The FARLINK FRL-W6A is an AP and supports mesh networking. The FRL-W6A has a 2.5GbE LAN port and uses Quantenna for the wireless chip, which is the first Quantenna wireless product I’ve come across.
About its wireless parameters and hardware configuration, in the disassembly inside and then detailed, a little complicated.
The appearance of the FARLINK FRL-W6A is as follows.
Compare the size with Xiaomi 10 phones.
There are only two network ports, both of which are LAN ports. One support port supports 1000Mbps and the other supports up to 2.5Gbps rate. There is no WAN port, because there is only bridge mode Internet access. This product is positioned to be an AP.
Power output specifications: 12V/2A. Standby power has 10W, no USB port, the power supply is considered adequate, if the USB port, at least 3A to be sufficient.
There is a physical button on the top of the top body, presumably the mesh networking WPS key.
Such a look is not out of place in the living room.
底The label information on the bottom is shown below. WiFI 6 SMart Mesh AP, model number is: FRL-W6A.
Second, disassemble the machine
Remove the bottom case.
Remove the top case and see the built-in PCB antenna, you can see that there are a total of five antennas.
After removing the power button, the motherboard can be pulled out.
There are plastic support frames on both sides. The main role I guess is to arrange the 5 antennas. Put it upside down, a bit of technology!
The antenna is defined as follows.
As you can see from the antenna, 2.4G is 2×2 MIMO and 5G is 5×5 MIMO.
Preparing to remove the plastic structural parts.
The other side of the motherboard does not seem to have the main chip, the last to remove this side of it, I call it the back of the motherboard:.
Disassembly from the front of the motherboard.
There are two chips on the half of the motherboard without a heatsink, and you can see that one is a power management chip and the other is a gigabit switch chip, as follows.
Quantenna QP8853 power management IC; detailed parameters are not available.
RTL8211FG is a single-port gigabit switch chip connected to a network transformer that constitutes a LAN2 gigabit network port.
Several 16V/470μF and 10V/470μF capacitors.
Remove the heatsink on the front of the motherboard, several main chips do not use thermal conductive silicone, but directly stacked with thermal grease, from the thermal conductivity index, thermal grease will be higher than the thermal conductive silicone gasket. It is also applied in other models. For me, it is easy to get dirty.
Collect the silicone grease and remove the metal shield.
A PHY chip near the network port, model AQR112R, with a maximum rate of 2.5Gbps for a single network port. 2.5GBASE-T/1000BASE-T/100BASE-TX Ethernet PHYs.
PI6C557-03 is a PCIe 2.0 clock generator, which is related to the operation of PCIe interface.
W25Q256JWEQ is a flash memory chip with 16MB capacity.
In the picture below, the memory chip model is M15T2G16128A, with 256MB capacity.
The model number of the main control chip is QT5GA-AX. (The “ON” logo on the chip represents ON Semiconductor ON Semiconductor, which acquired Quantenna)
The parameters of QT5GA-AX are as follows.
Dual-Band Dual-Concurrent Modem- ViSion Cloud ManagedUp to 5×5 5 GHz + 2×2 2.4 GHz
5 GHz Modem- Industry’s Highest Performance 802.11ax/ac/n/a/g/b WiFi 6 DBDC Baseband Chip- Up to 5 Streams in 5×5 MIMO Configuration
2.4 GHz Modem- Support for 802.11ax/n/g/b- Up to 2 Streams in 2×2 MIMO Configuration
Embedded Processors- DSP Processors- Support for Minimum Burden on the Host CPU
Hardware Datapath- Layer 2/3 Switching Supported,Including Full HW VLAN Support- Layer 3 Stream Identification
This is considered a DSP processor, integrated CPU, no CPU frequency parameters, and did not find other CPU-like chips. Only this explanation, DSP is a data signal processor, such as MT7911AN, MT7975AN these are also part of the DSP.
2.4G supports 2x2mimo, 5G supports 5x5mimo, my first contact with 5x5mimo.
QT5GA-AX will be the first chip in the industry to support the 5×5 with 5 streams and MU-MIMO.
If it is really 5x5mimo, then MU-MIMO will be better than 4x4mimio. How to say? Feel it!
5×5 and mu-mimo I will make a little more detailed test to see what!
The following chart has more information, the blue box is the 5G FEM chip, the model number is KCT8546Q, the yellow box is the 2.4G FEM chip, the model number is KCT8246Q, both are the products of Comhy Communications Technology Inc.
Five 5G FEM chips lead from the QT7510X chip and two 2.4G FEM chips lead from the QT6220X chip. There are two 5G and two 2.4G shared dual-band antennas.
The QT7510X is a 5G RF transceiver with the following parameters.
5Tx/5Rx Chains to Support True 5×5 MIMO System
5 GHz Operation Mode
Supports 20/40/80/160 MHz Bandwidth in 5 GHz System
It supports 5T5R, real 5x5mimo, with 160MHz support, but this hardware only supports 160MHz with Wi-Fi 5, and only 80MHz with Wi-Fi 6.
I don’t have a 5x5mimo NIC, so right now I don’t know what its maximum connection rate is!!!
QT6220X is a 2.4G RF transceiver with the following parameters.
2Tx/2Rx Chains to Support 2×2 MIMO System
2.4 GHz Operation Mode
Supports 20/40 MHz BW in 2.4 GHz System
IEEE 802.11ax/n/g/b Compliant
2.4G at 40MHz bandwidth, WIFI6 at a maximum rate of 573Mbps.
The relationship between QT5GA-AX and QT7510X and QT6220X is as follows.
These are the descriptions of the information found on the Internet.
5×5 5GHz 速度快 25% 到 50%
5×5 在 5GHz 频段更广的覆盖范围
2. 比 4×4 方案更好的 MU-MIMO
5×5 提供更好的MU-MIMO (1×1, 2×2 客户端)
3. Quantenna的 SmartScan
Point 3 of which talks about SmartScan DFS, using Background DFS, two effects can be achieved.
One is that the 5G signal will appear soon after power on, and will not appear until after scanning DFS like most Wi-Fi 6 routers, which is the reason why it comes out later than the 2.4G signal. This Background DFS can be used to make the fifth way to do DFS scanning, the other 4x4mimo can be first out for the receiving device to connect to use;.
Second, it can always be scanning DFS.
Reasoning, if not doing DFS when it becomes a real 5x5mimo can be used to transfer data? If always scanning DFS, it becomes 4x4mimo?
Add the information of the two FEM chips.
The parameters of KCT8546Q are the following screenshots: (general power)
The parameters of KCT8246Q are as follows: (Stronger power)
Finally removed the heatsink on the back of the motherboard, no important chips were seen. Below the heatsink there are a couple of thermally conductive silicone spacers, responsible for the thermal conductivity of the two RF chips and the seven FEM chips.
The disassembly is now complete.
Fourth, the highest speed test of 5G in close proximity
Here the 5G wireless channel is set to 44, with a bandwidth of 160MHz.
Combined with the 2.5GbE LAN port, you can fully test the wireless speed of 5G. Currently there is no high-performance WiFi 6 wireless card on the market, so the only way to play the role of “wireless card” is through the wireless router wireless bridge.
Here take the Asus AX86U to do the test, set the AX86U wireless relay, the object is the FRL-W6A 5G WiFi, wireless relay success, in the Asus routing interface can view the 5G wireless connection rate is how much, after slowly observe the highest value is as follows: 3446.7Mbps
3446.7 is exactly the wireless negotiation rate in 4x4mimo 256-QAM, 160MHz, 802.11AC mode, why not 4804Mbps because it FRL-W6A’s 160MHz bandwidth is not supported in WiFi 6 mode.
If you use two FRL-W6A wireless network, what will be the 5G wireless rate between them? The highest rate was observed with multiple refreshes: 3453Mbps. as follows.
Two simultaneous wireless relay rates are shown below.
It seems that the wireless bridging rates of the two are very similar. Immediately use IxChariot speed test to see, a number of consecutive tests, I did not count how many times, and finally take out the highest value, the results are as follows table.
It appears that the AX86U’s role as a “wireless card” is not as fast as the FRL-W6A, most likely due to the directionality of the AX86U’s fourth built-in antenna, as this test was conducted in close proximity.
From the wireless rate data and the 5G transmission speed measured here, the QT7510X’s 5G is 4T4R, not 5T5R as I thought.
The screenshots of the FRL-W6A test are as follows.
V. MU-MIMO test
In this test project, the 5G bandwidth is set at 80MHz, and the computer is connected to the 2.5GbE port on the FRL-W6A.
In the advanced settings page, there is a MU-MIMO switch, here the MU-MIMO on and off respectively to measure the speed to see if there is any effect. So let’s start with the wifi5 phone test.
I have in hand to support MU-MIMO cell phone test Xiaomi 5, LeTV X910, Xiaomi 4C, Xiaomi note 3.
The above single device connected to the FRL-W6A wireless, respectively, measured the speed of the following table.
Then two of these devices were selected to do MU-MIMO testing, and the results were tabulated as follows.
How strange that two 1x1mimo phones do not have the MU-MIMO effect.
while two 2x2mimo have a clear MU-MIMO effect.
If you use three 2x2mimo, if you do not see the data after turning off MU-MIMO, you really think it does not support, I used to think so. This is a 1.5 times gain before and after a comparison.
Attached are screenshots of the test with two Xiaomi 4C phones.
The next MU-MIMO test was performed with wifi6 terminals, and the MU-MIMO results tested after each two combinations were as follows.
Curious ah, two QCA6391 together instead of support, repeatedly measured several times are more than 600Mbps, but the same use of Qualcomm chips two Xiaomi phones have the effect, but the effect is only from the 802.11ac mode MU-MIMO gain.
The best performance is the Intel AX200 wireless card, I wonder if Quantenna and Qualcomm’s wireless chips have compatibility issues? So that the wireless speed is not fully developed!
From the MU-MIMO actual test data, and did not see 5x5mimo can make MU-MIMO (WiFi 5 mode) to bring a better feeling than others, compared to the Netgear R9000 is obviously worse. The following chart: two Xiaomi 4C can reach sudden gigabit speed after MU-MIMO is in effect.
Six, various receiving terminals in close proximity to the wireless speed measurement
Here with a conventional cell phone terminal to test, plus a 4x4mimo 802.11ac wireless network card (ASUS PCE-AC88), here still use the IxChariot tool to hit the stream speed measurement, ten processes, each device test multiple times, constantly adjust the position to achieve the best wireless speed, and the wireless connection rate of the terminal recorded in the table below, aspects of the control: ( Mbps)
It is a bit strange, strange is obviously set 2.4G 40MHz bandwidth, but all terminals are only 20MHz. plus 2.4G is only 2x2mimo, so the terminal device 2.4G rate in the above figure are relatively low.
Xiaomi 10 cell phones, which themselves only support 80MHz, with 5G downlink speeds of less than 800Mbps.
Xiaomi 11 cell phone, support 160MHz, in the cell phone WiFi connection interface can be seen in the rate of “2401Mbps” (the exact value of 2401.96Mbps, if the trailing number is different, because some of the zero rounding, some rounding). Under the 2.5GbE port, the 5G downlink speed reaches 1565Mbps, while the uplink is significantly worse, even if you adjust multiple positions, you can not see a better performance of the uplink speed.
Apple SE2, supporting 80MHz, 5G downlink speed is the same as Xiaomi 10, while uplink speed will be worse than Xiaomi 10.
Intel AX200 wireless card, supports 160MHz, in the wireless adapter interface shows a connection rate of 1201Mbps, but from the actual measured speed, it negotiated more than 1201Mbps, 5G downlink speed reached 1205Mbps. i think the transmission is maximized with the rate, using the WiFi 5 mode, so the actual wireless rate is The actual wireless rate is 1722Mbps, because AX200 also supports 160MHz in WiFi 5 mode, and I see the following connection rate in the router interface.
Why Xiaomi 11 can support 160MHz in WiFi 6 mode, but AX200 can’t?
Asus PCE-AC88 wireless card itself does not support 160MHz, in 256-QAM modulation mode, the maximum rate of 1733Mbps, if support 1024-QAM, the rate will be 2166Mbps. measured 5G downlink speed of 1123Mbps, not bad. If this card can support 160MHz, the wireless rate doubled is 2246Mbps, but this is only a theoretical value, the higher the bandwidth the higher the loss, the surface of the two FRL-W6A mutual transmission speed is 2100Mbps may have been very good.
VII. Wireless speed test of each terminal in different positions
Put the FRL-W6A in the “WIFI” position in the figure below, and put each terminal in four positions A, B, C and D for speed measurement.
The speedtest server is built to measure the speed, the result will be a little higher than the actual transmission speed. Each location is measured 6 times continuously, and the average value is finally taken as the parameter value. The tested terminals are Xiaomi 10, Xiaomi 11, Apple SE2, AX200 wireless network card. The network topology diagram is shown below.
1、Speed test with AX200 wireless network card
All the results are shown in the table below.
AX200 did not measure 2.4G, only 5G, AX200 support 160MHz, so there is an absolute advantage in close range, of course, provided that the wireless AP needs to support 160MHz and have 2.5GbE network port to work.
The 5G average is shown in the following chart.
At the location of point A, which is separated by a wall, a downlink speed of over 1600Mbps was measured . But as the distance increases, the speed slowly decays. A normal phenomenon!
2. Speed measurement with Xiaomi 11 cell phone
All the results are in the following table.
Xiaomi 11 phone also supports 160MHz, but it can’t break through gigabit like AX200 at the point A position which is separated by a wall.
The 5G speed averages are as follows.
3. Speed measurement with Xiaomi 10 cell phone
All the results are shown in the table below.
5G speed average.
Although Xiaomi 10 only supports 80MHz, the 5G speed at a distant location is faster than the 160MHz bandwidth of Xiaomi 11, which means that the WiFi reception of Xiaomi 11 is significantly weaker than that of Xiaomi 10.
4. Speed measurement with Apple SE2 phone
All results are tabulated below.
The 5G speed averages are as follows.
Apple SE2 is faster than two Xiaomi at a distance. Maybe this Quantenna chip is more compatible with Broadcom chip!
From the dismantling of the machine can be seen in several pieces of metal screen cover, a large radiator, give power supply design, workmanship and materials are very good. With a 2.5GbE is the biggest highlight, its role is designed to be a wireless AP, so it is ideal for accessing networks that already have a primary route. The downside is that the signal is weak and the speed is attenuated over long distances, but if you use more than one to network, this disadvantage can be ignored. There is also a point of compatibility with Qualcomm is not as good as Broadcom, I hope to improve in the future.