OC‘axe for the Bitaxe: optimal cooling or overclocking; your choice.

It is a case that was specially developed for the small open-source home miner Bitaxe. But it is not just a case, it is an overall concept of minimalism, aesthetics and unrivaled performance, all in one. OC'axe is in many ways special and currently unique, built for enthusiasts, aesthetes and plebs who simply love raw power. But also for people who like it absolutely quiet, without sacrificing additional power.

It is also a must-have case for mining companies, which must have at least one OC'axe with Bitaxe on their desks for presentation purposes.

The entire development took a little over 6 months with many adjustments and unique eye-catchers that are not recognizable at first glance. Be it the innovative cable management, or the integrated 70W power supply with the option of operating a real daisy chain Bitaxe swarm, or the unique closures in the housing, which are not only stable and secure, but also bring a high gamification factor with them. And of course the security of no open power cables and sockets, especially if children and/or animals live in the household. A simple and at the same time well-thought-out all-in-one concept that is second to none and requires absolutely no maintenance.

In addition, OC'axe has set several air-cooled hash records with a Supra and a Gamma that are still unrivalled today - I don't want to say it so directly, but it really is the most innovative overall concept currently available for Bitaxe on the market.

It all started with me setting up a small Bitaxe swarm. It was the day I had to clean up my old IT cabinet (everyone has one of those). I found an old Northbridge copper probe that was still new and in the box, which I bought in 2010/2011 but never used. The Ötzi - that's when I got the idea to mount it on the Bitaxe Supra.

To my amazement, it was so good that that I was able to operate a slightly overclocked Bitaxe Supra without a fan with the Ötzi. Later, it even managed 800Mhz at 1,375V with a 60mm 12V Noctua that barely moved any air.

From then on, the fever gripped me even more and I got completely lost down the rabbit hole. I'm not sure, but I think it was pretty much the only time that a Bitaxe Supra was overclocked back then. The community was still relatively small and there were hardly any reports or tests about settings, limits and optimizations. So I felt my way from day to day and exchanged ideas with many new friends.

From the small tests and discussions, some articles were created, such as the already well-known “Cracking nuts with the Bitaxe while following the white rabbit”. That's what I really missed at the beginning of my journey... So new insights were recorded and recorded in the form of one of the longest articles I saw on this topic at the time.

But back to the topic...

Overclocking the Bitaxe Supra worked so well that I actually managed to melt a DC plug. However, that happened before I overclocked the copper Ötzi, when I hadn't fully understood the problem yet.

However, I later realized that cooling the ASIC processor was not the primary goal. Since I didn't have a thermal camera back then, everything was “measured” with my finger - if you burned your finger, it was clearly too hot. This is how I found out that the DC connector, the top side of the PCB around the DC connector, but especially the V-regulators at the back became extremely hot and thus prevented higher overclocking. I actively shared all this information with the community, every small step

Yes, that was the very first attempt with OC'axe - at the time, I couldn't have imagined how it would continue and guide me.

But after I realized that this construction was not really optimal, I remembered a Shroud - which I used at the time to overclock my PC. So a 120 -> 92mm Shroud was taken, the Bitaxe provisionally placed on it and again fired with a no-name 12V fan.

But it worked so well due to the air compression that that I was actually able to run the Supra at 800Mhz with absolutely stable performance.

Even at that time, I was able to see completely crazy hashrate peaks that probably no one with an air cooling system could see at the time. As seen here in the picture, an incredible 1,584.8 Gh/s and that with absolutely stable operation for over 10 hours.

My first 1,000 Mhz on the Bitaxe Supra were achieved exactly with this construction. Even if the values were not optimal from today's perspective, it worked.

In fact, the idea and the basis of today's public Alpha OC'axe have remained the same. To compress and accelerate the airflow without scattering losses, to divide the airflow between two chambers and to cool both the top and bottom sides of the circuit board - but more on that later.

It quickly became clear to me that I wanted to pursue this idea and this path. Quite naively, I made a first sketch and had a first prototype printed at a 3D printing service. The prototype was not particularly beautiful and the service provider did not put much effort into the print quality. But I was able to carry out initial tests with the prototype and, above all, to control the entire air volume and the air flow 100 percent.

After all, I calculated at the time that about 20-30% of the air flow escaped with the shroud solution, resulting in a high level of scattering loss. But it's also clear that when the compressed air jet from the shroud hits a slanted circuit board, and you simply don't have any directional walls or the like on the left and right, a lot of the air flow escapes.

The first OC'Axe prototype, in combination with the copper Ötzi, simply thrilled me. At the time, I was able to overclock the Bitaxe Supra to an incredible 1048 MHz at 1500 V. It ran absolutely stable, no DC plugs melted anymore, I attached heat sinks to the hottest spots - everything was almost perfect.

But what currently looks like a 100% success story is actually a story of disappointment. There were so many failures that I almost gave up at times. I would even say that 99% of the way to the public alpha OC'axe consists of failures. But I'll probably have to cover that in a separate article, otherwise this article will just be way too long. After all, we want to concentrate and celebrate the 1% of the current state first.

However, during further tests, I realized that the internal geometry was not optimal. Initially, I thought it was a good idea to divide the air volume into 3/4 for the back of the board and 1/4 through a duct on the top of the board to direct cold and fresh air directly into the cooler and fan.

However, subsequent calculations, testing and flow simulations showed that the split and acceleration ratio was not optimal for optimal cooling performance.

So new designs, experiments and insights followed again and again. To avoid boring you too much, here is a list of the problems that had to be solved (I'm sure I've forgotten or suppressed one or two).

• Haptic: It was important to me that the case looked good. However, this meant that the case could not be printed in one piece using the FDM printing process.
• So the case had to be divided into many parts, and here, too, I experimented a lot. What looked good in the design did not work so well in the physical world.
• When you divide something, you also have to be able to connect it. All the crazy ideas were tried out, screws, rail systems, plugs and I even thought about gluing. The problem was not only to connect the parts of the box with each other, but also everything that was in the box.
• It was only much later that I realized that a removable base had to be developed for the Bitaxe. Otherwise, the Bitaxe could not be attached with screws.
• Similarly, the idea of cable management came to me much later, when I saw how unsightly it is to have a lot of cables hanging out of the case.
• The 120mm Noctua fan also had to be attached and, above all, decoupled from the case. There were countless attempts and concepts for this too... Until I finally came up with the idea of simply using the supplied rubber pins.
• And last but not least, of course, the integration of the power supply, which is safely stowed and sufficiently dimensioned – and above all, there are no additional cables hanging around.

As I said, there were really a lot of failures – that will surely be worth a separate article or even a series of articles.

Concept

Knowing that air pressure can compress air flow and therefore increase airflow, resulting in better cooling, it was my great desire to implement exactly this concept. My shroud test has already confirmed that it works perfectly with the Bitaxe. From the Bitaxe Supra onwards, the first problem with the very hot V.REGs on the back of the board was clearly visible, especially when overclocking outside of the AxeOS specification.

However, it was also important to me that the DC connector and the area around the DC connector were also cooled. So it made sense to place the Bitaxe at an angle in the case and to cool both sides with a strong 120mm fan. It is also worth mentioning that the airflow that is directed through the smaller secondary chamber is also sucked in by the fan on the ASIC processor, which further increases the cooling performance.

This idea and, above all, the overall concept was also confirmed by the physical prototype, which was further optimized in the course of development.

Additional ideas also arose from various optimizations, such as the removable Bitaxe tray, which can tend to increase compatibility for future Bitaxe generations. The same applies to the cable management, even if not many cables are needed when operating the Homeminers, the few cables are still a nuisance and a real dust magnet.

The icing on the cake and fortunately also the last idea was to create space for a 70 watt Mean Well LRS-75 power supply. The goal was to supply the Bitaxe with optimal voltage even with strong overclocking. Another advantage is that the open contacts are completely encapsulated and thus the operational reliability is significantly increased, especially if children or pets are in the household. Positive side effect, the power supply is automatically cooled by the air sucked in by the 120mm Noctua.

There is no substitute for safety.

Geometry

Due to the completely new concept, I was unable to draw on any comparative values. So everything had to be rethought, reconsidered and tested. Through simulations and calculations, I found out that the optimum angle of inflow of air on the underside is between 12-23°.

Likewise, it was taken into consideration how much air a fan can draw in at most, in order to construct optimal air intake holes. The honeycomb pattern, which has been proven in nature for thousands of years, was used for this purpose, offering several advantages: noise reduction, extreme stability and optimal air flow.

Both the primary and secondary air chambers are constricted, thus accelerating the air velocity at the targets to be cooled. In addition, the air in the primary chamber is not only accelerated, but also directed to the hottest spots on the circuit board. I always joke that it's like a little tornado in the OC'axe.

Likewise, the well-thought-out cable ducts, symmetrically arranged on both sides, to store everything optimally and invisibly.

A small feat was to use the natural geometry of the Noctual fan, which can be fixed with rubber pins. Not only that, the Noctua fan is pushed down and back through precisely calculated cutouts in the PSU tray, but the resistance of the rubber pins pulls the PSU tray to the cooling tray, automatically fixing both parts together. The fan is positioned so perfectly that both parts hold together without additional fasteners (even if these were implemented), which makes assembling and mounting the case much easier.

Measurements

Now for some theory and math, but it's important. At the moment, probably due to my articles and tweets, you can see a lot of setups with a fan at the back, even some attempts to position a 40mm or 60mm fan in a case and point it towards the back. In my opinion, a 40mm fan is completely ineffective at moving any significant amount of air, but it's certainly better than nothing and may look cool in some setups. The same applies to the 60mm fan, which is significantly more effective, but again, you simply can't build up enough air speed and pressure... Even if it works well for some concepts and can at least cool the hot v.regs on the back well.

My idea from the beginning was different: why mess around with small fans when you can use a 120mm model right away. Here you simply have a complete all-rounder, more than enough air pressure and air speed, a wide selection and you move an incredible amount of air with it. If you then calculate the optimum angle of incidence of the air, compress the air flow and explicitly direct it again in the compression, and also divide it into primary and secondary chambers - then the concept simply has to work very well based on these specifications.

Here are two calculation curves, based on a chamber division into two parts and concave airflow acceleration, to make cooling even more effective. Of course, a Noctua NF-A12x25 5V PWM with incredible values of 96.2 m³/h at 2.07 mm H2O serves as the basis – which is also used in physical tests. This is actually the first choice when it comes to operating the OC'axe.

Both diagrams show the effectiveness of cooling at different fan speeds. It is actually essential to choose sufficiently dimensioned outputs at certain concave angles so that the cooling effect of the underside but also the top of the PCB works optimally. 

Of course, other 120 mm fans can also be used; it's up to you. But you will actually get the best results with this Noctua fan. Whether your goal is semi-passive, silent or extreme overclocking... there's no way around Noctua.

But as with everything in life, it depends on measuring, testing and improving – or not. I would say that OC'axe is the best-calculated and optimized cooling concept based on air cooling that has actually been designed and built according to calculated data. There are just so many small cogs that simply mesh together... just like in a supercar.

Calculation & Simulation: Comparing the calculations between the provisional design with a shroud, the first printed prototype, the second prototype (where I thought I was done) and the current public alpha version. The simulations and calculations show that a single-directed air jet is okay (better than nothing), but it is far from optimal. For example, you lose about 20.25% of the airflow with the provisional design.

You can also see how the ratio has been increasingly adjusted and optimized. A 3/4 split for the primary space in the case is really not optimal. For this concept and the cooling performance, the calculated 64.10% is really the golden ratio - so that, of course, enough airflow is generated for the top side of the board.

But take a look at the data for yourself.

Particular attention should be given to the effectiveness of the volume flow and the air velocity in the primary and secondary chambers. These should be emphasized again separately to illustrate the formation of a tornado in the primary chamber.

Test environment and conditions

The tests were conducted indoors at 21-23°C. The fans were set to 100% to keep the test data comparable and not to complicate the analysis unnecessarily. A Bitaxe Ultra, a Bitaxe Supra and a Bitaxe Gamma were each equipped with firmware v2.6.0 b7. All three Bitaxe had no additional heat sinks on the circuit board, neither on the top nor on the back - this approach should show the RAW test data without the need to implement further stray-error measurements.

JLJ TP-133 thermal compound with a thermal conductivity of 13.3 W/mK was used for the thermal paste. This was for the simple reason that the large number of test series and the use of expensive thermal paste would have brought further improvement, but from a cost-benefit perspective, it would have been a waste. And with such high thermal conductivity, I think it's a legitimate replacement for thermal paste.

For the tests, I actually limited myself to three Noctua fan models, because an even larger selection of fans would have only unnecessarily prolonged the test series. Besides, OC'axe is a case for enthusiasts who probably want to build the best possible setup and certainly won't use $2 fans (although it's certainly possible) – although in many cases fanless operation is possible with the ASIC processor.

The case itself, the OC'axe, is a case that has been completely optimized for the Noctua NF-A12x25 5V PWM. The Noctua A4x20 5V PMW and the Noctua A6x15 5V PMW were used for the ASIC cooler tests. 

And yes, that's correct - the Noctua NF-A6x15 5V PWM is significantly more powerful than the x25 version due to the slat arrangement. So from now on, you should always buy this version when a 60 mm fan is needed.

In the upper table, the cooling performance is always given without a fan on the ASIC cooler. OC'axe is operated semi-passively with a 120mm Noctua fan. In the lower table, the test data is shown with a 40mm or 60mm Noctua fan.

In some cases, it was not possible to mount a cooler on an Ultra and/or Supra because no adapters were available. I saved myself the extra work of building the adapters, since most people run a lot more GAMMA's than Ultra or Supra. In general, it can be said that the Ultra and Supra are a good combination with the 52Pi ICE-Tower, for which I have developed a new optimized adapter.

Coolers

During the tests, I took into consideration the most common coolers. But I also included some exotic ones that no longer exist or are harder to come by. Due to the different shapes, materials and weights, rough estimates could also be made for coolers not tested here.

This cooler is probably the one most often found on a Bitaxe. Whether Ultra, Supra or Gamma generation, the cooler survives every time and moves up a generation. Even though it is actually no longer up to date from the Gamma generation onwards. It is the absolute flyweight among coolers, weighing only 21.5 grams and made of aluminum. Most of the time it is yellow anodized, but I have also seen it in black and blue. The dimensions are very compact at 40 x 40 mm and a height of 15 mm.

Its geometry makes it a perfect fit for the OC'axe, as the airflow can slide directly through the fins, ensuring optimal cooling. The test results looked quite exciting; no 625 MHz frequency was possible on the Gamma without a fan. However, it was possible on the Ultra and Supra – with a few heatsinks on the Gamma, overclocking within the AxeOS specification would probably also work on the Gamma.

This Northbridge cooler has been around for over 20 years. Today it is available in different colors for little money - in yellow and blue it is often on sale. It is a bit more expensive in black, but this is incomprehensible since there are actually no differences in the pigmentation when anodizing the aluminum cooler. It is actually the absolute all-rounder, quite massive for its size of 40x40x40 mm and a weight of 67.5 grams.

It cools the ASIC without any problems, only the attachment is a very nerve-wracking affair. But I have developed an adapter for that too, which the plebs can print for free.

Even with the cooler outside the AxeOS specification, I was able to overclock the Bitaxe Gamma. At 70.5°C, the Bitaxe ran very stably – although I would actually recommend lower frequencies here, up to a maximum of 65°C.

Now for an exotic one that I got from @PowerMining to test. It's actually really heavy, has a massive plate - the only problem is the very thin and delicate pins, which bend just by looking at them. Mounting a fan on the cooler is also not an easy task, you are more likely to break the fins – so the only option during testing was to place a 40 mm Noctua on the cooler. It's a shame, because good copper coolers are hard to find these days. It measures 40x40x35mm and weighs 105,0 grams.

In the test, it showed that copper is still the perfect choice when it comes to heat absorption and rapid heat dissipation. Overclocking outside the AxeOS specification, even without a fan, was absolutely no problem.

A really very massive and elegant cooler that looks really good on the OC'axe without a fan. It is really a cooler that I can highly recommend if you don't want to use a second active fan and want to cool your Bitaxe semi-passively. In winter, it really works like a small heater and you can overclock a gamma without any problems, even beyond the AxeOS specifications. With a 40mm Noctua, an average of almost 2 Th/s was possible.

Since I wanted to know the net weight of the cooler, I completely removed the steel bracket. It weighs exactly 111.0 grams of pure copper. It is fairly easy to get in the US, but more difficult in the EU. But I heard that solomining still has some in stock... so if you're quick, you can get it.

I don't have to say much about Ötzi, he is and remains a legend. With or without a fan, he simply delivers. 100% copper distributed over 139.5 grams and in combination with a fan, he lets the Gamma diligently search for blocks at 2.3 Th/s. It's just a shame that the geometry is not quite optimal for the OC'axe - a tower cooler, after all. But you can't deny the performance of the Ötzi in any case, it's just brutal.

The ICE tower is probably the most widespread cooler in the Bitaxe ecosystem and represents the first step towards optimized cooling with a heat pipe. Although it is not massive and does not contain an excessive amount of copper, it perfectly masters the concept of heat pipe cooling. Equipped with a 40mm fan, overclocking outside the AxeOS specifications is also possible.
 

For this cooler, I have developed two new and significantly optimized adapters that can now be used on the Ultra, Supra and, of course, Gamma. Thanks to the optimal adapter geometry, I was able to achieve results of -8-10°C on a Gamma. The adapters are available at Makerworld.

This cooler is in a completely different dimension and offers an almost perfect cooling solution. It comes with two heatpipes and weighs only 94 grams. With an adapter and a fan shroud, it is not afraid to put even the Bitaxe Gamma cooler in its place and to stabilize significantly higher overclocking absolutely stably.

A detailed test of the cooler itself can also be found here.

But even though the cooler is so impressive and delivers really good cooling performance - the cooling performance compared to other coolers is actually within reason. I would have liked more, but those are the facts.

Many have already read my review. If you haven't already, you can read it here.

There is a lot to say about the Low Profile Plus. With its geometry, it provides us with the absolute high-end cooling solution. It also proves that good cooling performance can be achieved not only through the choice of material and weight. 

But take a look at the detailed review, I don't want to repeat myself here :)

One thing became clear to me during the whole trip: better cooling reduces power consumption. This means that a significantly higher overclock is possible with overclocking or significantly higher efficiency with underclocking. During the tests, I noticed that the OC'axe cooling concept consumes an average of 5-7 watts less.

At first glance, that doesn't seem like much, but if you think further, of a Bitaxe swarm, then you already have a saving of 70 watts from 10 units. This means that you could operate three more Bitaxe in the overclocked state or you could reduce the electricity bill or the load on the PV system.

Although the test series were very extensive, there are still further optimization possibilities. For example, heat sinks made of aluminum or copper can be attached to the hot chips, as well as to the top side next to the DC plug. This measure further improves the cooling concept and makes it possible to overclock the Bitaxe even higher or, alternatively, to run quietly despite overclocking.

In addition, the possibility of overclocking has not yet been tested, but there is also enormous potential for increasing efficiency and the hash rate here.

It should not be forgotten that the test results also depend on the ambient temperature. What is possible on cold days in winter will probably not work in hot summer. Nevertheless, the OC'axe also offers enormous overclocking potential in summer without overheating. At least it will work significantly better than with the current enclosures that you can get or with DIY solutions that still involve air dispersion losses.

In my opinion, it has also become quite chic, it presents the Bitaxe beautifully and does not look as nerdy as it does with various cables and an external power supply dangling around it. A chic case for the Bitaxe, minimalist, functional and truly extraordinary.

Another advantage: you can start small: OC'axe with a Bitaxe, then gradually upgrade the components if you want. Another unique feature is the ability to operate up to three Bitaxe in a daisy chain (without overclocking, though). With the OC'axe, I have succeeded in developing a solution for power users but also for silence fans. It still leaves room to test other concepts and the gamification factor is not neglected either. Gamification - I think it's really successful and it's even fun to tinker with the OC'axe and the components and to further optimize the Bitaxe.

The result is a product that was developed entirely by a pleb without external financing. Even though it was sometimes very nerve-wracking, it has been a very nice journey so far. Many side projects have been created and made available to the community as open source. Whether it's new and optimized adapters, test data or cooling concept recommendations, I hope to be able to continue working in this direction in the future.

The case itself consists of a total of 11 individual parts that are perfectly matched to each other. Unfortunately, the elaborate printing process takes about 30 hours in total and the case weighs between 520-550 grams. I printed the first public alpha batch in black and orange, which symbolizes the connection to Bitcoin.

A few killer arguments should not be missing, without fabricating even more text:

• Use the next Bitaxe generation today
• With better cooling, lower power consumption
• Optimal distribution of the air flow to two chambers, which cool both sides of the Bitaxe
• Significantly more effective cooling, by transporting the cold air from below - keyword: Thermal air mixing.
• Semi-passive and noiseless cooling possible, whether Ultra-, Supra- or Gamma generation
• Extends the lifespan of the components and thus of the Bitaxe significantly
• Optimal power supply through a 70W Mean Well, with automatic cooling through this
• Encapsulated power contacts, significantly increases safety, especially with children/pets in the household
• Housing can be used as a quick test station for the Bitaxe by quickly and tool-free changing of the tray or the Bitaxe
• Well thought-out cable management, no more cables lying around
• Ideal for warm environments or warm countries, where good cooling is particularly important, defies overclocking
• Multiple hash rate records with the Supra and Gamma, unbroken under air cooling concepts so far
• You have a lot of stickers from the BTC ecosystem, but so far you haven't found the right places. OC'axe offers more than enough space for them ;)
• Gamification and inspiration for further optimization guaranteed
• Minimalist, aesthetic and highly functional - this combination is currently unique in the Bitaxe ecosystem.
• For enthusiasts, developed and manufactured in Germany in small series

I'm looking forward to the next stage of the journey, because we're just at the beginning

I now have to move on to the next phase, which is to find suitable ways to manufacture the case efficiently and as cost-effectively as possible. The companies I have contacted about manufacturing the case are simply too expensive. Furthermore, the manufacturing quality at three of the tested companies is really poor and does not meet my quality expectations.

But it should definitely not be a mass-produced product. I would like to offer it exclusively to Entusists in very small series. Like a supercar that you don't sell at the bazaar... It should remain something very special.

Further updates will follow here on the website, on Twitter or by e-mail. I am of course already grateful for ideas and comments - so feel free to send them.

At the moment, 21 public alpha OC'axe have been manufactured - so I depend on the alpha testers to help me eliminate as many teething problems as possible that I may have overlooked. Some units have already been promised and will soon begin their journey - and that already worldwide on 4 different continents.

I would be very happy if you would take part in the short survey and show me how much interest there actually is. Your information will, of course, be treated confidentially and will not be published. It will be used solely for internal analysis.

One of 21 public alpha OC'axes will be raffled off among all participants. So if you know someone who is interested in the topic, I would be happy to receive a recommendation. Just note in the text field how many recommendations you have made and where.

• Take the survey

Thank you very much for participating in the survey. A total of 314 votes were received, and number 42 was the winner. We are delighted that the winner comes from the Bitcoin country of El Salvador.

If you would like to be added to the waiting list for an OC'axe public alpha, please check here on the website to find out what you need to do.