrote:[From the "Snap a picture of your ADAFRUIT clock and post it here!" thread:] And solving the little problem I had with Q3 actually added to the fun. How boring if it had just turned on and worked the first time.

If you ever decide to have a bit more fun with the clock, there are plenty of alternative firmwares, hacks, and mods to play with. If that interests you, my xmas-icetube firmware might be a good starting point, as the documentation is decent.

jarchie
 
Posts: 557
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Ice Cube Clock - Low HV

by Barry914 on Thu May 15, 2014 12:40 am

Isn't the additional filament current mostly the base current's contribution? With an Hfe of over 200 (probably over 300) the transistor is well into saturation with just a few hundred microamps of base current. Using a lower value resistor won't really hurt anything, but it does force the MCU to source more current than it needs to. In any case, I'd recommend using a bipolar transistor over the MOSFET to anyone building the kit just to avoid problems. If the MOSFET it fully turned on and has an Rds of 10 ohms, you're going to get a voltage drop at least twice what you would expect from a saturated bipolar transistor. Other than the somewhat less than elegant way of adding the resistor, I don't see a down side to the substitution. Or am I missing something?

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by jarchie on Thu May 15, 2014 2:43 am

Barry914 wrote:Isn't the additional filament current mostly the base current's contribution?

The base goes low to turn the transistor on, so the base is pulling current from the supply--not providing current to the filament.

Barry914 wrote:I'd recommend using a bipolar transistor over the MOSFET to anyone building the kit just to avoid problems. If the MOSFET it fully turned on and has an Rds of 10 ohms, you're going to get a voltage drop at least twice what you would expect from a saturated bipolar transistor.

This assumes that minimal voltage drop is a good thing. The filament is intended to be driven with a full 5 volts of center-tapped alternating current. But if you drive the filament at a full 5 volts of direct current, there is a marked brightness gradient across the display at lower brightness settings. The apparent role of R3 is to reduce filament voltage which, in turn, reduces the brightness gradient issue. There's no consensus on how much voltage should be used to drive the filament with direct current, because the tube was not designed for direct current on the filament.

jarchie
 
Posts: 557
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Ice Cube Clock - Low HV

by Barry914 on Thu May 15, 2014 9:58 am

Yes, you are correct about the base current not directly adding to the collector current. I could come up with second order effects of high base current that could effect the CE saturation voltage and/or collector current, but I think I'd be reaching.

I have to assume the 22 ohm resistor was selected to provide optimum voltage/current to the filament. The original design used a bipolar transistor, but from what I have read it was changed to a MOSFET because the excessive base current w/o the series resistor caused the MCU to fail. The problems related to Q3 seem to be from variations in the device's characteristics, causing some MOSFETS to not turn on completely or because of high Rds. Using a bipolar transistor removes that variation, and adding the base resistor fixes the base current issue. At the very least, you would expect very little variation from kit to kit if you go back to the bipolar device. Perhaps changing the MOSFET would address that issue, but it seems to be a less reliable fix, at least in my mind. I would expect a larger variation in Rds from device to device and over temperature than the variation in saturation voltage of bipolar transistors, which is quite low to begin with.

I want to be clear that none of these opinions come from analysis of the data sheets or from testing. This is all from my gut and I would not be terribly surprised if I am way off base.

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by phild13 on Thu May 15, 2014 2:50 pm

At the very least, you would expect very little variation from kit to kit if you go back to the bipolar device.
I would agree with that and have actually made some minor design modifications to the Adafruit board design to accommodate the base resistor and it works very well. Good part about the modification is one can use either a transistor/base resistor or a FET/jumper for Q3 and everything still fits within the original case without any mods as the board has essentially the same footprint as the production Adafruit board. While the effort was mostly to play with Eagle and make a few boards, it does make it a bit easier with less chance of damage from heat than soldering the base resistor to the transistor. There is also much less chance of damage from static charges when handling Q3 when using a transistor.

I also agree that the first design had a transistor for Q3 and no base resistor which did cause too much current to be pulled through the processor pin. Easiest fix was to replace Q3 with a FET though the FET selected was very marginal for the application.

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

Re: Ice Cube Clock - Low HV

by jarchie on Thu May 15, 2014 5:23 pm

I think we're all on the same page here, although a PFET that turns on fully--like the ZVP2110A or BS250--still seems like an acceptable choice. A PFET might introduce more kit-to-kit variation, but that seems acceptable. The IV-18 filament works well over a fairly large voltage range. I generally recommend the ZVP2110A because it avoids cramming two parts into a space designed for one, and the ZVP2110A has worked well for the half dozen or so users who have tried it.

Also, I wouldn't worry about driving the chip or BJT too hard with a 1k resistor at the base. At 5 volts, the base current will be around 5 mA (5v / 1k), and the ATmega168v and ATmega328p pins are rated for up to 20 mA. The BJT will only need to dissipate an extra 25 mW or so (5v * 5mA), and TO-92s can generally dissipate up to ~600 mW. There's a comfortable safety margin with respect to both parts.

jarchie
 
Posts: 557
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Ice Cube Clock - Low HV

by Barry914 on Fri May 16, 2014 12:01 am

I agree we agree. And I did use a 1K base resistor even though I have a supply of other values in my parts bin. I was a bit surprised that the Q3 issue is relatively rare. From the number of posts on this subject I thought it was much more prevalent. A friend asked me about getting a kit for her husband and I was going to advise him to use the bipolar/resistor combo instead of the MOSFET. Am I overreacting?

One more thing. I am compelled to split hairs - the dissipation due to the base current is less than 3 mW. This assumes Vcc = 4.7 V and Vbe is 0.6 V.

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by adafruit_support_bill on Fri May 16, 2014 6:48 am

I was a bit surprised that the Q3 issue is relatively rare. From the number of posts on this subject I thought it was much more prevalent.

Looking at forum posts can give a somewhat distorted view. What we see here are mostly the ones that had problems. If we had a post for every clock that went together without a hitch, we wouldn't have time to read half of them.

adafruit_support_bill
 
Posts: 73190
Joined: Sat Feb 07, 2009 10:11 am

Re: Ice Cube Clock - Low HV

by Barry914 on Fri May 16, 2014 1:41 pm

I was discussing your company with a friend, mostly about your tech support and customer service. You guys aren't selling toasters, you're selling some pretty advanced components and kits to all kinds of people, from total neophytes to some obviously very knowledgeable and experienced people. I'm blown away by how well you deal with the inevitable problems that crop up. If there was ever a company that deserved to live long and prosper, you are that company.

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by jarchie on Fri May 16, 2014 7:43 pm

Barry914 wrote:A friend asked me about getting a kit for her husband and I was going to advise him to use the bipolar/resistor combo instead of the MOSFET. Am I overreacting?

For what it's worth, I generally recommend that people use a ZVP2110A instead of the ZVP3306A that comes with the kit. True, the ZVP3306A works fine for most people, but why not recommend something that you know is going to be trouble free (e.g., the PN2907A/1k or the ZVP2110A)?


Barry914 wrote:One more thing. I am compelled to split hairs - the dissipation due to the base current is less than 3 mW. This assumes Vcc = 4.7 V and Vbe is 0.6 V.

You're absolutely right!

The emitter is wired to the 5v supply, and the collector powers the VFD driver chip and filament. When the transistor turns on, about 4 mA (4v/1kOhm) flows from the emitter to the base. The 5v voltage drop is between the emitter and microcontroller--not the emitter and base. So the wasted energy would be on the order of 20 mW (4mA*5v), but the majority of that is dissipated by 1k resistor--not the BJT. My mistake!

jarchie
 
Posts: 557
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Re: Ice Cube Clock - Low HV

by Barry914 on Sat May 17, 2014 11:25 am

Of course, I have to ask myself what's the point of spitting that particular hair, but I have little self control and I just couldn't help myself. As I said, it's a compulsion. I guess since I now have 14 more PNP transistors that would all be fine, I'll give one to my friend, along with a 1K resistor and a link to one of these threads.

My clock is running a tiny bit slow. I don't have the right trimmer cap on hand, but I do have some twisted pair....

Barry914
 
Posts: 448
Joined: Sun Dec 08, 2013 1:26 pm

Re: Ice Cube Clock - Low HV

by jarchie on Sat May 17, 2014 6:56 pm

Barry914 wrote:Of course, I have to ask myself what's the point of spitting that particular hair, but I have little self control and I just couldn't help myself.

It was helpful to me, as I was thinking about the circuit incorrectly. Thank you.

jarchie
 
Posts: 557
Joined: Sun Jun 24, 2012 2:16 pm
Location: Santa Cruz, California, United States

Please be positive and constructive with your questions and comments.