The effect of deflected part cooling air on brass and steel nozzle temperatures.

1. Introduction

I have fitted a thermocouple inside a nozzle, as close as I can to the nozzle tip, in order to investigate the effect that part cooling air deflected off the build plate (or printed object), would have. Two sets of tests were run. One with a brass nozzle, and the other with hardened Vanadium steel nozzle.

2. Background

It is impossible to blow cooling air over a printed part without some of that air being deflected back onto the nozzle. One can do all sorts of things with fans and ducting to prevent air being directly blown at a nozzle, but one can never eliminate air being deflected back from the build plate or the part being printed (at least, I can’t think of any way to do that). A silicone “sock” will help but such a sock cannot completely cover the nozzle tip (for obvious reasons).

Whilst it is known that part cooling air directed onto a nozzle will reduce the temperature, as far as I am aware, nobody really knows by how much. In my opinion, it is not wise to assume that the temperature seen by the hot end thermistor is a true reflection of the temperature at the nozzle tip when there is some airflow passing over the nozzle. The reason being that the temperature sensor is usually situated inside the hot block, some distance away from the nozzle tip and surrounded by metal with a relatively large thermal mass when compared to a nozzle. In theory, if the nozzle itself is a good thermal conductor, then the fact that it has a relatively low thermal mass should be irrelevant because, once fitted to the hot block, it becomes an integral part of the whole assembly. But heat blocks tend to be made of metal with very good thermal conduction properties such as aluminium or copper alloys, whilst nozzles tend to be made from harder materials such as brass and even steel with lower thermal conduction properties. In still air, this may not matter because the nozzle will eventually reach the same temperature as the hot block (or close to it) – it will just take some time. But as I said earlier, it is impossible to blow cooling air over a printed part without some of that air being deflected back onto the nozzle. So what happens to the nozzle temperature when there is some (deflected) air flowing over the nozzle tip? This is what this investigation aimed to establish.

3. Test Preparation

This was a good time for me to conduct these tests because, due to the current Covid 19 crisis, I still have no heat breaks for my latest hot end design. So I have a hot end which has never had any filament passed through it, making it possible to pass a small thermocouple through the hot end from the filament inlet, right down to the nozzle. Here is a close up picture showing the thermocouple tip before the nozzle was screwed it.

This is the latest, untested iteration of my hot end design, so the next thing I did was to fit the heater and thermistor using the Boron Nitride paste that Slice Engineering kindly supplied, and tune the PID parameters. As I mentioned before, I don’t have any heat breaks at this time, so I’ll need to run that PID again, (and with filament loaded) before I attempt to print anything.

Finally, I adjusted the position of the fans so that very little cooling air was being directed at the nozzle tip. In another post, I described my part cooling solution which uses two 30mm blower fans which can be adjusted in 3 linear directions (XY and Z) and also rotated at any angle to the vertical. These fans are rated at about 1.24 cfm airflow, so they aren’t especially powerful. By heating the hot end and observing the thermocouple temperature, it was quite easy to see when air flow was passing over the nozzle. So I dropped the bed by about 300 mm to ensure that no air was being deflected back from the bed, then adjusted the fan positions until I noticed about a 2-3 degree C drop in temperature as measured by the thermocouple inside the nozzle. This means that the majority of the cooling air was being directed towards the part being printed, and only a tiny fraction was directly blowing over the nozzle tip. It should be noted that, using a brass nozzle, it was easy to see a drop in the nozzle temperature of 15 degrees C or more if the fans were blowing directly over the nozzle. (The use of a silicone “sock” would help but this is discussed later).

4. Test Methodology

Here is a close up picture showing the fans in the position where the airflow causes a (about) 2 degree C drop in nozzle temperature with the bed well out of the way (so no deflected air). Note that this picture shows the steel nozzle but the first tests were conducted with a brass nozzle.

Here is a picture showing the set up.

I heated the hot end to 200 degrees C (a nice round number), and noted the temperature of both the hot end as reported by the Duet electronics, and the thermocouple using a separate hand held reader. Previous testing on heat breaks had shown good correlation between this thermocouple/reader combination and the Duet but as a check, I initially withdrew the thermocouple so that it was inside the hot block, close to the thermistor. The temperature correlation was within 0.2 degrees C with the thermocouple reading slightly lower. Then I re-positioned the thermocouple so that it was inside the nozzle, touching the nozzle wall close to the tip.

The first tests were with a brass E3D, 0.5mm nozzle. With the bed about 300mm away from the nozzle and no fans running, I noted the hot end and nozzle temperatures. The difference would be due to thermal conduction in free air. Then I turned on the part cooling fans at 100% and measured the temperatures again. This time I noted the initial drop in temperature, and also what the temperature recovered to after time. In the spread sheets below, this initial drop and subsequent recovery are shown as the min and max temperatures. The difference in temperature was therefore a function of air flow over the nozzle due to the position of the fans. I did these tests twice. Once by turning on the fan and observing the hot end temperature as reported by the thermistor on the Duet Web Contol. Then I turned the fans off, waited a couple of minutes for the temperatures to stabilise, then turned the fans on and observed the temperature as recorded by the thermocouple inside the nozzle on the “hand held” reader.

Then I turned the fans off and moved the bed up until it was 50mm from the nozzle, waited for the temperatures to stabilise and repeated the tests with fans on. This was repeated again with bed/nozzle distances of 40mm, 30, 20, 10, 5, 2 and 0.3 mm. The latter being my usual “default” layer height. Obviously, the bed /nozzle distance is only part of the reason why air could be deflected onto the nozzle. The other reason is air being deflected from the part that has been printed, but this will vary depending on the shape and size of that part – every one will be different. With the bed/nozzle distance of 0.3mm, I did two further tests. One using 50% PWM fan “speed” and the other using 25%.

Finally, the E3D brass nozzle was exchanged with a Slice Engineering Vanadium steel nozzle, also 0.5mm and the test sequence repeated.

The ambient temperature was 28 degrees C during the tests with the brass nozzle, and 28.9 with the tests using the steel nozzle. The bed was unheated. Obviously, there was no filament loaded. No silicone “sock” was fitted. All of these factors will have had an impact on the test results, as well as the geometry of my particular hot end and the air flow characteristics of these particular part cooling fans.

5. Results presentation

I haven’t found an easy way to put spread sheets into posts with this free version of WordPress so I have used Windows snipping tool to make png images which I’ll put below, but I’ll make the spread sheets available to anyone who wants them.

6. Results discussion

Before looking too closely at the results, and making any conclusions, it is important to note that these tests were done with no filament in the nozzle. I would expect that the heat conduction through the filament itself would serve to mitigate the observed drops in temperature with fans running. But by how much is another matter. I have no way to fit a thermocouple inside a nozzle which already has filament loaded. Obviously, the wire itself cannot be inserted through the heat break and hot block. Another thing to note is that, even if the temperature of the nozzle drops, that does not necessarily mean that the temperature of the filament (which has been melted in the hot block) flowing through the nozzle would drop by the same amount.

As I mentioned before, the tests were done with an ambient temperature of between 28 and 28.9 degrees C. A lower ambient would result in larger temperature drops. Conversely, the tests were done with the build platform unheated. If the bed was heated, the deflected air might be warmer so the cooling effect might be less. Similarly, a heated chamber would help to mitigate the effects.

A silicone sock would also help. But it should be noted that this should cover as much of the nozzle as possible. The temperature drop of the hot block was minimal and mostly automatically compensated for by the thermistor sensing that drop in temperature and heater PWM being changed accordingly by the control system.

Another factor affecting these results is the design of the hot end itself, as well as the flow characteristics of the part cooling fans. But it should be noted that, because the hot block is made of metal which is a very good heat conductor, the position of the temperature sensor and heater within that hot block are largely unimportant. At first glance, one could say that the temperature sensor needs to be touching the nozzle, rather than being inside the hot block. Then if the nozzle temperature dropped, the heater would automatically compensate for it. But that might not be the answer because if the temperature at the nozzle can really drop by as much as 36.2 degrees below the hot block (as observed) then the filament inside the hot block block would end up being heated by that additional 36.2 degrees.

Keeping all that in mind, the results were as follows

6.1 Brass Nozzle

In free air, with no fans running, the temperature of the nozzle was 198.3 deg C. So we can say that the nozzle runs about 1.7 degrees cooler than the hot block. This is largely unimportant because one would normally run tests to determine the best print temperature for any given filament, and use that temperature. Also a tolerance of 5 deg C either side of the optimum temperature is often acceptable.

With the fans running at 100% but with the bed far enough away so that no deflected air reaches the nozzle, the temperature dropped a further 2 degrees. This is the effect of a small proportion of the cooling air flowing directly over the nozzle, due to the position of the fans. Again, this is largely immaterial.

When the fans were switched from an off state to an on state, with the bed closer to the nozzle, the hot end temperature dropped about 1 degree, before recovering back to the set temperature of 200 deg C.

The effect of deflected air reaching the nozzle did not become noticeable until the bed to nozzle distance was 20mm or so. This temperature drop increased as the bed was moved progressively closer to the nozzle. At a bed to nozzle distance of 0.3mm (normal layer height) and with 100% fan speed, the nozzle temperature initially dropped to 186.8 before recovering to 187.8 deg C. So the difference between having air deflected off the build plate and not having any deflected air, with 100% an speed, was about (196.3-187.8 = ) 8.5 deg C. Using 50% fan speed, this difference reduced to 6.0 deg C and with 25% fan speed, it was 3.8 degree.

6.2. Steel nozzle

These results were significantly different to the brass nozzle.

In free air, with no fans, the temperature inside the nozzle was 192.2 deg C with the hot block at 200 deg C, compared to the brass nozzle which was 198.3 deg C.

I will repeat here that this is with no filament and that with filament loaded, the temperature difference might not be so dramatic (but who knows?).

With the part cooling fans running but still with the build plate far enough away from the nozzle that there was no deflected air, the nozzle temperature dropped to 181.3 deg C compared to the brass nozzle which was 196.3 deg C. So we can say that a small amount of direct cooling air, which can reduce the temperature of a brass nozzle by about 2 deg C, will reduce the temperature of the hardened steel nozzle by a further 11 degrees or so. Here we can see the effect of steel being a poor conductor of heat compared to brass. When the conduction between hot block and nozzle is high, the cooling effect of air blowing over the nozzle is largely mitigated by heat conducting from the hot block into the nozzle. But when conduction is low, the cooling effect is not mitigated by the same amount.

As with the brass nozzle, the effect of deflected air starts to be most noticeable when the build plate to nozzle distance is 20mm or less, and gets progressively worse as that distance reduces (which is understandable).

With the brass nozzle, the temperature would drop but then recover by about 0.6 deg C or so, roughly in line with the hot block. With the steel nozzle, this temperature recovery was only about 0.1 to 0.2 deg C.

At 0.3mm (layer height) bed to nozzle distance, the temperature dropped to 164.0 deg C with 100% fan speed. At 50% fan speed, the temperature was 173, and at 25% fan speed, the temperature was 180.1. Overall, that is still 20 degrees lower than the hot block temperature and 12.4 deg C cooler than the brass nozzle under similar conditions. The ambient temperature was 0.9 deg warmer when the steel nozzle tests were conducted, so the deflected cooling air was slightly warmer.

7. Conclusions

Readers must draw their own conclusions based on the results I have presented, and bearing in mind all the caveats which might have an impact on these results, but here are mine:

I will certainly give serious consideration to making a silicone “sock” once I know the final size and shape of this hot end. Any such sock will cover as much of the nozzle as it is possible to do.

From the results using the brass nozzle, I will likely use a fan speed of 25% or less if I can get away with it. With this amount of cooling air, that which is deflected will only drop the nozzle temperature by 3.8 degrees C or so (perhaps less with filament loaded). If I use a larger nozzle, or if I need more cooling air for any other reason, I will consider using a slightly higher print temperature to compensate.

From the results using the steel nozzle, I think I will start by using a hot end temperature of 5 to 10 degrees more than I would for a brass nozzle, to compensate for the fact that in still air, with no part cooling fans running, the steel nozzle (without any filament loaded) showed 192.2 when the hot end was running at 200 deg C.

With the caveat that no filament was loaded, the effect on temperature of any air flow, deflected or direct, on a steel nozzle is dramatic. I did not imagine that the nozzle temperature could potentially be as much as 36 degrees cooler than the hot block with (mostly) deflected cooling air. For that reason, I will be very cautious of using any part cooling air whatsoever, if I start printing with a hardened steel nozzle.

8. Other thoughts.

Before the demise of my heat breaks, I had problems with filament leakage between the plates of this hot end. The cause appeared to be too much pressure, coupled with thin aluminium plates buckling. I have made new plates out of brass and removed the mixing chamber that I thought was too restrictive and which was the cause of the pressure build up. Due to Covid 19, I still don’t have any heat breaks and have yet to test my new design. But, when I had the leakage, was I using a steel nozzle? – Yes. Did I have part cooling fans running? – Yes. Were those fans blowing air over the steel nozzle? – Yes. Could I have been trying to print with a nozzle temperature of around 160 deg C or less (the hot block was set to 190)? – Highly likely. Might that low temperature have been the reason for the back pressure, rather than a too restrictive mixing chamber? ……………

…..Until next time……

Ian