A DIY splitter for the iBurst 2Mbps desktop modem (UTW)

[ginggs]

I designed a Wilkinson Power Divider for a Cell-C Huawei E1820 HSPA+ modem splitter some time ago.

Having recently upgraded to a 2Mbps iBurst modem and a friend finding an old iBurst panel antenna, I decided to design a splitter to allow me to attach the panel to the 2Mbps modem. The circuit is very simple:

Using this handy coaxial power split calculator, and a centre frequency of 1787MHz, I calculated a coaxial length of 28mm.

I had some RG59B/U cable lying around and I used a 1% tolerance 100 Ohm resistor (I measured a couple and found one that was 99.9 Ohm).
If you compare it to my Cell-C splitter, you'll see the length of the 75 Ohm coaxial section of the iBurst splitter is roughly half the length of the one in the Cell-C splitter. This is because iBurst's frequency is roughly double that of Cell-C's 900MHz.

The finished article:

With some heatshrink tubing it looks almost professional!

The foul weather has prevented me from installing the panel antenna on my roof so I haven't really been able to test the splitter yet.
With the panel sitting in the window, on the 2Mbps modem I'm getting the roughly the same signal strength with or without the antenna and splitter.
On a 1Mbps modem, I get about 3dB more with the panel. The expected loss from this splitter is 3dB, so this tells me the splitter is at least working as expected.

 

[h4tch3t]

I'm facing exactly the same issue. Went from 1MB to 2MB, and now I have ping spikes...

I have 100% signal, with a fair amount of packet loss. Ping would bring back one line with awesome ping of below 50ms, and the very next line would bring back over 300ms?!?!

I'm totally confused by this, however suspect that the modem architecture may be the cause. I was thinking the splitter could actually give the benefit of splitting the difference, and perhaps giving me a ping of 90/100ms at a more stable rate...?

Thoughts?

 

[ginggs]

I'm facing exactly the same issue. Went from 1MB to 2MB, and now I have ping spikes...

I have 100% signal, with a fair amount of packet loss. Ping would bring back one line with awesome ping of below 50ms, and the very next line would bring back over 300ms?!?!

I'm totally confused by this, however suspect that the modem architecture may be the cause. I was thinking the splitter could actually give the benefit of splitting the difference, and perhaps giving me a ping of 90/100ms at a more stable rate...?

Thoughts?

A splitter is the recommended way of connecting the 2Mbps modem to an antenna:
http://www.iburst.co.za/neuvo.aspx?link=service_support_faqs

My splitter is just a cheaper alternative to the one manufactured by Poynting:
http://www.poyntingdirect.co.za/pSPLT-A0014/Broadband-2-way-Splitter.aspx

I would definitely recommend getting a panel antenna. I was using an omni previously and found, using the eyeBurst software, with the panel I was able to aim at a different base station that was less congested than the one picked up by the omni. (see here)

 

[madrynn]

Also Experienced spikes while using:
http://th09.deviantart.net/fs71/PRE...itan_mace_morningstar_by_magmaton-d4ovpuv.jpg

My new splitter:
http://media.greenradio.topscms.com/images/93/20/c36805c24e47b534b69ae2a21272.jpg

 

[bdt]

I designed a Wilkinson Power Divider for a Cell-C Huawei E1820 HSPA+ modem splitter some time ago.

Tangential thread necro alert!

So I'm busy with a thing not a million miles off from what the 2MB modem does with its two radios, and needing a single HGA (high gain antenna) to feed the pair of them. Only I'm looking at this circuit up at 802.11a, for a nominal centre frequency of 5500 MHz. So given that this is established practise (and Poynting will gladly hawk you a Wilkinson-based device), it infers that the theory is just fine and I can reasonably hope to have this work up in that band, even though it looks weird as hell with the orange pieces above 9mm long! I can't find a whole lot of info about doing this up at this frequency, and it will take an age and a bunch of wonga to import a unit so, do you think this is viable? (Or, as is largely the case, am I on my own trying sh*t out here?)

 

[AntiGanda]

Does anyone still use iBurst? The pawn shops are full of their gear.

 

[bdt]

Does anyone still use iBurst? The pawn shops are full of their gear.

You'd be surprised, sometimes it's that or nothing ..hell, I've got a site in Craighall Park where, not for lack of trying, ADSL turned out to be so utterly abysmal that iBurst is the better alternative. But that era will end soon enough when Fibrehoods FINALLY get their act together and I can get them from < 1Mb iBurst to ~100Mb fibre.

 

[ginggs]

So I'm busy with a thing not a million miles off from what the 2MB modem does with its two radios, and needing a single HGA (high gain antenna) to feed the pair of them. Only I'm looking at this circuit up at 802.11a, for a nominal centre frequency of 5500 MHz. So given that this is established practise (and Poynting will gladly hawk you a Wilkinson-based device), it infers that the theory is just fine and I can reasonably hope to have this work up in that band, even though it looks weird as hell with the orange pieces above 9mm long! I can't find a whole lot of info about doing this up at this frequency, and it will take an age and a bunch of wonga to import a unit so, do you think this is viable? (Or, as is largely the case, am I on my own trying sh*t out here?)

I think that making the splitter using co-axial cable at 5500MHz, where the length of the quarter-wave transformers are only 9mm, wouldn't work very well.
Cut the cable 1mm too short or too long and the frequency is out by 10%.

Rather look at doing it in microstrip, e.g.

 

[bdt]

I think that making the splitter using co-axial cable at 5500MHz, where the length of the quarter-wave transformers are only 9mm, wouldn't work very well.
Cut the cable 1mm too short or too long and the frequency is out by 10%.

Yea, it utterly and completely failed to work, for the two attempts I made (though the second one was a LOT better looking than the first).

And I have a plan to get a PCB-based one, I suspect along this line (your attachment seems borked...)