APF Mark XII Calculator

The APF Mark XII (12) is an inter­est­ing cal­cu­la­tor. It is based on the Cal-Tex Semiconductor Inc. CT5005 FlexiChip, which was devel­oped in 1972. The cal­cu­la­tor is a four func­tion with sep­a­rate­ly address­able mem­o­ry reg­is­ter, all on a sin­gle IC.

The Mark XII is a 12 dig­it cal­cu­la­tor uti­liz­ing a gas-dis­charge dis­play sim­i­lar to a Panaplex dis­play. It is a bit odd in that it does­n’t have a sep­a­rate error, minus, or mem­o­ry indi­ca­tor, it instead uses the cen­ter dig­it seg­ment for the minus sign, and dec­i­mal points for the error or over­flow indi­ca­tion. The Mark XII uti­lizes lead­ing zero sup­pres­sion, and has a selec­table fixed dec­i­mal point selec­tion from 0 to 5. To select the dec­i­mal point posi­tion, the dec­i­mal point key is depressed and held, then the dig­it key from 0 to 5 is pressed to set the posi­tion. Default dec­i­mal posi­tion on pow­er-up is 2.

The Mark XII uses a case and key type iden­ti­cal to the APF Mark 1 with a dif­fer­ent bezel and key­pad frame insert to accom­mo­date the addi­tion­al keys. It also uses the same obso­lete 2 pin AC pow­er con­nec­tor as the Mark 1.
Estimated man­u­fac­tur­ing date for this unit is ear­ly 1973

When I received the cal­cu­la­tor I noticed a rat­tling noise when I shift­ed the cal­cu­la­tor posi­tion. Upon open­ing up the unit, I found one of the sup­ports sheared off the top cov­er, which was the noise mak­er. This sup­port had about 3/4″ of unin­su­lat­ed brass that very eas­i­ly could have short­ed out the pow­er sup­ply or oth­er crit­i­cal com­po­nent. This is why I almost always open the equip­ment up before apply­ing pow­er to the unit. Because you nev­er know what you might find inside.

The Cal-Tex Semiconductor Inc. CT5005 is a P‑channel MOS inte­grat­ed cir­cuit in a 28 pin ceram­ic pack­age. Cal-Tex also made a plas­tic pack­age ver­sion of this IC.
The CT5005 IC con­tains the equiv­a­lent of 3,000 tran­sis­tors and a log­ic cir­cuit with over 400 gates and 230 shift-reg­is­ter bits.
The arith­metic in the CT5005 is done with two, 52 bit, or thir­teen dig­it reg­is­ters. Timing for reg­is­ter cir­cu­la­tion is arranged in thir­teen groups of four, each one rep­re­sent­ing time for one BCD dig­it. At a clock rate of 25 kHz this results in a cir­cu­la­tion or “word time” of 2 ms. This unit has a clock fre­quen­cy of 16.8 kHz which is a bit slow­er than the 25 kHz typ­i­cal data sheet fre­quen­cy, result­ing in a cir­cu­la­tion time of 3 ms.

The CT5005 IC requires a two phase clock oscil­la­tor for oper­a­tion. The APF cir­cuit dif­fers slight­ly from the Cal-Tex rec­om­mend­ed cir­cuit, in that one resis­tor has a dif­fer­ent val­ue, and two capac­i­tors are 13.6% high­er in val­ue. The two 250 pF capac­i­tors vers­es the rec­om­mend­ed 220 pF are what most­ly sets the oscil­la­tor at the low­er frequency.

Another sig­nal required for the CT5005 Is is the “SW ON” sig­nal from the pow­er sup­ply, which is a neg­a­tive pulse from a R/C cir­cuit used to ini­tial­ize the IC.

The 12 dig­it dis­play uses a FLANDIPAK CD1201 mod­ule from Matsushita Electric. It has 12 dig­its with dec­i­mal points, and a tick mark above each dig­it which is not used on this calculator. 

While sim­i­lar to the Burroughs Panaplex dis­play, this gas-dis­charge dis­play uses seg­ments recessed behind a front mask result­ing in seg­ments with beveled ends in appearance.

Nothing too excit­ing with the pow­er sup­ply sec­tion of the cal­cu­la­tor. The AC line input uses a minia­ture rec­tan­gu­lar pow­er con­nec­tor, the same as the APF Mark 1. There is no vis­i­ble fuse for the 117 volt AC trans­former pri­ma­ry wind­ing that I could see.
The sec­ondary side of the trans­former has two wind­ings, one for the high volt­age 190 volt dis­play pow­er and a low volt­age wind­ing at 23 VAC for the rest of the board. There are three low volt­age pow­er rails which are all neg­a­tive when ref­er­enced to VSS ground.

All of the elec­trolyt­ic capac­i­tors were replaced with long-life Vishay and Nichicon units. Most of the Unicon capac­i­tors were in good shape with only the small­er capac­i­tors show­ing ele­vat­ed ESR read­ings. I don’t know the his­to­ry and use of this cal­cu­la­tor, but for 50+ year old capac­i­tors these seemed to have aged well.

The back side of the main cir­cuit board was less impres­sive as to the qual­i­ty of con­struc­tion. There were sev­er­al deeply scratched cir­cuit traces, along with many long com­po­nent leads that had not been trimmed after sol­der­ing. Some of the leads were per­ilous­ly close to short­ing oth­er board traces. After some trim­ming and re-sol­der­ing the board was in much bet­ter shape.

The key­pad has a red, white, and blue theme again sim­i­lar to the APF Mark 1, but with an extra row of four keys for the the mem­o­ry func­tion. Unlike the Mark 1 cal­cu­la­tor, this cal­cu­la­tor uses mechan­i­cal key switch­es instead of mag­net­ic reed switches.

The key­pad uses a 2 X 12 matrix for detect­ing key press­es, and is con­nect­ed to the main board with an 8 and 6 pin con­nec­tor. The CT5005 IC has built in switch de-bounce cir­cuit­ry to pre­vent false key­press­es from noisy switches.

During test­ing the cal­cu­la­tor was using between 2.9 and 3.5 watts of pow­er.
While it is an inter­est­ing 12 dig­it cal­cu­la­tor, and from what I can tell a fair­ly rare mod­el, I still pre­fer the APF Mark 1 cal­cu­la­tor with only 8 dig­its over this one. While not a bad cal­cu­la­tor, it lacks some of the niceties that oth­er cal­cu­la­tors of this era had. More dig­its is not always bet­ter. I did have a lot of fun reverse engi­neer­ing the cir­cuits on this cal­cu­la­tor though, and may do more of that on future projects.