Questions about IMB/ITR Master Clock power supply

[David Cole]

Hello All,

Thank you for this opportunity to post about a recent acquisition, an International Time Recorder Master Clock. The device appears to be in excellent condition although it did not come with the original power supply. I am not sure of the model – there are no tags or markings on the clock – but it was the time keeper for the San Francisco-Oakland Bay Bridge which was built in 1936 and I believe that it is a Model 13. I have poured over all the posts regarding power supplies and want to acknowledge the extremely generous contributions of many people including the late Harold Bain, eskmill, swolf and recently, Toughtool.

I am trying to use the existing impulse mechanism on the clock to create the simplest way of winding the clock. At this time, I have no intention of driving any slave clocks, though that could be a fun step for later on. My inclination would be to create a basic circuit with a 24 volt power supply where the power is switched on and off to the wind coil using the minute contact switch on the clock. But I have read in the forum that when the power is cut off from the winding coil, the electromagnetic field it creates collapses and causes the contacts (on the minute switch) to arc and wear out. The suggestion was made to use a "damping resistor" rated for 10x the resistance of the wind coil. I measured the resistance of the coil and it was 144.5 ohms. Therefore, I would like to use a 1500 ohm resistor. I have posted my design for a circuit and would sincerely welcome any comments or advise from this community. Also, I traced the wires from the contact switches to the terminal block on the top of the clock and have labeled them in the attached image. I have also included the clock face for help dating it. Thank you very much.

 

[Toughtool]

I think your resistor should be in series with the inductor, not across the power source. There are a lot of resources that talk about arc suppression. Here is one: Switch principles:Arc suppression I prefer the diode approach but others like the resistor approach. You need 0.132mA of current (0.273 mA for 12 V coils) through the coils to operate them properly. Series resistors will effect that current. That is why I like the diode version of arc suppression. Look at the graft of the first image of the above reference.
IBM switched the AC side of the rectifier. Seems that AC is easier on the contacts than DC current. I prefer a DC regulated power supply as the voltage stays at the rated voltage. Rectified peak voltage outputs of a 24 volt transformer will be 1.414 times the RMS value (@60 Hz) plus higher or lower primary AC line voltage. My line voltage is normally 128 volts AC. You really have no control of power companies line voltage and transits.

Another reference showing the resistor in series with the inductive load: https://www.acromag.com/wp-content/...Protection-When-Switching-Inductive-Loads.pdf

 

[David Cole]

Thank you so much, Toughtool, for your response. I totally get that I put the resistor in the wrong spot. In reading your response, I guess I am realizing that the electronics for this clock are more sensitive than I had imagined and shouldn't be taken lightly. I realize that I had thought that something from the 1930's would be a little cruder and I could come up with a simple work-around. Obviously, the clock mechanism is quite sophisticated and IBM was meticulous in everything they did, including the electrics. That being said, I am going to try building the circuit you has suggested, with the Trinket sending out a signal every minute and the MOSFET acting like the gate for the power supply. I'll let you know of my progress and can't wait to get this clock ticking again. Thank you.

 

[Toughtool]

OK, should only need to connect to the two wires labled A and C on top. Depends on what your clocks configuration is. These two wires should go the the terminals shown in the photo, which are the two wires connected to the wind coil. Notice the wires to the coils are disconnected for testing, in the photo. Photos of your master would be a great help. Also included is a current flow diagram of the Trinkit for you.

 

[Toughtool]

I guess I am realizing that the electronics for this clock are more sensitive than I had imagined and shouldn't be taken lightly.

I don't think the old relay logic is particularly sensitive, but IBM did build the systems robust. The contacts we are talking about is the minute impulse contact that operates the master relay. That is it's only job and the maximum current the contacts have to deal with is the relay coil's current.

The master relay has to supply switching for all the secondaries, wind coils and usually a program unit. Most systems were limited to a maximum of two ampere per circuit with additional relays and power supplies for larger systems. The contacts on the master relay are nearly a quarter inch in diameter and are probably alloyed with hard metals and very conductive metals, to handle the two amps and provide years of service before finally failing from pitting etc. There were some circuits IBM produced that have filters to reduce arc damage but most did not have these components installed. Cost may have been the reason.

They did have to work with limitations. The old copper sulfide rectifiers were only rated at 0.5 amps continually, 2 amps intermittently. The later selenium rectifiers were rated at 1.0 amperes continuously and 3.5 amps intermittently. Then came silicon diodes in the early 1950's
Notice the diode symbol inside the MOSFET symbol of the Trinket's transistor pulsing circuit. This is a "Back Electromotive Force" protecting device, or could also be said to be a diode arc suppression device.

This is an educational opportunity; I am claiming a fair use of copyright for educational purposes. Here is an extract from a page of the Customer Engineer Reference Manual of Electrical Characteristics, ca 1938, where they are discussing the problem.
See Photo.

Joe

 

[John Lippold]

Hi David,
Your master clock is not a model 13, it's a model 17, probably a 17-7. Judging by the face logo and the date you said the bridge was built I would guess a serial number range from late 1935 "488000" to early 1936 "493000". It's a shame it doesn't have the original serial number tag.
John Lippold

 

[David Cole]

Oh wow! That's great information. Thank you so much. Yeah, it's a real shame there is no tag or serial number. The only thing is a number 11 stamped in the wood at the top of the case. I only noticed it when I took that picture of the mechanism.

 

[David Cole]

Here is the clock with it's face back on. Now I'm waiting for a package of electronics so I can build the impulse generator for the winding coil.

 

[John Lippold]

The number 11 you are referring to is the case number, it should also be stamped someplace on the door. A unique case number was stamped on each piece of the clock case to insure all the pieces could be fitted to each other as the case was being assembled.

 

[Toughtool]

David,
I guess I should post the circuit and particulars of the pulsing circuit here since you picked it up from a different post. Notice the instructions in the software below that tells you how to upload to the Trinkit. The Blink program is an Arduino IDE training example program. Download the Arduino IDE from: Software
Additional notes: The off timing (in ms) should be adjusted to less then the 600000 ms in the text below so that more then 60 pulses are presented to the wind coils. The clock's A pulse is normally 80 pulses per hour; 60 pulses to keep the mainspring fully wound with 20 pulses, to catch-up after a power failure. I suggest you leave the ON count as is, to provide a 0.6 second ON time. You can also use the 3.3 volt logic version as the FQP30N06L will turn on with that logic level.

Here is the software:
It is the Blink example from the free Arduino IDE program. I only changed the ON time to 600 mS (milliseconds) and the OFF time to 60000 ms . Simple! You can very the time of OFF by milliseconds, and leave the time ON as is, to tune the overall pulse to give a total time of one minute. May take some time to adjust for accuracy. i.e reduce the 600000 ms to 590400 ms to compensate for the ON time of 600ms, then add or subtract to get to exactly one minute.
/*
Blink
Turns ON an LED for 0.6 second, then OFF for 60 seconds, repeatedly.

This example code is in the public domain.

To upload to your Gemma or Trinket:
1) Select the proper board from the Tools->Board Menu
2) Select USBtinyISP from the Tools->Programmer
3) Plug in the Gemma/Trinket, make sure you see the green LED lit
4) For windows, install the USBtiny drivers
5) Press the button on the Gemma/Trinket - verify you see
the red LED pulse. This means it is ready to receive data
6) Click the upload button above within 10 seconds
*/

int led = 1; // blink 'digital' pin 1 - AKA the built in red LED

// the setup routine runs once when you press reset:
void setup() {
// initialize the digital pin as an output.
pinMode(led, OUTPUT);

}

// the loop routine runs over and over again forever:
void loop() {
digitalWrite(led, HIGH);
delay(600);
digitalWrite(led, LOW);
delay(60000);
}

 

[David Cole]

Gosh, Joe, that is very kind of you to repost this valuable information. I'm sure I'll need a little hand holding once I have built the circuit and tired programming the Arduino but I'll do my best. What a valuable resource this forum is to a neophyte like myself. Thank you all.

 

[Toughtool]

Well I hope those sharper than me noticed my mistake. I just noticed there are too many zeros in the pulser writeup.
600 ms is 0.6 seconds. 60000 ms is one minute. So please disreguard the extra zeros where you see 600000 and 590400. Sorry for the goof. Joe

 

[David Cole]

Joe, I think you are that sharp guy Good eye.

Perhaps I can get advise on the case locking mechanisms too. I understand that this is pretty important to keep the dust out. I found some patterns for a key on the forums and I cut them out, perhaps a little crudely, with a jeweler's saw. I figured that these kinds of locks aren't particularly complicated. But the key doesn't really fit all the way in, even though the height appears to be accurate. It turns, but doesn't really make a positive action inside. I think it is the wrong pattern. Does anyone have better advise? Should I just make it more carefully?

 

[Toughtool]

David,
The diagram is the same as my key although I have not used my digital caliper to check the actual dimensions shown. I think the tolerances are very tight. Even my key does not turn smoothly sometimes. You may need to work in it a little. Try painting the key with layout dye and see if you can see where it is hanging up.

Also there is someone on the forum that is a locksmith (2014) and mentions he has blanks available. You should PM him.

See post #4 at this link:

https://mb.nawcc.org/threads/cabinet-key-for-ibm-master-regulator-clock.119130/

 

[David Cole]

That's a great suggestion with the layout dye. I will try it. Just to be safe and because it is available from a member, I did contact Peter regarding the key and he will be able to provide me with one. That is very exciting news. I sure appreciate the tip.

Be well,
David

 

[David Cole]

Quick update: I built the circuit on a protoboard and wanted to share the picture since I had some confusion with wiring the MOSFET. It is a bit perplexing with the Source, Gate and Drain leads but that is just because I am unfamiliar with basic electronics. It has been running perfectly for about a week and my next step is to solder the components onto a permanent board. Thank you so much for the design, Toughtool. It is certainly very easy and dependable.

I have also been using the time to slowly start getting the clock to run accurately. As you may know, there is small nut on the pendulum which raises and lowers the weight. I think the manual stated that each division is one second/day. So I've been using an iPhone stopwatch to get closer and closer to an accurate time and hope to soon be at the point where I can be making very subtle adjustments.

Finally, I got the original Corbin key from Peter Zasada and that was a huge plus. Not only did it fit perfectly, but being able to lock the cabinet is a great way to keep the dust out. He was very easy to deal with and I would highly recommend to anyone with one of these clocks to get this original key that fits so well. Thank you Peter.
(https://mb.nawcc.org/threads/cabinet-key-for-ibm-master-regulator-clock.119130/)