hckrnws

  I’m surprised that they chose to add a bunch of components to feed the AC line frequency to the microcontroller instead of just using a 32.768 kHz crystal. A single crystal oscillator seems like both the cheaper and more accurate option

[0] https://en.wikipedia.org/wiki/Utility_frequency#Stability

My microwave seems to gain minutes per month, my assumption was that it's due to incompetency of Eskom, the essentially sole producer of South African electricity. With the government and parastatals here incompetency is very common.

However, out of interest I just pulled yesterday's stats from my inverter on Sunsynk's website. It has the frequency of the grid at 5-minute intervals and the average over the whole day was 49.975Hz which doesn't strike me as particularly bad, so I have to wonder if the Microwave itself has an issue. It's a Samsung which is now 13 years old.

> the average over the whole day was 49.975Hz which doesn't strike me as particularly bad.

A day, having 86_400 seconds in it, is equivalent to 4_320_000 pulses at 50 Hz. At 49.975 Hz, it's only 4_317_840 pulses which is 2_160 pulses too few. Which, at assumption of 50 Hz, translates into discrepancy of 43.2 seconds, in this one day.

So, no, it's a pretty big discrepancy actually, over here anything over 0.2 Hz is legally declared to be "degraded quality", and it's been debated for years that this is actually a way too wide margin but the electricity providers/grid operators managed to successfully argue that they can't afford upgrades.

Moral of the story: don't get cute when designing electronics, just use AC/DC power supply and put a damn crystal oscillator as every other reasonable person.

I'm guessing you're being downvoted largely due to the "don't be snarky" rules.

You're right (by my maths too, which I only did now) about it being a discrepancy of 43.2 seconds per day, which as you say is quite high.

However, it is my understanding that most grid operators are actually very good about maintaining a 50Hz average over a day specifically for devices doing time keeping based their frequency, I've heard they intentionally run the generators faster or slower at certain points in the day in response to needing to get the average right over a day.

I used to have no issues with time drift on my microwave, only started in the last few years.

[dead]

> This control board uses the same microcontroller GPIO pin to both drive segment A of the LED display and sense the door switch.

Is it necessary to be so skimpy with a safety feature?

Later sections of the article detail how there are multiple different safety features integrated to guard against various different failures triggering the magnetron with the door open.

The post explains how the microcontroller sensing the door switch is not a safety feature.

I'm so glad that my ancient Moulinex microwave has no display at all and no keypad. Just a motorised dial to set the time and another dial that sets the duty cycle.

Articles like these are great to argue nonconformity which can get you your money back in EU. Even past the warranty period.

From what I can find, the guarantee period seems to be two years, after which the burden seems to 'flip'. Given that this microwave is at least five years old, I am not sure what standard one might cite to demonstrate 'non-conformity'. Do you know of a standard which says that a consumer microwave oven must work for more than five years?

https://europa.eu/youreurope/citizens/consumers/shopping/gua...

Yes, the warranty period for consumers in whole EU is the minimal amount: two years.

But some countries have a more lax law (Germany, for example, not).

For example, here in The Netherlands you can have more than two years warranty. If you buy a premium smartphone (say: the latest iPhone or Google Pixel), and it stops functioning after twenty-five months (two years and one month), then your warranty isn't exhausted because a premium / flagship device like that (costing that much) is to be expected to work longer than twenty-five months. So, your warranty is still active. Now, if you bought a budget smartphone, you're probably out of luck. This is also why, yes, sometimes Google Pixel devices are cheaper in Germany. But you'll have less warranty if you buy it from there.

Then there's non-conformity. A device like a microwave isn't supposed to stop working because of a LED display suddenly turning on due to a hardware failure. Especially not if a blue LED display has this issue far more often than a green one. You can argue non-conformity with the seller (so the company who sold you the product; not the manufacturer), and they have to figure out how to handle it with the manufacturer; such isn't your issue as consumer. Only issue is you need a lawyer to write the letter for you (but there are nice examples available online which you can copy/paste, and there are also some very nice lawyers who do this either for free or low fee).

For businesses, different laws apply…

[dead]

The warranty/guarantee is different from the average lifetime of the device, which is defined in the law as the average period a device must maintain its working parameters stated by the manufacturer.

In some countries a fridge is expected to have a lifetime of 10 years but a warranty of 2 years. So in theory if you could prove that the fridge cannot meet the expected average lifetime, the cooling ability decreases below the stated parameters much earlier, then it becomes the manufacturer’s responsibility.

> then it becomes the manufacturer’s responsibility.

Note that it becomes the _sellers_ responsibility - this might be the manufacturer if you bought it direct, but otherwise it’s the retailer

You are correct, I wanted to say the manufacturer is ultimately responsible. The consumer deals directly with the seller but the manufacturer still takes over when the seller doesn't exist anymore.

My guess is the LED's suffer reverse bias thermal runaway when they're hot from being in a steamy enclosure and then they get a reverse 5v across them and any leakage current turns into heat accelerating the process.

Wouldn't it be more likely to be reverse-bias degradation of the LED junction causing permanently increased leakage current?

All LEDs are photodiodes too, certain degredations of parts or poor circuit design could lead to the display turning into a switch.

Very impressive engineering on the door switches. On the display, not so much.

sort by price strikes again

168 points and 116 comments at the time: https://news.ycombinator.com/item?id=41480038

More proof blue LEDs are the devil and should have never been put into all of our electronics to be the shining beacon of "OW MY EYES" at 2 AM.

This is literally evidence of stuff being designed to fail. An extra diode costs less than a cent at production scale. This was a manufacturing choice, not an error.

My microwave mainboard failed because I changed the range light bulb without unplugging the whole microwave first, which I would not have thought necessary. It seems that, without unplugging the whole microwave, the act of changing a light bulb will cause catastrophic voltage to delicate parts. Turns out to be a common thing with this brand.

I ended up replacing the mainboard with a part from no apparent manufacturer with new features (the blue LEDs dim after inactivity so as not to illuminate the whole room at night) and no connection for the thermistor. Works like a charm. It feels very much like the original manufacturer wanted the board to fail and be replaced, while some random Chinese circuit-board maker sold me a better quality board.

nah, this is just not something designer would expect to fail like that. The LED has datasheet, the datasheet have leakage current, it has no data on increased leakage over years, you plan for what you have.

What would help is not randomly planning for some of the segments to fail (they are multiplexed with other things, you'd have to put more diodes), but to just get slightly better/less cheap LED display

Only "choice" made here was sorting by price when buying components for the cheap device.

Sounds like this is far more common a problem with blue LEDs than others, and that was certainly a choice.

As if I needed another reason to detest the eye-searing blue LEDs that have infested every device.

I miss the red led of my youth. Modern red and green diodes are both too harsh in comparison.

Amber is where it's at.

Eh, I don't agree.

LEDs are diodes (Light emitting diode). Certainly this was a cost saving measure, but it's not a bad assumption that the LED wouldn't allow reverse current flow.

It’s not exactly designed to fail, they just don’t care. If they could add a one-cent part that made it fail sooner, they wouldn’t do that either.

How is designing it to fail and not caring about part longevity not the same thing for the buyer?

Avoiding products where the manufacturer doesn't care about longevity means looking for products that are well designed. Avoiding products that are designed to fail is trickier, because even otherwise high quality products can be designed to fail. Random example, I had a nice Brother laser printer where the toner cartridges had a hard limit on how many pages they would print. This was done with a mechanical mechanism, so it was possible to disassemble it, reset it by turning a gear, and keep on printing. This took time and money to build just to make the product worse.

Intention.

Don't underestimate the appeal of saving one cent per unit. So long as the costs are externalised, anyway...

The correct calculation is what proportion of your profit margin will be saved by that one cent per unit. It gets more complicated if that control board was outsourced. If it costs 10 USD for the sub supplier to make and they sell it with a 10% markup then saving that one cent adds 1% to their profit.

what happens when that diode fails?

I don't think it's so much an issue of designed to fail as trying to get it as cheap as possible.

Theres further issues with everything coming out of china and a brand slapped on it. No one is left to take responsibility on the engineering front. This feedback I doubt will get to the correct people at best buy, let alone going back to the microwave manufacturer. And then there's the question of if they care, as they aren't a customer facing brand.

Capitalist profit motive strikes again. The invisible hand expands tech and the visible hand keeps making tech worse.

People usually respond to this by saying that it would be absurd to suggest the company did this for its own benefit, when anyone who engineers knows these are often caused by revising design to minimize costs... and increase profits.

Prev discussion:

https://news.ycombinator.com/item?id=41480038

Sounds like textbook criminal negligence.

A public prosecutor should take this on.

You can do an awful lot to make a device like a microwave safe with loads of failsafes...

But rarely do those failsafes protect reliably against 'the mainboard was splashed with salt water'.

Even with triple redundant relays, how do you know the salt water didn't just wet them all?

In almost every system with failsafes there will be conditions that can bypass them. The goal is not to make it impossible for the unsafe condition to happen, but to make it so that in the expected uses the failure will not happen.

In this case it's a domestic microwave and the mainboard is housed inside the electronics enclosure, so covering the whole mainboard in salt water is not an expected occurrence in a domestic kitchen.

But there are ~1 billion microwaves in the world... I'm sure it has happened somewhere. As a designer of a billion-sold device, your job is to make sure that the expected number of people harmed by your device is substantially less than one, which gets really hard when all the risks are multiplied by 1e9.

Your job is to make sure the number of people harmed _while using the device as intended in a reasonable situation_ is as close to 0 as possible.

A domestic microwave is for use only on land, indoors, in a domestic kitchen, and in an unmodified form. In these conditions there is no conceivable way that salt water could saturate the main board, or bypass all the interlocks in another way.

Yes there are ways that all the safety systems can be bypassed, but not while a reasonable person is using the device as intended.

> As a designer of a billion-sold device, your job is to make sure that the expected number of people harmed by your device is substantially less than one

Source? People take risk in their day to day life and should expect to take risk. Why would they expect their microwave to be completely free of risk?

I noticed when tearing down an old microwave for salvage that the light bulb was part of the power circuit. If the bulb burned out, so did the microwave.

Reminds me of a comment from a previous time this story was posted here:

https://news.ycombinator.com/item?id=41509748

> Even with triple redundant relays, how do you know the salt water didn't just wet them all?

The design typically includes a mix of normally open and normally closed switches. If everything failed in the same direction (closed) it wouldn't satisfy the failsafe.

If you're spilling conductive liquid on the board, it's going to blow fuses anyway. It's more likely to short to ground than to short only to the precise path needed to activate.

In that situation one of the switches should short the mains voltage and blow the fuse when the door is opened.