When sizing an SSR a common error is that bigger is better and bigger avoids the requirement for a properly scaled heatsink.

These images are from pg 4 of the LDG SSRs Data Sheet sold by Auber and referenced in La Pavoni + PID = better temperature control?



From the highlighted boxes we see the heatsink thermal resistance K/W requirement for the 100A SSR is only slightly less than that of the 25A for the same 10A @ 40°C ambient temperature :

 25A / 6.14
 50A / 6.11
 75A / 6.61 [interpolated]
100A / 7.49


The power dissipated by the SSR is the same regardless of its maximum current capability and the same amount of heat is generated within the device. Using a grossly oversize device is a waste of money.

It is also instructive to note that the Ambient Temperature specifications are FREE AIR temperatures. The inside of an espresso machine is an enclosed volume and the current capability / heatsink thermal resistance must be derated.

Thanks for the run down Ian. Perhaps I'll just run a heat exchanger from the boiler to the SSR and take some of the strain off the heating elements

Maybe I'm reading things incorrectly, but the chart below only really speaks to the thermal resistance requirements of the heatsink doesn't it?

The 100A SSR shows it has 22% lower cooling requirements in terms of what heatsink it need. Or, put another way, given the same conditions, the more expensive SSR would operate around 10C degrees cooler given the same heatsink (if I'm reading the chart correct).

I don't really consider that a waste of money, when cooling is a major consideration.

Ray

One doesn't necessarily even need to use a heatsink, per se, when an espresso machine is PID'd.

When I first PID'd my rotary DT1 Cimbali, I put the PID components into a project box, which can be seen to the right of the grinder in this picture:



Within the project box, the SSR was mounted on top of a suitable heat sink.

Later, I decided to revise the PID and to install it onto the machine. I contacted the manufacturer of the heatsink and spoke to a tech in their tech support dept. He told me that I could mount the SSR directly onto the case of the machine, and I could test with a thermocouple that the heat generated was being adequately absorbed, which I did.

Here is a picture of the machine with the revised installation:



The SSR is screwed directly into the bottom case on the underside of the machine. A plastic cover is on top of the SSR to isolate the electric wiring.

Where you mount an SSR in a PID installation will be a function of your machine's size, design, and clearance off the counter. There is no need, however, to mount the hardware inside of the machine, and from the standpoint of heat dissipation, it would be preferable not to.

ken

Thanks for the info Ken. I was also thinking about mounting the SSR on the machine itself. However, in my case, the machine becomes very hot all over. The base plate may be an exception, but I will have to test and see.

Also, in speaking with the manufacturer, they indicated that the cooling requirements were for constant load, which this SSR won't really be under. It will most likely only be operating about 25% of the time, so I would imagine the cooling requirements may not be quite as high as some of the printed charts might lead one to believe.

We'll see though

Ray

another_jim wrote:Several semi-commercial machines use SSRs to interpose between a 10amp pressure stat contact and a larger than 1100 watt heater. The one on my Isomac died, and I discovered it had been glued to the case at the metal surface designed to mate with the heatsink, a mistake. When I replaced it, I used heatsink compound on the metal, and glue strips on the sides. In that arrangement, the case provided adequate heat sinking.

If that is impossible, Din mount SSRs, speced for industrial controls, can run at 80C ambient inside enclosures since they have integrated heatsinks. These will work inside an espresso machine. They cost around 2.5 times as much as regular SSRs, around $30 to $45 for a 20 amp model, but may be worth it for the avoided hassle. Cooling fans are another possibility.

If you are using a PID, the load will be constant (or cycling inside a 1 second window), so it may be necessary to look at heavier duty solutions than for an SSR that interposes a pstat.

The bottom case of my Cimbalis was painted with some sort of anti-rust compound and paint. I sanded all of that off, then applied heatsink compound (as Jim did above) before installing the heatsink to the bare metal. The SSR is going to get as hot as it will get when the espresso machine warms up from a cold start. I taped a bare TC next to the SSR baseplate where it affixes to the bottom of the case. On warm up, it never got more than a few degrees higher than ambient temp, eons below the heat level that would be concerning, which I found in an online chart for that particular SSR on the Mfr's website. Further monitoring showed that at steady state the temps at the baseplate/machine junction were even lower. The Cimbali Jr. is a single group commercial machine with a heating element rated at around 1300 watts. Smaller home machines will often have heating elements that draw less current, but on the other side of the ledger their cases will not have as much sheet metal to dissipate the heat.

ken

another_jim wrote:If you are using a PID, the load will be constant (or cycling inside a 1 second window), so it may be necessary to look at heavier duty solutions than for an SSR that interposes a pstat.

As Ken says, the biggest stress on the SSR is heating the machine. Once the machine is heated, the same power is generated by PID or pstat as it's the heat loss that controls the heat input, ignoring minor differences due to temperature swing. If the SSR w heatsink can handle start-up, it can handle temperature maintenance, regardless of control method.

RayJohns wrote:The 100A SSR shows it has 22% lower cooling requirements in terms of what heatsink it need. Or, put another way, given the same conditions, the more expensive SSR would operate around 10C degrees cooler given the same heatsink (if I'm reading the chart correct).

The chart is not the temperature the device runs at for a given current, but the maximum current at an ambient temperature when mounted upon a heatsink with a given thermal resistance. The larger SSR has a bigger chunk of silicon mounted to a heavier base, so it can take more abuse. Electronic specs can be off by 25%, so if the heatsink is undersized / non-existent as in the case of Jim's Isomac, it's gonna fail. Ray is correct that the large device will run cooler and therefore should last longer. However, only by a factor of 2. Proper cooling will extend that several factors of 2 and precludes the possibility of crossing into the red zone.

From the 100A SSR data sheet :

The SSR dissipates ~16w. Without a heatsink, the junction temperature is Tj = Ta + P x °C/W. Say the datasheet is very conservative and the Rth_j-a is only 5 and the machine internal temperature a sultry 40°C : 40 + 16 * 5 = 120°C or very close to the maximum allowable junction temperature. A rule of thumb : the life of an electronic component is doubled for every 10°C drop in temperature, so keeping the device cool is paramount. Note that the Rth_j-a does not change with the increase in current capacity, but the Rth_j-c and Rth_c-s drop dramatically. These number are only effective when connected to a heatsink. With a heatsink like the RHS300, the 25A SSR has a Tj of 40 + 16 * (5 + 0.8 + 0.2) = 136°C while the 100A has a Tj of 40 + 16 + * (5 + 0.3 + 0.1) = 126.4°C. Too hot in either case. Increase the machine temperature to a more likely 50 or 60°C and the device is in distress. $15 for a properly sized device + $10 for a heatsink ensuring optimal performance makes much more sense than spending $40 for an improperly cooled, grossly oversize device with marginal reliability.

Ken Fox wrote: I taped a bare TC next to the SSR baseplate where it affixes to the bottom of the case. On warm up, it never got more than a few degrees higher than ambient temp, eons below the heat level that would be concerning, which I found in an online chart for that particular SSR on the Mfr's website.

Measurement of this sort require specialized sensors to get accurate readings. The fact that the sheet metal is not warm maybe of concern. It could indicate a poor contact between the SSR base and the pan. Poor contact may be due to a dished pan, swarf, mounting holes not deburred, etc. A better test is to test the temperature of the pan between the mounting bolts on the outside of the machine near the end of warm up.. The temperature should be quite high, indicating good heat transfer from the case, declining evenly with distance from the center.

Obviously the machine sheet metal can act as a heat sink. As long as the surface mates evenly with the SSR, is of sufficient mass and large enough temperature differential, it's fine. The thicker the material and the larger the surface area, the better. Stainless steel [k~16] is about 1/3 as effective as steel [k~50] which is about 1/4 as effective as aluminum [k~200] in dissipating heat.

Very rough calculation for a Vibiemme Domobar Super :
pan where SSR mounted .25m x .15m x .0015m
Stainless thermal conductivity ~16
Steady state thermal resistance about .0025 K/W
Horizontal surface thermal radiance efficiency to still air about .1%
Bottom surface efficiency derating 50%

.0025 / .001 / 0.5 = 5 K/W - Borderline.
A steel pan with the same dimensions would have a thermal resistance of about 1.7 K/W, but paint could decrease the heat transfer to the air. A very slight forced air current drastically reduces heat buildup, but a heatsink is dead quiet and never fails.

I did exactly what the tech from the SSR mfr. told me to do, to put the TC right at the junction between the TC and the sheet metal of the machine. I do not think you need any more sophisticated measurement than that. Maybe you would if this was being installed in the Space Shuttle, but in a large metallic cased machine in your house, I do not think so.

Perhaps the proof of the pudding is that I used this setup for several years, 24/7, on my rotary Cimbali Jr. and never had any sort of a failure in my PID system. The originally installed SSRs are still in service, even though I have a few spares in the drawer. I used an identical setup in my vibe machine for 2 or 3 years, however that was used intermittently. It never failed, either, and in fact has been in service the last 2 weeks also (on a timer, shudder the thought!) while I'm awaiting parts for my GS/3.

I'm sure that a lot of the heat being produced is going off directly into the air. The exterior of the bottom sheet metal of my Cimbalis is not much hotter than ambient room temperature, and there is lots of air circulation around the SSR as it is mounted.

I did not choose to mount the SSR without a heatsink because I wanted to save $10 on a heatsink. The reason that I did this was that there was no room for a heatsink, e.g. not enough clearance if I interposed one between the bottom of the machine and SSR. The other options were an ugly very visible installation on the side or back of the machines, or installing the SSR inside of the body of the machines, a choice I considered grossly inferior in every way. The case seemed to me to be massive and to have quite a bit of capacity to dissipate heat, which is why I thought of this approach in the first place, and why I called the SSR manufacturer's tech support line.

ken

Relax, Ken. I'm not criticizing your project. You can't argue with success.
I'm simply trying to assist DIY'rs in using the data, upon which the manufacturers expend a great effort to create, to ensure their project survives cycle 1 with a reasonable expectation of a life exceeding the MTBF.

Thanks again for the information Ian. I'll study what you have provided and try to gain a better understanding of all the charts, etc.

I'm probably going to end up milling a heatsink of some sort to go with this. I'm not sure exactly what yet. My first step is going to be to wire everything up on the counter and start taking some temperature readings, etc. The sealing nut for inside the boiler will be here on Tuesday or Wednesday, so I will probably get back on things then.

Ray