E1820 External Antenna Connector

[GRES]

Similar to what this guy did:
Solder a new antenna to Huawei E1820 E182e

Yes. Idea is from there. It is betteer to bring cable from the side - no need to damage original modem antenna.

 

[ginggs]

I have always use my E1820 with an inductive coupler to an external antenna. But I have a second external antenna, so with the inductive coupler still in place, I connected my second antenna to the external antenna connector. Now with two high gain external antennas connected to my E1820, I seem to be seeing 25% faster downloads.

I'm seeing a similar improvement using the inductive coupler and the pigtail connector together.

RSSI: -69dBm (Poynting LPDA0044 / ADPT-026)


RSSI: -71dBm (Poynting LPDA0044 / ADPT-026 / CRC-9 Pigtail)

 

[Glipsie]

Hi All

I have both modems once again (also have the 8ta sim). Here are some tests I just did:

E367

Pigtail -53dBm 19175


Inductive
-79dBm 19175



Bare -87dBm 19175



E1820

Pigtail -97dBm 19175



Inductive -81dBm 19175



Bare -93dBm 19175


I forced both modems to 2100mhz as I believe 900mhz is either broken or congested.
The E1820 inductive speed test isn't really accurate. It was hovering at 7Mbps the whole way and then suddenly spiked to 14 at literally the last second. All these tests are done on the Cell C test server.

I will run the same tests tonight forcing the modems 900mhz.

Let me know if you would like me to test something else.

/Glipsie

 

[sajunky]

Thanks Glipsie.These are impresive results with signal around -90dBm for both modems. I wonder if someone can repeat ginggs dual cell operation tests with 8-ta E367. Here I am more interested with 900MHz results, as it is more common for Cell-C network.

 

[Glipsie]

Here we go, 900Mhz


E367
19172 >= 51dBm Pigtail


19172 -61dBm Inductive


19172 -87dBm Bare



E1820
19172 -83dBm Pigtail


19172 -67dBm Inductive



19172 -83dBm Bare



Conclusions: The inductive coupler works best on both modems right at the tip (i.e. furthest from the USB connector). It seems best when the PCB is on the top (i.e. the side with Cell C/8ta written on it). This is not a lot on the 8ta modem but is much as 10db on the E1820.

The E1820 tests are NOT smooth. They graph is very jagged (1Mb, 7Mb, 2Mb, 14Mb, 0.5Mb, 18Mb etc), whereas the 8ta modem is smooth (basically the same speed from start to finish).

My E1820 is now almost a year old (bought November 2010) whereas the 8ta modem is in its first month.

Speed wise: I don't have a clue, though I believe 19172 is oversubscribed (hence the big discrepancies in speed tests). I'm personally using the E367 with the pigtail on 2100mhz (see my previous post). It provides the lowest latency and generally past 11pm I can get 21Mb. I was having problems before with the E1820 and the inductive coupler (< 1Mb) but I'm guessing it was a problem with the tower.

Anyone want me to test in some other way (apart from chopping up my connectors like ginggs did )

/Glipsie

 

[Glipsie]

Here are the ping tests. Limited to 8 images per post
E367

E1820


*Edit*
I did the E1820 test first, then switched modems and could not perform a ping test to ANY server (I could ping 8.8.8.8 though). I then noticed that I had a 197 IP address, so I disconnected and reconnected. The ping test then worked first time.

I used the inductive coupler with the E1820 and the pigtail with the E367. Honestly the E1820 + pigtail is a waste of time.

 

[sajunky]

Anyone want me to test in some other way (apart from chopping up my connectors like ginggs did )

/Glipsie

Nobody expect you to follow insanity... Besides, in my opinion ginggs's approach is dangerous, as he is routing half of the power transmitted by the modem back to the diversity receivers. It can potentially damage receivers - immediately or after the prolonged use. Note that receivers of the main channel are always protected by duplexers.

I asked you for additional tests, as these operator customised modems might have limited functionality. Even in many general market dual-band modems receive diversity feature is limited to a single band only. Here the operator can chose on which frequency receive diversity should work.

In other words, it is expected that Cell-C E1820 works faster in 900MHz and 8-ta E1820/E367 is faster in 2100MHz. In this case (if we take out pigtail tests - for the obvious reason), 8-ta E367 is slower in all tests in 900MHz band. It has updated revision of chipset, newer and more stable (MIMO) firmware, but it cannot keep up with its older friend. In 2100MHz band (where receive diversity is expected to work on this device), 8-ta E367 kicks-out competitor by a hudge margin in bare test.

 

[ginggs]

...in my opinion ginggs's approach is dangerous, as he is routing half of the power transmitted by the modem back to the diversity receivers. It can potentially damage receivers - immediately or after the prolonged use.

That is not true. Half of the signal does go back towards the diversity receiver, but because it passes through two 1/4 wave sections, it arrives 180° out of phase with the transmitted signal and is dissipated in the resistor.

 

[Glipsie]

Nobody expect you to follow insanity...

LOL!

I asked you for additional tests, as these operator customised modems might have limited functionality. Even in many general market dual-band modems receive diversity feature is limited to a single band only. Here the operator can chose on which frequency receive diversity should work.

In other words, it is expected that Cell-C E1820 works faster in 900MHz and 8-ta E1820/E367 is faster in 2100MHz. In this case (if we take out pigtail tests - for the obvious reason), 8-ta E367 is slower in all tests in 900MHz band. It has updated revision of chipset, newer and more stable (MIMO) firmware, but it cannot keep up with its older friend. In 2100MHz band (where receive diversity is expected to work on this device), 8-ta E367 kicks-out competitor by a hudge margin in bare test.

Ok, but then I am a pretty poor test case considering the ongoing tower problems (see the relevant thread, I've actually stopped complaining because it seems to fall on deaf ears).
I will however disagree with your assessment. While the E1820 got a very good result, the actual speed was VERY choppy (jumping from 1Mb to 7Mb back down to 1Mb). It was only really in the final seconds of the test that the result popped up. The E367 on the other was a stable speed throughout the test. In my opinion, stability trumps speed.
Then there's the ping test. 20ms is also a rather big deal.

What I should probably do is run 5 speedtests in a row on each modem, pigtail with the E367 and inductive couplar with the E1820. This will give a better overview of the situation. My feeling is that the E1820 is going to have widely ranging values while the E367 will be consistent. However for that to happen, I need to buy some airtime. Its easy to go through 5Gb at these speeds

/Glipsie

 

[sajunky]

I agree that E367 is better modem, better overall speed and stability of the connection. No question about. What I take is your results (including comments) and maintain my assessment regarding crippled receive diversity in these modems (working in one UMTS band only). If you make another tests, forget pigtail tests, this matter is closed. It is like comparing apples to oranges.

 

[UnUnOctium]

That is not true. Half of the signal does go back towards the diversity receiver, but because it passes through two 1/4 wave sections, it arrives 180° out of phase with the transmitted signal and is dissipated in the resistor.

+1. There's a reason why the resistor impedance is 2*Zo and each of the lines are sqrt(2)*Zo by design.

 

[sajunky]

There's a reason why the resistor impedance is 2*Zo and each of the lines are sqrt(2)*Zo by design

Correct, as it is impedance converter, this way all 3 ends of the splitter can see the same 50 Ohm impedance. You are on the rigt track, indeed.

 

[Glipsie]

Ok, more speed tests
Both modems forced to 900Mhz, both connect to Cell ID 19172

E367 with pigtail -53dBm





E1820 with inductive couplar -61dBm





Based on this, I don't think the E367 has been crippled. This is also representative of my general browsing experience and why I was ready to give Cell C the boot.

/Glipsie

 

[sajunky]

@Glipsie
Still comparing pigtail tests?

Based on this, I don't think the E367 has been crippled.

EDIT: OK, keep cool... I can tell you, so you won't be upset. Unless:

1. modem has negotiated dual carrier MIMO R8 42Mbps, (but modem is not capable of) or

2. modem has negotiated R7 28.8Mbps MIMO (which modem is capable, but unlikely network does)

You can say nothing about crippled receive diversity having signal level around 53dBm

 

[sajunky]

Half of power is wrong

When I wrote about "half of the power", I was about my general experience with splitters. The picture from Wikipedia posted by ginggs is very good in detais, but it gives obscured wiew on design principles (see below) and I decided to let others to defend themselves. So, ginggs did.

Now I take a look at Wilkinson design, it is very simple in principle and (indeed) gives separation between P2 and P3. However the picture is not exactly Wilkinson design. One end (P1) of the 70 Ohm waveguide shield is left floating. It gives complication to the design and I am not sure about consequences. When signal is applied to P1 there is no problem as both waveguides are driven with the same magnitude and phase, so there is no transformer effect between waveguides. Such an effect comes up when signal is applied to P2 or P3. In consequence separation between P2 and P3 is not like in the books.

As usual devil is in details, including inductance of 100ohm resistor with its leads. or whether it is right or wrong when talking of grounding the floating end or removing outer PVC insulation of waveguides and wrap-shrinking them together.

Finally, the waveguides should be tuned-up to the uplink channel: 1950MHz and 898MHz respectively (prioritizing protection of receivers). One user mistake of connecting modem not retricted to single frequency band can be costly experience.

 

[ajax]

Finally, the waveguides should be tuned-up to the uplink channel: 1950MHz and 898MHz respectively (prioritizing protection of receivers). One user mistake of connecting modem not retricted to single frequency band can be costly experience.

Isn't the diplexer protection enough? After all, when using just the modem's own internal antenna it transmits and receives with the same antenna. The diplexer with its very sharp band reject filters should protect against receiver damage? What is different then in ginggs's setup?

 

[sajunky]

Duplexer always protects primary receivers. Splitter routes some of transmitted signal to diversity receivers, and those have no duplexers. Yes, in current design there are analog filters in receivers. You are right, filters help protecting receivers of different bands. These should be safe, some other can be damaged. By example splitter is tuned to 900MHz and lets assume it gives sufficient separation in this band. But only in this band, it is how it works. Modem is allowed to transmit 2100MHz. Here damage can be done to 2100MHz diversity receivers.

 

[T-Man]

2. modem has negotiated R7 28.8Mbps MIMO (which modem is capable, but unlikely network does)

My 8ta E367, CellC sim, no external antenna, connects at 28.8 Mbps, at first I thought it was 28.8 Kbps! Haven't done any speed tests yet.

 

[ginggs]

Now I take a look at Wilkinson design, it is very simple in principle and (indeed) gives separation between P2 and P3. However the picture is not exactly Wilkinson design. One end (P1) of the 70 Ohm waveguide shield is left floating. It gives complication to the design and I am not sure about consequences. When signal is applied to P1 there is no problem as both waveguides are driven with the same magnitude and phase, so there is no transformer effect between waveguides. Such an effect comes up when signal is applied to P2 or P3. In consequence separation between P2 and P3 is not like in the books.

I don't believe the one end of the shielding is supposed to be left floating, it is probably an omission by the artist. Similarly, the diagram shows the grounds of the 1/4 wave transformers being connected to P2 and P3, but not to each other. In my splitter, all three shields are tied together on the input (P1) side and all four shields are tied together on the output (P2, P3) side.

 

[sajunky]

In my splitter, all three shields are tied together on the input (P1) side and all four shields are tied together on the output (P2, P3) side.

Right! Not many people think what what they are doing, most just use the existing drawings.