http://cdn.macom.com/datasheets/DS-MAMX-007238-CM25MH_V6.pdf AR9283 2.4GHz TRX feeds this mixer https://www.anaren.com/sites/default/files/X3C09P1-03S_Datasheet_RevB.pdf used for splitter/combiner into 2 parallel amps http://www.vectron.com/products/saw/pdf_saw/TFS915A.pdf <-- main RF port filter http://www.cel.com/pdf/datasheets/upg2009tb.pdf <-- final RF switch, with cheap prices: http://www.aliexpress.com/item/UPG2009TB-E3-UPG2009TB-SOT363-G2U-RF-Switches/954776235.html http://www.cel.com/pdf/datasheets/upg2163t5n.pdf <- RX-side mixer switch near the AR9283 http://www.cel.com/pdf/datasheets/upc8172tb.pdf <-- RX side up-converter Si4113-D-GM PLL https://www.silabs.com/Support%20Documents/TechnicalDocs/si4133.pdf LO freq = 1.545GHz RX-side switch, right next to the AR9283 http://www.cel.com/pdf/datasheets/upg2163t5n.pdf SKY65111-348LF RF amplifiers: http://www.skyworksinc.com/uploads/documents/200428E.pdf If the interface is disabled, it doesn't do RX mixing, you gotta put it at least into scan mode. LO freq stays fixed. RX of 929.7MHz gets up-converted via 1.545GHz LO to 2.4747GHz. http://cdn.macom.com/datasheets/DS-MAMX-007238-CM25MH_V6.pdf AR9283 2.4GHz TRX feeds this mixer * EO_ wants to know why that specan feature in it sucks so bad ... <EO_> ok you can do spectral-scan with an interface that's disabled! <EO_> it wakes it up in read-only mode to scan <nigelvh> Interesting <EO_> so now I know it's safe to send signals into this modem and see wuddup <EO_> wow, looks like ~3MHz RBW on spectral-scan! <EO_> http://i.imgur.com/XWlgi3W.png <-- this is a screen shot of a pure sine wave <EO_> LOL <EO_> @ 910MHz <EO_> http://www.bartk.us/dump/9HPn/910MHz-60dBm.png <-- same 910MHz pure sine wave @ -60dBm <EO_> So now I wonder if that -40dBm or whatever 931MHz signal is causing the crazy floor <EO_> -41dBm @ 929.7MHz <-- real world power QRM <EO_> let's SEE how she reacts! <EO_> bwahahaha YUP <nigelvh> FLOOR DESTROYED <EO_> http://www.bartk.us/dump/9HPn/929.7MHz-41dBm.png <EO_> YUP <EO_> worse than a porn star on efukt <nigelvh> LAWL <EO_> so now I know where my -60dBm noise floor came from <EO_> question is, does it actually affect comms? <EO_> or are these just false specan readings * EO_ is not sure how to simulate that in the lab <EO_> I know I did it before though. <EO_> @ 5.8GHz <EO_> need me a splitter <EO_> for 900MHz, with good port isolation <EO_> yikes $90 <EO_> [eo@Base900] /interface wireless> set 0 tx-power=0 <EO_> note to self: this is minimum 9HPn power setting, and is actually +5.3dBm <nigelvh> Alright, it's been about 45 mins: https://digitalnigel.com/wordpress/wp-content/uploads/2015/10/70CM_MaxHold_45mins.gif <nigelvh> Granted, it's later in the evening. <EO_> hmm yeah that's not very satisfying <EO_> what's the huge peak @ 450? <EO_> commercial? <EO_> also the only isolators I have are for 5.3GHz :( <nigelvh> That'd be my guess. It's there all the time. <EO_> even with +5dBm through a 30dB attenuator, I'd still be shooting -25dBm back into the signal generator...which it may or may not like <EO_> would feel a lot better with another 20dB isolation there so it's only -45dBm going back into the generator <EO_> http://www.ebay.com/itm/RF-Isolator-800-MHZ-900-MHZ-Dual-section-SMA-high-isolation-SWEPT-Tested-/351321240799?hash=item51cc60b4df:g:-s0AAOSwZjJU3mct <EO_> we have a winner <EO_> now need a spleeetor <carrar> someone is WINNING? <allenp> a winrar is you <EO_> http://www.ebay.com/itm/Mini-Circuits-ZAPD-21-Power-Splitter-0-5-2-0-GHz-SMA-Connectors-10W-50-Ohm-/271875040552?hash=item3f4d03e528:g:vU4AAOSwBahVW1-a <EO_> we have another winner! <EO_> looks like about 25dB of isolation <EO_> should be all set now to do a simular QRM test <EO_> with 9HPn <EO_> simulated <EO_> Si4113-D-GM PLL https://www.silabs.com/Support%20Documents/TechnicalDocs/si4133.pdf <EO_> LO freq = 1.545GHz <EO_> fascinating. if the interface is disabled, it doesn't even do RX mixing <EO_> you gotta put it at least into scan mode <EO_> LO freq stays fixed <EO_> 929.7MHz gets up-converted via 1.545GHz LO to 2.4747GHz <EO_> looks like only 6dB conversion loss in the whole chain, including my coax! <EO_> -70dBm sig gen becomes .. uhhh -64dBm after the mixer? <EO_> whoa this up-converter chip has positive conversion gain! <EO_> 8dB @ 2.4GHz <EO_> damn. the IF compression point of that mixer isn't until it is fed with -10dBm * EO_ verifies <EO_> yeah, it's all linear <EO_> So it's just the Atheros TRX chip freaking out? :( <EO_> hmmm harmonic @ 2.1603 <carrar> odd <EO_> http://hamwan.org/t/RouterBoard+Metal+9HPn?structure=HamWAN new lab page! * EO_ goes to take some medium-rez pix * EO_ fails @ taking medium-rez pix <nigelvh> EO_: responded to your email <EO_> nigelvh: responded to your email <EO_> I am encouraged by the fact you can see the individual solder paste balls in your footage <nigelvh> Really? Features on the edge of resolution from what you saw in my video? <EO_> Although the camera seems to be defocused <EO_> yup <nigelvh> That's suprising. <EO_> They used some extremely tiny parts here <nigelvh> And yes, the focus is adjustable. <EO_> ok cool <EO_> I should really own a usb microscope <EO_> the tiniest labels are laser-etched into metal <EO_> I can't make them out by eye, or with my magnifying glass <EO_> also wanna stitch a high rez front+back pair of photos for documentation <EO_> 4.7GHz fT measured! <EO_> thayward: basically there are 2 strong carriers around 930MHz <EO_> thayward: when you feed a -40dBm carrier @ 930MHz into the 9HPn, it gets seen as a broadband noise floor rise when running scan-spectrum <EO_> thayward: I checked the analog front-end, and it's NOT clipping. The RX mixer is good up to like 0dBm <thayward> so, it turns out the 9HPn doesn’t make as good of a spec-an as your spec-an? <EO_> It's looking like the signal spread is taking place inside the AR9832 chip <EO_> So then the question is, is the spreading REAL, and actually affecting comms below 930MHz? <EO_> Or is it just an artifact of the way scan-spectrum works and it doesn't affect normal comms at all? <EO_> I have hardware on the way to answer this question. <thayward> and if yes, can we filter it? Lots of high power stuff on both sides of 900 MHz <EO_> yeah, it'd be tough <EO_> and it'd raise cost even more <EO_> 900MHz is already gonna be way more expensive :( <EO_> (station cost that is) <EO_> base modem is $108 vs $96 for 5SHpn <EO_> so that's +$12 <EO_> Then you'll need an amp ($100-$200) <EO_> then if you add a filter that's another $100 <thayward> can’t do ham radio without an amp <thayward> antenna? <EO_> impossibru! <EO_> antenna is reasonable. About $50 for the tenna, $50 for mag mount. <EO_> If you want yagi/stationary, those are reasonable too <thayward> FWIW, I’ve got one of these and it kicks ass on LTE. I haven’t tried 900. http://smile.amazon.com/Wilson-Electronics-700-2700-Directional-Connector/dp/B00J14YEHQ/ <EO_> So yeah, looks like top spec station would cost you $400, plus antenna. <EO_> I'd raelly love to avoid the filter <EO_> drop it to $300 <EO_> Some Say(tm) it achieves 50dB isolation <EO_> Some Say(tm) 9HPn can take 0dBm RX without damage <EO_> Some Say(tm) +46dBm - 50dBm = -4dBm :D <thayward> add a bit of insertion loss <thayward> you’ll be fine <EO_> yeah, I'll have a 2nd set of TRX switches on the low power side that can further shunt that -4dBm to ground <EO_> or to a 50 ohm load <EO_> even better, less distortiony * EO_ wonders how to measure 900MHz-carrying switching events with a sampling scope <EO_> heh, the filter in the 9HPn is rated for a max of +10dBm <EO_> holy cow, 3dB insertion loss on it too! <EO_> So it's a 2W radio with 1W output. Silly. <blueben> Yikes <blueben> Isn't that the only TDMA capable 900mhz radio too? <EO_> RX pad is 3 second period 50% square wave between 0V and 2.5V <EO_> which seems to be altering the duty cycle of the TX pad <EO_> between 4ms on / 2ms off and 10ms on / 2ms off <EO_> tdma period size is set for 4ms <EO_> the actual TRX switches are running at a steady 4ms period though <EO_> with about 500us taken during each TX <EO_> s/steady/jittery/ <EO_> the size of each frame seems to be varying a bit. modem probably adding extra crap into the data once in a while. <thayward> FEC? <EO_> nah, FEC is constant <EO_> also booo <EO_> there is no phase correlation between TX pad and actual TRX signal <EO_> cursed hidden layers <EO_> can't trace signals <EO_> carrar: you should find me the AR9283 spec sheet. <EO_> zee pinz are too small to probe! <EO_> allenp: sicence, bitch!!! http://bartk.us/dump/9HPn/DSC_1298.JPG <allenp> what is this? <EO_> part of an 9HPn modem <EO_> it would be swell if someone could find the data sheets for those ICs * EO_ is finding it hard to do <allenp> wonder what those dots mean <EO_> good question <EO_> I also don't know those package names <EO_> apparently you have a better time finding the specs if you punch in marking code + package name <EO_> allenp: wanna see the most insane text on this whole modem? <allenp> yes <allenp> I'll try anything at least 3 times <EO_> http://bartk.us/dump/9HPn/DSC_1287.JPG <-- check out AKq1 <EO_> at the bottom <EO_> would totally love to know what that part is too <allenp> took me a while to find it <EO_> it's so small you can't even find it! <EO_> never mind its text! <EO_> you can also see the stray solder ball in that photo :) <EO_> it's a where's-waldo! <allenp> well it was worth a shot http://i.imgur.com/zCyv7QS.png <EO_> nice idea! <EO_> the part # is F303 <EO_> so it's probably a Filter ? <allenp> seems like a reasonable guess <EO_> plus it's running along the RX signal path trace <EO_> so it's seeing 900MHz signal there <EO_> it's right before the upconverter chip <EO_> RX signal goes left -> right along that trace <allenp> you should do a live stream of reverse engineering circuits <allenp> video inlays of the scope n shit <allenp> mashup of eevblog and protonvids <EO_> allenp: I was thinking of video'ing the QRM injection testing <EO_> GOOD NEWS, EVERYONE <EO_> ooh wow it's 1:30AM <EO_> anyway, got it all on (boring) tape <EO_> and the results are GOOD <EO_> the spectrum-scan feature is just bogus <EO_> the real comms are not nearly as affected by the interference signals as spctrum-scan would have us believe! <EO_> yay!