From what I read there is also omnipolar, such as the Honeywell ss451a, and from what I understand, that would be perfect for my application.

The unipolar, bipolar, and latching ones, it seems to me, would all act the same, namely closing and opening the circuit (or turning on and off - I’m unsure about the correct wording) with the same frequency as the AC power supply switches direction, due to the magnetic field reversing polarity at that same frequency.
Whereas the omnipolar Hall effect sensor would remain “on” whenever the solenoid is activated, as it is indifferent to the polarity of the magnetic field, only turning “off” when there is no magnetic flux present at all.

Would you agree?

The answer depends largely on your overall strategy. If you're thinking of using a microcontroller, then it can handle the brief off periods and/or polarity reversals with pretty simple code, so most versions are ok. Latching might be problematic, but the others would all be easy to interpret in code.

If you want to stick with your current schematic as closely as possible, then you'll either need some analog filtering or you'll need to do the switch to DC coils and rectifiers like Max suggested. If you go DC, then all but latching work again. If you keep the coils AC, I would think bipolar or omnipolar would be better because you'll have twice as many pulses available to filter, making the filter circuit a little easier to implement.

P.S. Gotta get back to work now, but I'll respond to the grinder question this afternoon.

 

I’ve been watching this conversation between you guys and it’s been very informative. Thank you.

As for the timer project: I’ve ordered attiny85 Microcontrollers and Honeywell ss495a Hall effect sensors. There’s some wait but I expect they’ll arrive beginning of next week. In the meantime I will study to learn more about these sensors. will post updates when I’ve made any progress.

Ebeowulf, thank you also for your insights and personal experience regarding scales. We have (semi-)reliable scales at all our venues. The problem is, though, that not everyone who works behind the machines is as involved with quality as we’d like, and it’s been my experience that often the weighing and timing is passed over due to this.

Our products are of high quality (we import and roast our own coffees), and it is my wish to make the measuring of parameters as easy as possible - hence these projects - so that even the least passionate and most lazy “baristas” will be able to serve high quality coffee.

We use Mazzers, on which one programs the grinding time. The point of my envisioned scale is that it will ultimately be connected to the grinder (I’m aware this will need to be “invasive”) so that it will stop the grinding process automatically at a set dose, rather than at a set time.
More clearly: I want to portafilter holder on the mazzers to be equipped with strain gauges, and then when the grind button is pressed, the scale will first reset to zero, then start the grinding process, and interrupt it when the desired dose is reached.

From what you tell me, this is far-fetched. You mention accuracy and durability as main obstacles. Could you elaborate? I’d imagine one would put the strain gauges somewhere where they cannot take any direct hits, and just calibrate them with a microcontroller. This may be incredibly naive, and if so, I’d like to know why.

Melvin

I'll back off a little from my earlier statements. It's certainly not hopeless. Since you want it to be integrated with the grinder, both physically and functionally, it makes much more sense than I originally thought for you to try to make your own custom system. I was thinking you just wanted to make your equivalent of the Acaia, Hario, etc.

As for what I've seen fail, it's just about anything that could. Lots of them die after getting wet, which is unavoidable given the environment. If you're mounting this to a grinder (which is tall,) you can probably get the sensitive electronics up out of the drip/splash zone.

Most of them would gradually drift over the course of minutes or hours, sometimes a gram or more. In a situation where we didn't expect to be taring them frequently, this was annoying. Since you plan an automatic tare step at the start of each grind, this slow drift may not matter to you.

Some of them had noisy readings even in the best of conditions, randomly jumping quickly +/-0.3g or more without any outside disturbance or explanation. Some others failed completely after being hit with too much weight. These two problems are related, sort of. It has to do with resolution vs. total capacity. It's easy to get 0.1 gram resolution if the most you'll ever weigh is 10 grams, or 1kg resolution if your max is 100kg (100 steps, 1% increments, in both cases,) but it's hard to get 0.1g resolution if your scale must handle 5kg weights (50,000 steps, 0.002% increments.) Load cells that can handle large weights without being damaged put out very small signals for lighter loads. This means you need very stable reference voltages, high resolution ADC, numerous safeguards against pickup of external noise, etc. All of this is expensive and requires careful design. So, generally people cut things really close as far as max physical capacity vs max expected load. Then, any inadvertent overloading is enough to break it. There's little margin for error. On the other hand, if you over spec for safety, you either sacrifice precision or price.

There are apparently mechanical means of protecting parts from overload, but I've tried to research them a little and gotten nowhere so far. Any scheme I can think of requires precision machining, manual calibration, etc. It would probably be worth looking at how La Marzocco and Nuova Simonelli / Victoria Arduino handle this. They've integrated scales into their machines with the resolution you need, and I presume they've taken steps to protect them, but I don't know how it's done.

Sorry I've rambled so much. Technically this is off topic for the current thread, which is interesting enough on its own merits. It would probably be worth starting a separate thread for the grinder scale project. You'll get more useful advice that way, from people who know electronics a lot better than I do (I came to electronics from a coffee background only about 4 years ago, so I'm still a novice compared to most contributors here.)