ZPM: The Kickstarter Failure That Revolutionised Espresso

Just wanted to follow up on a few comments here. Firstly: did not expect this many Stargate references in the comments. Secondly: the PID. I want to add a couple of bits of detail. Firstly, I was put off from excess tinkering by the fact that any changes to settings didn’t save beyond switching the machine off. Secondly, I did mess around with them a bit (having done something me reading to try and get some guidance) and nothing I did had any real impact on performance. (Though, like I said, I don’t really know much about programming PIDs.)

I had no idea this existed. And something about the Construction Zone Orange stamped metal panels screwed together with the cheapest fasteners really really appeals to me

I'm so glad we now have a sound bite of James saying "back that ass up". Marvelous.

I've said this in other videos, but I'd like to reiterate that I appreciate seeing CC (not auto-generated) on your videos. It made a big difference to me (and likely those who rely on CC for their media consumption). Very interesting video, thank you.

"I had two buisinesses that were failing so i gave these two random dudes another million" Man i did something wrong in life.

"Back That Ass Up" Admittedly, was not on my James Hoffman Bingo Card

James' little moment of wide-eyed Gollum-esque greed when he said "A patron asked if I wanted to buy it off him" had me choking on my lunch, I am increasingly convinced the man is a cartoon character brought to life.

James Hoffman saying “back that ass up” was something I had never expected to hear in my life and made me choke on my drink

As a programmer, I can confirm there's loads of immaturity in closed betas. It's never meant to be seen by the public so we frequently add in our own thoughts and humor.

08:19 - "Turing Compliant Espresso Machine" needs to be on a tee shirt.

As someone who works in videogames, the "back that @$$ up" prompt feels like something that, was, just for the prototypes 😂 sometimes it's the stupid little things that can keep you going during a repetitive testing phase.

ZPM backer here. Sucked at the time, but no regrets. I appreciated the Decent coupon, and it's been a wild ride with my DE1+ v1.0. It just keeps getting better -- I'm making tasty espresso every day, and I just started making great filter coffee with it too.

I'm a control systems engineer with a specialty in building automation, but the concepts are very similar. Tuning the PID is somewhat straightforward, and I'll get into the basics later if anyone wishes to read. What I really wanted to point out was the fact that this system is not at all new and the sensors have likely drifted in accuracy over the years, especially if they were cheap sensors to begin with. We genearlly must certify accuracy of sensors annually to ensure a properly controlled loop. This machine, I'd guess, has never been recalibrated. As far as the PID tuning, some basic concepts first: Proportional gain should get you most of the way to the finish line - this is likely the largest gain of the 3. Proportional looks at how far you are away from the setpoint and applies force in the opposite direction to push you toward setpoint. The integral portion of the control loop looks at how far you are away from setpoint for how long (that extra time component is the key difference between P and I). It pushes a little bit if you're away from setpoint for a little bit of time and pushes ever harder the longer you stay away from setpoint. Derivative looks at how fast the control variable is moving toward or away from setpoint and attempts to limit that rate of change. It's like a cop telling the loop to slow down because it's heading toward the setpoint too quickly. Adding derivative allows a rapid response to a step change while limiting the overshoot. To begin tuning, zero out the other gains and focus on tuning the P first. There is some P gain value above which the loop goes unstable - stay well below that point. Too high proportional gain and you get uncontrollable osscillations. Too low proportional gain and there is slow reaction to stimulus. Properly tuned proportional-only control will result in a steady state offset error. This is expected and fine. So...move on to the second part of tuning: Integral. If you have a steady state offset from the proportional, you fix that with a little bit of integral. But much like adding spice to a dish, there is definitely a point where you've added too much. Too much integral results in steady state osscilation (imagine integral says "TOO LOW - INCREASE TEMPERATURE" immediately followed by "TOO HIGH - DECREASE TEMPERATURE" and just repeats that over and over and over. So, in the software I used, it was typical to do something like I=P/20. Derivative is the last thing to tune in, but in my line of work, derivative gain was always set to zero because it only works well if you have a very direct, accurate, and fast feedback of where you are at any point in time or space. Our systems always had a time constant that was a tad too long to effectively use D gain, so we didn't. Of course, there's math you can do to theoretically calculate the gains, and most literature says those of us who tune by feel are dumb and should at least feel bad for doing so. But I find the math depends highly on the application. If there's liklihood of dangerous results, do not do this by hand because things can go wrong very quickly. I always had multiple layers of software and mechanical safeties in place so if the loop were to go unstable, nothing catastrophic would occur. That's my way of saying be careful and by all means, research the math behind it if you're so inclined.

Anyone familiar with Stargate knows that they always had problems with a lack of functional or fully charged ZPMs. An ominous sign, I would say!

What's really mind-blowing though is the idea that people STILL, in 2022, seem to think Kickstarter is a shopping website and not a crowdfunding website.

I think this really puts emphasis on Breville/Sage Bambino machines and how they can build and ship such a budget machine that works really well.

Plot twist: the patreon supporter who wins the machine is the patreon supporter that sold him the machine.

I remember this very well. I almost backed this device. in the end, I decided against it... I did, however, follow the entire case study from start to finish... a very sad finish indeed... Very shortly after this product failed, everyone was reminded that Kickstarter was NOT a store, and you weren't purchasing a product, you were backing an idea, sending them money to fulfil that idea, and likely getting one of the product devices, if that idea came to be a good one... Kickstarter was never a web store, and most people forgot that point...

I'm quite amazed that the ZPM allows the user to fiddle with the parameters of the PID controller(s). PIDs are very simple to write (about 30 lines of code, even with diagnostic logging) but take an enormous amount of skill, knowledge, time/multiple iterations to tune via the parameters. PID parameters really aren't something that users will be capable of successfully choosing - this suggests that perhaps the ZPM founders weren't able to step outside their own technologist/engineer way of thinking shoes into the shoes of a more regular user.

I was hoping that talk about using cheap components in the Decent machine would lead to a low (at least sub $1k) price point, but lo and behold, the machine cost $3.5k. I guess I'll have to keep dreaming...