Galvanic Corrosion on external connector

[Ginettag20]

Hi every one

I am repairing some old data logging equipment that has an external connector for transferring the data via a dongle to a PC. The connector is a proprietary type built into the data logger enclosure and has about 6 small gold pins (pips) sticking out.
Three of the pins are used for TTL level data transfer via serial UART (RX TX GND).  Two of the other pins are used to detect when the mating half of the plug on the download cable is pressed against the connector.

The way this works is that the plug-in the cable short across two of the pins in the connector.  The first pin is an input to the data logger processor and has a 10K pullup. The second pin is at ground.   When the connection is made, the cable detect input pin is pulled to ground and this initiates transfer of data from the logger to the PC.

The problem I have found is that when the logger is installed in wet locations we get galvanic corrosion occurring between the ground pin and the cable detect pin. This is obviously being caused by the constant DC level on the input pin from the 10k pull up.   These connectors last indefinitely when installed inside but when wet only last 6 months and the connector then need to be changed. Strangely it’s only the two cable detect pins that become corroded, I assume the data pins must be driven low when idle so don’t corrode.

I have spoken to the company that makes the equipment and apparently, it’s a known problem but they don’t appear to be receptive to resolving the problem and just change out the connector. This is understandable given this equipment is now old and obsolete anyway. 

This data connector sits on a small daughter board and it would be relatively simple for me to modify the board whenever I swap out the corroded connector and this is my plan.My thoughts are to perhaps install one of those small 8 pins PICs on the daughter board.

I would connect the two cable detection pins to the PIC inputs via pull down resistors.  I would then allow the pic to alternative drive one pin high and set the other pin as an input, then do the reverse.  So both pins would swap between 0 and 5 volts. I would only poll the cable detect pins occasionaly and this would further reduce the time that either of the two input pins is powered.   

I could probably do something similar using a 555 and capacitively couple this to the  cable detect pins but to be honest I’m more comfortable using a PIC given I have more control over the configuration and pulse time shape etc.  Anyway, my question is has anyone encountered similar problems working with wet connections and how were they resolved. Does my solution sound reasonable? Any other ideas to extend the connector life etc.

Thanks   Chris


 

[Jeff_Birt]

I might try some dielectric grease on the connector.

[Ginettag20]

Deoxit have a grease formulated for protecting gold but I was worried this will be wiped away and would also atract dirt and grit. The connector is quite a tight fit with the housing.

Chris

[Fraser]

I prefer to resolve the problem n hardware and would approach yer problem from a different angle. I would make an adapter 'tail' that left the custom connector packed with grease and connected with its mate. At the end of the tail I would use a waterproof connector such as used with CCTV telemetry. These connectors have waterproof covers for when not connected to a mate. The download activation remains the same but you can choose pins that are spaced further apart and with the cover fitted, they should stay dry. No mods to the logger itself are needed so they are interchangeable with standard spare units if such is needed. Boats use all manner of waterproof connectors. Farnell and RS sell a good selection. I feel sure you can find something suitable.

I would use self amalgamating tape around the custom logger connector as that will seal it well against the weather. Such tape is used for radio aerial connectors and performs very well as it mounds tightly to the connectors cases forming an excellent weather seal.

I tend to modify equipment only if essential as non standard builds can be a nightmare to manage on the spares and support front. One spare that fits all scenarios is normall more economic and convenient and manufacturer warranty is maintained. I know that is not an issue in your case though.

Fraser

[salbayeng]

Yep its a pretty common problem.
I have a unit with pushbutton switches where there is 15v across the contacts when the unit is asleep, a little moisture in the switch and the unit goes crazy!.
Basically the anode corrodes, the way to minimise this is to make the anode bigger than the cathode, so the shell of the D type should be at 5v , obviosly unworkable.
Good old telephone cables run at 50v, but they are used with positive earth, so the dirt corrodes, and the copper wire at -50v is unharmed.

With connectors the only real way to get some success is to fill the connector with silicone grease (RS sells a really thick one) .  One of my products is used next to a swimming pool, and we conformally coat the PCB's and laboriously apply silicone grease to every single pin on all the connectors.

For connector presence detection , like in your application, simple capacitively couple to the sense pin (that gets shorted to ground when plugged in). If you use the "soft pullup" feature on your CPU , you can tell whether there is a capacitor present by setting the port low as an output, then make it an input, then set it "high", wait 10uS, read the pin, if its 1 there's no capacitor, if it's still 0 there is a capacitor.  With a  bit of programming you can distinguish between a capacitor, a 10k resistor or a short circuit. We use this method to check if a remote switch is plugged into a 10m cable, and can reliably detect which end the cable is unplugged , or localise a cut cable to about 2m , by timing how long it takes to float back up again. (It's poor mans TDR) very helpful where said cable runs through the ceiling space and down a wall cavity!

[Ian.M]

Mod the logger to add a reed switch in series with the cable detect connector pin so its volt-free while logging.  To read the logger, connect the interface cable and apply a magnet next to the connector to activate the reed switch.

I'd also 3D print some dummy connectors that could be packed with silicone grease and used to cap and protect the contacts.

[salbayeng]

Adding to Frazers notes, you can also get a variety of mastic putties, I have some called "AK22"  from Omega, it's like plasticine, you just squish it into your connector to fill the voids, then wrap with self amalgamating tape. Refrigeration mechanics , and fibreglass suppliers also have similar mastic materials. Actually poster putty works in a pinch "blu-tac".
The local electrical store (Lawrence and Hansen (Australia)) has recently starting stocking this blue gel that looks like snot in a tube, you just squirt it in to your housing.

[Ginettag20]

Hi Thanks for all the comments.

All good suggestions but unfortunately not ideal for my particular application and sorry for not fully defining this.  As they say in the movies “Some of names have been changed to protect the innocent”

It’s not practical to make any changes to the connector itself or the plug-in part. The connector data recovery aspect is done frequently many times a day.  The application its more like application of the Dalas touch key if anyone can remember that.
 Applying some grease may help but would have a cost due to the ongoing maintenance of sending out engineer to apply this grease.  So I need a fix under the cover hidden from the end users.

I remember that they used to make plant moisture meters that used AC fed electrodes in the soil. So I want do something similar.  I don’t think I can use a simple series capacitor with the existing circuit as I need to know the cable remains plugged in and when its removed.  A simple pulse would not be reliable enough under various weather conditions etc.

 The existing input to the processor is input only so If I want to do something toggling it would have to be by some additional hardware on the daughter board. Such as feeding the detection pins with AC then rectifying the returned signal before passing it the normal input circuit.

So do we think this approach will prevent the corrosion etc ?

[Fraser]

I see your problem.

Mmmmm now I also see why many modern loggers are IR or RF wireless downloadable.

An AC signal does reduce corrosion. In TV head amplifiers that were powered up the coax it was not uncommon to use 12v 50Hz AC on the coax centre to avoid electrolytic

An  AC output pin that is connected to a sense pin by the download cable would be a workable idea. The sense pin would connect to a Diode and small smoothing capacitor to generate a DC output to the microprocessor sense pin input. Is this effectively what you were thinking of doing with a PIC ? If so, it should work.

Fraser

[Ginettag20]

Hi Frazer

Thanks for the reply. The function of the two pins would effectively be swapped over constantly. On the first cyle input(1) would be pulled hi and input(2) would act as the cable detect input. For the next cycle input(2) would be pulled hi then the input(1) used as the cable detect input.  I can configure the PIC pins as inputs or outputs on the fly and just alternate between the two. The other advantage that the PIC gives me is that I can con figure the duty cycle to minimise the time either of the pins is pulled hi. For example, i can hold both pin low the every hundred ms do a 1ms scan.   So the majority of the time both pins would be a 0v.

 Hopefully this makes sense

Chris

[Ian.M]

You need to capacitively couple the sense pin to block all DC current.  If you simply alternate driving the sense pin and the former sense ground pin high, both will corrode at half the rate due to leakage to the data pins.   It is quite complex to do right, especially when you have to consider ESD  protection, hence my suggestion to use a reed switch to isolate the sense pin and add a magnet to the docking connector.  However that won't work if you need 100% compatibility with unmodified docking connectors.

[Ginettag20]

Hi Ian

Are yes, I see what you’re saying I’m still just pulling both pins to 5 volts.  I wonder if this would still improve things enough with the reduced duty cycle over the constant DC condition I have at the moment. Perhaps I should just couple a 555 timer capacitively and half wave rectify the return signal.  I guess I’m just happier programming a PIC than trying analogy electronics.

Yes spot on - I cannot modify the docking connectors but i can modify the internal circuit  on the small daughter board feeding the docking connector.

Thanks

[Ian.M]

A PIC is still a viable option.  Connect a capacitor to the sense pin and drive the other end of it from a toggling output via a 10K resistor.  Connect another 10K resistor to a PIC comparator input.  If the sense pin is floating there will be minimal phase shift between the toggling output and the comparator input.  If its grounded, there will be a significant phase shift due to the RC network.  The 10K resistors + the PIC's input protection diodes provide ESD protection.

[Ginettag20]

Hi Ian

Can you run that by me again. Not quite sure I understand what your proposing.

Thanks

[Ian.M]

With nothing connected to the external connector, the sense pin is O/C so the comparator input will follow the toggling output, as there are only 2x 10K resistors between them.  Ground the sense pin, and the capacitor must charge or discharge via the 10K resistor from the toggling output so te level on thew comparator input lags that on the toggling output

If you sample the comparator shortly after toggling the output, if it has the previous state, the sense pin is grounded.  If it has the new state, its floating.  Pick the capacitor value to get a suitable time constant for your detection routine.

You can spend most of the time asleep, with the comparator disabled and set as an output with the same level as the toggling one, and wake up using the WDT to enable the comparator and its Vref chain before toggling the output to minimise the current consumption.

[Ginettag20]

Hi Ian

I will schetch this up when I get a minute and try to knock up a prototype.

Thanks for the idea.

Chris