Black Lives Matter - Action and Equality. ... Adafruit is open and shipping.

Ice Tube Clock Doesn't Light. MAX6921 is Underpowered. (SOLV
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Ice Tube Clock Doesn't Light. MAX6921 is Underpowered. (SOLV

by jgros on Sat Aug 09, 2014 2:24 am

Hello Everyone,

I recently decided to put together the Ice Tube Clock I bought a few years back. The clock is now working, but I went through quite a bit of trouble to figure it out. It was a good trouble shooting experience for me, so I think it was worth while. If you are not as patient as I am, perhaps you will Google onto this page and save yourself some time.

I apologize if someone has already discovered this. I don't mean to duplicate post. I did do some cursory searches for keywords like "DOUT" and didn't see any hits.

I followed all the directions diligently and performed all the various voltage tests (highly recommended). Then I got to the part where I needed to connect the daughter board with the tube to the main board and apply power. The directions say that the tube should be flashing an incorrect time (like a VCR after a power outage). At most only 1 segment ever appeared lit (flickered), but mostly nothing lit at all.

I did some probing and saw a half voltage at the gate of Q4, so I chopped the PFET off and ordered replacements. The Q4 I had must have gotten zapped (I admit I wasn't wearing my anti-static wrist strap). I then reinstalled a new ZVP3306 (this time making sure to wear the strap) and I saw 0 V at the gate and I saw Vcc at source and drain.

However, this didn't fix the issue. The tube still didn't light up. So I did some more looking around. I started poking around in the code. I didn't spend too much time in there, but I wanted to understand how the code commanded the MAX6921. I spent some time looking at how the SPI traffic to the MAX6921 was formatted.

The MAX6921 SPI acts as a big 20 bit shift register. You shift in the bits you want, and when you have all the bits how you want, you pulse the LOAD signal to latch that data into the MAX6921 logic. Any bits that fall out the end of those 20 bits go out the DOUT pin. This allows the MAX6921 to be daisy chained together to control more segments.

At this point I broke out the logic analyzer to confirm the SPI data looked correct. The DIN/CLOCK/LOAD data all looked good. No problems there. The next step was to ensure that the MAX6921 understood the bits, and the best way to do that was to look at the DOUT pin. After finding a pinout for the PLCC socket (search Digikey for "PLCC socket", I had a pretty hard time using Google to find the pinout), I took a look and found that most of the time DOUT was not changing at all. When I triggered on DOUT, I did see some glitchy looking pulses. Looking with an oscilloscope, I found that these glitchy pulses were half level, and appeared to crash mid bit. A very strange result. Consequently, I'm also seeing nothing out of any of the segment control pins from MAX6921.

Now let's look at the voltage measurements concerning Q4, MAX6921 and the tube: Q4.gate = 0V, Q4.source = 4.65V, Q4.drain = 2.96V; Other side of R3 (22 ohms) is 1.88V. I measured 13.6 ohms from pin Tube.vbias to tube.gnd.

According to the MAX6921 datasheet, VCC min is 3 V. Surely they gave us a little bit of safety margin on that 3 V minimum, right? Is 2.96V really going to cause an issue? That's only 0.04V below spec!

Other folks on the forum recommended shorting out Q4 as a troubleshooting measure. I shorted Q4 drain to D2 anode. When I did this, the MAX6921 perked right up, echoed out the correct bits on DOUT, and I saw signals coming out of the MAX6921 segment control outputs.

It appears that the Q4 PFET ZVP3306 has too large of a R_DS(on), which results in a voltage drop that starves MAX6921.

I did some looking around on Digikey, and found that the Diodes Inc. ZVP2106A looked like a good fit. I ordered some and swapped it in (making sure to wear that wrist strap). My measurements are now as follows: Q4.gate = 0V, Q4.source = 4.65V, Q4.drain = 4.12V, Other side of R3 (22 ohms) is 2.774V. With this part, DOUT looks good, and so do the segment outputs.

Tonight I plugged the tube in and it lit up just fine! Everything looks great.

What does that say about the MAX6921? Does the part really work at 3V? I'm not sure. However, I know it works at 4.12V.

I hope this helps!


Posts: 1
Joined: Sat Aug 09, 2014 1:19 am

Re: Ice Tube Clock Doesn't Light. MAX6921 is Underpowered. (

by phild13 on Sat Aug 09, 2014 6:56 am

The subject of the supplied ZVP3306A for Q3 has been very well discussed and analyzed in the forums. Here is a link to some testing that was done by forum member Russell 27.

Generally the consensus is that the ZVP3306A (Q3) supplied in the kit does not fully turn on and with some clocks this results in a too low of a voltage condition that prevents the display tube driver (MAXX6921) from working. Shorting the outside pins of Q3 (source and drain) is a quick test as it will bypass Q3, providing the available voltage of about 4.7 volts to the MAXX driver chip and R3. If the tube works normally, then there is an issue with Q3.

I recommend (as do others such as jarchie, and Russell 27) using either a ZVP2110A P-Channel FET as a replacement or use the PN2907A PNP transistor modified with a 1K base resistor soldered to it. See this post for how to do it:
Either replacement has been extensively tested and will work just fine.

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

Re: Ice Tube Clock Doesn't Light. MAX6921 is Underpowered. (

by adafruit_support_bill on Sat Aug 09, 2014 8:43 am

Thanks for posting your experiences with the clock. As PhilD13 mentions others have experienced similar issues. In a small percentage of clocks, Q3 does not turn on fully and causes these issues. Usually, replacing Q3 takes care of the problem. Alternative parts as suggested by PhilD13 seem to work well also.

Posts: 78210
Joined: Sat Feb 07, 2009 10:11 am

Re: Ice Tube Clock Doesn't Light. MAX6921 is Underpowered. (

by phild13 on Sat Aug 09, 2014 3:06 pm

I didn't mean to sound as if something was wrong with the components sold with the clock kits.

The MAXX6921 chip has a minimum VCC (logic supply) voltage of 3 volts. Each chip has it's own unique characteristics and and some probably can be run a bit out of spec. As Maximum specs 3 - 5.5 volts operation for VCC, many chips would likely also be able to run at some voltage level below 3 volts just fine. As seen in the chart by Russell 27 the ZVP3306A supplies around 3.02 volts (subtracting about .30 as he used 5 volts in test) which is right at the minimum spec on the Maxx. Due to manufacturing tolerances of various components, a ZVP3306A in some cases may not be able to supply enough voltage to the Maxim chip for proper operation.

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

Please be positive and constructive with your questions and comments.