The solenoid valves from China are fabulous and show great potential for automated brewing. The problem is that they have a direction of flow arrow on them. Without really understanding why this might be, I had configured my valves to fill the hot liquor tun and mash tun with mains water against this directional arrow. The design choice was made because it saved me a valve.

After some testing it was quickly apparent that there were troubles with this approach. Without knowing why, whenever I turned on the solenoid that supplies mains water into the machine, the kettle solenoid, mash tun solenoid and hot liquor tun solenoid would all leak water in against the arrow flow.

At first I though I'd been supplied valves that were normally open rather than normally closed, as powering the valves slowed down the reverse flow (but didn't fully stop it).

I was determined to get to the bottom of this problem so I pulled a valve apart. It was a simple issue. The pressure against the arrow was providing the same mechanical action on the valve that the solenoid does - that is it opens the valve. If the valve was the oriented the other way around, the solenoid would have to work against the water pressure to open the valve. These things are designed to work against a significant back pressure, not a significant forward pressure.

It seemed a possible 'fix' would be to stiffen the spring that holds the diaphragm in position. I figured there must be some latitude here because the tuns won't hold a lot of pressure compared to the mains water. If that is so, the solenoid should have some spare power to act against a stiffer spring.

The simplest way to stiffen the spring was to lengthen its action. A nylon thread (4mm) fitted neatly. With the valve still apart, it was tested with power to see if the solenoid could pull against the stiffer spring.

So far so good.

So the next test was to reassemble the valve and try some back pressure against it. And even with a low pressure trickle on the outlet side of the valve the diaphragm still opened. The same test was done with stiffer springs until the solenoid would barely open under power and it still would not hold against the back pressure. These valves are simply not designed to be used this way.

In order to fix the water fill path, we will need an extra solenoid and need to move the path to the other side of the mash tun and hot liquor tun drain valves. A picture is better than a description here:

Above is the old fill path via the solenoid (circle) at right. The pressure against the three arrows at left create the problem. The proposed fix is below:

The single 'water in' solenoid is replaced with discreet solenoids for both mash tun and hot liquor tun, and now there is no back pressure from mains water against the existing solenoid valves.

I have an extra valve (and a spare) on order but it looks like they are 6-8 weeks away. In the meantime I intend to move the existing fill solenoid into the new HLT fill path and test the fix.

It has been a frustrating few weeks and I'm beginning to wonder if I will get the cellar and bar sorted before Christmas.

With the poor diagnosis on the glycol chiller I thought I would try a cheap new freezer as the chilling plant. I found a cheap brand 80 litre upright freezer and bought one via ebay as a factory second.

When I looked inside I was surprised to see that the shelves were part of the cooling coils - that is there was no cooling in the walls of the freezer, but rather each shelf was fixed with coolant lines. It meant that it was never going to work. Anyway, it gave a starting point to test the theory that a freezer could really be used as glycol chilling plant.

The bottom plastic tray had a capacity of around 12 litres, so a glycol/water mix was put into the freezer with the controller's temperature probe. It took a day or so for the temperature to drop to around 0 degrees C, at which point I turned on the font pump to see what would happen.

The font got cold and began to drip with condensation, but before long the font had sucked the cold from the glycol and the freezer could not cope with such a simple demand.

I decided it was time to move to a chest freezer, so another ebay search ...

... and I ended up with a nice little 150 litre chest freezer. So this was loaded with 22 litres of glycol/water and I waited to see what would happen.

I am not a refrigeration mechanic so I don't really know how much cooling power units like this should have - for me it is simply guesswork. I also don't know how a freezer might compare to an average refrigerator, although I have successfully used a glycol bath in the freezer compartment of a keg fridge to flood the font. So I waited and watched the temperature slowly lower on the glycol bath in the second freezer. But one day ran into another and then after half the week had passed the glycol had only dropped to about 10 C. I began to wonder if the thermal mass of glycol was too great for a domestic freezer to deal with.

One evening I had a few spare minutes (a bit of a luxury at this time of year) and so I installed a fan to recirculate cold air and an agitator to stir up the glycol. I was shocked the next morning to see that these things had added sufficient heat for the glycol to be sitting at 22C next morning

Ok, so something is really wrong here. Finally yesterday I had the chance to do a proper test on the chest freezer. I bought some bags of ice and placed them in the freezer with the temperature probe to see what would happen. The glycol bath was removed. The temperature settled at -1.7 degrees C and then over the next 12 hours settled out to near enough 0 C or freezing point. Damn, the second freezer is also a lemon

Oh well, there is at least some good news. I was given an ebay refund the same day - much better than other ebay experiences I've had recently.

So I'm on the lookout for a chest freezer that actually works to test again if this is at all feasible. I'm getting ebay fatigue at this rate ...

If all else fails, I will pack my beer coils with ice for our new year party - a giant miracle box - so at least the bar will be functional for the patrons.

The 40x2 LCD has not yet been functional in HERMAN6. When I ordered it I had not realised it needed a negative bias on the contrast line. I had knocked up a less than elegant solution to this and modified my P. H. Anderson serial drive board to accommodate the solution. For whatever reason, the modification worked on the test bench but not on the machine.

I left that part of the project to focus on other things, and in the meantime relied on the PC connection to show important information.

It often happens that when I take a step back I find a better solution. In a recent blog entry I wrote about a simple circuit based on a 7660 chip that is commonly used to provide the negative contrast voltage using a single voltage supply.

While my 7660s were on order, it was time to re-check that the LCD was ok. So using a known working serial drive board and adding a AA battery below zero to supply the contrast voltage, I was happy to see all was well.

That was yesterday, and I was pleasantly surprised when the 7660s arrived on the doorstep this morning (Farnell are fast).

So with just a couple of electrolytics, a 7660 and a trimpot, I had all I needed to test the display off a single supply. And presto, all is good!

I think I will build this little interface on strip-board. I need an adaptor between the serial board and the display anyway.

As an aside, I will reverse the mods on the Anderson serial board and use that in another project. The AXE-033 board from Revolution Education has a few advantages for the HERMAN project.

• Being a complex project, having the programming editor simulate the AXE-033 will be an advantage. (The Anderson board uses different code commands.)
• The AXE-033 has provision for a real-time clock. If it seems worthwhile, we have this as an upgrade option.

We never did get around to the last test before brewing - but that is something we can do this weekend.

But a HERMS system WILL NOT WORK if there are flow issues!

Ok, so we brewed a Helles last Monday. I've been wondering about the grain bill and getting too adventurous. 10.5 kg of grain should be ok in a 38 litre igloo cooler ...

At this point I'm not sure if there were problems of trapped air on the pump inlet, or if the grain bed was set too tightly. If only I'd taken one or two educated measurements, I would be enlightened at this point.

The liquor did flow, but only very slowly in the recirculation loop. I know from experience that it is simply not worth the heartache to try and step from a saccharification rest to mash out. The sparge water pretty much did this anyway, albeit taking the entire kettle fill time to do it.

What I should have done was let gravity do its work, draining a sample of mash liquor and working out what this rate was compared to through the pump working against gravity. That would have given me the info I needed. Same rate by gravity as by pump would show a compacted grain bed. A greater rate draining by gravity would suggest a pumping issue, and in my case almost certainly trapped air.

With more than 10kg of grain in the tun, it means there is not much room for rice hulls. I put some in, but would have liked to add more. At least while this continues to plague me, I must remember not to overload the mash tun.

I guess the main reason I did not collect the data was I was distracted cleaning out a cupboard while I was brewing. It was satisfying to get that sorted, but the brewing mayhem continues. Must remember to focus on brewing while brewing ...

A pleasing thing about the brew day was doing the pre-warm of mash tun like I did last time. I hit the target of 65C precisely (65.1 seems close enough to call precise ). So I will continue this method.

The change to the cleaning routine has meant that any flaky matter from the heat exchange coil no longer clogs the plate chiller, so that is a step forward. The Helles is fermenting away strongly, and it promises to turn out well.

All things going well, tomorrow I will get back to the HERMS pre-heat testing that I wrote about last time. But I'm also getting the urge to construct the new HERMAN control panel. And there are a few leaks to repair on HERMAN.

So many things to do, and so little time.

I ran the test outlined in the last blog:

• stablise mash at 62C
• preheat HLT to 69 + 2*7 + 5 = 88C
• set mash target to 69C
• set HLT target to 69C + 5 = 74C
• log the results with particular interest on final HLT temperature and time taken for mash step.
• work out what to do next.

OK, some interesting observations ...

First, with a target of 62C and the HLT at around 67C, the mash liquor settled nicely at 62C with pretty much a flat line - nice.

The blue line is the HLT temperature, and the ripple shows that there is no stirrer or pump in it to mix the liquor around. I have a mixer that came with my glycol chiller that I'm currently not using, so it might be good to add this to the HLT. Incidentally, the time scale on the graph says hours, but the markers are actually the minute values of the second hour. The graph shows 8 minutes of stable mash liquor temperature.

From this point, I set the HLT to 88C to perform a pre-heat, with the intention of doing as rapid as possible a step up to 69C. It took 22 minutes for the HLT to step up 20 degrees. During this time it was interesting to observe that the mash was not as stable as it had been.

The graph shows that the mash temperature began to oscillate, just dipping below target and then peaking around 1 degree higher. This makes sense, because the temperature probe in the mash liquor is about half-way down the tun. By the time the heat has found its way to the probe, too much hot liquor has already been dumped into the tun so the temperature continues to rise around the probe as this hot liquor slowly finds its way down.

If the pre-heat is short, this may not be of any real concern, but I will think a bit more about this. It might be fixed by pumping for a short time, say 30 seconds, waiting for same, and then pumping again if needed.

The next graph shows a result that is very close to what I'd expected - that the extra heat in the HLT can speed up a step, and that the amount of extra heat needed is quite predictable, meaning it is not too hard to do a quick step and rapidly return to a stable mash temperature.

The HLT is shown here rapidly losing heat as the mash liquor rises very quickly initially. In about 4 minutes, the mash temperature has jumped from 62C to 67C, an impressive figure for this setup. Although not shown on the graph above, the recirculating liquor temperature did not rise above 75C and settled quite quickly to 70C. This is good, because it means that enzyme activity will not be radically affected by this process.

After the quick rise, things tail off quite quickly, as the difference between HLT temperature and mash liquor ceases to be high. It is interesting to note that the mash liquor does not ever reach 69C, but rather settles at an even 68.

It seems like the system heat losses rise as the target temperatures rise, which is understandable, although I'd hoped it would not have been that great.

The graph above shows that during minutes 40 to 55, the HLT was set to 74C. Between 55 mins and hour 4, the HLT was raised to 75C. During the next 15 minutes, the mash liquor temperature barely rose, settling in the low 68s. The HLT was set to 76C at about minute 15, and the mash liquor slowly made its way to the 69C target, albeit still a little sluggishly.

It is obvious that the HLT did not have sufficient heat mass to perform the required step, although it was close. It is worth tweaking the algorithm, so maybe something like 2.5 times the step difference and analysing the result. It also seems that the system heat loss can vary significantly between cooler and higher temperatures. It would be worth testing the system with a generous heat loss figure (say 8 degrees) and testing stability at both 62C and 69C.

A further test would probably reveal if the algorithm needs a changing system heat loss, or if a single figure can cope with the entire range of 40C to 76C for mashing. So tomorrow some more testing, and then on Monday we will brew a Helles.