Need to create a 16 to 1 switch box with ICs.. any help?

[auto_turret]

Hello, I am brand new to this board my name is Keith I'm from the Detroit area. My problem is, I need to create a switch box that will allow me to read several resistances with an ohm meter. What I'm testing is several connections on hundreds (possibly thousands) transmission control units for the correct ohm values. Basically we're looking for shorts. I drew up the schematic with several 4016 bi-lateral switches to switch the pins together. But now I need to have the ohm meter connect to these one by one and I want to use an IC for this tester. I'm just not sure on what the correct IC would be. I'm thinking it's an analog multiplexer that would do the job. I know that a digital multiplexer will just give me an on-off status.

Guess my question boils down to, would an analog multiplexer pass the connection untouched along so I can read the ohm value? Any help would be much appreciated!

Reason I'm doing this is because I'm a lowly parts sorter trying to break into the electronics field, and if I can build this and show it to the vendor maybe they'll offer me a job! We've already received testers with dial switches but thats boring and it's lame to sit there and click these dials all day. I want to accomplish this with a single push button and a reset button and a 7 seg display to indicate which position i'm at. Got all that down pat just not sure about the analog multiplexer part since I've never worked with one.

Edit: and I'm using 4017 decade counters to activate the bilateral switches one by one. Funny coincidence how those 2 IC's are right next to each other (4016, 4017)

Edit again: Sorry about all the edits but it's more like 33 connections .. 16 pins but 33 connections to be tested.

 

[Mosaic]

So u have 16 resistances to check for a short.
You are cycling each resistance circuit to an ohmmeter to test for a short or not.

I'd suggest a parallel approach for efficiency...basically u test ALL 16 at once and report on which are not good.

Something like an 18f23K22 MCU with 16 ADC channels can do it instantly for u and report on which circuits fail with a single 7 segment digit showing hexadecimal numbers representing the failed circuits numbered (0,1,2,3,4,5,6,7,8,9,a,b,c,d,e,f)

I'd also suggest pogo pins as the interface to your device to test unless u have a standard connector to use.

This eliminates the ohmmeter and the need for switching altogether.

It's even possible for the MCU to report on the actual resistance of each of the 16 circuits if u wish.

Basically u set up a voltage divider using a precision voltage ref....eg regulated 5VDC as the supply can be used as Vref.

The ADC channels can read the voltage drop across your resistance circuit loads as compared to a precision resistor in series with the test circuit and you can develop the actual resistance of each of the 16 circuits instantly. This can be reported in 7 seg. displays if u wish as opposed to a good/fail indicator. The functionality is your choice.

 

[auto_turret]

Oh how I would love for it to be that simple.. however the vendor wants us in addition to looking for a short, to make sure each connection has the correct ohm value. We have a list to go through and we click through all these connections all day with a multimeter looking for wrong ohm values. Haven't found one yet but they pay us to do it. So testing in parallel isn't an option sadly.. It's just how the vendor wants it done.

Edit: We do use a standard wire harness to connect to each transmission's TCU, they're all laid out on the floor and we have about 8 of us, and I'm the only one with a very slight knowledge of electronics and I had to train everyone on the use of a multimeter. A good number of the connections are supposed to read O.L or >1 M ohm, (seems some of them are either/or) yet we caught our day shift setting the auto range meter to under 400 ohms and sped past all those high ohm value connections thinking they only had to read the solenoids which read 5-6 ohms.. we were wondering how they did so many so fast )

 

[auto_turret]

I would also love to do your microcontroller idea, yet I have no experience with those as of yet, however it is a subject I have been looking into since I'm handy with C. Sooner or later I will buy a microcontroller kit to learn with. Just don't have the time to figure all that out, so i guess my only option now is this quick and dirty (and probably needlessly complicated) circuit.

Edit: I'm afraid the sort will end before I can learn anything new so Im just trying to figure out if the analog multiplexer thing would even work for what i intend to use it for

I meant to say 33 to 1 switch box, I was thinking of the 16 pins on the wire harness. We're testing like, pins 1-2, 1-4, 7-8, 7-10, etc.

 

[KeepItSimpleStupid]

You need a crosspoint switch such as this one: **broken link removed** controlled by a GPIB or USB and software such as National Instruments - Test and Measurement LabVIEW. You have to pick an economical mainframe and the appropriate switch card.

Many moons ago, I worked in a facility where it was manually done. There are way too many interconnects.

If you really want to do it right you need a system voltmeter with 4-wire ohms. This removes the contact resistance of the switching elements.

I specified two such model to do a 16 x 4 crosspoint switch and another smaller scanner to do a 4 x 4.

Not cheap, but will work.

Adv: Reed Relay Matrix Card

 

[auto_turret]

Eh... thats kinda what I'm after.. and I really thank you for your input.. but it doesn't really fit in well with what I'm trying to do unfortunately. What i'm looking for is a 33 x 1 switch (doesn't need to be that exact number, i can stack multiple devices to reach that number) and I'm looking for this in a DIP package.

Couldn't find a crosspoint switch IC that does just that. Guess I'll probably just go on and buy the analog multiplexer IC's since looking at the schematic for it seems like it may the best choice. Or maybe i'm totally wrong and will fail miserably. Oh well. Maybe what I'm talking about doesn't exist. My current job doesn't depend on it. I just want to put the guy from the vendor's mechanical switch box to shame just to show off and maybe get a job with these guys. I'm not too worried if the values are off by a few ohms due to the IC's circuitry being in the way since we aren't interested in the exact value, just whether or not it's within the ranges specified.

I've only had 9 months of formal electronics training at a lame tech school. I'm not very far out of high school. This is not nearly enough to get a real job in electronics so I've been looking into electronics engineering at Eastern Michigan University so I can play on the big playground with the big kids.

Thank you all for your help anyways!

Edit: This device would have to be used by unskilled parts sorters, (i am one of them!) so what I was trying to build was a 1 button solution.. press the button, it advances the switch position and they read the ohm value, the switch position is on the 7 seg led, then hit the button and see the next value, until it gets to the end then resets to 0. Then they plug into the next trans and do it all over again for 8-10 hours a day Seems simple enough haha...
no it really isn't i've found out.

Edit 2: Here is the pdf datasheet on the IC i kind of had in mind or something similar: https://www.electro-tech-online.com/custompdfs/2010/12/dg406b.pdf
The DG406B is what I was looking at. Wondering if I'd be able to read my ohms through it, I have some good ol 74ls157's that I know will not work because they only show whether the input is high or low. Can you see how much of a dummy I am on this subject

Edit 3: I'd be using a 74ls193's binary up/down counter to count on the multiplexer. Also to run the 7 seg with 74ls47's to drive it. Some 4017 decade counters to select the different connections on the 4016 bilateral switches. Maybe i'm already showing that I'll be a horrible engineer with my awfully complex design. But I love this stuff! Its fun!

Edit 4: I'm using whatever parts I have on hand that I got to learn with from that tech school and whats available at my local electronics store. Wish I had a scanner to show my schematic so far, all I'm missing is a way to hook up the ohm meter to it.

 

[auto_turret]

I took a pic of what we have now with my cell phone so you have an idea what we're doing: <a href="http://s1219.photobucket.com/albums/dd432/auto_turret1/?action=view&amp;current=593995656_2114613695_0-1.jpg" target="_blank"><img src="http://i1219.photobucket.com/albums/dd432/auto_turret1/593995656_2114613695_0-1.jpg" border="0" alt="Photobucket"></a>

if that didn't work here's a direct link: https://i1219.photobucket.com/albums/dd432/auto_turret1/593995656_2114613695_0-1.jpg
took the pic a few days ago to show the cup holder to my supervisor who got a kick out of it

Now here is the list we have and also explains why there's 2 rotary switches:
Names of connections withheld due to I'm sure Ford would fire my ass:

I didn't create the setup.. the "resident engineer" at the company who makes the TCU's made the switch boxes and brought them to our plant. I wanted to do one better. I'm just a contracted unskilled parts sorter.

Rotary Switch 1:

Position Pin Pin DMM Value
1 1 4 Open (> 1M ohm)
2 1 5 Open (> 1M ohm)
3 3 4 Open (> 1M ohm)
4 3 8 24 ohm
5 4 5 Open (> 1M ohm)
6 4 8 Open (> 1M ohm)
7 5 6 20k ohm
8 5 8 Open (> 1M ohm)
9 5 9 Open (> 1M ohm)
10 6 7 Open (> 1M ohm)
11 6 9 Open (> 1M ohm)
12 7 9 5 to 6 ohm
13 8 10 24 to 24 ohm
14 8 11 Open (> 1M ohm)
15 8 12 Open (> 1M ohm)
16 9 12 Open (> 1M ohm)
17 9 13 10 to 11 ohm
18 9 14 10 to 11 ohm
19 10 11 Open (> 1M ohm)
20 12 11 Open (> 1M ohm)
21 12 13 Open (> 1M ohm)
22 12 15 Open (> 1M ohm)
23 12 16 Open (> 1M ohm)
24 Open Open

While on postion 24 on rotary switch 1, we move to rotary switch 2

Rotary switch 2:

1 Open Open
2 13 14 10 to 11 ohm
3 13 16 10 to 11 ohm
4 15 11 Open (> 1M ohm)
5 7 3 5 to 6 ohm
7 7 10 5 to 6 ohm
8 7 13 5 to 6 ohm
9 7 16 5 to 6 ohm
10 7 8 17 to 18 ohm
11 8 9 22 to 24 ohm
12 Open Open

Edit: Eh.. looks like the neat spacing i made won't show up in the post

 

[Reloadron]

What you are after isn't impossible but not an inexpensive task either.

If you were to look inside the existing switch box at the rotary switches and find their part numbers you would likely find those switches were chosen because they offer a very low contact resistance. Obviously contact resistance of a set of contacts is not a big issue when looking at test that are short or open but when measuring 5 to 6 ohms it comes into play. Measuring resistance is relatively easy but to accurately measure low resistances gets a little more complex. This is where problems arise using electronic switches like quad bilateral switches as the switch offers up added resistance. Not that it can't be compensated for but it is a problem. My guess is the original rotary switches were chosen for a reason and thus the box designed the way it was.

From a practical point you need to do a little homework. That begins with note of time involved to test a single TCU. Time is money and even testing 1,000 TCUs the way you currently are does not mean there would be a savings using a more complex nd expensive automated system.

Personally I would look at a design using reed relays in a matrix configuration along the lines of what Keep It Simple suggested. I would run a low current of a few mA possibly through and look for voltage drop and measure that. That will give you a GO or NOGO for each test point. The reed switches are clocked, switch after switch.

There are off the shelf commercial systems to do this but they aren't cheap.

Just My Take
Ron

 

[Mosaic]

I don't understand why the OP doesn't think the MCU will work in parallel. You can get the Precise OHM reading that u need.....as I said before.

This process can be automated to a single operation. All at a cost of under $15 USD.

The MCU will report each connection and it's ohm's sequentially so u can fill out your log entries quickly. It can even 'remember" some of the last few devices tested limited only by the installed eeprom.
It's really not hard, nor expensive.

 

[MikeMl]

Either I missed it, or it hasn't been discussed, but what is the voltage range being switched? Any given resistor being measured must first be isolated from an external circuit, and then connected to an "Ohmmeter". Presumably, during testing, the external circuit is not powered?

 

[KeepItSimpleStupid]

The ON resistance of your switch is 45 ohms. This adds to the series resistance. Two switches would add 90 ohms.

 

[auto_turret]

Thank you all for the help. I am going to scrap what I had in mind at first and explore the options you've all suggested. You've all given plenty of things for me to read up about. And yes, I do need to do a little more homework to figure all this stuff out!

 

[KeepItSimpleStupid]

I don't know exactly what your trying to do, but I assume it's to check for shorts and opens in a cable, lets say 16 wires or so.

In order to get really low resistance values, like in milliohms, you need a 4-wire ohmmeter. With a 4 wire ohmmeter, the contact resistance of the relay will be removed.

So, you would need a 16x2 for one end and a 16x2 in the other end. Your switching 2 wires to each end.

Milliometers can only compensate out a small amount of lead resistance.

So, with two such matrix card, you can read the resistance between any wire and any other wire.

To reduce test times for wire shorts, you can do things in groups. wire 1 to (wire 2-16) and then wire 2 with (1,3-16) etc.

I do believe that there are commercial wire harness testers available too without having to roll your own.

Good luck and come back.

 

[Mosaic]

He didn't seem to require milliohms. Perhaps tenths, 2 orders of magnitude easier , and quite doable with a 10bit+ ADC and a correctly selected precision resistor voltage divider.

 

[auto_turret]

I liked the reed relay matrix card idea, looking at the schematics it seems like the most logical choice. But it needs to be plugged into a pc and we can't really cart a pc around to do these sadly. I went ahead and ordered myself a microcontroller kit to learn with, then I was going to get one that I'd actually be able to use for the purpose. And maybe i'd even be able to use the kit one, since it has an ADC onboard. You all have been so helpful. Thank you.

 

[auto_turret]

I've chosen the ATMega168 as my platform for the box. I chose it because the people I bought it from sold it with a beginner's guide, they fully support everything that came in the box, and it's reprogrammable. I'm sure most of you know who I'm talking about . It has a 10 bit ADC like the MCU K.I.S.S. suggested. Last night I got the package when I came home from work and stayed up most of the night playing with it. Everything works great so far.. I've gotten my first program (well not mine, it came with lots of example code) loaded on it to display words on a 4 line 20 character LCD. I can tell I'm going to be spending many sleepless nights with this thing. If I ever get this figured out to work in my intended project, I'll let you guys know. For now I've got alot of tinkering to do. This is the perfect way to marry my slight programming knowledge to my slight electronics knowledge. I love it. Again thank you all for the suggestion!

Edit: Oops.. he said 16 channel ADC.. ah well I'll figure something out eventually! durrr

 

[Reloadron]

A word of caution. Getting into a MCU can lead to many sleepless nights. They are addicting creatures.

Let us know how things progress and the forums have an excellent micro controller section.

Ron

 

[KeepItSimpleStupid]

Now, you could have asked about a suggestion. Why not program the MCU in BASIC? These kits look very promising: **broken link removed** I'm thinking of a possible project to use them on.

Also check out CPUStick™ and StickOS™ -- Embedded Systems Made Easy. The BASIC can run in a DOS window for free so you can PLAY without any hardware attached. It's the same BASIC.
Also look at youtube. Let me google that for you