Hi everyone,

I've been thinking of adding some sort of pressure profiling mechanism to my Elektra and I know that some of the forum members here already did this, for example AndyS or gregs. Their systems use rather sophisticated and expensive elements, like air driven pumps, VFDs, 3phase motors and suchlike, which aren't very easy to find/buy. I've been tinkering with this idea for a while now and I was curious whether an additional bypass valve would allow me to get similar results without having a NASA-like budget. Placing the bypass valve (pretty much an OPV) just after the pump would give me the ability to lower/rise the pressure at it's output depending on the position of the valve screw. Throw in a servomechanism, some gears, a uC and you've got yourself a rather simple setup. Has anyone tried anything like it?

Stuff I'm a bit worried about:

- pressure fluctuations when changing the setting of the bypass valve
- mechanical wear on the valve
- inaccuracy of the method when working in an open-loop config (without a pressure transducer)

Am I crazy to think this might work? any feedback from all you engineering-minded home-baristas is greatly appreciated.

Regards,
dsc.

The OPVs I've seen aren't really suitable for real time control.

I've experimented long ago with a dimmer designed for ceiling fans on a vibe pump. I could get down to 6 bar but no further without the pump dropping out. I think rotary pumps are low wattage enough that the same dimmers would work on them, but I do not know what the result will be. You do need an in-line pressure gauge to give you a vague clue of what is happening.

Andy and Greg were looking for precise and programmable systems that went from 0 to 15 bar. What I'm describing is a knob you turn during the shot and that will go from around 5 to 6 bar up to whatever you set the OPV at.

I sometimes feel a bit frustrated because I've posted this link several times and nobody seems to pay attention

Simple Fully Adjustable Brew Pressure Modification

When the valve is all the way open there is no pressure, and by making small adjustments to the valve I was able to make fine changes to the brew pressure- so any pressure between 0 and OPV setting is possible.

If it was combined with your idea for a programmable servo it should be able to get repeatable pressure profiles. You have a good point in your question about pressure fluctuations... I only used it manually so I cannot answer from experience, but...

Would it act differently when pulling a ristretto than a double? Would it require some kind of feedback about the actual pressure or flow? Would it require re-calibration as the pump aged and weakened?

I can't really wrap my mind around these questions...

Regards, Henry

Henry, your scheme is simple and will work. I use something very similar to tune variable speed pumps so that they produce slightly more than the maximum desired pressure when running at full speed. The same idea can be used to make a cheap, manually-operated profiler for rotary pumps.


For rotary pump machines, the simple solution is to install two tee fittings, with one tee just downstream of the pump, and the other just upstream of the pump. Install a cheap throttling valve on this bypass leg between the two tee fittings. A 1/4 inch diameter ball valve is a cheap solution. If the pump is activated while the ball valve is open, the pump will recirculate all of the water thru the bypass leg. Closing the ball valve progressively will increase the downstream brewing pressure. Adjust the pump pressure relief valve so that the pump produces 9 bars at the group when the ball valve is fully closed. Now you can manually control pre-infusion and pressure ramping as you wish.

-Greg

Hi guys,

thanks for the tips guys!

I was thinking about an OPV, but one can of course use a normal valve of any type (although I've heard ball valves aren't the best, dunno why). I wasn't sure what to do with the water that comes out of the bypass valve, although I had the 'back to pump' idea in my head. Now that Greg confirmed this can be done without any problems I will give it a go. All I need to do is find the right fittings which might be quite hard because they only seem to have compression ones here in the UK.

Regards,
dsc.

For precisely setting the pressure I think in theory a needle valve would be best (maybe the steam valve described by hbuchtel is of this type ?). The ball valves I know are quarter turn valves and it is not as easy to find the right position as with a multi turn needle valve, but perhaps it works ok (I did not try any of this).

Some other thoughts (that of course do not apply to a manually operated valve):
- I also tried using a light dimmer with a vibe pump and the pump also dropped out below 6 bar. As the pump went completely silent my impression was that the problem was caused by the dimmer which could not be set low enough (the usable range was e.g. only 50-100%). Using (or building) a dimmer that can be set from 0-100% might be a solution, but i did not yet try this.

- Wouldn't it be much easier to control the pump than using a servo to control a valve designed for manual operation?

- If only 2 (or perhaps even 3) different pressure levels are needed it is possible to add another OPV and a solenoid valve. Or you could used 3 OPVs set to 1/7, 2/7 and 4/7 of the desired maximum pressure and three solenoid valves (one across each OPV) to digitally select the desired pressure .

Regards, Frank

gscace wrote:Henry, your scheme is simple and will work.

Thanks Greg, I may now rest in peace

Frank, the valve in the link posted above is a needle valve as you thought. I think it is a lot harder to make small changes with a ball valve and they tend to be a lot tighter as well.

Regards, Henry

Hi guys,

I know a needle valve will probably be the best, the problem is finding one over here and mounting a servo on it.

Controlling the pump is more expensive and creates some problems at lower speeds. Using a servo controlled valve is quite cheap (if you build your own), effective and (hopefully) won't cause too much trouble. I guess we will have to wait and see:)

Regards,
dsc.

Actually electro-mechanical control of a needle valve is not too easy because you have to find a reference point and that costs money (encoders?). Other posters are correct that ball valves are not the best throttling valve. But... ball valves are cheap, quickly adjusted and have very low flow resistance when fully open. It's very easy to visually observe the position of the valve, which is hard to do in multi-turn valves such as needle valves. One only has to look at the position of the lever / handle. As an inexpensive solution, I can envision mounting the valve to a plywood base, then fastening a mechanical stop to the piece of plywood, positioned so that the valve handle would only open part-way. It would be fairly simple to install a bolt into the stop that allowed some adjustability to the position. One could adjust the stop so that the pump / bypass system produced 2-3 bars when the lever was against the stop. Now brewing could reproducibly start at the same low-pressure preinfusion, with duration controlled by the barista. Smoothly closing the ball valve would provide smooth pressure ramping to full pressure. In order to get repeatable full pressure it's important to adjust the internal pressure relief valve to produce full pressure when the ball valve is entirely closed. I bet with practice one could become very proficient and reproduce profiles pretty well during preinfusion and pressure rampup. Easily controlling declining pressure at the end would be harder. IMO it's the preinfusion and pressure ramp-up that is most important.

A private email on the subject pointed out that the system behavior could be complex, considering the fact that the internal pressure relief valve lifts off of its seat when the pump is developing full pressure. It really doesn't matter that much because the internal pressure relief valve only begins to lift off its seat when the pump pressure builds substantially, not during the preinfusion step.

For DSC - look on the net for your local Swagelok folks for fittings. Swagelok is a little pricey but is really good for this sort of stuff.

-Greg

Hi Greg,

I agree that multi turn valves are harder to control (and too actually mount the servo so that it works), but one can always use a stepper motor to eliminate the use of encoders.

I looked at Swagelok fittings but they charge £14 for an ordinary brass 3/8" tee :O I can get all the compression/ordinary fittings from my local Screwfix for that amount of money!

Thanks for all the tips, I'll update this topic with some photos and results when it's up and running.

Oh one more question, do I need a one-way valve on the input of the pump, just before the tee to prevent backflow into the inline pipe (if I set the brew pressure to 3bar the 'other' 6bar will flow back to the input of the pump, won't it 'try' to flow against the inline water supply?).

Regards,
dsc.

Although the thread is for a bypass valve, the used Bodine motor + controller would be simpler to control than a servo needle valve.
Procon pump replacement. Coming up with the DC supply shouldn't be too hard for you DSC since you know electronics.

Mark

dsc wrote:Hi Greg,

Oh one more question, do I need a one-way valve on the input of the pump, just before the tee to prevent backflow into the inline pipe (if I set the brew pressure to 3bar the 'other' 6bar will flow back to the input of the pump, won't it 'try' to flow against the inline water supply?).

Regards,
dsc.

There's pressure drop across the valve. Pressure on the outlet side of the pump will be at whatever the pump / bypass combination produces. Flow across the bypass valve experiences pressure drop to essentially the pump inlet pressure. So no additional check valve is required. I should point out, however, that there should be a check valve between the pump and the group, installed by the manufacturer. The bypass loop we're talking about must be installed upstream of the check valve (between pump and check valve). You will have to get creative with your plumbing if the check valve is mounted directly to the pump outlet. Shouldn't be too much trouble.

-Greg

dsc wrote:Hi Greg,

I agree that multi turn valves are harder to control (and too actually mount the servo so that it works), but one can always use a stepper motor to eliminate the use of encoders.

I looked at Swagelok fittings but they charge £14 for an ordinary brass 3/8" tee :O I can get all the compression/ordinary fittings from my local Screwfix for that amount of money!

Thanks for all the tips, I'll update this topic with some photos and results when it's up and running.

Oh one more question, do I need a one-way valve on the input of the pump, just before the tee to prevent backflow into the inline pipe (if I set the brew pressure to 3bar the 'other' 6bar will flow back to the input of the pump, won't it 'try' to flow against the inline water supply?).

Regards,
dsc.

We've done the stepper motor thing here at work for a variety of motion control apps, but I really hate it when folks here don't wanna pony up for an encoder because stepper motor systems have a habit of losing their zero position (losing steps) and ya gotta go through all kinds of contortions to re-initialize them without an encoder. WRT the stepper motor / encoder / cobbling it together thing, by the time you fool around with it this much you should just buy a fluid-0-tec TMFR pump with analog variable speed controller, a pressure transducer, programmable process controller and do the damn thing right. You could write your own Labview code and leave out the process controller, but that costs time and you'll have to have something to read the transducer and send the data to the accompanying computer. Personally I think all that sh** looks good in any modern kitchen, but my coffee stuff (except the roaster) is in debasement - at Espresso Research LLC corporate headquarters. The roaster is in the laundry room.

-Greg

Hi guys,

I might try the motor approach sometime in the future, although now I'd like to go down the bypass route and simply test it out.

Out of curiosity, will the size of the fittings affect anything in the system? I need to use some 1/2" compression/threaded fittings and will have to use 1/2" - 3/8" reductions. Would that be a problem? It is possible to use dedicated 3/8" threaded air fittings, but there's no easy way to connect various parts of the bypass system using flexible hoses. With 1/2" fittings I can simply use flexible tap connectors which are cheap and easily available.

Regards,
dsc.

Just as a note, I was in my electronics store yesterday. They just had gotten in a box of rotary encoders with 25 pulse per revolution, with a 1/4 in shaft fitting, just 5 bucks. I am not sure if 15deg would be accurate enough for a needle valve pressure profile. But it's just a thought.
-Cecil

another_jim wrote:Andy and Greg were looking for precise and programmable systems

Yes. For home barista use, manual profiling systems can be practical, fun and educational. Home baristas have the time to obsess over every moment of every shot.

But my original emphasis (and I think Greg's, too) was for commercial operation. In a cafe you want to find a profile that gives good results, program it into your system, and let the electronic brain repeat it time after time. No cafe barista can afford to stand there nursing each shot through a manual profile.

Hi guys,

of course a manual system is simply a way to see how pressure profiling affects the espresso, it's not really a long term solution as one cannot easily recreate the same profile. But if this bypass approach works I will try to add a uC based HMI to allow programming, using pre-set profiles and full manual (via a knob) operation.

Regards,
dsc.

Hi guys,

I got the bits today and will be testing out the manual operation after work. The only thing I'm a bit worried is the ball valve itself which seems to be lubricated with some rather smelly grease/oil. Do you think it's going to be a problem, ie. the water is going to taste different when passing the valve? I'd say no as it is hot/cold water approved, but maybe I should switch to something more food safe?

Regards,
dsc.

Seeing as half of taste is smell would you want to risk your machine getting a smell as you describe it? Not quite boiling water at 9 bar may not have registered to the Mfg as 'hot water'. Just a thought.

CRCasey wrote:Not quite boiling water at 9 bar may not have registered to the Mfg as 'hot water'. Just a thought.

The water would be "cold" tap water- hot water doesn't go through the pump.