I've been going over Sylvain's report on removing the RX filters, http://www.246tnt.com/gsm/rx_filter.html, and I had some slight differences on the c139 that I'm hoping someone can help me with.
Similar to Sylvain's experience, the schematic doesn't match reality.
Just as Sylvain describes, the signal from the balanced output appears to be going through inductors rather than capacitors and there isn't a bridging inductor.
However on my c139, the unbalanced input is missing the inductor to ground on the EGSM path and is missing the capacitor to ground on the DCS path. (Sylvain had a capacitor to ground on DCS where the schematic has an inductor to ground.)
You can find a picture of what I'm looking at here: http://thre.at/c139/c139rxfilters.jpg . I've placed the c139 schematics in that directory as well, http://thre.at/c139 .
My guess is that the filters in the schematics had a different unbalanced impedance than the ones they ended up placing on the c139. But I'm not an EE, and that is just a guess.
I can't read the markings on the filters I have, so I can't check the unbalanced input from a data sheet. I also don't have a way to measure the components that are there now so I can't try and compute the value.
My question is, does this matter? That is, should I use the same balun that Sylvain chose or should I find one with a different unbalanced impedance? Alternatively, should I use the same baluns and just install an inductor (or capacitor on DCS) to ground so that it matches Sylvain's c123?
Any ideas?
(As an aside, the baluns that Sylvain chose aren't perfect for the US frequency range, but they should work okay.)
Hi,
My guess is that the filters in the schematics had a different unbalanced impedance than the ones they ended up placing on the c139. But I'm not an EE, and that is just a guess.
Mmm, I doubt that. The filters probably had 50 ohm impedance as spec. But the circuit before that / the antenna maybe was a bit off and they tuned it, or it might just be cost optimized, or they just tested and it worked better like that.
But anyway, it's probably tuned for 50 ohm unbalanced, so you can use it as is.
My question is, does this matter?
Truth is ... not that much. If you look at the end of the page, you can see I tried the most botched up job _ever_ and it _still_ worked pretty good !
That is, should I use the same balun that Sylvain chose or should I find one with a different unbalanced impedance? Alternatively, should I use the same baluns and just install an inductor (or capacitor on DCS) to ground so that it matches Sylvain's
Just use the same, it'll work fine.
(As an aside, the baluns that Sylvain chose aren't perfect for the US frequency range, but they should work okay.)
The hi band one if the one recommended for both PCS and DCS. For the low band it's indeed E-GSM but the graphs are virtually flat, insertion loss varies by less than 0.5 dB. Return losses are a bit higher but I doubt that it matters much.
So yeah, not perfect but certainly not gonna be noticeable.
Cheers,
Sylvain
Quoting 246tnt@gmail.com (246tnt@gmail.com): [...]
(As an aside, the baluns that Sylvain chose aren't perfect for the US frequency range, but they should work okay.)
The hi band one if the one recommended for both PCS and DCS. For the low band it's indeed E-GSM but the graphs are virtually flat, insertion loss varies by less than 0.5 dB. Return losses are a bit higher but I doubt that it matters much.
Yes, not "baluns" but "balun". The HHM1523C1 E-GSM side balun is slightly better in the 900 band. It looks like the HHM1523B3 is exactly the same balun with the same pin-out except slightly favoring the GSM-850 range.
You can find a picture of what I'm looking at here: http://thre.at/c139/c139rxfilters.jpg . I've placed the c139 schematics in that directory as well, http://thre.at/c139 .
As a side note, you should use the quadband branch (soon to be merged) and replace (in board/compal/rffe_dualband.c)
uint32_t rffe_get_rx_ports(void) { return (1 << PORT_LO) | (1 << PORT_DCS1800); }
by
uint32_t rffe_get_rx_ports(void) { return (1 << PORT_LO) | (1 << PORT_PCS1900); }
because the hi band input is plugged to the PCS port of rita and not DCS port.
Cheers,
Sylvain
Quoting 246tnt@gmail.com (246tnt@gmail.com):
You can find a picture of what I'm looking at here: http://thre.at/c139/c139rxfilters.jpg . I've placed the c139 schematics in that directory as well, http://thre.at/c139 .
As a side note, you should use the quadband branch (soon to be merged) and replace (in board/compal/rffe_dualband.c)
I've been using a merge of your quadband and your burst_ind branches.
uint32_t rffe_get_rx_ports(void) { return (1 << PORT_LO) | (1 << PORT_DCS1800); }
by
uint32_t rffe_get_rx_ports(void) { return (1 << PORT_LO) | (1 << PORT_PCS1900); }
because the hi band input is plugged to the PCS port of rita and not DCS port.
Ah, thanks for the tip. My local tower only has GSM850 channels so I haven't noticed a problem. I'll make that change though on my local branch.
Hi,
On 26.05.2011 22:06, Joshua Lackey wrote:
I've been going over Sylvain's report on removing the RX filters, http://www.246tnt.com/gsm/rx_filter.html, and I had some slight differences on the c139 that I'm hoping someone can help me with.
this site is currently unavailable.
Does anyone has a mirror?
Thanks, Philipp
baseband-devel@lists.osmocom.org
-
246tnt＠gmail.com -
Gerard -
Joshua Lackey -
Philipp