The Coffee Must Flow: Replacing A Spent Lithium Cell In A Coffee Machine

When [hacky] bought a used Douwe Egberts Gallery 200 all-in-one coffee maker, the machine was known to have a ’empty battery’. Being one of those fancy coffee makers that handle everything from the grinding of coffee beans to the application of hot water and steam, it relies on instructions for each coffee recipe. Unfortunately, it turns out that this machine stores these on battery-backed SRAM, as [hacky] found out with help from friendly folk over at the Dutch Tweakers forum.

The Douwe Egberts Gallery 200 is a rebranded machine that’s also sold in Scandinavia as the Wittenborg FB 5100. These machines have an ST M48T58 TimeKeeper module that combines 8 kB of persistent SRAM with a real-time clock. Being powered from a single coin cell (lithium carbon monofluoride chemistry), their lifespan is limited.

Replacing the coin cell in an M48T58 TimeKeeper module with AA cells.

Fortunatley, a DE-9 connector is provided on the back to provide service/maintenance access to to the hardware. Using a conveniently available programming guide for the hardware, it was easy to figure out the pinout and baud rate (9600, 8 bit, ignore parity, no flow control). This allows for reprogramming the SRAM, but without replacing the battery this data would be gone again on the next start.

Based on the ST M48T58 datasheet, it’s not clear that the clip-on module containing the coin cell and crystal can be replaced, though one could simply plug in a new M48T58 module. Or, as [hacky] did, it’s also possible to cut open the ‘SNAPHAT’ top section and wire in a replacement battery module. With two 1.5V AA cells providing the 3V to the module, it was operational again.

Next up: working out what to write to the SRAM to make the coffee flow again.

37 thoughts on “The Coffee Must Flow: Replacing A Spent Lithium Cell In A Coffee Machine

    1. From the datasheet : “… The SOIC and battery/crystal packages are shipped separately in plastic anti-static tubes or in tape & reel form. .. “.

      So, just a case of being more fun to devise a different way to fix the thing than just replacing a part.

    2. Not really, this is a commercial machine and if it is kept plugged in the battery would have lasted almost 50 years according to the datasheet. Even unplugged it should have lasted 10 years with the smaller battery installed. And they still make both sizes of batteries for them. https://www.st.com/resource/en/application_note/cd00004080-predicting-the-battery-life-and-data-retention-period-of-nvrams-and-serial-rtcs-stmicroelectronics.pdf

      I would have just put the the proper battery replacement back on there than rising a couple AA’s to start leaking at some random point in the future.

      1. Almost 50 years is still a planned date of obsolescence. Why didn’t they use non-volatile memory instead of SRAM? Being a commercial machine has nothing to do with any of this. Making batteries for it NOW doesn’t matter either when you need to replace the battery in 50 years. If they’re making replacements now, the maker isn’t expecting them to last 50 years. Your claim against planned obsolescence is far weaker than the claim for.

        1. I just finished reading that datasheet. If anything, it tells us exactly why they chose these parts to store the brewing profiles: Once the battery dies, the machine becomes useless. The only advantage these have over EEPROM is the data WRITE speed, which in the case of storing coffee profiles, is useless.

        2. Almost everything will be obsolete in 50 years. Even a EEprom probably wont hold data that long, the only thing that would is probably core memory. Even if the electronics survived good luck on most of the mechanical. There is not planning for most obsolescence, it just happens because technology is a moving flag.

          1. I don’t see how a coffee maker becomes obsolete. If it can still brew coffee on its 50th birthday, it’s not obsolete. Any coffee maker that uses degrading memory is built with the mindset of it being thrown away, not because it has broken due to use, but because it was never intended to last. If something can be repaired or parts replaced, that is not planned obsolescence, but if replacing the broken part leaves a device that refuses to work, that counts as planned obsolescence.

            So, to prevent a useless debate, you deny the existence of planned obsolescence, do you not? If so, there’s not point to continuing this conversation because I do believe it happens and your opinion will not sway me from the evidence in front of my eyes.

          2. EEPROM, Flash, and EPROM all degrade eventually – especially at higher temperatures, as one might find in, say, a coffee machine?

            See https://www.st.com/resource/en/application_note/dm00436563-cycling-endurance-and-data-retention-of-eeproms-in-st25dvxxx-products-based-on-cmos-f8h-process-stmicroelectronics.pdf, chapter 3 – 40 years retention at 55 degrees C becomes 5 years at 125 degrees C.

            Electronics (and especially semiconductors) don’t last forever. Transistors do age, and the problem is worse at smaller geometries. You can make tradeoffs between performance, cost, and longevity, but you’ve always got to pick a solution. That doesn’t mean people are intentionally making things expire at a certain point – it just means that they don’t prioritize making it last a century over making it small or cheap enough to actually buy.

        3. I think, given the form factor, that they have used the all-in-one RTC+SRAM+Battery pckages used for personal computer. To have an RTC one needs a battery and some register to save the current time.

          I suppose also that given the service manual is possible to enter in the “BIOS” setup mode and having the automatic coffee maker bak in working order.

        1. Why is an AA more likely to leak than the lithium battery this module used originally? I have seen a LOT of articles (here on Hackaday and elsewhere) about RTC batteries leaking. Therefore it seems that a quality AA can’t bee any worse.

    3. Bull…

      These “timekeeper RAMs” are a standard module for battery buffered memory (with RTC). Fast, reliable and unlimited write cycles.

      So what, it has a built in battery that, of course, will go flat eventually. Then you buy another one, stick it in, reprogram your machine (which is no secret, but standard operation).

      “planned obsolescence” is when a part fails that is not replaceable without reverse engineering or other unobtanium means…

      And for a tinkerer, replacing the battery is not really a hassle here. I did ist some time ago for all my old Sun and SGI systems.

      (I didn’t bother to be careful not to hurt the 32kHz crystal, though. I ripped it all off and soldered/glued on a CR2032 holder and a newly bought crystal.)

  1. If you Google Wittenborg FB 5100 you will see that this is a large pedestal mounted, commercial drinks machine – the sort that you might find in an office canteen. It weighs 82kg including the pedestal – which holds the water tank. It’s somewhat more substantial than your kitchen counter “Mr Coffee”.

    With an original price tag of several thousand dollars, these machines would be maufactured for the lease market, where restocking and servicing would be done by (possibly franchise) service engineers, under a lucrative service contract arrangement.

    The machine would be expected to deliver several hundred cups per day, and would be kept operational by monthly service visits. If it was located in a region with hard water, there would probably be build-up of scale on the heating element – again needing regular attention. Valves, pumps and dispensers would also need regular attention to keep them clean and working.

    Under these conditions of use, a machine might only have a 3 to 4 year expected lifetime, and this would all be factored into the business model of the leasing and manufacturing companies.

    The customer would be under annual pressure to upgrade to the latest model, again providing a constant income flow for the leasing company.

    When the machine was designed, the engineers probably were well aware of the expected machine lifecycle, and so the battery backed SRAM module was considered to be sufficient to meet the machine’s anticipated service life.

    Built in obsolence has been a main driving force behind manufacturing probably for the last 70 years or more. It doesn’t matter if it’s cars, smart phones, photocopiers, laser printers, computers, washing machines, household appliances or consumer electronics.

    Everything is designed to have a minimum acceptable lifetime – as judged by the manufacturer. If gadgets didn’t break, the manufacturer would struggle to sell the next generation and stay in business.

    1. As an OBM (Original Brand Manufacturer) I cannot agree more: everyday products are designed with planned obsolescence in mind (or as mentioned, minimum acceptable lifetime).
      Because of this free-market, designing product with a higher than minimum acceptable lifetime WILL cost more to the manufacturer, either from BOM point of view or lost sales of next generation.

      This view is accepted from both the manufacturer and customer and the only way to change it is to and another variable: resources and energy.
      But this change can only be induced by lawmakers within your country: when appliance will have 20years warranty by law, be sure manufacturer will deliver. Sure price will increase, but energy and resources will be saved, and customer will save money on the long run.

      1. Dunno, but the GDR could produce long life kitchen appliances, check the AKA electric RG28 as a prominent example. Unless you break it by force it won’t die and even a 35+ year old used one you pick up on eBay will last you a long long time.

        So yes, it is perfectly possible to create machines that would last a lifetime, even for good prices.
        It won’t happen under our current unregulated form of capitalism tough.

          1. No, but if you understand German, Volker Pispers Bis Neulich is a very enlighting cabaret act. You can find it on YouTube and you should watch it.

  2. I was asked one time to look into changing the lifespan of a motor from 500 hours to 250 hours in a product I was design engineer for.
    I blame US , the consumer. Give a choice between a 20 year product at twice the price of a 5 year product we will buy the 5 year one even though long term it’s more costly.

    1. We the consumers are part of the issue, but not all of it. I would (and do) happily pay twice as much for a tool that lasts four times as long when it’s something that will have a corresponding useful life (hand tools, flatware, pots/pans). I would also happily pay more for some things that have less features (‘dumb’ televisions, infotainment systems in vehicles) if/when it is available. I’m not paying twice as much for a lot of IT-related things though b/c the useful life is usually measured in two-digit months and not two-digit years.

      I do think manufacturers should be on the hook more for recycling ‘old’ products instead of putting the full burden on the consumer. I think that would do the most good in seeing quality in products rise, and better overall use of materials/resources.

    1. Yes… the all in one Dallas “module” (DS1287) that also contained a battery . There have been several resourceful souls that have hacked these modules to allow for an external battery to be used. Google dallas rtc repair or check out this page that provides a great pictorial overview.

      https://www.classic-computers.org.nz/blog/2009-10-10-renovating-a-dallas-battery-chip.htm

      Lastly…this hack does work, as I have done it myself on an older (1998) industrial machine that used these Dallas modules on their motherboard.

      Cheers

  3. I don’t know why people would rather destroy a IC such as this to replace a battery when you can simply replace the IC. If anything install a Socket and make it easier to swap it out again in 10 – 20 years. Also why put in AA batteries when a coin cell would of worked better, AA batteries will leak over time.

  4. I work on older systems from the early 1990s that use this exact part. The battery is NOT replaceable without cutting it out. It doesn’t “snap-on” to the top of the SRAM IC. There are tiny wires that you have to avoid for the battery and the crystal that are embedded in the epoxy The first time this part gets powered on, it enables the battery which starts draining at a faster rate than if it were on the shelf because of the way the logic in the chip is designed. There is a flag that gets set when the battery depletes, and if you don’t have access to the OEM code to disable the routine looking for the flag or try to reverse engineer the binary file to disable that feature, or try to change the code to work with newer NVRAM, this silly chip locks R/W in order to protect whatever is in the SRAM, but then you can’t access the data!!!. That’s why you end up resorting to this hack and hope you have copy of whatever data was there and a means to upload it…
    I don’t think this chip was designed with planned obsolescence in mind, I think that was just the state of the art of SRAM at the time it was designed.
    As far as batteries leaking, just use AA lithium batteries, they last way longer than alkalines and don’t leak…

  5. All the ‘what about leaking batteries’ haters. When I do this change (which I have done hundreds of times over the years) I always make sure the battery compartment is accessible to change over the batteries. I do this because I figure something has run down the battery to start with and it probably will again, although the AA’s have substantially more staying power compared to a button cell. This also means, if there is a leak, you can change out the batteries and clean up the mess or, just clip off the compartment and solder in a replacement easily. I would assume they use their heads and make sure this is easy to get to.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.