Ericsson Tower Mounted Amplifer Teardown and measurement

[Da_Stier]

I recently came accross a TMA or Tower Mounted Amplifier on ebay. It was sold as non working and since the price was okay, I decided to buy it and take a look at it, since I never had a TMA to play with.

First of all, what exactly is a TMA. A TMA is an amplifier used in cellular communication. Other than a power amplifier it is used to give more sensitivity to a basestation in the RX path. It adds a low noise amplifier as well as an additional filter to the path between the antenna and the basestation. It is mounted directly on the antenna tower (therefore the name TMA). A TMA is typically used if you have a very noisy environment, for example with a lot of other signals that are close in frequency or if the distance to the mobile devices is very high, for example if the celltower is mounted on a mountain outside a city. In this scenario you might have a few kilometers before there  are any mobile devices in reach.

The particular amplifier I bought is an Ericsson KRY 112-14/14. It was manufactured in 2000 and is made for GSM. The RX uplink frequency is 890-915MHz. This means that it was built for either P-GSM or E-GSM applications. I found this product overview on the web:

KRY-112-1414.pdf

The gain of the unit is meant to be around 12dB.
This gain can be set anywhere between 4dB and 14dB as I will find out later.
This is the unit as received:







After removing a few screws, the bottom mounting plate can be removed. Under this plate is another plate that also comes off, reveiling the inside of the unit:



The connector leading to the basestation is also used to feed the TMA with DC. The port has an integrated Bias-T, consisting of a piece of coax, which is propably used as an impedance trasformer, however it doesn't seem to be a lambda/4 stub. After this follows a LC filter.
After looking at this section, I found the first problem, the TVS diode on the input was shorted and even had a small hole burnt in it.
Maybe the unit was hit by lightning at some point in its life?
My quick and dirty fix for this was to simply remove the diode from the unit.





Now for the main PCB. The LNA itself consists of two MMIC amplifiers, configured in a balanced topology by using Xinger hybrid couplers, just like in most poweramplifiers.
The unit also has two high power RF relays, to bypass the LNA completely if the unit is not supplied with DC, which I found rather interesting.
The power supply consists of a LM2941S LDO, which is expected, since you don't want a switching power supply in a LNA. The LDO is configured to deliver 10V.



After taking a good look at all the components, I connected the unit to some power. Since I don't have a Bias-T with 7/16 DIN connectors, I just soldered on some wires on the PCB to supply DC.
The unit takes around 300mA at 12V, which sounded about right.
Next I connected my VNA to the unit.



I took S21 measrements of the TMA, both in RX and TX direction and with and without DC supplied, these were the results:







As can be seen in the plots, the unit has a really nice filter in both directions. There is also really nice seperation between the RX and TX band. (The dip in the middle)
This is achieved by a cavity filter, similar to the Duplexer units found on power amplifiers. Other than a Duplexer, this filter acutally has its RX and TX path combined into a single port through the rest of the TMA.

After confirming the function of the unit I went on to take a look at the cavity filter.
After removing around 2 milliion screws and some bleeding fingers, I finally had the lid removed:









I hope some of you find this look at a TMA of interest.


Here are some more datasheets of some of the key components:

AA210-25(LF).pdf
lm2941.pdf
MAX480.pdf
um1.pdf

[Da_Stier]

I just wanted to give a quick follow up, incase anybody is wondering what I did with the unit. 

Since I don't have a use for a TMA (obviously), I tore it down.
I salvaged around 200 stainless steel screws, the two 7/16DIN connectors and the resonators from the filter unit.
(They are made of silver plated brass and pretty nice to machine into spacers and such)

For the PCB I hacked it into pieces and converted the main part into a general purpose LNA for lab use.









So after all I still managed to get a few usable bits and pieces from the unit.