Reverse Engineering a Clock Pulse Giver
Over the years my professional watchmaker friend brought me several of his professional devices when they stopped working or when he needed some advice. And I am always curious about the electronic products he brings me to investigate. Last and recent one was about his Gent of Leicester XC408 Master Clock which was fixed and works again like a sunshine! And another very big repair was about the Exachron DCF time pulse receiver that completely was examined and even got simulated circuit parts that worked in the Tina designer/simulator. Of which also several Youtube videos were added to that article on Jestine’s great Blog. And this article gets a Tina circuit too.
This one is about reverse engineering a Clock pulse giver he in the past had bought that no longer was available because the manufacturer and/or designer of this product no longer existed or worse past away. And my friend (and also old colleague of a previous job we both worked) therefore wanted to make sure it was examined and saved in case he needed to repair it or duplicate it in future projects.
The designer and/or manufacturer of this Clock pulse giver however had made sure that his product couldn’t be duplicated by removing all markings on the IC’s.
Above photo shows the boxed Pulse giver circuit. The red and blue wires are the about 9V DC + and – battery inputs. And both black wires are output to the Clock.
It worked splendidly but as said no longer could be bought from the designer/manufacturer. And neither could the schematic of this circuit be found anywhere. Mainly because the seller simply no longer existed.
Finding the used but unmarked IC’s is no problem if they are the standard digital TTL/CMOS types.
Most of my IC testers are capable in recognizing the IC function to find what type it is. If they are normal or the special Schmitt triggered type is however not shown. Per example: the 7404 hex inverter is also available as a 7414 hex Schmitt trigger inverter but they behave almost identical in the tester if the different trigger level is not checked. Some testers then do give more IC types as result. And we have to select the right type from the given results. Schmitt triggers are used if we want to eliminated any erratic trigger pulses influencing its functioning.
And as was concluded in my previous Gent of Leicester XC408 Master Clock, 2 of my portable universal digital IC testers had problems with disapproving bad counter IC’s like the 4024 7 bit binary counter that was bad but tested good anyway! My other PC controlled IC testers from Elektor March 1998, and the ELV PC slotcard controlled IC tester had no problem with disapproving the bad chip. To my surprice the TL866A nor the Genius G540 could handle all counter IC’s. Because the TL866A simply lacked types CD4024, and types 7490, 7492 completely! The Genius G540 universal iC tester did better with recognizing the 4024, but lacked also the 7490 and 7492 counter/divider IC’s. (Counters are used as dividers when only a selected output or a combination of outputs are used).
Next photos show the solder side of this Pulse giver.
So please bare in mind that most (portable battery operated) IC testers only recognize the IC function but not fully test them. Especially when they are the complexer counters!
To complete this reverse engineering article also the in Tina designed circuit of this Pulse giver is added.
Above the Tina circuit with explanation. Below some additional info.
Sadly simply measuring the exact frequency on the counter outputs in the Tina simulation failed if the integrated Multimeter was used. The integrated virtual Multimeter only gave the frequency of the used frequency generator and refused to probe on the signal outputs I wanted to measure. I first thought it was a bug in my version 11 of the Classic Tina Designer with the extra HDL package. I therefore asked the Tina creators why it can’t be used as the probe button on the Multimeter suggests but clearly fails. Because when connections are probed the previous running simulation just Halts and the Multimeter is of no use.
But after a quick and confirming helpful answer from the 24 hours a day available email support from the technical Tina designer staff they told me that the Multimeter couldn’t work in the Transient simulation mode because it only works in the AC/DC simulation mode. And checking the frequency on the counter outputs was still possible but only by using the virtual oscilloscope and its Wave Diagram in Transient simulation mode. And by using its cursors to find the pulse width and also the frequency of the counter signal outputs is very helpful. And the Diagram can be exported like all simulations can be saved as jpg, gif or as other screen snapshots.
Why do we need to know and double check the frequency on output Qc? (See Tina schematic). Because the counter which is used as a divider is confusing, and it is wrong thinking that counter U1 with only output Qc used (output active when 0100 is reached) is dividing by 4! Because it isn’t! Counter U2 output Qd (output active when combinations 1000 up to 1111 are reached) which is 16 combinations counting starting from 0000. Which means that it devides by 16. But because transistor T1 disables combination 0000, counter U2 starts at 0001 after combination 1111 was reached and output RCO became ‘1’. Which means that Counter U2 divides the frequency of the input signal by 15! And for Counter U1 is difficult to see that it divides by 8 but apparently it does!
Because it is confusing to check the real dividing factor of a Counter I simply use the simulator and the given Tina circuit and then input a signal with a known frequency. And check the result. In this case a 200KHz signal and examine the output signals at QD from U2 and at Qc from U1. See the result in the Wave Diagram and the signals on the Virtual Oscilloscope further below. VM5 is the signal at output QD from Counter U2. And VM1 is the signal at output Qc from Counter U1.
And we notice that we get a 5.42 uSec pulse at VM5 with a pulse length of 75.81 uSec. And we get a symmetrical output pulse of 299.64 uSec with a pulse length of 2 x 299.64 uSec = 599.28 uSec.
And Frequency = 1/T <=> 1/75.81 uSec gives a frequency of about 13.19087192 KHz. And for 1/599.28 uSec <=> gives a frequency of about 1.66866907 KHz.
So at an input frequency of 200 KHz we get an output frequency of about 13.19087 KHz at Qd Counter U2. Which is dividing by 15! And the frequency of the signal at Qc Counter U1 is about 1.66 KHz which is compared to the frequency at output Qd of Counter U2 dividing by 8!
The total dividing factor of both counters therefore is 15 x 8 times = by 120 !
With this knowledge we finally are able to determine the correct output frequency and the shape of the signal at output Qc at Counter U1.
The original Pulse giver input frequency in our original Tina circuit was 32768 Hz divided by 16384 is 2 Hz (by U3 the used 4060 cmos IC).
And output signal Qd at Counter U2 is <=> 2 Hz divided by 15 because the frequency slows down a factor 15 <=> frequency at Qd becomes 1/7.5 Hz.
Following the given dividing factor we previously found in the Tina simulator for Counter U1 at output Qc which is another 8. We get a frequency of 1/7.5 Hz divided by 8 = 1/60 Hz.
The pulse length at Qc Counter U1 therefore is T = 1/F = 1/( 1/60 Hz) = 60 seconds. And according to the findings we gathered from the Tina simulation that signal is symmetrical, so it also must be a square wave with 30 seconds positive going signal and 30 seconds negative.
Below the Wave diagram from our Tina simulation (in Transient mode) when a 200KHz signal was used at the CLK input of Counter U2. And showing result VM1 at Counter U1, and the output signal VM5 at Counter U2.
Above the Tina Counter test circuit I used in Transient mode with Oscilloscope and Wave Diagram to check the output signals of both Counters.
Although checking circuits with Counters can be very confusing , and also a problem when most IC testers fail to test them correctly, above mentioned simulation method is maybe the best way to examine these circuits fast and easy.
* Testing of the unmarked IC’s was of course only possible after carefully desoldering the IC’s from the very thin copper tracks of the circuit board first. And soldering them back afterwards.
By publishing this circuit we respect the work that was done by the original author/designer and or seller. And the only purpose of the reverse engineering done was to save this product that no longer can be bought.
This way making sure this great battery operated Pulse giver circuit as project will not get lost in time.
I guess that this exact 32768 Hertz crystal operated Pulse giver circuit will also be of good use to other Professional Watchmakers. And hopefully this article is a very useful document if you need help when Counter circuits need to be examined.
Note: Both e-caps on the board are 1000uF 16V. The transistors are plain BC547/BC557 npn/pnp’s. (See circuit). The circuit needs at least a DC voltage higher than 1.5V. (else the crystal simply won’t oscillate, which also depends on the clock’s coil used). I used my digital IC-testers (several of them to be sure) to determine their function successfully. In the Tina simulation circuit some other component values were used to speed up the simulation.
Albert van Bemmelen, Weert, The Netherlands.
Please give a support by clicking on the social buttons below. Your feedback on the post is welcome. Please leave it in the comments.
P.S- If you enjoyed reading this, click here to subscribe to my blog (free subscription). That way, you’ll never miss a post. You can also forward this website link to your friends and colleagues-thanks!
Note: You can read his previous repair article in the below link: https://jestineyong.com/gents-xc408-master-clock-repair/
(46)Dislikes
(0)
22 Comments
Leave a Reply
Parasuraman S
April 29, 2020 at 3:43 pm
Another amazing service cum research work! Hat's off to you!
Albert van Bemmelen
April 29, 2020 at 9:06 pm
PS: the serie resistor in serie with the + of the about 9V battery was not mentioned in the Tina circuit but as can be seen on the component side of the Pulsegiver pcb it is a 2k2 5% resistor.
Mark
April 29, 2020 at 3:45 pm
Wow Albert,
Looks like you have been busy!
Thanks for such a detailed and informative article.
Albert van Bemmelen
April 29, 2020 at 9:12 pm
You probably agree Mark that doing nothing is very hard to do (lol).
Henrique J. G. Ulbrich
April 30, 2020 at 1:18 am
Some cases of reverse engineering are amazing. Like this present article. Very, very good, Albert.
amir
April 30, 2020 at 12:17 pm
thanks, great information
Waleed Rishmawi
April 30, 2020 at 1:20 pm
Super informative repair article. Man that is a lot of work presented here. About the UC tester you mentioned where can I get information about? Have a blessed day
Albert van Bemmelen
April 30, 2020 at 6:57 pm
If you mean the Elektor and the ELV IC tester Waleed?, I can send you the information through Jestine or to you if you send me your email adres. And maybe you can still find them somewhere on the internet as they also were published in Elektuur and ELV Journals. And my Xeltex Super Pro/L universal programmer was bought new in the 90's. The other IC testers like the TL866A and the Genius 540/840 have new versions these days. The old TL866 is now the new TL866 II version (it seemed that because the old Chinese TL866C version was cloned, hacked and upgraded to the the better TL866A ISP version why the manufacturer chose no longer to support the old trusted IC tester by not upgrading the device list with newer IC's). But as I wrote they didn't support all counters anyway.
Albert van Bemmelen
May 1, 2020 at 9:02 pm
https://de.elv.com/elv-ic-tester-204452
(software for PC used was IC-Pascal for IC's upto 20 pins including counter IC's 74163, 7490,7492,4024, RAM2114, PCF8574/A etc.etc., and works perfectly starting with Windows 95/98. And with the integrated editor new tests can be added easily. The program also contains an IC Info database in German, and a IC function simulator).
About the perfect working stand-alone Elektuur cmos/ttl IC-tester from March and April 1998 Elektuur magazine, I will upload the info files to Jestine. I made this great working IC-tester somewhere in 1998 that outruns the Genius and the TL866 testers on most if not all special IC's like the mentioned counter IC's. And you can also add new testvectors to add new tests to the existing already great working hardware.
Albert van Bemmelen
May 1, 2020 at 10:45 pm
Here another link to software of the ELV-IC tester with PC ISA slot card:
http://www.sax.de/~zander/zubehoer/k1520/ELV-IC-Tester/
Albert van Bemmelen
May 5, 2020 at 4:11 am
Or this one Arduino based complete with PDF instructions and software:
https://www.electronicsforu.com/electronics-projects/hardware-diy/arduino-based-digital-ic-tester-truth-table/3
Albert van Bemmelen
April 30, 2020 at 1:26 pm
Using a simulator like the Tina circuit analyzer from Designsoft helps me a lot when I need to understand or create a circuit. Using any other simulator/designer like Proteus or Multisim probably will do the same but since I bought a licensed Tina version with the VHDL option I stick with that program.
Calculating on pulse length and pulse time using counters is never easy and can be confusing but as a rule we also can think of the dividing factor by looking at all output counting combinations possible as being the max dividing factor. Like when we only output Q3 on our counter (QD in our pulse giver) we just can say that we have a 4 bit output range from 0000 to 1111. And the MSB bit which is Q3 therefore outputs the input pulse divided by 16. Although bit Q3 only is active from 1000 to 1111. In case of outputting Q2 (QC in the Pulse giver) that would be dividing by 8 counting from 0000 to 0111. And although Q2 is only active at combinations 0100 to 0111 its dividing factor is by 8. If more outputs are used combined it gets more complicated but simulating the circuit always makes it much easier to check the results.
Albert van Bemmelen
April 30, 2020 at 1:48 pm
Or in other words: a pulse has a negative going half (or just inactive being 0V
with TTL outputs) and a positive going half. Using output Q3 is half of the combinations from 0000 to 1111 active going positive from combination 1000 to 1111 (8 to 15 decimal) the other half of the signal from 0000 to 0111 (0 to 7 decimal) it is inactive. So making a complete output block pulse shape.
And for Q2 it is active during outputs 0100 to 0111 (4 to 7) and inactive from 000 to 011 (0 to 3). In time exact pulse lengths.
Albert van Bemmelen
April 30, 2020 at 8:51 pm
In the article the pulselength and the signal of VM5 at the base transistor T1 was shown and that naturally had a asymmetrical shape showing the moment when the outputs all became logic ones (combination '1111' counter U2) at about 1/15th of the total pulselength time. Of course the pulselength of the signal of VM4 at output QD was identical, and also had an almost symmetrical shaped block signal like the signal at output QC. But only almost because only the block signal on QC was perfect as both positive and the negative signal half where 300uS. The block at QD counter U2 was not exact symmetrical because the positive going half was 40.24uSec and its negative half 34.76uSec. The second counter U1 after counter U2 is therefore beside a divider also a much needed shape improver for being an exact pulse giver.
Albert van Bemmelen
May 1, 2020 at 3:00 am
This is why above is true: As explained in the article Transistor T1 loads the first output combination being 0001. Why all outputs QA to QD on counter U2 are asymmetrical block signals because in the range 0000 to 1111 combination 0000 is missing. Making the negative going pulse half shorter then the positive half pulse length. That is not the case with counter U1 that still counts the entire combination range from 0000 to 1111 creating a perfect symmetrical block output pulse on all outputs.
Paris Azis
May 2, 2020 at 3:27 am
Hi Albert
Very interesting analysis of this time base generator. I knew that the crystals of the ordinary electronic watches work on this frequency of 32768Hz, which then the circuit divides by 2^15 to produce the time unit of 1 second, but I never entered deep in further details...
Albert van Bemmelen
May 3, 2020 at 3:01 am
Your professional info and story in the post after Parasuraman's recent jinxed-amp-with-torroid-transformer-failure article concerning the bad inrush current about adding the necessary safety time delay components was an interesting read too Paris!
And I'm glad that you are still an active blog visitor in these dangerous times.
About the 4060, it sadly is only a 14 bit counter/divider as you know.
[http://electro87expert.blogspot.com/2007/05/24-hour-timer-by-ic-4060.html]
And the clocks that run on this pulse giver apparently need the exact 30 sec positive and the 30 seconds negative block pulse in every minute. The designer was a real genius in the way how he made this 15 bit divider/counter followed by a 8 bit divider/counter. Making the exact timed block output pulse every 60 seconds.
If it had been a 74(LS)92 12 bit counter/divider as is used in the Exachron DCF receiver repair, that chip contains a 2 bit counter/divider followed by a 6 bit counter/divider when they are placed after each other.
Albert van Bemmelen
May 5, 2020 at 3:03 am
Here another complete DIY IC-Tester from epe magazine. I haven't made this one myself but it sure looks very promising to build!
http://oldcomputer.info/software/ictester/index.html
el shaddy
May 7, 2020 at 6:28 pm
your articles seem very complex for the ordinary, self taught DIY electronics lover.
try to simplify next time.
Albert van Bemmelen
May 8, 2020 at 1:47 pm
That may not only look complex at the beginning,el shaddy, it most likely also just is. It can be rather complex and confusing to explain as I wrote in the article. It was also for me at the start trying to calculate the correct pulse times and pulse shapes. And Reverse engineering an unknown circuit almost always is a bit more complicated. But in the end when you finally grasp the details of using these counter/divider chips, it probably will be very clear to you next time.
I also added more about the circuit in the comments after the article which may be of additional help. Thanks for your appreciated comment
!
Ulises Aguilar
July 25, 2020 at 1:35 am
MR Bemmelen, great job Sir
Ulises Aguilar
July 29, 2020 at 10:24 am
Mr Bemmelen grate Job Sir