OC'axe

Manufacturer

The king is dead, long live the king; Low-Profile Plus on Bitaxe Gamma

13. November 2024, 00:40

Hannover,

Germany

Analysis

In fact, I didn't think much of it when I ordered the 52Pi Low-Profile Plus. A quick test and the test data in the already published article “Ice Tower, Argon THRML 60mm and the ‘Ötzi’ on the Bitaxe Gamma, what can they do?” But, during testing, it turned out that the cooler deserves its own article.

Introduction

In the meantime, there are a lot of articles, which was not planned at all in this form. I'll have to think of a strategy for how to view all the information in one place, especially the test data, which is now spread across four articles. Either way, I saw by accident that someone mounted a 52Pi Low-Profile Plus cooler on a Bitaxe. I was interested in the geometry because it is a raised top blow cooler for the Raspberry Pi 5. Although I'm not a fan of top blow coolers, confirmed by the many test series, I wanted to take a closer look at this cooler.

Unfortunately, not much information about the cooler is available on the internet yet, even in the manufacturer's wiki, the product is labeled as weighing 220g. My thought was that if the cooler really weighs 220g, it offers a massively increased mass over all coolers that can be used with the Bitaxe.

The cooler was quickly ordered and arrived quickly, but when unpacking and looking at it, it turned out that the cooler does not weigh 200g under any circumstances. Rather, it is the total mass of the delivery, which, in addition to many screws and cooling pads, also includes a huge aluminum cooling plate for the Raspberry Pi 5. However, for our Bitaxe cooling solution, we only need the mounting screws and the cooler itself.

For this reason, I wondered what the actual mass or weight of the coolers we use to cool the Bitaxe is.

The first disillusionment came here, but it was surprising that the Argon THRML was significantly larger but not significantly heavier. Nevertheless, I was convinced that the 52Pi Low-Profile Plus would be between the Ice Tower and Argon THRML in terms of performance.

Weird geometry

The only differences of the 52Pi Low-Profile Plus compared to the already tested coolers were the four copper heat pipes and the extremely massive base cooling plate, which is over 1 cm thick. The base plate is really huge compared to other cooler candidates, 30x22 mm, there could easily fit four ASICs on it.

Cooler name	Number of heat pipes	Measured weight
52Pi Ice Tower	1	36,0 g
Argon THRML	2	94,0 g
The 'Ötzi'	2	139,5 g
52Pi Low-Profile Plus	4	73,6 g

But the non-square shape of the lamella arrangement is also unusual, which is why a fan of an unconventional size of 50x50mm is supplied with the cooler. This is not only noisy but actually has only 3 connection pins in micro-plug format; which means that compatibility for fans that you would like to use is not possible. In this case, you naturally want a good fan, possibly quiet and effective, so Noctua – but you can't get the Noctua mounted on the cooler.

Likewise, four copper heat pipes are a very interesting feature, but due to the size of the ASIC surface, only two heat pipes can be achieved. Two other heat pipes are thus without function or have very little effect. To test the full potential of the cooler, two 0.3mm thick copper plates were placed on the base, which fit perfectly. The connection point to the ASIC lies exactly in the middle between these two copper plates, which can distribute heat evenly across all four heat pipes. Of course, it is important here to distribute a good amount of thermal paste over all contact surfaces to avoid cold spots.

Test Environment

During testing itself, Thermal Grizzly Kryonaut Extreme is used as a heat transfer compound to ensure the optimal connection between the 52Pi Low-Profile Plus and the ASIC.

The new reference in the form of the air cooler is the Noctua NF-A6x15, which has a significantly increased air pressure compared to the A6x25 model, thus making cooling significantly more effective.

To keep the voltage stable and ensure a proper supply to the gamma, the MEAN WELL LRS-200-5 5V 40A comes into play as a power supply with sufficient reserve, connected with 16 AWG power cables.

The fan speed was manually set to 100% in all test series so that comparable and verifiable test results could be created.

Adapters

As is so often the case, 3D-printed mounting adapters are needed to attach a cooler to the Bitaxe. Although an adapter already exists for the 52Pi Low-Profile Plus, in my case it did not fit the cooler at all and was very fragile. When I screwed the adapter to the cooler, it unfortunately broke.

So I developed another version that is much more stable and, in addition, the screws supplied can be used to attach the adapter.

Furthermore, the 50x50mm fan gave me a slight headache; I really wanted to mount the Noctua NF-A6x15 on the cooler to achieve the maximum possible effectiveness. So here, too, an adapter had to be developed that could be used to securely and elegantly attach the Noctua to the original mounting brackets of the cooler.

In fact, a small cooling boost has been created due to the geometry of the Noctua adapter, which is a shroud with a reduction from 50 to 60 mm. Does that mean that the air is compressed even more and that the cooler can dissipate the heat even better?

Both adapters are free for private and non-commercial use and can be found here:

Testing data

I have reorganized the tables a bit to make it easier to compare the previous test data for the tested coolers with the 52Pi Low-Profile Plus cooler. The frequency and power scheme was retained. From now on, it will be continued in this way, sorted by the set speed in AxeOS up to the maximum speed of the test series.

400 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	400	1,15	0.816	13,39	12,8	1,15	1,11	43,4	55,1	0.5 h run
Argon THRML, NF-A6x25 PMW	400	1,15	0.816	15,95	13,0	1,15	1,11	27,9	44,0	0.5h run
The 'Ötzi', NF-A6x25 PMW	400	1,15	0.816	14,14	11,1	1,15	1,12	26,5	51,0	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	400	1,150	0.816	16,02	13,2	1,15	1,13	18,9	43,0	0.5h run

525 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	525	1,15	1.070	14,73	16,5	1,15	1,11	41,2	69	0.5 h run
Argon THRML, NF-A6x25 PMW	525	1,15	1.070	13,25	16,3	1,15	1,11	33,2	51	0.5h run
The 'Ötzi', NF-A6x25 PMW	525	1,15	1.070	12,99	13,9	1,15	1,12	41,4	56	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	525	1,150	1.070	12,79	15,6	1,15	1,12	19,8	49,0	0.5h run

596 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	596	1,2	1.220	13,72	18,8	1,2	1,16	43,8	70,2	0.5 h run
Argon THRML, NF-A6x25 PMW	596	1,2	1.220	13,52	19,6	1,2	1,17	37,6	62	0.5h run
The 'Ötzi', NF-A6x25 PMW	596	1,2	1.220	14,43	17,6	1,2	1,17	45,3	65	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	596	1,200	1.220	14,60	18,4	1,20	1,17	25,6	55,0	0.5h run

625 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	625	1,25	1.270	16,34	22,3	1,25	1,22	48,2	78,1	0.5 h run
Argon THRML, NF-A6x25 PMW	625	1,25	1.270	14,9	23,1	1,25	1,22	44,3	67	0.5h run
The 'Ötzi', NF-A6x25 PMW	625	1,25	1.270	16,3	20,7	1,25	1,21	49,4	82	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	625	1,250	1.270	15,60	20,9	1,25	1,22	29,8	61,0	0.5h run

650 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	650	1,3	1.330	18,99	26,2	1,3	1,26	63,9	89,1	0.5 h run
Argon THRML, NF-A6x25 PMW	650	1,3	1.330	16,73	25,6	1,3	1,27	48,7	75	0.5h run
The 'Ötzi', NF-A6x25 PMW	650	1,3	1.330	13,15	24,2	1,3	1,26	51,3	86	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	650	1,300	1.330	16,29	23,3	1,30	1,27	32,8	69,0	0.5h run

675 Mhz Test series

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
52Pi Ice Tower, NF-A4x20 PMW	675	1,325	1.380	20,99	29,8	1,32	1,29	74,9 (fail)	99,2	5 min run
Argon THRML, NF-A6x25 PMW	675	1,325	1.380	19,86	27,8	1,32	1,3	59,8	85	0.5h run
The 'Ötzi', NF-A6x25 PMW	675	1,325	1.380	18,87	28,3	1,35	1,32	55,7	93	0.5 h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	675	1,325	1.380	16,27	25,7	1,32	1,30	38,1	74,0	0.5h run

Analysis of results

That's right, I was just as surprised as you are. The much smaller cooler, which is not only slightly but also smaller, outperformed the 'Ötzi' by miles. That was a complete surprise to me, but on closer analysis it suddenly made sense.

The 'Ötzi' is made of 100% copper, weighs a whopping 139.5g, and has two heat pipes. And that's exactly where the problem lies: there are only two heat pipes and not four, like on the 52Pi Low-Profile Plus. The 'Ötzi' absorbs the heat of the ASIC extremely quickly due to the material properties, tries to transport it to the cooling fins via two heat pipes, but this does not work as quickly as it could. A kind of “heat build-up” occurs, so the fan also cannot remove the heat from the cooling fins quickly enough.

Furthermore, the 'Ötzi' does not have a Noctua Shroud due to its geometry, so something like that could certainly be created, but whether it would really be effective remains questionable. After all, the Noctua A6x25 fits perfectly on the 'Ötzi' in terms of height and width.

So I really have to admit that a cooler that is over 13 years old and made entirely of copper has to beat better and newer cooling concepts made of four copper heat pipes and aluminum fins. Another advantage of the 52Pi Low-Profile Plus is that the base is over 1cm thick and thus has more heat absorption volume than the slightly thinner copper base of the 'Ötzi'.

Stability assessment

In fact, I thought it couldn't be that the near-perfect 'Ötzi' is so inferior to such a low-priced 52Pi Low-Profile Plus. For that reason, the measurements were double-checked and verified, and it was true.

As the ultimate stability test, a light to medium overclocking on the OC'axe (Prototype 3). I have to say that I tested a Bitaxe Gamma that didn't win a silicon lottery and didn't perform that well, and didn't have an aluminum or copper heat sink on the PCB, which surprised me even more about the results with the 52Pi Low-Profile Plus.

733 Mhz (1.5 Th/s)

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
The 'Ötzi', NF-A6x25 PMW	733	1,250	1.480	17,68	24,5	1,35	1,32	33,5	53,0	0.5h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	733	1,250	1.380	16,35!	23,2	1,25	1,22	18,4	47,0	0.5h run

1.000 Mhz (2.04 Th/s)

	Frequency Mhz	Core Voltage (mV)	Avg. Th/s	Efficiency J/Th	Power in W	ASIC Volt. req.	ASIC Volt. mes.	ASIC Temp °C	V. Reg Temp °C	Hint
The 'Ötzi', NF-A6x25 PMW	1.000	1,4	2.040	17,01	41,5	1,4	1,36	53,3	83	2h run
52Pi Low-Profile Plus, NF-A6x15 5V PMW	1.000	1,4	2.040	17,58	40,3	1,4	1,36	32,4	81	2h run

Particularly impressive

All test series were carried out with the fan pushing air towards the cooler. However, the front side of the PCB is also cooled in the OC'axe, which means that cold air is transported from below over the housing so that the air can flow out above the PCB at the top. By changing the pull direction, the ASIC temperature could be reduced by a further 3.5°C when overclocking to 1,000 Mhz. Thus, the cooling performance is significantly increased by the Noctua shroud and the pull direction, with the result that the fan can work much quieter or the Bitaxe Gamma or BM1370 ASIC can be cooled even better on warm days or in warm environments.