Creating a replacement pcb

[dknguyen]

I don't know who designed the originals but just about every bit of space between tracks is filled with areas connected to ground. Absolutely nightmare when re-tinning everything as whiskers catch me out every time.

That's pretty standard. Is faster to produce because less copper needs to be etched and makes less waste and more environmentally friendly because less etchant is consumed. Also, it helps keep the copper on both sides of the board even to prevent the board from warping, but this is apparently no longer an issue in modern boards.

I don't get why it's tinned so heavily either. It doesn't look like there are high currents anywhere. However, the planes don't cover the entire board so it looks like they were intentionally put there for some reason.

 

[Western]

That's pretty standard. Is faster to produce because less copper needs to be etched and makes less waste and more environmentally friendly because less etchant is consumed. Also, it helps keep the copper on both sides of the board even to prevent the board from warping, but this is apparently no longer an issue in modern boards.

Ok, thanks ... that makes sense.

I don't get why it's tinned so heavily either. It doesn't look like there are high currents anywhere.

That's my doing. Once moisture gets under the original coating and starts eating away at the copper, I scrape the coating off, polish and re-tin them to protect the surface ... then use a conformal coating.

You're right about no high currents also.

However, the planes don't cover the entire board so it looks like they were intentionally put there for some reason.

It's never really made sense to me either ... the majority of the boards from this manufacturer seem to be designed with most of the spaces between tracks filled. I find them much harder to work on than most other brands.

 

[Pommie]

If you trace the connections from the ICD socket to the 5450 you can then hook it up to something like an Arduino and work out what LED is connected to what output. I'm guessing that there is some multiplexing going on as you have more LEDs than outputs - you'll probably have to find the transistors that are doing the multiplexing so you don't turn more than one on at a time. Do you have any old boards that are scrap? Are you able to program? Would it be easier to just trace the whole board?

Mike.

 

[Western]

I'm guessing that there is some multiplexing going on as you have more LEDs than outputs

The 5450 only requires 5V, GND, DATA, CLK_IN from the ICD input ... the rest of the pins drive all pins of the 7 seg displays. Plus an extra one for brightness adjustment. Pretty much a stand-alone part of the board.

The other 10 leds and 16 x switches are handled by the other 2 ICs with some minor, odd multiplexing.

This section uses 5 In and 2 Out to the ICD connector ... where the micro on the rear board does the interpreting.

Do you have any old boards that are scrap?

No, so far I've fixed them all. I probably will need to sacrifice one eventually to use as a template perhaps.

Are you able to program?

I can ... wouldn't say I'm good at it ... but have some experience and do enjoy it.

Would it be easier to just trace the whole board?

Yes, that part's done. Has actually taken me years to complete the whole board ... had just done the parts I needed at the time. Some of the finer points I didn't understand until just recently.

Like two separate switches would not work (sometimes) ... turns out a led was leaky ... ~200 ohms each way on both analogue and a digital meter ... yet the led worked perfectly ... same brightness as the others. It is part of the "odd" multiplexing I mentioned above.

 

[kubeek]

I think you could easily take off all the components and scan both sides on a scanner. Then import it, do some corrections on the shapes and you are basically ready to produce new PCBs.

 

[Western]

I think you could easily take off all the components and scan both sides on a scanner. Then import it, do some corrections on the shapes and you are basically ready to produce new PCBs.

Yes ... that will give me an exact size and physical locations for mounting holes, switches, leds, displays ...

But then I want to relocate all the ICs and components to the back as smd's ... otherwise sooner or later I'll be facing the same issue with corroded boards.

If there's next to no tracks on the front ... particularly around the bottom half ... most of my issues will be gone.

Any tracks needed for the switches can be large and spaced far apart ... compared to the existing layout.

Then I just need to use a better conformal coating and they have to be 100's of percent better protected than what we have now.

 

[Nigel Goodwin]

I don't know if you have milk on your cereal for breakfast ... but these things are used to collect it.

That's not BS on the front of those boards ... that's CS.

Reason they want them fixed is that many of them are used on farms with 40, 50, 60 ... maybe as many as 120 of them. To put in something newer they have to buy a complete new batch ... along with the associated hardware. Absolutely megabucks.

Industrial use, lot's out there, expensive to buy - looks like a reasonable proposition to make a replacement board - have you checked to make sure you can source all the parts you need?.

Have you also considered completely redesigning the board? - do you know what the connections are from the front board to the back one? - if you do, or can work it out, you could redesign the board to use a micro-controller, which would probably simplify things somewhat.

 

[AnalogKid]

when I have peeled away some of the tracks I have been amazed at how thick they are compared to many modern boards. Maybe twice the thickness at least.

Sounds about right. Today, 1/2 oz. copper is common. Back then, 2 oz. copper was standard and 1 oz. was increasing in use as it became more trusted.

Depending on the age of the design, the 40 pin part could be an 8051, 68xx, or an early PIC. Any date codes on the old chips or on the pc boards?

ak

 

[Nigel Goodwin]

Depending on the age of the design, the 40 pin part could be an 8051, 68xx, or an early PIC. Any date codes on the old chips or on the pc boards?

He's already mentioned that it's not a micro, it's an M5450 display driver - interestingly it's listed at MicroChip as an MM5450, and they have 6337 DIL versions in stock

 

[Western]

have you checked to make sure you can source all the parts you need?.

Yes, all the bits are available still. The MM5450 is the dearest part and is available for $3.80 ea for a tube of 26. Remember I want to fit it on the back ... so I'd use a PLCC44.

The switches are ~$1.91 each .... and the last tube of displays I bought cost $2.39 each.

The other two ICs are a 74LS145 (1 of 10 Decoder/Driver) and a CD4093 (Quad 2 Input NAND Gate)

Have you also considered completely redesigning the board?

Mmmm ... hang on ... by redesigning the board do you mean redesign the board ... or redesign the circuit ??

Certainly I want to redesign the layout on the board ...


... but redesigning the circuit ... I did want to avoid that ... but I have considered it.


All these great questions have forced me to think it through even further.

I guess if I used a micro to read the input for the existing display driver ... and multiplexed the outputs to the 7 segment displays ...

... and then use further inputs to read the keyswitches and send key presses to the rear board micro ... and lastly provide outputs to drive the 10 individual leds.

Then I could certainly leave out the two separate ICs.


I can't see it being all that much less complicated track layout wise ... still have to have tracks to all the displays and all the keys.


Again ... thinking about it some more ...

Multiplexing the 4 x displays would cut it down from 32 traces up to that area ... down to just 12 ... mmm!

do you know what the connections are from the front board to the back one?

Yes, pretty much ... there's 11 of them.

There's 5 volts and GND for the board ...

Then DATA and CLK_IN for the M5450 ... I need to read up on the datasheet to see how it's formulated.

At present there's 3 x BCD lines to the 74LS145 Decoder for reading the key switches ...

... and helping switch the 10 x individual leds (split into 2 groups)

Plus 2 more Inputs to select which of those 2 groups to turn on ...

Then 2 output lines back to the rear board micro to tell it which switch has been pressed.

 

[AnalogKid]

He's already mentioned that it's not a micro, it's an M5450 display driver - interestingly it's listed at MicroChip as an MM5450, and they have 6337 DIL versions in stock

OK, that just means that the microcontroller is somewhere else. Any idea what it is? Based on the Microchip display driver, I'd guess it is a PIC.

Separate from that, why does the TS assume that SMT would be more reliable than PTH? If solid/liquid contaminants are a recurring problem, I would think the wider pin spacing of PTH would me a bit more immune. Unless the idea is that SMT allows relocating components away from the contaminant ingress points.

ak

 

[Western]

OK, that just means that the microcontroller is somewhere else. Any idea what it is?

Yes, it's a Motorola MC68HC11F1CFN2

Separate from that, why does the TS assume that SMT would be more reliable than PTH?

That is a good question ... and it is not so much a belief that an SMT device is any more reliable than a PTH ... my reasoning is because 95% of the time, the areas of the board that get corroded are on the front ... across the bottom.

Having an SMT on the upper back where it virtually never suffers from corrosion makes the most sense to me ... plus that keeps all the fine tracks that drive the 7 segment displays on the back and up the top as well, all out of harms way.

I would think the wider pin spacing of PTH would me a bit more immune. Unless the idea is that SMT allows relocating components away from the contaminant ingress points.

I definitely do agree about the wider pin spacing ... but the second sentence is spot on.

 

[dknguyen]

you can't conformal coat the problem areas?

 

[JonSea]

There will be some cost in either time or money to lay out a new board, but if you have the boards fabricated and assembled in China, they will cost little more than you can buy the parts for, even for a small lot of 10 or 25.

 

[hyedenny]

I like DipTrace .
It's very intuitive, it has a HUGE library, new components are easy to make or customize. it has built in and customizable design rules, it has a pretty good auto-router, it has exportable and importable gerber, DXF, NC, ASCII, etc files, AND IT'S FREE (for up to 200 components).

 

[Western]

you can't conformal coat the problem areas?

They are already coated from the factory and I have tried several different types. It holds up for quite a while but eventually breaks down. I don't seem to have major problems with any other boards.

I've just started using a new one (to me) ... Electrolube FSC. When it's dry you can not use probes on it to measure anything ... you have to scratch and gouge just to get a reading. Makes fault finding a real chore.

There will be some cost in either time or money to lay out a new board, but if you have the boards fabricated and assembled in China, they will cost little more than you can buy the parts for, even for a small lot of 10 or 25.

Yes, that was the conclusion I had come to as well, and initially at least those numbers make sense.

I like DipTrace .
It's very intuitive, it has a HUGE library, new components are easy to make or customize. it has built in and customizable design rules, it has a pretty good auto-router, it has exportable and importable gerber, DXF, NC, ASCII, etc files, AND IT'S FREE (for up to 200 components).

I hadn't heard of that, I shall have a look. Have been watching various tutorials on You Tube on software tools. Some of them make it look really easy. Perhaps I can learn one of them.

 

[Western]

Hi guys, I thought I should update this thread seeing as I received a lot of help earlier.

I read most of the recommended material … decided to download DipTrace … and have spent a lot of time on it.

I managed to draw out the circuit on two sheets … main thing I haven’t worked out is how to share the header connector between the two sheets.

Otherwise very happy with the result so far.

I learned how to create patterns for some of the components … particularly the PLCC version of the M5450 … so the correct pads line up with the IC pins on the circuit diagram.

I’m still struggling with the way they operate their libraries … hopefully that will make sense eventually.

I even managed to go through the verification process and tick off all 30 or 40 errors!!!

I opened the circuit up in the pcb layout section … and have placed the parts that must be in certain physical places (to remain a drop-in replacement board) … all the switches, leds, displays, mounting holes etc.

So I’m at a point where I need to work out how to do the routing.

I understand it’s a no-no to use auto routing … and my attempts so far have shown me why … but I’m wondering what is the best way to approach it.

The main criteria being that it fits with my aim of having very few tracks on the lower half of the front of the pcb.