09/20/2011 EDIT: This is not a new thread. It was originally posted under a different title. I changed the title because it may have given some readers an unintended negative impression of the GS/3 that doesn't reflect my opinion of the product. I started the OP with the phrase, "I love my GS/3", and after two years of daily use I still do. The choice of title had more to do with the frustration I felt shortly after the problem occured than with reporting on the problem for the benefit of HB readers. I have since found the source of the water that got into the brain box (it wasn't from the vacuum breaker, as I thought), and as soon as I can find the time I'll post a final report here about how I determined the cause and will post a separate thread, with photos, on how I did the repair.

I love my GS/3, but as anyone who frequents this board knows, the machine can sometimes try one's soul.

I've had lots of issues with my GS/3, but I've always been able to repair the machine without missing a day of espresso. This time my GS/3 is dead in the water, and it's likely to be out of commission for a while.

Long story short, it looks like water got into the brain box and fried the CPU board. Before I get into the details, Dan thought it was important for me to put this note of caution at the beginning of my post:

Every GS/3 is vulnerable to water ingress into the brain box, regardless of whether it's a volumetric or paddle version, "firesale" machine or not. I'm not the first GS/3 owner to find water in the brain box, and probably not the first to need a replacement because of it. It seems to me that we GS/3 owners need to find a way to protect the brain box from water ingress. I will go into some ideas for that at the end of this post. Additional suggestions are most welcome.

Now for the gory details:

Normally, I leave my GS/3 on 24/7 due to a well-known problem with the vacuum breaker valve. If the machine is cycled off and on daily, mineral deposits form inside the valve, eventually causing the valve to not close completely or to stick in the closed position. If the valve doesn't close completely, water sputters out of it and wets the inside of the machine. If, on the other hand, the valve is stuck closed, air isn't forced out of the boiler as the water heats, and you get a false high pressure reading. First time you operate the steam valve, the air comes rushing out with a forceful pop.

The first time the vacuum breaker in my GS/3 failed, it didn't close completely. I cleaned the valve with vinegar and that solved the problem. After that, I left the machine on 24/7. Recently, when I had occasion to turn the machine off during a vacation, I got a high false pressure reading on the steam boiler when it reheated, leading me to suspect that the valve was stuck closed. I ordered a replacement valve, but have seriously considered just blocking off the opening and putting up with having to manually bleed the air through the steam valve after the steam boiler reheats.

In spite of the vacuum breaker problem, I don't like leaving the machine running when I'm gone for more than a day, especially during the summer when there can be lightning storms. So on Monday afternoon I turned off and unplugged the machine before leaving for a couple of days at the Jersey Shore. When I returned Wednesday night, I plugged in the machine so it would be ready for morning coffee. Knowing that the vacuum breaker might be stuck open or closed, I hung around the kitchen to make sure the machine reheated OK.

I should note that this point that my GS/3 is set to HALF heating mode. This mode minimizes the overall current demand when the machine is run on a 15A circuit by heating the brew and steam boilers alternately (not simultaneously, as in FULL heating mode.) The brew boiler always gets priority. This means that when the machine is first started, the brew boiler must come up to full target temperature before heating of the steam boiler begins.

When I saw that the brew boiler had reached its target temperature, I started watching the steam boiler pressure gauge and listening for sputtering from the vacuum breaker. Thinking that the breaker might be stuck in the closed position, I briefly opened the steam valve to let some air escape. When I did so, the motor/pump came on. It's not unusual for the pump to come on to top up the steam boiler when the heating cycle begins, so I wasn't concerned. But when the pump continued to run, I realized something was wrong. I turned off the machine and turned it on again several times, and tried letting more air out through the steam valve, but the pump came right back on and wouldn't stop. I must have let this go on too long, because water started flowing out of the steam boiler OPV into the drain box, indicating that the boiler was full.

At this point I figured something was wrong with the steam boiler level probe. I had a similar problem a year or so ago: one time after draining the boilers the pump kept running and overfilled the steam boiler. When that happened, I cleaned and reseated the spade lug on the level probe and that fixed the problem. There hadn't been another instance of the probe failing since.

When I went to remove the side panels to get at the probe, I noticed a small pool of water under the machine. There were also a few drops of water on the chassis floor inside the machine, and some water evident on both the left and right sides of the boiler. I wasn't terribly surprised to see this. If the vacuum breaker was in the "stuck open" mode, water would come spewing out of it when the boiler overfilled. I've seen that happen before. In fact, when I later inspected the vaccum breaker, I found that it was open and pretty-well fouled with mineral deposits that would keep it from fully closing. I was a little puzzled about water being on the left side of the machine, though, and am still not sure how it got there.

I mopped up the water, drained the steam boiler through the ball valve used for that purpose, and cleaned/reseated the level probe contacts. But when I plugged the machine back in and turned it on, the display indicated that the CPU had lost contact with the brew boiler temperature probe and the machine refused to come on. This is very unusual, and made me suspect that water had gotten into the brain box.

I immediately unplugged the machine and pulled out the brain box. I saw no evidence of water in the opening where the cables enter the box, which is where I would have expected to see it. I did notice, however, that water had filled several of the recessed screw holes in the brain box cover. I used a rolled up paper towel to wick the water out of the holes, and opened the brain box. I saw some drops of water on top of the CPU board. Then I unplugged the wires from the board and removed it from the box. There was at least an ounce of water sloshing around in the compartment underneath the board, and the bottom of the board was wet. There was no water in the adjacent compartment where the motor-start capacitor and solid-state relays are mounted.

I should pause here and say that this isn't the first time I've removed the CPU board. When I first got the machine a couple of years ago, it would randomly reboot. When the problem didn't go away, I removed the CPU board, wiped it down with alcohol and dried it with a hair dryer. That fixed the problem. At the time, the speculation was that a small amount of water had entered the brain box during shipping (the water being left over from testing at the LM factory.) I also had occasion to open the brain box last November when I outboarded my pump and motor and added a motor cutoff switch for preinfusion. However, that didn't require removing the board, just splicing some wires (and, no, there's no evidence that the outboard motor/pump modification had anything to do with this problem.)

I mopped up the water in the board compartment with paper towels, then proceeded to dry the board with a hair dryer. After doing so, I noticed a number of areas where whitish deposits had been left on solder joints under the board. I used an alcohol-soaked Q-tip to clean these areas, wiped down the board and dried it again with the hair dryer. I then remounted the board in the brain box and plugged in the cables (I photographed the cable positions before removing the board so I would know where to replug them, and also had several photos on hand from earlier excursions into the brain box.) Finally, I remounted the box in the machine.

When I replugged and turned on the machine with the switch in the back, the display backlight came on, but no characters appeared in the display. When I tried turning the machine on with the hot water button, nothing happened on the display, but I heard clicking similar to what I would normally hear. I figured perhaps one or both display cables had come loose, so I unplugged the machine, removed the brain box, and reseated the cables (though the display cables appeared to be properly seated.) This time I left the box out of the machine for testing. Once again, the display backlight came on, but there were no characters in the display. Even worse, I started hearing a buzzing sound from the CPU board. I think it's the PCB-mount relays chattering, but I'm not sure. There was no burning smell or anything like that.

At that point I was virtually certain that the CPU board was fried. I called Roger at Chris Coffee the next morning, and he got a price for me on a replacement board: $520! I realize it's a low-volume part, but I find it a bit outrageous that LM charges so much for the board when it's obvious that the board is very vulnerable to water damage due to a couple of serious design flaws. The first design flaw is that the vacuum breaker valve inevitably gets fouled and allows water to get into the machine. This can happen whether the level probe fails or not. The second design flaw is that any water escaping the boilers or tubing is likely to find its way into the brain box. The electronics should be sealed and completely isolated from any water source, period. This is not only a matter of expense, but a matter of electrical safety. I suspect that if NSF really understood the design of the GS/3 they would not have certified it for sale in the US.

Although repair time will be extended, I decided to send my CPU board to Chris Coffee so they can test it in one of their GS/3s to make sure that it's indeed fried. I went this route because the replacement board can't be returned to LM if it turns out there's some other problem causing the malfunction. I'm pretty sure CC will find the board is dead and will order me a new one. Talk about the high cost of owning a GS/3!

Incidentally, it's occured to me that the steam boiler level probe may not have been the cause of the original problem. It's possible that that a problem with the CPU board might have been responsible for failure to detect the water level. Once I get a working board, I'll be able to determine if the probe is OK or not.

As mentioned at the beginning of this post, it seems to me that we GS/3 owners need to find a way to protect the brain box from water ingress. Since there was no water in the right-hand compartment of the brain box, I think the water in the CPU board compartment must have gotten there through the screw holes in the top cover. You would think the screws and washers would seal the holes, but maybe not. Anyway, it shouldn't be too hard to cover the holes with some sort of waterproof material. Waterproofing the cover seam might be a good idea, too. But I can't think of a way to keep water from getting into the large opening where the cables enter the brain box. Maybe a rubber flap to deflect water away from the hole? Part of the challenge in sealing the box is ensuring adequate ventilation for the electronics.

This is certainly a disappointment. Like I said, I love my GS/3, but there have been a long list of problems with the machine. Most of them are design-related. Given the cost of the machine, we GS/3 owners deserve a lot better.

[BTW, CC has been great about this and every problem I've had with my GS/3. They always go the extra mile to help their customers, and there's no better dealer in the espresso business, period. But even CC can't undo the GS/3's design flaws.]

Dick - Even though I do not own, nor will ever own a GS3, I read your entire post. I feel for you, man. That is just sad. You may remember what I went through with my Vibiemme and the rust in the drip tray area causing the spot welds for the tray's supports to rust out and the supports to basically fall off. Still, that was an easy fix and a miniscule problem compared to what you are facing.

You would (IMO, rightly) assume that when you are buying a $6500 espresso machine that a good part of that is for R+D, some of which is done to avoid your problem. There are so many situations where this sort of thing is dealt with (in nautical electronics, automotive electronics, pro photography equipment, etc.) . When you start wondering things like, "I wonder if duct tape would work to protect the $520 electronics in my $6,500 espresso machine from water?" then something is wrong.

Whether you spent $500 or $6500, unsealed electronics at the bottom of a device inside of which are parts that are actually designed to spew water out? Ridiculous...

Silicone tape may be one possible solution.

There is a product from 3M called Scotch Kote and many clones of it. I used to fill wire nuts with it in damp places, never had a failure. Very stinky, lots of solvent in it but when dry quite water proof. Not sure about temperature rating. How hard would it be to move all the electronics to a remote box? That would be dry and cooler but might be quite a bit of work.
KDM

Randy G. wrote:Dick - Even though I do not own, nor will ever own a GS3, I read your entire post. I feel for you, man. That is just sad.

Likewise. You have my sympathies, and I hope you can get your problem sorted out and modify the box to eliminate any chance of it re-occurring. A $6500 machine should not have this kind of issue.

Even if they couldn't move it- There are many machine type computers that are encapsulated with a waterproof compound that does not damage components. And also protects against water exposure. I'm sure to the manufacturer its an additional cost. But then again they wouldn't sell as many replacement brains either..

Dick,
I had a similar problem with the motherboard sitting in a puddle of water that had leaked thru the 4 screwholes.

I did as you did, drying with a hairdryer but not fooling around with any solder points etc. After drying everything on the machine, it worked fine.

I put waterproof tape over the 4 screwholes. As for the entrance way of the cables, it might be possible to use silicone as it would grab the cables and plastic at the same time. If at a later date you want to remove it, then you can just peel it away.

Had you checked the rear of all your LED's and button contacts as a short circuit anywhere might affect it. I had water on the contacts behind the readout and dried everything via hairdryer and it works.

Drag to have to layout $520 for a motherboard and I agree that you would think that there would be a better watertight box ie: gaskets maybe, just closed. It is at the bottom and water falls to the bottom....

Potting: Kize reminded me that there is also the concept of "potting" where any type of apparatus is totally enclosed in poured epoxy. Rock solid but you can never get in again to test or repair. I have no connection to 3M but their products were always dependable in industrial use.
http://www.shop3m.com/80610533509.html?...otting-Kit

The main problem is sealing the opening through which the cables enter the brain box. It would require very careful application of sealant between the many cables in the bundle to prevent water wicking in between the cables. Second, it would eliminate the only opening in the box, which may (or may not) be required for ventilation.

I like the idea of remoting the brain box, but that's much easier said than done. Most of the cables would be relatively easy to extend, but it's possible that the sensor inputs have been calibrated for the resistance of their respective cables, and that lengthening the cables would throw off the calibration. Also, extending the display and button cables would require getting custom-made ribbon cables with connectors.

Dick -

There would be no technical problem with increasing the length of the RTD temperature sensors - this is done all the time as the resistance of the wiring is essentially zilch as compared to the sensor itself.

As a side note, the design of the GS3 vacuum breaker valve system has been greatly improved as shown in their latest parts breakdown. It essentially duplicates (but in a better way) the valve offered by Chris' Coffee on their Quickmill line. If you ever need spare o-rings, I have them - Rehabilitating the La Marzocco GS/3 vacuum breaker.