FreeCalypso > hg > freecalypso-docs

comparison Quadband-ideas @ 28:3799892b1a79

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Quadband-ideas article rewritten for the new situation since Tango discovery
author Mychaela Falconia <falcon@freecalypso.org>
date Sat, 25 Jan 2020 22:41:11 +0000
parents 00216b7cfc4d
children 6d7486db31cb
comparison
equal deleted inserted replaced
27:f7ed8c45a8ec 28:3799892b1a79
1 Triband status quo 1 I, Mother Mychaela, hold the belief that any newly designed FreeCalypso hardware
2 ================== 2 made in 2020 or later needs to be fully quadband, supporting all 4 possible GSM
3 frequency bands of 850, 900, 1800 and 1900 MHz, nothing less. Furthermore, we
4 do not need to invent or innovate anything in order to produce a quadband
5 Calypso phone or modem: TI already had a quadband reference design back in the
6 early 2000s, called Leonardo or Leonardo+ (the exact proper designation is
7 unclear), we just need to resurrect it, adding some updates of our own like we
8 always do.
3 9
4 Our current Openmoko-based Calypso+RF modem core is very very good, but it has 10 Our current FCDEV3B modem board is triband rather than quadband because of
5 one shortcoming compared to TI's Leonardo+ reference design: it is triband 11 historical circumstances: while I had *always* desired a quadband solution
6 rather than quadband. This triband restriction stems from OM's use of discrete 12 since the very beginning of FreeCalypso (since I started gathering TI Calypso
7 antenna switch and SAW filter components as opposed to an integrated FEM (front 13 docs from Chinese sites in 2011), back in 2015 we did not have the necessary
8 end module) like on Leonardo+. In addition to the band restriction, our current 14 know-how to confidently resurrect TI's lost quadband reference design by doing
9 triband RF design suffers from one other very unpleasant problem: we have no 15 a new PCB layout on the basis of the available surviving documentation, which
10 datasheet for the antenna switch component which we have to use. We know from 16 consists of just schematics and a floorplan drawing - instead going with
11 Openmoko's BOM data that the manufacturer is Darfon and that the part number for 17 Openmoko's PCB layout and their triband RFFE was the only viable option at that
12 this antenna switch component is ASM4532T0P06-1, we are able to buy this part 18 time. We have now successfully produced both 900 MHz and 850 MHz versions of
13 from our Chinese grey market suppliers, we build our boards with these parts and 19 our FCDEV3B (different SAW filter parts populated on the same PCB footprint)
14 our boards do work perfectly fine when we get a good batch, but we have to do 20 and both work flawlessly as verified with our CMU200 instrument, so we do
15 this entire process blindly, without any datasheet or other documentation for 21 already support all 4 GSM frequency bands in a way - but only 3 at a time with
16 this mystery part. 22 our current hardware.
17 23
18 While we lack any official documentation for our triband ASM, we know its basic 24 A major breakthrough happened in December of 2019: we found a certain extremely
19 logical function: it has two Tx inputs (low band and high band Tx coming from 25 obscure historical commercial Calypso modem module that is almost a verbatim
20 the PA), 3 Rx outputs going to 3 separate SAW filters for the 3 supported bands, 26 clone of the core section of TI's Leonardo+ quadband reference design, and the
21 and 3 switch control inputs. Two of these switch control inputs are Tx controls 27 knowledge gained from examination of this obscure historically-made modem module
22 (low band Tx enable and high band Tx enable) which appear to be the same across 28 (both physical PCB reverse eng and evaluation of its RF performance with our
23 all common RFFEs, whether they are 2-band (single region), triband or quadband. 29 CMU200 instrument) has given us the necessary confidence boost with this
24 The 3rd switch control input gets a logic high voltage applied to it during PCS 30 particular way of implementing a quadband GSM MS.
25 band Rx, thus it appears to be a switch that diverts the high band Rx path
26 between DCS and PCS SAW filters. The same arrangement is found in most other
27 triband phones and modems from that era, i.e., they also have two Tx switches
28 and one Rx path switch selecting between DCS and PCS, plus 3 discrete SAW
29 filter components (outside the ASM) for the 3 bands.
30
31 This article outlines some ideas for how we may be able to move from this RFFE
32 to a different one, replacing our current mystery ASM with something less
33 mysterious and better documented, and improving our radio capability from
34 triband to quadband at the same time.
35 31
36 Epcos M034F 32 Epcos M034F
37 =========== 33 ===========
38 34
39 TI's Leonardo+ and E-Sample boards used a magic component made by Epcos (the 35 TI's Leonardo+ and E-Sample boards used a magic component made by Epcos (the
40 canonical SAW filter manufacturer during that era) called M034 or M034F (the 36 canonical SAW filter manufacturer during that era) that was called M034 or M034F
41 exact proper designation is unclear). It was an integrated quadband FEM, 37 at the time of TI's designs in question. Epcos later gave it a completely
38 different name when they released it into volume production, thus the name that
39 is needed in order to buy this part and that is physically marked on the
40 component package bears to resemblance to M034 - but we will not be able to
41 disclose this other name until we physically produce our first FreeCalypso
42 hardware product with this quadband FEM in it, meaning until and unless someone
43 pays for it.
44
45 The component in question is an integrated quadband FEM (front end module),
42 integrating the antenna switch and SAW filters in one component package, with a 46 integrating the antenna switch and SAW filters in one component package, with a
43 special twist. The special twist is that even though there are 4 separate Rx 47 special twist. The special twist is that even though there are 4 separate Rx
44 band SAW filters inside that M034 "chip" module, corresponding to its advertised 48 band SAW filters inside that M034 "chip" module, corresponding to its advertised
45 quadband capability, only 3 Rx signal path differential pairs come out of it, 49 quadband capability, only 3 Rx signal path differential pairs come out of it,
46 neatly corresponding to the 3 LNA inputs on TI's Rita transceiver. This twist 50 neatly corresponding to the 3 LNA inputs on TI's Rita transceiver. This twist
47 is important because even though the Rita transceiver itself is fully quadband 51 is important because even though the Rita transceiver itself is fully quadband
48 internally, it has only 3 LNA inputs, with GSM850 and EGSM bands sharing the 52 internally, it has only 3 LNA inputs, with GSM850 and EGSM bands sharing the
49 same LNA input while each of DCS and PCS get their own. 53 same LNA input while each of DCS and PCS get their own. Thus this M034 FEM is
54 special in that it facilitates building a quadband GSM MS using transceivers
55 like TI Rita or Silabs Si4200 with only 3 LNA inputs, and the name M034 itself
56 reflects this specialness: the digit 3 in the name refers to the 3 differential
57 Rx signal paths, while the digit 4 refers to the 4 SAW filters inside which
58 provide full quadband capability.
50 59
51 We do have an M034F.pdf datasheet for this magic component (came along with 60 We do have an M034F.pdf datasheet for this magic component (came along with
52 Calypso and Leonardo docs), and the block diagram on page 6 shows the magic 61 Calypso and Leonardo docs), and the block diagram on page 6 shows the magic
53 quite clearly: there is a baseband-controlled switch selecting between EGSM Rx 62 quite clearly: there is a baseband-controlled switch selecting between EGSM Rx
54 and GSM850 Rx (in addition to the two usual Tx switches), this switch directs 63 and GSM850 Rx (in addition to the two usual Tx switches), this switch directs
55 the low band Rx path toward one of two different SAW filters, and the outputs 64 the low band Rx path toward one of two different SAW filters, and the outputs
56 of those two filters are then joined. The high band Rx path always goes to both 65 of those two filters are then joined. The high band Rx path always goes to both
57 DCS and PCS band SAW filters, and each of those high band Rx SAW filters gets 66 DCS and PCS band SAW filters, and each of those high band Rx SAW filters gets
58 its own output going to its own dedicated Rita LNA input. Note the lack of a 67 its own output going to its own dedicated Rita LNA input. Note the lack of a
59 baseband-controlled switch between DCS and PCS in the high band Rx path: this 68 baseband-controlled switch between DCS and PCS in the high band Rx path!
60 switch is present in all triband RFFE designs I have seen, thus a big question
61 is raised as to how this magic M034 component functions without one. I can
62 think of two possibilities:
63 69
64 Possibility 1: perhaps they do a 50/50 split of the total incoming energy 70 For a long time prior to the 2019-12 breakthrough I was concerned about this
65 between DCS and PCS Rx paths, with each path suffering by 3 dB as a result. 71 lack of a baseband-controlled switch between DCS and PCS in the high band Rx
72 path: this switch is present in every triband RFFE design I am familiar with,
73 including the one we got from Openmoko, and I was concerned that feeding the
74 high band Rx path to both DCS and PCS SAW filters without a switch would
75 introduce a 3 dB penalty into these high band Rx paths. But when I got my
76 hands on the newly discovered Tango modem modules, looked at the GMagic numbers
77 written into their FFS as part of the manufacturer's production calibration and
78 then independently verified them with our CMU200, all fears were dispelled: the
79 Rx performance of this M034 FEM in all 4 bands is exactly the same as our
80 current Openmoko-based triband RFFE, with GMagic around 200 half-dB units.
66 81
67 Possibility 2: perhaps by virtue of integrating the ASM and the SAW filters 82 Now that we have physical proof that our desired quadband RFFE based on this
68 into a single monolithic FEM, Epcos found some way to have unswitched DCS and 83 Epcos M034 FEM and other aspects of TI's Leonardo design that go with it really
69 PCS Rx without incurring that 3 dB penalty. Perhaps they successfully 84 works exactly as we would like and has been used successfully by a historical
70 implemented some form of frequency diplexer such that out of the total incoming 85 commercial modem module manufacturer (albeit an obscure non-mainstream one),
71 energy picked up by the wideband antenna, DCS downlink frequencies go through 86 the way forward for FreeCalypso is clear: instead of continuing with Openmoko's
72 the DCS Rx SAW filter, PCS downlink frequencies go through the PCS Rx SAW 87 PCB layout and triband RFFE for our future hardware products, we can confidently
73 filter, and no needless losses are incurred. This hypothesis is supported by 88 go with a new PCB layout based on Leonardo/Tango, using Epcos M034 as our FEM
74 the observation that the available M034F.pdf document gives approximately the 89 and achieving the full quadband capability we have always wanted.
75 same insertion loss numbers for all 4 Rx bands, i.e., the same between the
76 switched low bands and the unswitched high bands. Note that they could not
77 have similarly eliminated the GSM850 Rx switch: both EGSM Rx and GSM850 Rx need
78 to go to the same LNA on the transceiver, thus a switch is needed somewhere.
79 90
80 I (Mother Mychaela) would absolutely love to play with an M034-based quadband 91 The matching networks ("RF black magic") that would need to be placed between
81 Calypso+Iota+Rita board in my lab with the trusty CMU200 instrument, and to see 92 M034 Rx outputs and Rita LNA inputs were also a big area of concern for a long
82 how well it actually performs, especially in comparison with our current 93 time - I did not feel confident with blindly going with the matching networks
83 OM-based triband version. However, in all of my years of searching I have never 94 depicted on Leonardo schematics (or the slightly different E-Sample version)
84 found a physical Leonardo board (any version), nor have we ever found any 95 with no ability to actually understand them - but the new Tango find has once
85 Leonardo PCB layout files which would allow us to build a modern recreation - 96 again saved the day. Tango uses exactly the same matching networks as depicted
86 thus the magic of M034 remains elusive. 97 on Leonardo+ schematics (not the slightly different E-Sample version), or more
87 98 precisely, it uses exactly the same topology (PCB layout), but with slightly
88 Unless a miracle happens and we are able to obtain either a physical Leonardo+ 99 different component values populated. The safest approach for FreeCalypso is
89 board or a PADS PCB file for one, there is no quick or low-effort way to "just 100 once again clear: copy Leonardo+ matching network topology and Tango component
90 try" this M034 RFFE. Instead if we wish to build a FreeCalypso board with this 101 values.
91 RFFE, it would have to be "the full 9 yards": a full-blown PCB design and layout
92 effort. There is no way to just "drop" the M034 into our existing PCB design
93 in the place of our current triband RFFE, it would have to be either a very
94 disruptive RF section layout change or an entirely new PCB layout, making this
95 idea very open-ended - an open-ended venture with quite uncertain outcome, but
96 with a high dollar cost attached to it. Given the massive effort required and
97 PCB layout labor costs, I currently have no active plans to pursue this idea
98 beyond hypothetical.
99
100 Commissioning a new custom RF FEM
101 =================================
102
103 Here is a wild thought: what if instead of twisting over backwards trying to
104 hammer an existing RF FEM like M034F into our not-quite-fitting PCB design, we
105 were to get an entirely new FEM made specially for us, made exactly the way we
106 need it? If we were to venture that way, I would ask for a FEM very similar
107 conceptually to M034F, but with a few changes:
108
109 1) Instead of diplexing between DCS and PCS SAW filter inputs with a 50/50
110 energy split, implement another switch (just like the GSM850 Rx switch) for
111 PCS Rx, exactly the same way how it is done in classic triband designs like
112 our current OM-based one. This change should eliminate the extra 3 dB
113 penalty which I assume (for lack of experimental data) must happen with the
114 existing M034 FEM. Or as an alternative to making this change, if someone
115 who is more knowledgeable than me in this area can explain to me why it isn't
116 necessary, I would accept that option as well.
117
118 2) I would ask for a rearranged pinout: the existing M034F pinout does not fit
119 at all into our OM-based PCB layout, but it would fit much better with some
120 rearrangement.
121
122 3) The hypothetical M034-like FEM would fit into our OM-based PCB layout a lot
123 better if it were made a little smaller than the 8.2x5.5 mm size of M034F.
124 Considering that the original M034F was created some 15-16 y ago, I assume
125 that it should be possible to make a smaller version in 2020 or 2021 or
126 whenever.
127
128 Timeline sequentiality
129 ======================
130
131 All of the above ideas will be considered on a less hypothetical level after we
132 get our already-committed FCM40 product built. FCM40 will be a modem module in
133 the same 56.5x36 mm form factor as Huawei GTM900 (with a mostly-compatible FPC
134 interface with only a few changes), featuring the same OM-based triband modem
135 core as FCDEV3B V2. The reason for this sequencing is that our current FCDEV3B
136 suffers from a couple of issues which FCM40 is expected to solve:
137
138 1) FCDEV3B has a very tight PCB layout: not only do we have the tightly laid out
139 core from GTA02, but also the whole board is quite small for the implemented
140 peripheral complexity, imposing further constraints from all sides. This
141 tight and complex layout makes a poor choice of starting point for bold
142 experiments like RFFE changes.
143
144 2) FCDEV3B is locked into Altium. Layout data migration from Altium to FOSS
145 appears to be much less feasible than migration from PADS to FOSS, thus
146 freeing our PCB layout from the clutches of proprietary software will most
147 likely require giving up (or rather setting aside) all of FCDEV3B new layout
148 and going back to the GTA02 starting point, which is in PADS format rather
149 than Altium. Redoing all of FCDEV3B anew does not sound appealing at all,
150 but the much simpler FCM40 board offers a perfect opportunity for a fresh
151 start.
152
153 FCM40 will have exactly the same OM-based triband RFFE as our current FCDEV3B,
154 but it will be a much simpler board, and if we can get it done in FOSS instead
155 of continuing the Altium track, then we would have a very solid reference and a
156 good starting point for potential RFFE change experiments.
157 102
158 Firmware compatibility 103 Firmware compatibility
159 ====================== 104 ======================
160 105
161 Our current FreeCalypso firmwares drive TSPACT RFFE control signals as follows 106 Our current FreeCalypso firmwares drive TSPACT RFFE control signals as follows
167 TSPACT5 = Rx GSM850 band 112 TSPACT5 = Rx GSM850 band
168 113
169 The driving of TSPACT1, TSPACT2 and TSPACT4 matches the way these signals have 114 The driving of TSPACT1, TSPACT2 and TSPACT4 matches the way these signals have
170 been assigned by Openmoko and thus the way they function on our current OM-based 115 been assigned by Openmoko and thus the way they function on our current OM-based
171 triband RFFE, whereas TSPACT5 is a new signal which is not wired anywhere on 116 triband RFFE, whereas TSPACT5 is a new signal which is not wired anywhere on
172 our current FCDEV3B. This signal driving arrangement is expected to be 117 our current FCDEV3B. If and when we actually produce a new FreeCalypso hw
173 compatible with all 3 RFFE hw possibilities under consideration: 118 product with the newly-confident M034 FEM, we will need to wire its control
119 signals (going through logic-inverting PNP transistors as usual) to TSPACT2,
120 TSPACT4 and TSPACT5 as mapped above, leaving TSPACT1 unconnected as there is no
121 switch between DCS and PCS in the high band Rx path. This new control signal
122 wiring is an original FreeCalypso invention, different from TI's original
123 Leonardo and E-Sample TSPACT signal assignments, but in my maternal opinion
124 having the same fcmodem firmware build run on both legacy FCDEV3B and future FC
125 quadband modems would be more valuable than paying tribute to the historical
126 Leonardo.
174 127
175 * On our current OM-based triband RFFE it works as is. 128 Other ideas
129 ===========
176 130
177 * If we use Epcos M034 or a semi-clone thereof that has the two Tx switches and 131 In a previous version of this article I entertained the idea of getting an
178 a GSM850 Rx switch but no PCS Rx switch, then we will need to connect TSPACT2, 132 entirely new M034-like FEM custom-designed and made specifically for us, asking
179 TSPACT4 and TSPACT5 per the table above, and leave TSPACT1 unconnected. 133 someone with the necessary capabilities to produce a new FEM for us that would
134 be very much like the historical M034F, but with just two key differences:
180 135
181 * If we get a new M034-like FEM custom-made with a full set of all 4 switches, 136 1) A rearranged pinout so that the putative new FEM could be just "plopped"
182 then all 4 TSPACT signals will need to be connected per the table above. 137 into our current OM-based PCB layout in the place of OM's triband RFFE;
138
139 2) Better-understood Rx output impedance specifications so that simpler and
140 more intuitive matching networks like OM's could be retained as well.
141
142 That desire was driven by the mystery of Epcos M034 and the lack of empirical
143 test data for it, stemming from the lack of an already-existing historically-
144 made physical specimen that could be examined and tested: I simply could not
145 stomach the idea of expending a monumental effort on a new PCB layout followed
146 by thousands of dollars of cost to physically produce the new experimental board
147 batch while doing it as a completely blind stab-in-the-dark with the unknown
148 Epcos FEM. But the discovery of Tango modems in 2019-12 has drastically changed
149 the situation, and all mysteries and unknowns surrounding that M034 FEM have now
150 been cleared.
151
152 In the new reality that exists from 2020 onward that idea of commissioning the
153 design of a new FEM to replace Epcos M034 no longer makes any sense and can be
154 taken off the table: while it is true that using the existing M034 FEM will
155 require doing an entirely new PCB layout (no more reuse of OM's version), it is
156 now just a matter of doing the layout labor, no more risks or uncertainties,
157 and the cost of this PCB layout job will certainly be much less than having a
158 new FEM custom-made for us.
159
160 Of course there are also myriad other, completely different ways of implementing
161 a quadband GSM MS without using an M034-style FEM. The approach used by TI in
162 the days of Calypso, using FEMs that combine the antenna switch and SAW filters
163 in one component while the PA is separate, fell out of fashion shortly
164 afterward, replaced with a competing approach where the PA and the antenna
165 switch are combined into one component while Rx SAW filters are external - the
166 latter approach is featured on TI's I-Sample (LoCosto) board, and it is
167 quadband - LoCosto has 4 separate LNA inputs, not 3 like Rita. One could even
168 implement the same approach while keeping the Calypso+Iota baseband, perhaps
169 replacing TI's Rita transceiver with Silabs Si4210 (Aero II) that has 4 LNA
170 inputs like LoCosto. But such ideas are far outside the scope of FreeCalypso,
171 so if anyone feels like pursuing them, feel free to do so on your own, not
172 involving FC - my mission in life is NOT to invent or innovate anything new,
173 instead I am all about resurrecting and bringing back to availability those
174 perfectly good and perfectly working technical solutions which already existed
175 once before, but which have been wrongfully discontinued and thus taken away
176 from us.
177
178 In the present situation, the goal of producing a quadband GSM MS while reusing
179 as much of the already existing FreeCalypso IP and know-how as possible would
180 be accomplished most efficiently by using the newly-confident M034 FEM and doing
181 a new PCB layout with it, hence that is the approach I am currently pursuing.