Old Herring-Hall-Marvin safe - typical direction & numbers?

[JohnKoz]

Thank you Squelchtone & MartinHewitt, I'm so glad you two brought that up, i was thinking about adding to the efficiency by trying to move the last disks first, and working backwards towards the first disk. There was just so many things I had to get working fist, but it's purring like a kitten now, so this is the time!

...you do not need to start at 0 for each combination it tries.

Of course, but it happens that the dial is already at 0 from the last attempt, so the next attempt always starts at 0 by coincidence. This all changes if I do partial attempts of course.

...25 goes past the dialing index at 12 o'clock 3 times, and stop at 25 the 4th time around...50 goes past the dialing index at 12 o'clock 2 times, and stop on 50 the 3rd time around...75 goes pas the dialing index at 12 o'clock 1 time, and stop on 75 the 2nd time around

It seems you have 1 more rotation than I do everywhere. This also results in more than a full turn on the last number before going back to zero - I'm having a hard time picturing the disks, I wish I had a disk-pack in my hand, but if that's the way I'm happy to add it.

If all combinations with 25-50-* should be tested, the first number would be 25-50-55, then all around to 25-50-95, 25-50-0, 25-50-5 and only when we get to 25-50-45 or 25-50-50 we should decrement W2 to 25-45-50.

Are you adding to what Squelchtone said, or was this comment completely new. It seems like you're pushing the last disk in 2 directions instead of 1, is that right?

By the way, you're both suggesting increments of 5 - wow, i was thinking of going to 2 but 5 is even better!

[Squelchtone]

This video will show you why you have to dial 4 turns this way and 3 turns that ways etc.. The rotation of the drive cam is needed until it starts what is called "picking up the wheels"



** your lock is a bit different inside than this modern example.

You may have already watched this video on youtube but here is a professional safe auto dialer in action in case you want to compare against how yours is set up.



a home brew safe cracking robot


Squelchtone

[MartinHewitt]

You have chosen to do 4L-3R-2L-R, so the "natural" movement for W1 and W3 is to increase and for W2 to decrease. To go from L20-R40-L60 to L20-R45-L60 means you need either move W2 around left to something like 50 (nearly a full turn of the wheel), and than right to 45, or you need to dial the complete combination. To go from L20-R40-L60 to L20-R35-L60 means you only need to push W2 5 numbers. That's 2 to 3 turns less than the first option and something like 6 turns less than the full redialling. Going 95 - 0 - 5 is only a reason to increment the previous wheel if it was set to a location there. Lets say you are at L60-R10-L90 with the dial at 0 for an opening test, the following turns would be: L95, R0, test, L0 (full turn), test, L5, R0, test, L10, R0, test (L60-R10-* fully checked), R10 (full turn to pickup), W3, R10 (full turn to pickup W2), R5 (move W2 by 5 numbers), L10 (slight turn), L10 (full turn to pickup W3 and move it slightly to 10), R0, test.

With the 5 increment I just followed Squelch's example. A simple lock like that has often wide gates, but I don't know how wide it really is. I probably would indeed use 2 to be on the safe side.

[JohnKoz]

Ok Squelchtone, I've determined that the dial instructions I provided are actually the same as yours, just described differently. I know because I programmed my dialer to try 25-50-75 at a real slow speed and watched it, it passes 25 3-times and stops on the 4th...etc. So I'm pleased I have the correct full-combination sequence.

Now, I have to improve the efficiency by skipping numbers (I'll try 2 for starters, worried about being as aggressive as 5) and then cycling through the last combination, and working my way to the first. Let me wrap my head around what you and MartinHewitt suggested, it may have to wait until the weekend to code.

The videos were fantastic, that professional dialer screams (a lot faster than mine), which is good for MartinHewitt's caution about inertia and going to fast.

I'd like to post a clip of my dialer, but don't want to use Youtube, can anyone suggest an alternative?

[Squelchtone]

I'm at the edge of my seat man.. One of the reasons I mention going in 5's is because most people are lazy and they have been since safes were invented so quite often you will combinations ending in 0's or 5's such as 10-20-30 or 25-50-75. The gates on the wheels are also several numbers wide compared to modern safe locks, so you could honestly dial 22-52-73 and it will still open, but to be fair, if you dial 25-50-75 but the real combo is 25-50-80 you may not get the handle to turn without some aggressive jiggling. The great thing about doing combinations every 5 numbers is it is only 20x20x20 or 8,000 total combinations which your machine should be able to get through in a day.

If you have a way to start the machine at a certain number, people like putting in years such as 1900 or 2000, so it may be worth setting up wheel one on 19 and then having the machine do wheel 2 and wheel 3 with an exhaustive search. Of course your combination could be 5 - 73 - 41 in which case my theory goes right out the window.

As far as sharing a video or photos of it.. we could keep it private if you want to keep it under wraps for now, and if you want to email me at squelchtone @ lockpicking101 dot com I can share it with Martin and anyone else in this thread who wants to take a look without posting it in public on here or anywhere else.

I have a drop box account but not sure how I would accept an upload from a visitor.. if you have a google drive or drop box that wont leak your real name or other private info in the URL, you could use that to share the video or photos.

Thanks and can't wait to see your safe open,
Squelchtone

[JohnKoz]

VIDEO

Ok, let's see if this link works.
I couldn't embed the video with the youtube BBCode, but here's a link to the embed-ready video.
Click on the image once its visible to start.

This is an earlier video when I was still going clockwise first, I now start left.

https://onedrive.live.com/embed?cid=7B8570BDFA089677&resid=7B8570BDFA089677%21369282&authkey=ABxBAAh-Hj2ozow

Let me know if there are any trouble viewing.

[Squelchtone]

VIDEO

Ok, let's see if this link works.
I couldn't embed the video with the youtube BBCode, but here's a link to the embed-ready video.
Click on the image once its visible to start.

This is an earlier video when I was still going clockwise first, I now start left.

https://onedrive.live.com/embed?cid=7B8570BDFA089677&resid=7B8570BDFA089677%21369282&authkey=ABxBAAh-Hj2ozow

Let me know if there are any trouble viewing.

Fantastic! Great design and I love that linear actuator with the switch above it to signal if handle moves far enough to stop the program. Are you using magnets to hold it on the door or do the angles hook onto the safe from the top or do they wrap around the back of the safe like a belt?

Your stepper motor turns so nice and smooth. You can probably speed it up some and it will still dial without risk of adjacent wheels slipping or not stopping where you want them to.

Are you using an Arduino to control the stepper motor and switch IO?

Very clever setup, bravo!

Squelchtone

[Squelchtone]

Check out this guys belt driven dialer:

[JohnKoz]

Are you using magnets to hold it on the door or do the angles hook onto the safe from the top or do they wrap around the back of the safe like a belt?

Your stepper motor turns so nice and smooth. You can probably speed it up some and it will still dial without risk of adjacent wheels slipping or not stopping where you want them to.

Are you using an Arduino to control the stepper motor and switch IO?

Thanks Squelchtone, I see now I need to make a blog post with the whole design, but that's something I can do after I reprogram for redial and I'm waiting.

But till then, I'm using an Arduino Due, JRK G2 (actuator driver), SpartFun Big Easy Driver (stepper driver). The stepper is doing 1/4 steps, using the AccelStepper library.
I have the whole thing clamped to the safe, more like your "belt" analogy, except it's angle iron on both sides and long threaded rod.

My only complaint is that I'm using I2C to speak with the JRK and it's intermittently failing, so the door doesn't pull. But I have code in there to stop the dialing if I detect the actuator isn't moving; and the whole thing writes to an SD card for the progress, so at anytime I can just reset the whole thing and it will start from where it left off.

I saw the belt-drive video, looks cool. Much easier on a fatter/flatter dial. Still, I'd be concerned with belt slippage over time (there's no 'teeth') and would need some way to occasionally re-calibrate to zero.

[JohnKoz]

Update:

I changed my algorithm to advance the last wheel first and work backwards towards the first. I started running it yesterday.
To make sure I was on the right track, i made myself a model of a wheel-pack. For the purposes of simplicity consider the first dial the same as the back "wheel", i think the result is the same.
I don't know if I have the interpretation correct. Specifically I'm not sure about the 'nubs' that bump between wheels. (I could use a good exploded view with terminology!)
I'm not even sure if the placement of the screws matter, do they?

Here is what my simulation looks like:
https://1drv.ms/u/s!AneWCPq9cIV7lsUVh_iHe3TKgZmxEg

-John

[Squelchtone]

Update:

I changed my algorithm to advance the last wheel first and work backwards towards the first. I started running it yesterday.
To make sure I was on the right track, i made myself a model of a wheel-pack. For the purposes of simplicity consider the first dial the same as the back "wheel", i think the result is the same.
I don't know if I have the interpretation correct. Specifically I'm not sure about the 'nubs' that bump between wheels. (I could use a good exploded view with terminology!)
I'm not even sure if the placement of the screws matter, do they?

Here is what my simulation looks like:
https://1drv.ms/u/s!AneWCPq9cIV7lsUVh_iHe3TKgZmxEg

-John

I have to say that is really creative of you, bravo!

I am not sure if is 100% accurate to your lock though..

If the right most wheel that has numbers written on it represents the dial on the outside of the door, then you are missing a wheel inside the safe.

I wish I could mock up a quick drawing for you, I may have to hand draw it and scan it in.

In a nut shell, it's this order

Dial > spindle (the threaded rod going through the door itself) > drive cam wheel > combination wheel 3 > combination wheel 2 > combination wheel 1

The dial on the outside of the door and the drive cam wheel inside the safe are connected by the spindle like a toy car's wheels connected by an axle. If you pick up the toy car and turn the passenger side wheel with your finder, the driver side wheel turns at the same rate and number of degrees because they are connected together under the car by an axle (metal rod in case of a toy car)

This is what transfers the motion of the dial to the combination wheels inside the safe.

I know pictures are worth a 1000 words, so I'll try to draw something later this morning and post it here.

You're almost there!
Squelchtone

[JohnKoz]

Dial > spindle (the threaded rod going through the door itself) > drive cam wheel > combination wheel 3 > combination wheel 2 > combination wheel 1

The dial on the outside of the door and the drive cam wheel inside the safe are connected by the spindle like a toy car's wheels connected by an axle.

I think I get exactly what you mean, but I thought I accounted for that. If the Dial is directly connected to drive cam wheel, and it turns as it does, then isn't that the same as the dial and drive cam wheel being 1 wheel? Or, put another way, picture my simulator as not having a dial, but instead I'm directly turning the drive cam wheel. That in-turn turns wheel 3, 2 and 1.
Put still another way, don't I have a 3-disk pack plus a drive cam in my model?

[L4R3L2]

Considering how much work you are putting into this project, I'm more than a little surprised you have not obtained a lock to have in-hand to see the construction and inner workings.

[JohnKoz]

Considering how much work you are putting into this project, I'm more than a little surprised you have not obtained a lock to have in-hand to see the construction and inner workings.

Agreed, where can I get one?

[Squelchtone]

Dial > spindle (the threaded rod going through the door itself) > drive cam wheel > combination wheel 3 > combination wheel 2 > combination wheel 1

The dial on the outside of the door and the drive cam wheel inside the safe are connected by the spindle like a toy car's wheels connected by an axle.

I think I get exactly what you mean, but I thought I accounted for that. If the Dial is directly connected to drive cam wheel, and it turns as it does, then isn't that the same as the dial and drive cam wheel being 1 wheel? Or, put another way, picture my simulator as not having a dial, but instead I'm directly turning the drive cam wheel. That in-turn turns wheel 3, 2 and 1.
Put still another way, don't I have a 3-disk pack plus a drive cam in my model?

Yes, it sure seems we are on the same page. I follow what you're saying and yes, the dial and drive cam '"turns as it does", so your model works to simulate operation.