Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
- BITX20
- Messages
Search
|
Re: 80 meter bandpass filter
Jim, it sounded strange to me as well. But most qrps have just a final transistor and all the harmonics have to be filtered out. In the uBitx the push pull configuration of the PA attenuates the even harmonics so the corner can be shifted some Mhz allowing more bands to pass (eg.? 80 and 60m). Il 20/ago/2018 23:46, "Jim Tibbits" <ab7vf1@...> ha scritto: Aha!!!! photo of "80 meter" bandpass filter on my ubitx ...L20, L21, L22 as swept? ..corner frequency is closer to 6 mhz than 4 |
|
Re: coding
Hi Jack,
Yes, most programing optimization is in the design and the chosen solutions to the problem you are trying to solve. That is a bigger picture than the smaller optimizations that the complier can make. Your RDC example is good case in point. Once and a while I'll be trying to optimize some critical tight piece of code or unique algorithm, but my work is mostly, now, working with people to define their business or system problem, breaking that down to a design and coding it. Most of the efficiency and robustness comes from the overall design, efficiently representing the data for manipulation, and having a good sense of what established libraries perform the low level stuff the best. But, you can't do that if you haven't understood the basics of good code design and what you might be relying on the complier to optimize for you. Great, thanks for that information, I'll look for the assembly file, these tiny processors are a wonderful back to basics opportunity, I'll enjoy looking at that code. Was thinking I'd use an object dumper and hope the symbol table was included in the elf file.? I should dig deeper into the Arduino tool chain, I imagine there must be a config file where I can change the complier flags or there may still be a make file somewhere, it would be fun to shut all the optimizations off and compare the results. And force myself to go back to when it was necessary to do the optimizations yourself as a fun exercise. Your examples for your students sound like they worked to leave an lasting impression of what you were trying to get across. It would probably be necessary to be more politically correct nowadays. But the off-the-wall examples are fun. Now days with algorithms determining who's product is shown first, whose app gets shown first in an App Store, or what ads go to the top of the page, gaming the system is prevalent. While thankfully way more subtle than your eliminating poverty example, attempts to skew the data are involved in gaming the system. Tom, wb6b |
|
Re: Intermodulation Performance
Warren, you did quantitative tests that surprise me to a certain extent. Surely the uBitx is not an IC-7300. Even my SWL ears find the IC-7300 transmitted audio one of the best ever. But as I continously monitor my transmitted audio with an IQ SDR receiver I have not found the TX audio that terrible. Mine may sound tinny, the BFO is almost 1khz far from the passband but the opposite sideband seems to me quite well rejected. Does the audio sound distorted? Not really, reports were fine. I imagine that the IMD products may give more problems within the filter passband rather than outside. Maybe it would be worth to test the IMD with digital modes like PSK and see the reports by others. Il 22/ago/2018 17:23, "Kees T" <windy10605@...> ha scritto:
> > Hello Warren Allgyer, > > I want to thank you for all the effort you have put into the uBITX spur/harmonic problem definition and the excellent data presented. I realize your project is shelved ....but, would you be interested in trying out an 8 relay switched LPF board with 4 plug in LPFs using the parts off the uBITX board. You ran an experiment like that earlier with a perf. board. I have boards (75mm x 62mm) coming in soon and can supply the relays and misc parts (no charge). > > 73 Kees K5BCQ |
|
Re: coding
Jack Purdum
This entire journey has been a real tickle for me. Now if I can just get JackAl out the door! Jack, W8TEE
On Wednesday, August 22, 2018, 3:24:59 PM EDT, Des Watson <desw@...> wrote:
Jack, > Today's compilers are smart enough they might generate the jump table > anyway! Exactly so. Having spent many years studying and writing compilers and optimisers, I see that today's compilers can produce astonishingly good code, usually *far* better than any hand-written assembler. The implication is that it's almost always better to code applications (even time-critical real-time applications) in a high-level language. > That will allow you to see the assembly code for the program by load > that file into any text editor. Good plan. I always asked my compiler students to look at the code generated by a good compiler with an optimising backend (like gcc or llvm). It's a most educational exercise! And thanks for all the contributions to this list - such a help in getting my uBitx going and teaching me so much about RF design. Best wishes Des, G3YXO |
|
Re: coding
Jack,
Today's compilers are smart enough they might generate the jump tableExactly so. Having spent many years studying and writing compilers and optimisers, I see that today's compilers can produce astonishingly good code, usually *far* better than any hand-written assembler. The implication is that it's almost always better to code applications (even time-critical real-time applications) in a high-level language. That will allow you to see the assembly code for the program by loadGood plan. I always asked my compiler students to look at the code generated by a good compiler with an optimising backend (like gcc or llvm). It's a most educational exercise! And thanks for all the contributions to this list - such a help in getting my uBitx going and teaching me so much about RF design. Best wishes Des, G3YXO |
|
Re: Wow... 15 volts in and a bunch out..
Arv,
>>>From tests by Allison, Warren, Kees, and others it seems that part of the problem? may be in drivers for the IRF510s.? As these devices run out of gain at higher?
frequencies they start generating garbage(?).? If that is true then part of the solution?
would seem to be fixing what gets into the IRF510s before it gets there.??
Receive side has not received much attention other than recent IMD comments, but?
Maybe fixing the transmit drivers is worthwhile to look at.?? Yes, but also over simplified.? The IRF510s or RD for that fact have a lot but not infinite gain.? So at some point you have to supply the needed drive or things decay. Its also true if the various stages have inadequate frequency response the result is over driving to get some power.? it is unclear if its totally in the power amp or in the earlier stages, likely both. You have to think about what you say when the word "clean" is applied to the power amp.? Clean means what ever you put in is reproduced faithfully at a greater level.? So spurs in means spurs out.? That is what filters solve for us. When applied to a practical amplifier it means minimum intermodulation distortion, low harmonic content for a given design, and also stable under mismatched loads. >>>I have not talked with Farhan?about this but maybe there could be an upgraded
uBITX version to replace the present?design and incorporate what we have learned.? Have we learned enough yet for that to be?a viable direction? I think so.? It clearly indicated skimping on filters for simplicity is a false economy. of birdies increases but they are otherwise annoying but harmless.? On the transmit side its unacceptable. Allison |
|
Re: Wow... 15 volts in and a bunch out..
Jerry,
For some that is a mountain. Others maybe an afternoon. I tried to balance it with very few new parts, no involved measurements, maximum HF functionality with improved filtering. For the "but 10M" I offer with straight though using the mentioned filter will solve the 10M issues both harmonics and spurs.? That filter type would likely work for 12M and 15M with the appropriate frequency filter. Allison |
|
Re: Intermodulation Performance
Allen and David,
Right now this is just a test to see what the results look like and "if there is a pony in the room". We may just provide the Gerber files, the LPF Relay boards, A complete LPF upgrade kit (not sure I can handle the potential demand after the AGC/Click activity), or just give it to someone else to run with ...? 73 Kees K5BCQ |
|
Re: Wow... 15 volts in and a bunch out..
I use a 78s05 voltage regulator , at 2 amps it gives a bit more headway and runs cooler .
toggle quoted message
Show quoted text
|
|
Re: Wow... 15 volts in and a bunch out..
Jerry, Allison, etc... From tests by Allison, Warren, Kees, and others it seems that part of the problem may be in drivers for the IRF510s.? As these devices run out of gain at higher frequencies they start generating garbage(?).? If that is true then part of the solution would seem to be fixing what gets into the IRF510s before it gets there.? Receive side has not received much attention other than recent IMD comments, but Maybe fixing the transmit drivers is worthwhile to look at.? The original BITX was based on fixed gain amplifiers.? This eliminated level variations due to frequency.? Maybe we can only get a few DB from a 2Nxxxx but by fixing the gain per device at some level that will not have it pushing up against its maximum gain-frequency limit at any frequency where it is bing used, we could have the same drive at all frequencies. This may require that several lower gain stages would be needed in place of one higher gain stage, but still could provide necessary gain and clean output at all frequencies.? We have learned a lot since the uBITX became available.? All this is probably of limited value to the present uBITX without extensive modifications.? I have not talked with Farhan about this but maybe there could be an upgraded uBITX version to replace the present design and incorporate what we have learned.? Have we learned enough yet for that to be a viable direction? Arv _._ On Wed, Aug 22, 2018 at 11:50 AM Jerry Gaffke via Groups.Io <jgaffke=[email protected]> wrote: Allison, |
|
Re: Intermodulation Performance
Warren Allgyer
Hi Kees
I would be honored to test and record data, publish if you wish, any filters you would be willing to send. I would of course return them after testing is complete. If you would like to proceed please contact me privately for shipping info.? Thank you Kees.? WA8TOD |
|
Re: Wow... 15 volts in and a bunch out..
Allison,
toggle quoted message
Show quoted text
That summary is much appreciated. Looks like a reasonable to-do list, anything much more than that and we may as well rebuild from scratch. Thanks! Jerry, KE7ER On Wed, Aug 22, 2018 at 10:21 AM, ajparent1/KB1GMX wrote: On Wed, Aug 22, 2018 at 08:31 AM, Jerry Gaffke wrote: |
|
Re: boosting the power on 28 MHz
#ubitx
As a 6M radio I'd give it a, your not serious..
At a bare minimum with enough mods to meet the FCC -50dbc for spurs and trash. * band pass before power amp. * better output filter. *Then dump the whole power amp as its maybe .3W at 6M or better said a lot of ?effort for little power.? (at a minimum replace all the 3904s with BFR106 and 5109s) ?The IRF510s can make good power there (assuming short leads). *Add a RF LNA as by weak signal standards deaf as a stump.? As a single band radio doable. Allison |
|
Re: Wow... 15 volts in and a bunch out..
Jeffery,
One last step to the last #57649.. Once the relays are rewired the code must change. From this: /**
?* Select the properly tx harmonic filters
?* The four harmonic filters use only three relays
?* the four LPFs cover 30-21 Mhz, 18 - 14 Mhz, 7-10 MHz and 3.5 to 5 Mhz
?* Briefly, it works like this,?
?* - When KT1 is OFF, the 'off' position routes the PA output through the 30 MHz LPF
?* - When KT1 is ON, it routes the PA output to KT2. Which is why you will see that
?*? ?the KT1 is on for the three other cases.
?* - When the KT1 is ON and KT2 is off, the off position of KT2 routes the PA output
?*? ?to 18 MHz LPF (That also works for 14 Mhz)?
?* - When KT1 is On, KT2 is On, it routes the PA output to KT3
?* - KT3, when switched on selects the 7-10 Mhz filter
?* - KT3 when switched off selects the 3.5-5 Mhz filter
?* See the circuit to understand this
?*/
?
void setTXFilters(unsigned long freq){
??
? if (freq > 21000000L){? // the default filter is with 35 MHz cut-off
? ? digitalWrite(TX_LPF_A, 0);
? ? digitalWrite(TX_LPF_B, 0);
? ? digitalWrite(TX_LPF_C, 0);
? }
? else if (freq >= 14000000L){ //thrown the KT1 relay on, the 30 MHz LPF is bypassed and the 14-18 MHz LPF is allowd to go through
? ? digitalWrite(TX_LPF_A, 1);
? ? digitalWrite(TX_LPF_B, 0);
? ? digitalWrite(TX_LPF_C, 0);
? }
? else if (freq > 7000000L){
? ? digitalWrite(TX_LPF_A, 1);
? ? digitalWrite(TX_LPF_B, 1);
? ? digitalWrite(TX_LPF_C, 0);? ??
? }
? else {
? ? digitalWrite(TX_LPF_A, 1);
? ? digitalWrite(TX_LPF_B, 1);
? ? digitalWrite(TX_LPF_C, 1);? ??
? }
}
To this: /**
?* Select the properly tx harmonic filters
?* The three harmonic filters use only three relays
?* the three LPFs cover? 18 - 14 Mhz, 7-10 MHz and 3.5 to 5 Mhz
?* if none are selected the path is straight through and use of external bandpass filter required. ?*?See the circuit to understand this
?*/
?
void setTXFilters(unsigned long freq){
? if (freq > 20000000L){? ? ? ? ? //? ?for 20 mhz and up its straight through no filters.
? ? digitalWrite(TX_LPF_A, 0);
? ? digitalWrite(TX_LPF_B, 0);
? ? digitalWrite(TX_LPF_C, 0);
? }
? else if (freq >= 14000000L){? // KT1 is picked and the 14-18 MHz LPF is selected
? ? digitalWrite(TX_LPF_A, 1);
? ? digitalWrite(TX_LPF_B, 0);
? ? digitalWrite(TX_LPF_C, 0);
? }
? else if (freq > 7000000L){ // KT2 is picked and the 14-18 MHz LPF is selected
? ? digitalWrite(TX_LPF_A, 0);
? ? digitalWrite(TX_LPF_B, 1);
? ? digitalWrite(TX_LPF_C, 0);? ??
? }
? else {? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? // KT3 is picked and the 14-18 MHz LPF is selected
? ? digitalWrite(TX_LPF_A, 0);
? ? digitalWrite(TX_LPF_B, 0);
? ? digitalWrite(TX_LPF_C, 1);? ??
? }
}
?
?
|
|
Re: Wow... 15 volts in and a bunch out..
On Wed, Aug 22, 2018 at 08:31 AM, Jerry Gaffke wrote:
Yes, the whole thing needs a re-think.OK this is radical by some belief. First give up on the idea of 20mhz and above too many tough fixes needed. Both fixes are not fun but make sense for believable accomplishment. NOTE:? it does nto disable 20mhz and up.? An external BANDPASS filter could be used with 10M and would work better with this change. see?http://www.kitsandparts.com/W3NQN_May_June_1998_QST.pdf Power amp help: *Replace Q90 with BFR106 or SMT 2N2369A *C81 470 to 1000pf to keep the 80m power in balance. *Replace all of the 3904s with 2n2222A? (predriver and driver) Optional mod: The IRF510s are mounted with .45" leads and is bad practice in power amps. Its possible to shorten them to under .09" Required materials, two insulator kits and suitable heat sink for case anticipated. It improves power out some? at 10M. * Remove IRF510 and remove heatsinks. * bend leads 90 degrees up at the case. * insert leads into board from bottom so that IRF510 case touches. * use heatsink insulator kits and heatsink.? Heatsink must be grounded ? for better stability. Filter help: * remove the relays KT1 though KT3 an K3.? (replace if damaged). * disable the use of K3 for audio switching it causes issues. * remove all the interconnecting traces for RF paths for all. ? DO not cut the traces to the relay coils on the RF ones. * Using COAX jumper from K3 to K1 for RX path. Ground braid ? ends of the underside ground plane. * Reinstall relays on the underside! * using short wires from the input side and output side of the fitlers ? to the relays NO contacts. ? KT1 is? 20/17M filter ? KT2 is? 40/30M filter ? KT3 is 80/60M filter. ?K3 is antenna transfer from jack to RX or TX fitlers.? * jumper the NC contacts for each relay. * daisy chain the wiper contacts from one relay to next. * mod the code to select the right relay for band. This is a tested set of modifications.? With it you loose 20mhz (unless you use external filter) and up but the filter issue is FIXED for the lower bands.? If some one wants to write it up I can help. Allison |
|
Re: QRP SWR meter recommendation?
#ubitx
A little off-topic, but looking at disassembly is important for the big processors too if you want to squeeze out the best performance. In particular, vectorizing compilers can be pretty finicky (easily derailed).? For DSP filter implementations properly vectorizing inner loops makes a huge difference.
73 Mike KK7ER |
|
Re: Wow... 15 volts in and a bunch out..
Jerry,
Did you reduce the gain on the amp's driver and pre-driver stages when adding the MMIC stage? Yes.? I cranked them down to 11db range as the goal was about 60db total using 2n2222a and BFR106.? The plan was assume 11db for the irf510s (force it with feedback) and all the prior stages at 10db.? So the added 10db of a mav11 was tolerable in the overall scheme.? I tried before q90 and after at the drive pot.? However the systemic ground loops became obvious. Cutting the?line form the amp out put to the filter area and driving the load from there helped greatly.? Never was satisfied at any time with the stability.? The only reason that kept me from frying finals was variable current limit power supply set at 2.5A.? If I didn't have that many devices would be destroyed. Allison |
Marc Jones
to navigate to use esc to dismiss