Since you control PID, you could put the steam boiler in standby mode (96-99C) prolonging the life of the components attached to it like the plastic thermostats and the check valves (I'd say also prolog the life of the boiler but it is pretty bullet proof anyway - how many lifetime's do you want it to have anyway?).

I don't record and playback profiles anymore. If I was a coffee shot I would do so - but at home I just use the paddle.

Awesome work!

@Pepersass: As UI I use Bootstrap webpages hosted directly on the ESP32. Nice screenshots of your UI - I like it!

I wonder if it makes sense to use the following "mode" as the pressure ramps up very slow in the pure flow profiling shots:

1. PI (Slayer like)
2. Pressure profile to a predefined value (e.g. 9 bar).
3. once pressure reached: switch to flow profiling.

Any experience from your side about pressure/flow ramps? Which work best?

@Assaf: As usual very good ideas from your end. Standby function sounds really good and will prolong lifetime

"flow profile" as long as headspace still has air voids is PI.

Once the pressure gets to 9bar - high flow will result in excessive pressure and thus can't really be done. Low flow will cause the pressure to dissipate. As to which one of the two happen would be entirely dependent on grind.

On the nether parts of the extraction, you can set a decay pressure a-la spring decompression in a lever. I've been doing that for a while. Hard to tell if it is better or just a wee bit different.

Since espresso is a function of a bounded range of flow and not really pressure (arguably - but it is easy to get a very good espresso at 6 bar and adequate flow; than to get a stalled 9 bar - or a gusher - to work at all) and since flow depends on PI - perhaps PI can be tailored somewhat by looking at the PI to flow relationship, backing off on pressure until a set flow range is achieved. That is hard to do because the target ranges are unclear.

Assaf is right: In the Slayer sense, flow profiling only applies to the preinfusion phase, before the basket has filled and full pressure has been reached.

All you need for that is a needle valve. There are many ways to hook up a needle valve, but the simplest is the Jake Valve. With it, you can do manual flow control. To do a Slayer shot, you preset the Jake Valve to the desired preinfusion flow rate (e.g., 100-120 ml/min, like the Slayer.) Once pressure has reached the peak set by your pump (nominally 9 BAR), you open the needle valve for full flow. If you leave it that way, it'll be like a Slayer shot (though Slayer optionally allows you to enable the needle valve, still at its preinfusion setting, at the back end of the shot to slow the flow.) Optionally, you can back off the needle valve during the back end of the shot to reduce the flow and pressure.

My setup is a little more complicated. I have a Jake Valve, but I can swap it in and out of the coffee circuit with a solenoid bypass valve. I preset the Jake Valve to the desired preinfusion rate for the coffee I'm pulling. Before the shot begins, I swap in the Jake Valve. I set a timer for the desired preinfusion time, usually 10-20 seconds. I determine the time by how long it takes for pressure to peak when dialing in. When the timer expires, the needle valve is bypassed by the solenoid valve. When the flow rate (which the Arduino monitors via the flow meter pulses) begins to increase to about 40 ml/min (20 ml/min for a single), I start backing off the gear pump speed to keep the flow rate constant.

IMHO, the effect of reducing pressure/flow during the back end of the shot is to increase contact time. This can be help to fully extract light and light-medium roasts. None of this -- the needle valve, solenoid bypass or gear pump -- is necessary if you drink medium-dark or dark roasts exclusively. However, some feel that longer PI can be beneficial for a dark roast when pulled Ristretto. I don't drink dark roasts, so I've not tested that.

The needle valve for PI is already inside. I did the same modification like Assaf and steer it with an additional solenoid. The question was more about the ramp up after PI.

I assume there is still room between coffee and shower head - so the pressure ramps up very very slow. Even after 15sec of PI. You can see this in the following profile.

After PI (with needle valve) - its the light grey box, 15sec - the pure flow PID kicks in. Flow is constant at 60ml. However no coffee drips from the basket, yet. I believe with low rpm, the space inside the basket is slowly filling up. At 30sec its filled, and pressure builds up...

I had a similar experience when I first installed the Jake Valve and solenoid bypass valve. I was opening the bypass when I saw the first drops of espresso on the bottom of the basket. This resulted in a big jump in flow rate, which made no sense -- at peak pressure the flow rate should decrease from the PI flow rate.

Turns out there were two reasons for this: The important reason was that drops can and do appear at the bottom of the basket before peak pressure is reached. I'm not sure if the basket still has some headroom at this point or if the puck hasn't fully compressed, but either way opening the bypass allowed water to rush in, increasing the flow rate and gradually increasing pressure. The solution was to extend the PI time until max pressure is reached and stable for a second or two. Typically on a light roast my PI times can be on the order of 22 seconds with a flow rate of about 120 ml/min. This is several seconds past when the drops first appear. I think Slayer users were the ones who recommended setting PI to end when drops first appear, but they weren't able to see the spike in flow rate.

The less important reason is the way flow rate is averaged by my Arduino. I took that code from an early version of Assaf's Arduino code. It turns out that the averaging algorithm doesn't deal with a zero flow rate. I think it was written that way because it can't tell the difference between the flow rate being zero because the motor is turned off and the flow rate being zero because the shot is stalled. Assaf added some code to detect a stall (flow rate zero for more than a certain amount of time), but I haven't implemented it. I've been meaning to try accepting zero flow for averaging if and only if the motor is running. I think that might work but I haven't thought it through.

Hi Dick,
you gave the right hints.

Flow rate during PI was too low ... adjustment of the needle valve did the job (now 80ml/min.). Basket is filling faster now.

I couldn't resist to combine the pressure profiling with flow profiling. I did some experiments - and couldn't believe the good results:
10 sec PI
PP with 9bar until weight of 5g is reached.
FP for the rest at 60ml/min.

I assume the fast pressure ramp at beginning with the 9bar setting is "good" for the coffee puck. Still some spikes due to change in PID from P->F (hope this doesn't affect taste too much). Still there is a high ramp in flow, but it's fine as PI of 10sec was not too long and there is for sure still some space between coffee puck and shower head to fill. The graph looks like this:

You might try waiting a little longer after max pressure is reached before switching out the needle valve. If you still get the big spike, and you're averaging the flow rate, check the algorithm to make sure it can deal with zero flow rate. I wouldn't worry about the tiny spikes in flow rate. Very doubtful they're affecting taste.

80 ml/min is on the low side compared to Slayer recommendations, which are 100-120 ml/min. That's the flow rate I use most often, and with most light roasts it results in about 20 seconds of PI. Typical total shot times, including PI time, are in the range of 50-70 seconds. That's good for the light roasts I like.

Of course, the choice of PI flow rate depends on the roast you're pulling and how fine you have to grind to fully extract them. At 80 ml/min, PI in 10 seconds suggests you're not grinding super fine. That may not be necessary if you're pulling a medium-light or medium roast or even a light roast that extracts well. For difficult-to-extract light roasts, a higher PI rate with a finer grind may be necessary. That seems to be the type of roast for which Slayer's recommendations are made.

Grind and flow rate interactions can get complicated. For example, right now I'm pulling a medium-light Colombian that has a touch of oil in it that clogs my grinder's burrs if I go too fine. But at the finest setting that doesn't clog the burrs, the shot gushes when I set the PI flow rate to 120 ml/min. I was able to prevent the gush by increasing the PI flow rate to 240 ml/min (flow rate without the needle valve is in the 325-350 ml/min range.) The faster ramp compresses the puck sooner, preventing excessive breakup of the puck by the long, slow PI.

I took a look at the LM "Connected Machine Retrofit Kit" instructions (https://home.lamarzoccousa.com/wp-conte ... -FINAL.pdf) and it seems that the board connects to the wifi daughter board via the serial I use to get the button used to pull the shot.

So that serial now has the temp, the PI state, pulse count and many other params. The board looks like a standard 3d5 dosatura. Is the firmware for it available?