Reducing machine dirt buildup

Most things in coffee get a little easier the more you do them. Dosing consistently, understanding grind adjustment, understanding the taste of an espresso extraction, tamping simply and properly, the list goes on…

One thing that doesn’t get much easier is dealing with the build up of unpleasant flavours in an espresso machine. There are no shortcuts, you just have to deal with it regularly and to be honest it annoys me a lot. That taste is so obvious, so distracting, so unpleasant, that it ruins a lot of otherwise well made espresso for me.

I was thinking about the build up of dirt inside the group head caused by the release of pressure after you stop a shot. To better explain we have the following photo:


Photo courtesy of Dan Kehn of Home Barista

In a La Marzocco like this the route the water takes is a little confusing. It goes out of the far side of the neck of the group, through a flowmeter, back underneath the group head to a valve. When this valve is closed no water can leave the boiler, and when it opens water is pushed through and heads back inside the group head into that little tube you can see and towards the group head, until it drops down into the screw and showerscreen and over the coffee. 1 Other machines have a similar tube, of varying lengths, that run between the valve and dispersion mechanism above the coffee. (Not all LM’s have this long of a pathway outside the group – machines that don’t have flow meters and more modern models have eliminated this pretty much.)

Dirt builds up here because that last tube is a two way street. At the end of the shot the pressure shoots back up the tube until it gets to the stop valve. This prevents anything from getting back into your boiler, and this valve is a three way valve allow this pressure to exit through a drain tube, usually ending up in the drainbox under the drip tray – though some deposit straight into the drip tray in the E61 style.

The only machines that don’t have this problem are lever machines. Lever machines only release the pressure when the spring has finished expanding. This is why you can’t interrupt a lever machine shot without making a mess. (I believe the technical term is ‘portafilter sneeze’)

This tube is very hot. Any liquid in this tube will likely evaporate and leave behind whatever it had dissolved. At the end of the shot this can mean dissolved coffee gets dragged into this tube where it will quickly deposit and start to taste unpleasant. Some of that unpleasantness will be picked up when you next pull a shot and fresh water is heading down the tube towards coffee. The same unpleasantness builds up pretty quickly underneath the basket of a portafilter, and we know from sticking our noses in there how bad that would taste. 2

I was wondering if getting into the habit of flushing immediately after pulling a shot would dramatically reduce the build up of dirt in that particular area. As a flush builds up no pressure it would mean that water would only travel one way out of the group, and hopefully drag with it any coffee before it had the chance to dry out and deposit.

Some people would argue that you should leave the puck in for temperature stability. Even the WBC references this idea by no longer looking inside the portafilters at start up as you can leave pucks in as you wish. I don’t think temperature stability is a concern, but I do think dirt is. I’d rather lose a little temp at the start of the shot than have to deal with the residue of a puck sat in a machine for 10 minutes. I’m not sure I’ve ever seen any proof that having a puck in promotes thermal stability, and if it is out there then a link in the comments would be great!

Also – would using a rampdown in pump pressure tofinish the shot have a similar effect, as there would be little to no pressure to release back into the group?

  1. If none of this makes any sense then leave a comment and I will try and find a better way to explain.  ↩︎
  2. This build up is why I would use Cafiza on a domestic espresso machine every day, even if I had just pulled one shot.  ↩︎

29 Comments

  1. I agree with you on temp stability – somewhere along the way everybody got overly obsessed with it, and forgot about everything else. I would much rather taste the effects of a slightly cooler portafilter than dirty machine, and I hate scraping old, dried out coffee from the bottom of a filter basket. I find portafilters are generally pretty stable, and I don’t think keeping pucks inside them makes a massive difference (I’m more than willing to be proven wrong on this. I’m speaking from pure speculation).

    Can you honestly taste the difference after every shot, though? I can taste dirt on a machine that has been sitting for a while without being used, and I can taste it if no one has flushed for a few hours on a busy day, but I guess my palette just isn’t sensitive enough to taste dirty machine in between two shots. In trainings, I always tell people to try to flush every hour.

  2. That is a really interesting photograph.  I hazard that your idea of the immediate flush is a good way to mitigate build up…In the future!  I wonder how much soaking it would take to clean the tube fully with the grouphead removed from the boiler…

  3. Since using stainless steel portafilters there is an incredible reduction in the amount of residue and build up compared with standard portafilters. Perhaps machine manufacturers could look at using stainless steel or another material with similar properties for the water to travel along on its final route to the coffee and back to the waste to reduce the possibility of a dirty build up.
    Obviously this doesn’t solve anything in the short term but I do like your idea of a post flush + I’ve always been a fan cleaning out the portafilter sooner rather than later over keeping the puck in to apparently improve temp stability.
    Good to have you back after your digital sabbatical :-)

  4. Sorry if I was a little confusing there.  The rate of build up is really dependent on machine use and pace and but once you can taste it, it becomes all I notice in a shot and sort of smothers the good stuff.  Plus the finish gets really rough.

    I’d love to see if there is some puck related data out there, but I’ve never seen any…

  5. Tried to work with the post flush some years ago. Didn’t do it for me. Too much hassle in a n already busy environment. Puck in filter is great but not for too long. Cold filters for us equals sour shots, therefore we always make a seasoning shot if the filter has been empty for a while. Flush every hour and clean your filters and I believe there is very little difference between an opening hour shot and a closing hour shot. But cleaning the grinders should possibly be done more often. We do it once a day, but maybe it needs to be done more often. This is really not a problem as long as you have good cleaning routines. It is as basic as keeping your kitchen and cooking ware clean while you are cooking a shift in a restaurant.

  6. Oh, and welcome back to cyberspace, James. So glad you quit living in the real world and started blogging again! 

  7. Completely with you Tim on sour shots, especially working with milder roasted coffees. When first using Square Mile this was a major issue of mine and moving towards reducing the time the portafilter is out of the group and keeping the puck in the basket between shots immediately reduced/eradicated sourness…….. as long as flushed regularly i found the shots much better, but the one drawback is dirt build up on the screens, which is hard to get around, not sure if you have any thoughts?

  8. Re: the backup of coffee solids into the brew water delivery tube, you say, “The only machines that don’t have this problem are lever machines.
    Lever machines only release the pressure when the spring has finished
    expanding.”

    That’s not strictly correct. In 2004 we saw the Versalab espresso machine at the Atlanta SCAA show. Brew water entered the group through one delivery tube, and when the shot ended waste water exited the group through a separate exhaust tube. In theory if not in practice, this solved your problem.

    But like most of Versalab’s innovations, the coffee world didn’t pay much attention.

    James, you can do a fairly easy experiment that may answer your own question. If you cap the three-way valve exhaust tube on one of your groups, it will act like a lever group: the dirty coffee water will not be able to back up, it will only be able to exit through the puck. I did this on my home machine in 2005 to see if I could detect “cleaner” flavors. I could not detect a difference then, but I used to home roast a lot of Sumatras, so what did I know about clean flavors anyway?

  9. No, you weren’t unclear. I’m just not convinced that flushing after every shot to prevent build-up is any better than flushing every hour or so. Like Mr. Wendelboe said, I think it’s fairly impractical in a busy cafe situation, too. 

    I just flicked through Rao’s espresso book to be sure he hadn’t secretly done some tests on pucks and temp stability that I had failed to notice or remember, but no luck. I’m struggling to see how a bunch of soggy coffee grounds could make a significant difference to heavy chrome-plated brass. I’ve certainly never noticed sour shots as a result of knocking pucks out.

  10. When I get a lull in service I will clean out the pf and remove the basket and wipe it down and allow the group to run during this time and using that water to rinse as needed. When time allows I leave my pf empty of grounds.  I haven’t tested this but I just count on it making a small yet worthwhile difference in quality/taste.

    I will also grab a blank filter sometimes when there is a lull and backflush with no cleaner (leaving the screen installed), before this I let the group run while scrubbing with a brush to remove excessive grinds.  Also running the group and inserting and removing the empty/clean portafilter will lossen some grounds stuck up with the gasket allowing them to fall into the basket that can than be wiped up with a cloth.  

  11. When busy: flush between shots. When not busy (I mean a matter of few minutes) I would clean out the portafilter, and put it back in the grouphead, flushing water through it!
    That’s massively important to prevent dirty old coffee oils building up inside the portafilter (if its not naked one)

    I had a bad experience of cold portafilters when making coffees for tasting purposes….

  12. I’m thinking, even if left for a few minutes, the pre-shot flush (if at least 5 sec) would be enough to dissolve and purge any evaporated solids left in the piping.  At least enough to generate a perceptible change, that is, perceptible to the average palate I guess ;)

    Easy enough to test I would’ve thought:  check straight group water TDS post chem clean and rinse backflushes, then re-test flush water TDS from pre-pull flush after letting sit a set time.  No?

    Might take the MoJo in to work tomorrow, but right now (midnight here) I’m not feeling particularly scientific.

    That off-taste might be coming from elsewhere.  I’ve noticed that even after two chem cycles and two rinse cycles on our machine (each programmed cycle doing four manuals), when we take the dispersion screens off there is a surprising (read: disgusting) amount of the day’s grounds still hanging about in there drinking up all the tasty and not paying any bills or doing the dishes.

    On a related note I want to be the founder of a Please Clean Your French Press Properly movement – while my friends are enjoying the otherwise well extracted brew, all I can smell (and taste to an extent) is gob loads of rancid oils.  No spank you.

    (I’m a knock-out right after shot guy myself, for reference.  I can see the old oils, whilst I’m yet to see any data backing the temp stability claim.)

  13. I did a little experiment today at the shop. Our machine is Aurelia, and as you guys may know, each grouphead has brass dispersion blocks. So what I did is, before opening the shop in the morning, I took the picture of the dispersion block. Then for 3 hours, during busy service, I tried to keep 1 group head clean as possible ( flushing and throwing away used coffee bed after extraction) and for 1 other grouphead, leaving  spent coffee beds inside the portafilter unless I was going to use that grouphead again. 

    so here is the picture of the dispersion block before service, and after 3 hours of service. 
    Visually, it is quite obvious.

  14. It makes sense to flush after each use, and back-flush for sure at shift change. I like the idea of keeping the groups clean to avoid contaminating the next shot. Great discussion! Welcome back, Mr. Hoffman. You’ve been missed. 

  15. I’m wondering your thoughts as to hot water tank depletion using pre and post extraction flushes. I fear running water from the group heads too frequently might cause a ‘tap’ on the system. Are the quality of the machines we brew on (I use a Synesso most often) able to keep up with this water demand while maintaining consistent temperature delivery?

  16. Good cleaning often needs agitation therefore I would like to see group heads which allows an easy and very fast disassembling (like quick release skewers on bike wheels). Perhaps this is generally the time to reinvent the shower screen and water dispersion system. Using the VST baskets there is often a center dead spot at the beginning of the extraction which seems logically because there is also a dead spot in the shower screen ( caused by the screw)…?

  17. I too await an improvement of the shower screen when it comes to an easy disassembling and a final removal of the screw : we’ve notived considerable change with the VST baskets, and no other brew method -from AeroPress to Pour Over-  has no weak spot like right under the dispersion screen screw of an espresso machine.

  18. Are you getting flushing and back-flushing confused Hadassah? I think James just means  running some water through the group when he says flushing, also called purging.

  19. Yes the machines in my experience cope fine with this.
    I use a GB5 in a busy cafe, and immediately after each shot we: knock out puck, purge group whilst rinsing portafilter under it, lock back in, and stop the group. I find that combined with backflushing in downtimes this eliminates grubby tastes for the most part.
    This has been my technique for a fairly long time and doesn’t seem to adversely affect the temperature or related shot flavour.
    To those who are suggesting that removing pucks gives a sour taste: surely your enitre calibration – extraction speed, weight, temp, etc – is all based on having the puck left in, probably you can correct the flavour by increasing the temp a little.
    I certainly don’t find this an issue, as all my calibration is based on no puck, rather, if pucks are left in, i find the taste sometimes dirty and most often a little bitter.

  20. In terms of your idea of the rampdown do you think a similar result could be achieved by overlapping the point at cut off? ie open the exhaust valve whilst keeping the pump running for a very short period of time? I would lead to a higher volume of water passing through the valve and a smaller diffentiation in back pressure, if I understand the system correctly?

  21. Pingback: Anonymous
  22. Hey I work on Kees’ new machine the “spirit” and he has eliminated the screw on the shower screen; it’s awesome. Also if you are really busy, a flush while you’re wiping the basket and a back flush every hour or so is completely sufficient to remove any dirty taste. In terms of this post though isn’t this why we back flush our machines with chemical at the end of each day?

  23. I initially wanted to write this off as machine error as opposed to human error.  Played with my distribution – no dead spot.  VSTs, LM Strada (has screws), with diligence in dosing and distributing perfect horizontal plane of initial drops, no preinfusion.  Had to break old habits with the VSTs – due to the better design there is less friction on the side walls of the basket, hence less flow rate difference from centre to walls, hence less need for a denser dose in the centre – dosing for a Synesso basket for instance leaves me with a dead spot.  Are you using Robur Es?  That little mound drives me nuts now.

  24. Off topic, Im curious as to how you are finding the Robur E and Strada combo? ie how much more of an issue are slight dose inconsistencies? The cafe I work at is going to be running this combination soon. We currently have a Linea that works amazingly efficiently with the Robur E.

  25. Depends on your definition of ‘slight’ ;)

    I still strike-off even with the E, admittedly this is only as accurate as my eye, but I still manage to catch most of the Robur’s capriciousness (I believe, perhaps naively).

    Like the VSTs the Strada is simply less forgiving – maybe it’s mechanical (but the only thing I think it could be then are the magnetic pumps (EP)), more likely I think it’s my greater attention; honestly the temperature stability (which I’m not saying is desirable or un-) is no better than a new FB-80: getting +/- 0.3C in-between and during shots.  Either way, to ‘know’ I’m sending out something worthy I’m adjusting my shot cut-offs individually based on time, flow rate, volume (sight) and temp (the real-time temp and pressure readouts are quite a nice feature).  Again, I must stress the accuracy here is only as good as my eye – drip tray scales anybody?  I’m impatient on that one I must say.

    I plan on doing a detailed write-up on the EP once I find the time; short version:  it’s a great and fun machine to experiment with (e.g. I’m pulling 40 second shots that taste rather nice) and would be perfect for a roastery, but at the end of the day I prefer a Synesso (or FB-80) for a busy cafe.

Submit a comment