Updated IceTube Clock Design

After fin­ish­ing the IceTube clock ver­sion 1.31 build for myself, I was ready to work on the board design for some IceTube clocks as gifts. I could have used the exist­ing V 1.31 boards, but I want­ed to use an up to date design that incor­po­rat­ed all the hard­ware improve­ments over the past 10 years while still stay­ing with a thru-hole design.

IceTube clock Version 2.1 mount­ed in laser cut acrylic enclo­sure. March 2020

This updat­ed design is based on jarchie’s (John Archie) xmas-ice­tube clock revi­sion D board. He had incor­po­rat­ed most of the hard­ware fea­tures that I want­ed for my new batch of clocks, along with a vast­ly improved firmware update.
One of the most impor­tant hard­ware improve­ments for me was the abil­i­ty to prop­er­ly dri­ve the IV-18 tube to spec­i­fi­ca­tions (pro­vid­ing more even dis­play illu­mi­na­tion and pre­vent­ing cath­ode poi­son­ing). Some oth­er fea­tures of the design are the inclu­sion of an one wire tem­per­a­ture sen­sor near the oscil­la­tor crys­tal to pro­vide high­ly accu­rate tem­per­a­ture com­pen­sa­tion of the crys­tal fre­quen­cy, reset pull-up resis­tor, micro-con­troller pow­er decou­pling capac­i­tor, GPS con­nec­tion pads, along with an improved board lay­out while still main­tain­ing phys­i­cal com­pat­i­bil­i­ty with Ver. 1.X board switch­es and connectors.

IceTube clock ver­sion 2.1 main bare board top view

While there are so many things that I real­ly liked about jarchie’s Rev D board, there were also a few areas that I thought need­ed some improve­ment.
The first area I worked on was increas­ing the size of all pow­er traces, with spe­cial atten­tion to the micro-con­troller, high-volt­age, and fil­a­ment sec­tions. I then worked on increas­ing sig­nal con­duc­tor width and improv­ing spac­ing between con­duc­tors and pads. With a few small moves of com­po­nents, con­duc­tors, and added ground via’s, I was able to sig­nif­i­cant­ly improve the exist­ing ground pour on both top and bot­tom planes. Next I added com­po­nent val­ues and adjust­ed font sizes to fit the avail­able space. The side board also ben­e­fit­ed from trace width and spac­ing adjust­ments, along with a small increase in height to accom­mo­date the need­ed changes. 

IceTube clock ver­sion 2.1 main board with all com­po­nents added
IceTube clock ver­sion 2.1 main board bot­tom sol­der side with addi­tion­al decou­pling capac­i­tor on IC5

After receiv­ing my ver­sion 2.1 boards from OSH Park and adding the com­po­nents, I checked that my mod­i­fi­ca­tion did­n’t break any­thing. All was well, but I was see­ing some oscil­la­tions on IC5 which is the LM317 volt­age reg­u­la­tor for the fil­a­ment dri­ve cir­cuit.
I end­ed up adding a 0.22 uF 50 volt ceram­ic capac­i­tor on the back side of the board direct­ly from the LM317’s input to the ground pin of the pho­to-resis­tor R6, which cor­rect­ed the issue.
That prompt­ed ver­sion 2.2 of the main board which added C12 the LM317 decou­pling capac­i­tor, chang­ing the foot­print of IC3 to the TO-220 ver­ti­cal ver­sion, reduc­ing the foot­print size of R4 which is part of the pho­to-resis­tor cir­cuit, reduc­ing the pad size and adding a TX pad to the GPS pads, along with some oth­er minor improve­ments.
When I get the updat­ed V2.2 boards, my plan is to try a TO-220 style 5 volt switch­ing reg­u­la­tor sim­i­lar to the one I used on the V 1.31 build. The cur­rent V 2.1 boards draws 110 mA at 9 volts, while the V 1.31 board draws 59 mA at 9 volts which is a sig­nif­i­cant ener­gy sav­ings. Part of the extra cur­rent draw oth­er than the lin­ear reg­u­la­tor is the fil­a­ment dri­ve cir­cuit­ry oper­at­ing at the prop­er volt­age and current.

IceTube clock ver­sion 2.2 main cir­cuit board ren­der top side

The firmware that jarchie pro­duced is very robust and works extreme­ly well. It sup­ports both the Adafruit ver­sion 1.X boards and the advanced hard­ware Rev D / 2.X boards.
Settings for the type of board and options are set in a very well doc­u­ment­ed config.h file and are easy to mod­i­fy before com­pil­ing the firmware. The config.h file is well worth a full read thru to under­stand what options are avail­able. I end­ed up defin­ing the fil­a­ment as 3.3 volts instead of the default 5 volts as my spe­cif­ic IV-18 tubes had an extreme­ly bright fil­a­ment glow at night when run­ning at the default 5 volts, and a just notice­able glow when oper­at­ing at 3.3 volts. This can vary between IV-18 tubes, and was pos­si­bly enhanced by the low­er volt­age drop due to the increased trace width and low­er resis­tance on the AC1 and AC2 board traces.

Latest ver­sion of the IceTube clock in front of the ver­sion 1.31 clock in it’s 9 year old 3D print­ed case

The enclo­sure was laser cut using 1/8″ (3mm) green edged glass tint­ed Acrylic, which gives it a slight­ly green col­or when look­ing at the edge of the acrylic but is most­ly clear when look­ing thru the sheets. I also used some clear poly­car­bon­ate screws and nuts to hold the pieces togeth­er which were pur­chased from McMaster-Carr.

Original doc­u­men­ta­tion with updat­ed EagleCAD 7.7 schemat­ic and board files for ver­sion 2.2 main and side pcb’s IceTubeClockV22.zip

OSH-Park shared projects Main PCB Rev. 2.2
OSH-Park shared projects Side PCB Rev. 2.2

John Archie’s GitHub orig­i­nal documentation

7 Replies to “Updated IceTube Clock Design”

  1. Great write-up! I built this accord­ing to your schemat­ic and it worked per­fect­ly. The only minor issue I ran into is that C6: 80v 56uf cap isn’t cor­rect in the BOM. The cap list­ed in the BOM is 33uF 100v. Minor issue but thought I’d pass the info along.

    1. Hi Sean,
      Congratulations, good to hear that your project was a suc­cess! If you get a chance, post a pic­ture on Twitter @breviceps @oshpark
      I was able to find a Rubycon elec­trolyt­ic capac­i­tor with a high­er capac­i­tance rat­ing at a slight­ly low­er max volt­age rating.
      The 33 uF @ 100V capac­i­tor works fine, but the addi­tion­al 22 uF of the 56uF unit pro­vides addi­tion­al rip­ple reduction.
      The reduced volt­age rat­ing should not be a prob­lem, as the cir­cuit is lim­it­ed to around 56 volts by Zener diode D5.
      The Rubycon 56 uF capac­i­tor part num­ber is 80ZLH56MEFC8X11.5 and has a 8000 Hrs @ 105°C rating.

      Greg (Barbouri)

  2. Thank you for the superb work you have done to improve the Ice Tube Clock and its PCBs. Also for shar­ing your inter­est­ing projects in such detail !
    Please could you add a link to your Rev 2.2 Gerber files if pos­si­ble, or per­haps email the files to me? I would real­ly like to build the Rev 2.2 ver­sion of the clock. The OSH-Park links have stopped working.

  3. Hi, I just mount­ed my tube clock by your design, great v22 update, thanks.
    Just one note so far that I found out, on the PCB mask the C1 and C2 are swapped.
    Also R4 on your design is a 10K but BOM says is a 5.6K, I installed the 5.6K since I ordered all parts from the BOM list, should I changed it for the 10K?
    Because of this my C6 is a 33uF, is it impor­tant to be a 56uF instead as you replied before to anoth­er com­ment here?
    Just to final­ize… the J1 jumper is real­ly for what, since it dis­ables the R2 and C10?
    I can’t find any doc­u­men­ta­tion about it.
    Thanks again for all this.

    1. For C1 and C2, they are in par­al­lel between the same two points. It should not make any dif­fer­ence which goes where.

      The jumper dis­ables (pulls to ground) the bat­tery sen­sor. You would need to look at the firmware to see what the proces­sor does with that input. 

      I am in process of build­ing one of these and will be look­ing at the firmware code in the next few weeks.

Leave a Reply

Your email address will not be published. Required fields are marked *