-
Teardown of a Nokia Siemens Networks Flexi WCDMA Power amplifier
Posted by madsbarnkob on 01 May, 2017 09:17 -
Nokia Siemens Networks Flexi Multiradio BTS is a GSM/EDGE, WCDMA/(I-)HSPA, and LTE base station for use in mobile telecommunication antenna networks. A network that you use daily on your cellphone.
This is the 4th part of a video series on this system, all high resolution pictures at the website http://kaizerpowerelectronics.dk/teardown/nokia-siemens-networks-flexi-wcdma-teardown-power-amplifier-part-4/
Part 1: System station http://kaizerpowerelectronics.dk/teardown/nokia-siemens-networks-flexi-wcdma-teardown-system-station-part-1-of-3/
Part 2: Power Amplifier http://kaizerpowerelectronics.dk/teardown/nokia-siemens-networks-flexi-wcdma-teardown-power-amplifier-part-2-of-3/
Part 3: Antenna http://kaizerpowerelectronics.dk/teardown/nokia-siemens-networks-flexi-wcdma-teardown-antenna-part-3-of-3/
The following 25 minute video shows the part 4 teardown step by step with explanations, high resolution pictures of content can be found on the links above.
-
Nice teardown!
It's been a while since I left Nokia Siemens. The hardware was always impressive there. I see you have relatively new WCDMA system module, this one AFAIR can support some LTE functionality. The previous generations used much more ASICs and older PowerPC.
One might think that this is specialist equipment, hence the prices and the excellent build quality but in reality base stations, especially at the time of 3G to 4G upgrades were manufactured in quite large quantities, not unlike consumer products.
Also, surprisingly to me at a time, BTS is not mission critical, no massive redundancy, failover etc, like on a project I worked before the LTE base stations. The reason is simple - if BTS fails then the financial losses are acceptable for the operator, in the cities sometimes nearby BTSes can take over (depends on network planning/capacity). However systems that are above the BTS in the 3G/4G networks, core elements like MSC/RNC etc are designed with higher reliability (as the failure can affect millions of customers).
Since you have all the modules, the system module, the RF outputs and the antennas you'll just need 48V 100A PSU to power the whole thing and try running your own femtocell
-
Thank you, took quite some time to research all that stuff
It was a fairly new system, it could run 4G, but then you had to combine all the amplifiers, so it was thrown out for a upgrade.
I did also have 4 power supplies, but I sold those to radio amateurs, they were all in good working condition.
Do you happen to know anything about the design of the power amplifier itself? A user on youtube suggested it is a Doherty amplifier, but I think its a little more complex than that.
The NXP BLD6G22L is integrated doherty amplifier. But as two of these are used, is this really a doherty amplifier inside a doherty amplifier?
It is clear that there are identical mirrored tracks for each transistor, for its carrier and peak amplifier input, but there is a difference from the upper transistor to the lower transistor. The combined output also clearly shows the quarter wave transmission line to phase match the outputs again. So if they are just used as carrier in the upper and peak in the lower, or each of them is carrier and peak amplifier.
My best guess is that the upper is the carrier, due to the differences in the bias circuitry connected to the transistors.?
-
Do you happen to know anything about the design of the power amplifier itself? A user on youtube suggested it is a Doherty amplifier, but I think its a little more complex than that.
Sorry - I worked on the system module and on the software side. I have some friends that worked in the RF module division, but also on the DSP software and they are bounded by NDAs.
For me this is all evil RF voodoo
-
Since you have all the modules, the system module, the RF outputs and the antennas you'll just need 48V 100A PSU to power the whole thing and try running your own femtocell
4800W input, a femtocell
Nice teardown to the OP BTW, the quality of the equipment and the effort that goes into designing it always brings me down to earth and shatters any delusions I may have had of being able to design 'stuff' -
Quote
A user on youtube suggested it is a Doherty amplifier, but I think its a little more complex than that.
That's me!
I'm not an RF guru, so I may be completely wrong :-) But it does seem like a Doherty inside of a Doherty
Also, I can't find the specific datasheet for the BLD6G22L-150BN. The datasheet for BLD6G22L seems to be for a different device than seen on the board. -
Wow that's beautiful black magic
Always impressed by these cavity filters and mechanical tuning still needed for RF in this day and age where most stuff has become lifeless solid state and software. -
I can't find a datasheet for the BLD6G22L-150BN/2, but the BLD6G22L-50 datasheet is easy to find. My guess is the -150BN/2 is basically two -50 parts integrated into a single package, for better thermal matching.
The BLD6G22L is a Dougherty amp, specially for the 2110MHz to 2170MHz WCDMA and LTE band. The main signal path is easy to see in the picture, but the "peak" path isn't. The main path is simply through a driver stage at the bottom, then the power is split between four parallel paths in the main power amp stage. Finally, the four power outputs are combined. -
I can't find a datasheet for the BLD6G22L-150BN/2, but the BLD6G22L-50 datasheet is easy to find. My guess is the -150BN/2 is basically two -50 parts integrated into a single package, for better thermal matching.
The BLD6G22L is a Dougherty amp, specially for the 2110MHz to 2170MHz WCDMA and LTE band. The main signal path is easy to see in the picture, but the "peak" path isn't. The main path is simply through a driver stage at the bottom, then the power is split between four parallel paths in the main power amp stage. Finally, the four power outputs are combined.
Last night I started filming for a look inside the doherty amplifier ICs, that will be the next video going online
I think each of the integrated doherty amplifiers are used for carrier in upper and peak in the lower. Look at the track width entering the upper transistor, its wider than the lower, so there is a difference there. The combining outputs also suggest that the lower is the peak amplifier as it has a quarterwave length filter from the longer tracks. -
That argument sounds mostly reasonable, especially when you consider the high PAPR of LTE, which doesn't favour using a small secondary channel. However, the BLC6G22L-40 at the bottom appears to be the driver for the final PA, and there is only one of them. What provides a similar amount of drive for the other half of the Dougherty amp in the upper half of the picture?I can't find a datasheet for the BLD6G22L-150BN/2, but the BLD6G22L-50 datasheet is easy to find. My guess is the -150BN/2 is basically two -50 parts integrated into a single package, for better thermal matching.
The BLD6G22L is a Dougherty amp, specially for the 2110MHz to 2170MHz WCDMA and LTE band. The main signal path is easy to see in the picture, but the "peak" path isn't. The main path is simply through a driver stage at the bottom, then the power is split between four parallel paths in the main power amp stage. Finally, the four power outputs are combined.
Last night I started filming for a look inside the doherty amplifier ICs, that will be the next video going online
I think each of the integrated doherty amplifiers are used for carrier in upper and peak in the lower. Look at the track width entering the upper transistor, its wider than the lower, so there is a difference there. The combining outputs also suggest that the lower is the peak amplifier as it has a quarterwave length filter from the longer tracks. -
It drives both of them, through the Anaren Xinger hybrid coupler, phase shifting the signal for the peak amplifier at the bottom and the original signal goes to the upper carrier amplifier.
-
I've seen these boxes as I work in a CO, but never got to see inside one, quite impressive stuff. Crazy the RF engineering that goes into this not just the electrical but even physical stuff too. Imagine the precise calculations and machining required to get all the cavities right etc.
We used to monitor CDMA for our region, so I've "logged in" to these boxes for various reasons such as cycling the radio sectors etc. There are usually 3 antennas (sectors) per frequency per tower to cover 3 different directions. For HSPA the radios (the part being teared down) are actually up in the tower as the HSPA signals are greatly affected by coax length. Kinda a pain when you have to send a guy up a 500ft+ tower to reseat a card or do a hard reset.
-
I did a further part on the integrated Doherty amplifier IC itself. Teardown of a NXP BLD6G22L-150 integrated Doherty amplifier from a Nokia Siemens Networks Flexi WCDMA base station.
-
I can't find a datasheet for the BLD6G22L-150BN/2, but the BLD6G22L-50 datasheet is easy to find. My guess is the -150BN/2 is basically two -50 parts integrated into a single package, for better thermal matching.
The BLD6G22L is a Dougherty amp, specially for the 2110MHz to 2170MHz WCDMA and LTE band. The main signal path is easy to see in the picture, but the "peak" path isn't. The main path is simply through a driver stage at the bottom, then the power is split between four parallel paths in the main power amp stage. Finally, the four power outputs are combined.
NXP** device naming conventions are
BL is for silicon, GaN devices start CL
D is for Doherty
6G is the 6th generation of silicon. Higher numbers indicate later generations of doping profiles and other stuff done by people with PhDs in physics.
22 is for frequency, indicating it's a design for 2.2GHz
-XX is peak output power, so -50 is 50W, -150 is 150W
There then follows some letters which always seems a bit random
** now Ampleon, as NXP had to sell its RF power division off when it took over Freescale to avoid competition issues.
-
I also see a Doherty construction,
just to point out some cool parts for those that does not immediately recognize them,
1 - First hybrid with a 0º and a 90º phase shifted output, input is in the lower corner and in the left corner there is a load resistor for the hybrid.
2 - a cute directional coupler probably for monitoring RF output level and modulation.
3 - a isolator, "RF roundabout" the dump resistor for reflected power is visible just below port 3
-
Also note the 1/4 wave stubs used to supply the DC to the transistors, the location of that cap to ground really matters...
Regards, Dan. -
Agree, also the clearence, via-stitching and trackwidth
proper RF constructions are a work of art.
..but.. there is too much solder and residue left on the transistors
-
Great with even more details pointed out, so the quarter wave length trace from decoupling capacitor to the transistor, is this to prevent noise to ground?
Agree, also the clearence, via-stitching and trackwidth
proper RF constructions are a work of art.
..but.. there is too much solder and residue left on the transistors
The huge amount of solder and flux has been mentioned many times by others, it is not from me trying to desolder the transistors and the unit had not been opened up before I got it in my hands. At least the warranty void stickers was replaced then. -
Great with even more details pointed out, so the quarter wave length trace from decoupling capacitor to the transistor, is this to prevent noise to ground?
A quarter wavelength line transforms the RF short from the capacitor to an RF open circuit at the other end of the line, hence the transistor doesn't see any RF affect from the bias lines.