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: Dead Or Just Sick? (P.S. - It was just sick!)
Curious, that's two bad 0.1uF 1206 caps found shorted recently.
toggle quoted message
Show quoted text
Both of them on the 3.3v supply rail from Nano to the Si5351. The Raduino has two such caps on the 3.3v rail, and three identical caps for DC isolation from the three clocks going from the Raduino out to the uBitx main board. Makes me wonder how many shorted DC isolation caps we have out there, and what the symptoms would be. I doubt these caps failed because of abuse, they are operating at a tenth of their rated voltage. I'd guess poor quality control on the part of whoever manufactured those caps. Possible that the auto-insertion equipment was handling them roughly or the boards are getting flexed somewhere, but those both seem doubtful. I personally found a shorted 0.1uF cap on a Bitx40 main board built at the end of 2016. The uBitx main board has about three dozen of the same cap, and the rig won't work if pretty much any of them decide to be shorted.? Some of those would be hard to find. If we suspect there might be a shorted cap, a good start would be to compare ohmmeter readings across each cap for a good board compared to the bad board (with power off). That may not identify exactly which cap if it's on a power rail with 10 such bypass caps plus bunches of other parts, but could help point the direction to go. The old thread for this particular story can be found here: ? ??/g/BITX20/topic/81064470 Jerry, KE7ER On Tue, Apr 6, 2021 at 03:37 PM, secondchancesailor wrote:
I finally got back into the shack after almost a month.? And I am glad to say, the uBitx lives!? ?I want to thank everyone for the excellent advice!? ?I started by removing C1 and replacing it with a small discrete capacitor.? ?My plan was to work my way through the capacitors until I found the problem.? ? After replacing C1, the radio was fixed!? ?Other than re-reading all the posts here and taking a few minutes looking at the schematic, the whole repair took about 10 minutes!? ?Thanks again for all the help from so many knowledgeable individuals!? ? This is truly an amazing group! |
|
Re: Dead Or Just Sick? (P.S. - It was just sick!)
I finally got back into the shack after almost a month.? And I am glad to say, the uBitx lives!? ?I want to thank everyone for the excellent advice!? ?I started by removing C1 and replacing it with a small discrete capacitor.? ?My plan was to work my way through the capacitors until I found the problem.? ? After replacing C1, the radio was fixed!? ?Other than re-reading all the posts here and taking a few minutes looking at the schematic, the whole repair took about 10 minutes!? ?Thanks again for all the help from so many knowledgeable individuals!? ? This is truly an amazing group!
|
|
Re: A simple question, I hope...
Bob
I am agreeing with the others. If you don't have a power meter, well you could try a cq in cw and see what RBN spots you get. If you get a few spots on a band with signals, likely the transmit chain beyond the mixer is intact. Testing ssb even with a power meter is harder. Injecting a steady audio oscillator or if you can do a steady loud whistle would work. As there isn't ALC in this rig, some microphones may not have enough audio drive. We can't blindly determine your microphone is adequate. Component failures out of the box are believed to be very rare.?? Because the cw and digital paths are different on transmit in the ubitx, best to test both of them. Requires a power meter, key or piece of wire, and an audio oscillator. A pc can produce an audio signal, in fact Ashhar discovered something like 1khz.com? Curt |
|
Re: Many uBitx's have reduced power on 40m CW
Given the variation in impedance as seen in the simulations of ideal components,
toggle quoted message
Show quoted text
it's amazing to me that the output power manages to be as flat as it is. Fun to figure out what all is going on here, more than it might seem at first. From a practical standpoint, Curt's bigger C81 strikes me as the best fix, for those that care about the slightly reduced power when operating 40m CW. But with C81 too big, the difference in output power between 3.5mhz and 30mhz may become too extreme. For the v3 configuration with C81=0.1uF and R83=2.2ohms, my simulation shows that the Q90 stage has an input impedance of 51 ohms to 45 ohms, at 173 to 165 degrees Voltage gain ranges from 4.5 to 3.65. These values are for a range of 3.5mhz to 30mhz, in that order, Q90 stage has a load of 50 ohms. On v4, R83 was moved to 2.2 ohms for more gain.? ? Input impedance dropped to a range of 18 to 19 ohms, 170 to 183 degrees. Voltage gain ranges from 5.5 to 6.7.? On v5,v6 the cap at C81 was moved to 470pF so high frequencies have more gain than low frequencies, this compensates for losses at high frequencies in the various power amp stages. Input impedance now ranges from 200 ohms to 45 ohms at 150 to 100 degrees. Voltage gain for the Q90 stage runs from 1.0 at 3.5mhz to 5.5 at 30mhz. The v3 configuration is likely the best as far as giving the filter and mixer a proper 50 ohm load. But we needed more gain at higher frequencies to get sufficient power out, without too much drive on 3.5mhz. I haven't analyzed the effect of the Q90 impedance phase angle on how the filter or mixer might behave. Jerry, KE7ER On Tue, Apr 6, 2021 at 03:25 AM, Evan Hand wrote: I also looked at changing C81 to .1uf and find that the impedance is better across the spectrum of interest.? I do not know how to see if there is an impact on the rest of the amp without building it. |
|
Re: A simple question, I hope...
The problem of low SSB output where CW output level is good, might be something as simple as a defective microphone.? Some of the microphone cases have the sound hole covered or miss-aligned with the electret mike element.? Other SSB transmit level problems arise when trying to use the wrong style microphone (dynamic or crystal) in place of an electret element. You can use other style microphones or mike amplifiers but you will have to alter the microphone input circuit to provide proper voltage levels and to isolate power from the audio path. _._ On Tue, Apr 6, 2021 at 5:01 AM Evan Hand <elhandjr@...> wrote:
|
|
Re: A simple question, I hope...
Bob,
|
|
A simple question, I hope...
Here is a general question: If Q90 is bad, will there still be a CW transmit signal? In other words, will Q90's health affect both the SSB transmit signal AND the CW signal...? The reason for the question is that I seem to have a low SSB transmitted signal (although I don't yet have a 'scope or power meter to verify this) while I have a healthy PTT (in manual mode) CW signal on one of my V6's. (Yes, I have more than one... Ha) If this is a question that should be boring to the majority of the group, please reply in a private message. Thanks. Bob ¡ª KK5R |
|
Re: Many uBitx's have reduced power on 40m CW
On Mon, Apr 5, 2021, at 09:26 AM, Jerry Gaffke wrote:
This one is is more like 37mhz, too high?Jerry, I did pick the cutoff at 35MHz in the RF Tools filter design because of the large variation in impedance right at 28HMz if a lower frequency is picked.??I have gone back and looked at a 31MHz cut off and there are then issues at 28MHz. I also looked at changing C81 to .1uf and find that the impedance is better across the spectrum of interest.? I do not know how to see if there is an impact on the rest of the amp without building it. I need to learn how to do transformers in LTspicw to simulate Raj's suggestion of a transformer in front of Q90.? I see that it is in the tutorial link that you sent. Yet again, thank you for your time and knowledge sharing. 73 Evan AC9TU |
|
Re: Many uBitx's have reduced power on 40m CW
This message got stuck in my "Drafts" box.
Raj, The added transformer might be another test in my search for the root cause of the problem.? I will need to rethink my order of tests.? Thank you for the suggestion. Jerry, I got the attribution of the capacitor fix wrong.? Thank you for the correction!? My apologies to Curt. 73 Evan AC9TU |
|
Re: Hust put together the kit and have a question
Jack, W8TEE
John: Is the tuning encoder using polling or interrupts? It sounds like polling. Jack, W8TEE
On Monday, April 5, 2021, 12:11:16 PM EDT, John Cunliffe W7ZQ <n2nep@...> wrote:
Thank you for your input, I will look at Reeds display. My problem is that the stock units frequency readout is so small and tuning is buggy it skips between 100 and 300khz unless one tunes very. very slowly in the fine tune mode. I already got rid of the original case, its too small to add more of the features I am envision to install. The original case will be modified into a signal generator lol. I bought the whole kit because it was available in the US with 2 day shipping from Giga Parts for~ the same price as from India. John -- Jack, W8TEE |
|
Re: Hust put together the kit and have a question
John
The stock firmware may still be designed that faster rotation of the encoder produces much larger tuning jumps, since the rig doesn't have another way to quickly qsy. Other firmware designers have tweaked this.? Remember this is a global rig design, at a lower price point than the alternatives. Yet many add whatever add ons they wish. The previous generations came with a simple lid, and two different color display types of scaleable size he been used. I am not sure ubitx.net is still that active, but there and other places will discuss display tradeoffs and available firmware. Evan has good advice here. Curt |
|
Re: Hust put together the kit and have a question
John Cunliffe W7ZQ
Thank you for your input,
I will look at Reeds display. My problem is that the stock units frequency readout is so small and tuning is buggy it skips between 100 and 300khz unless one tunes very. very slowly in the fine tune mode. I already got rid of the original case, its too small to add more of the features I am envision to install. The original case will be modified into a signal generator lol. I bought the whole kit because it was available in the US with 2 day shipping from Giga Parts for~ the same price as from India. John |
|
Re: Many uBitx's have reduced power on 40m CW
Evan,
That "Better" filter does look better. The original was starting to reject rather seriously at 29mhz, probably too low. This one is is more like 37mhz, too high? You're right, when evaluating a filter we should drive it from a 50 ohm source impedance. But measuring the filter impedance should be done at the node after that 50 ohms, not before it.? This can be done by plotting V(n006)/I(Vin), with R1 set to 50 ohms. The name of your wire between R1 and the filter is n006.? ?We could use V(n006)/I(R1), but it then complains when we set R1 to 0 ohms. Evaluating as above with a 50 ohm source impedance at R1 shows only a slight change from evaluating with a 0 ohm source impedance.? But changes in the filter load impedance at R2 bring on a far greater change in filter input impedance. For this reason, connecting your new filter to Q90 instead of the 50 ohm load at R2 shows a large?variation in system input impedance.? You can flip between any two settings easily with the undo/redo icons and then just hit the running man. Whether those settings are in the plot window (perhaps V(n006)/I(Vin) vs 1v/(I(Vin) )? or in the schematic window (perhaps R1=50 ohms vs R1=0 ohms) depends on which window is selected. Jerry, KE7ER |
|
Re: Many uBitx's have reduced power on 40m CW
Jerry,
I used this tool to calculate an LPF: Adjusted to the inductance values for 9 and 10 turns on the T30-6 toroids that are part of the current design to get the values from this site: That provided the filter in the attached LTspice schematic. Based on this information, I will build that later this week to see if it improves the CW response. 73 Evan AC9TU |
|
Re: Many uBitx's have reduced power on 40m CW
On Sun, Apr 4, 2021, at 10:09 PM, Jerry Gaffke wrote:
However, when evaluating the input impedance of the filter,Jerry, In the other filter evaluations that I have seen using LTspice, they modeled the source as a voltage source with a series resistance that is supposed to be the characteristic impedance of the source.? In the example that you gave in message #87433 that referenced the diplexer10mhz1.asc file in message #2504 the 50ohm series resistance was in the V1 voltage source (the Rser = 50 in the ).? I had seen it used outside of the source as I had it connected. I think I know now what you are modeling with your first schematic; you were measuring just the impedance of the input to the transistor at C80.? For that to work, you need the 1volt AC signal to be divided by the current from Vin.? To do the voltage measurements that I was used to you would add in the 50ohm parasitic to get them.? To measure the input of the filter with the transistor connected you need to move the source to the filter input and connect the filter output to C80. Thank you again, I am learning a lot about LTspice. 73 Evan AC9TU |
|
Re: Many uBitx's have reduced power on 40m CW
Evan,
toggle quoted message
Show quoted text
Yup, I missed the change on R83. Thanks for the correction. Was 10 ohms on my v3, went to 2.2 ohms with v4 and beyond. However, when evaluating the input impedance of the filter, you don't get to add a 50 ohm series resistor to the model for the mixer source impedance. The mixer expects to see a 50 ohm load presented to it at the mixer output pin. At 7mhz, the impedance seen looking into that filter really is down around 7 ohms. >? Also, not sure why the impedance does not change with the dc bias point of the 2n3904 I think it does, that does get modeled. R83 only affects small signal AC gain, not the DC bias point. Adjusting R82 does change the DC bias point, also the gain and the input impedance. Try adjusting R82 from 100 ohms to 10 ohms, that makes a significant change to the input impedance of the Q90 amp. The DC currents are getting out of hand but LTSpice doesn't care if you want to send 100 Amps through your 2n3904, LTSpice does make it easy to probe those voltages and currents when in Transient Analysis mode and check against a datasheet. When in AC Analysis mode I don't think you can probe DC currents and voltages, all you can see is the AC currents and voltages. Jerry, KE7ER On Sun, Apr 4, 2021 at 07:28 PM, Evan Hand wrote: Jerry, |
|
Re: Many uBitx's have reduced power on 40m CW
Jerry,
I edited the LTspice schematic that you sent me changing the C200, C204, and R83 values.? C200 and C204 changed to 100pF.? R83 changed to 2.2ohm per the v4 and v6 schematics.? I also added a 50ohm series resistor to represent the mixer impedance and connected the voltage source to the LPF and the LPF to C80. I have attached the corrected LTspice file. When I do the analysis that you suggested it seems that the input impedance is between 56 and 71 ohms for the range that I have been looking at (3.5 to 10.5 MHz).? Looking at the voltages at both the input to the filter and the output of the filter seems to match what I have seen for both in the real world, at least from a relative perspective for the voltage at the input to the LPF. I also tried looking at the filter with a 50ohm input and 50ohm output with both 300 and 360 nH inductors.? The model does not really show much difference. I have been using AC analysis with the LTspice models.? I was not aware of the ability to change the measurement scale to read ohms.? That is a cool trick. I am still not convinced that the model is accurately following the real world.? It seems close yet cannot explain the significant drop in power.? Also, not sure why the impedance does not change with the dc bias point of the 2n3904.? I assume that is because we are only looking at AC characteristics and the DC values are not being modeled. I still plan on putting my v4 back together and changing C81 to be 0.1uF 73 Evan AC9TU |
|
Re: Many uBitx's have reduced power on 40m CW
The input impedance of the amp at Q90 is much better behaved
toggle quoted message
Show quoted text
with Curt's larger cap at C81 (51 to 45 ohms) than it is with the 470pf cap ((200 to 60 ohms). On Sun, Apr 4, 2021 at 06:56 PM, Jerry Gaffke wrote: So apparently a number of ways to fix it. |
|
Re: Many uBitx's have reduced power on 40m CW
I had the wrong value for C81 when writing up that last post.
toggle quoted message
Show quoted text
With 470pF there, impedance phase is not exceptional at 7mhz and 18mhz. However, impedance magnitude at 7mhz is at a minimum, around 7 ohms. Up around 150 ohms at 14 mhz. So apparently a number of ways to fix it. But my guess is that the root of the problem is in the filter at L1,2,3,4 Jerry, KE7ER On Sun, Apr 4, 2021 at 06:28 PM, Jerry Gaffke wrote: Having edited C100 and C104 in my simulation to 100pF, the impedance of filter plus Q90 |
|
Re: Many uBitx's have reduced power on 40m CW
Interesting.
toggle quoted message
Show quoted text
Moving CW-KEY to unbalance the second mixer at T4 instead of the first mixer at T2 does fix the issue of low power out on 7mhz CW. The nine bands are grouped into four transmit low pass filters as: ? ?3.5 5.3? ---? 7 10? ---? 14 18? ---? 21 25 28? ? (in mhz) Since 60m is at the high end of the 80m LPF,? it is starting to see some attenuation. And 17m might well be down some in power for the same reason. The low power on 7mhz CW is definitely the outlier here. Having edited C100 and C104 in my simulation to 100pF, the impedance of filter plus Q90 looking into L4 from that first mixer varies between 30 and 80 ohms.? Not too bad. Impedance at 40m is around 35 ohms, and probably a bit higher than it is for 80m and 60m. But the phase of that impedance on 40m is around 180 degrees, a maximum. There might be a hint of lower power on 18mhz CW than on SSB in that first test, perhaps 18mhz CW has a problem similar to 7mhz CW? My simulation shows a second peak in the phase of the impedance at 18mhz,? similar to the one at 7mhz. Of course, the simulation assumes ideal parts and no board layout issues, there may be more going on. But it does seem that unbalancing the first mixer is disturbing how the mixer drives L4. When Evan replaced L4 with a 50 ohm resistor to ground, he got a very clean 40m signal there. So it may be an interaction between the unbalanced mixer and the out-of-phase filter impedance when operating a 7mhz, not just a problem of one or the other. Curt's extra capacitance at C81 seems to fix this, though the simulation does not show much of a change. John's unbalancing of the second mixer seems to fix it, but would be a significant hack for some. A slightly different filter design may also fix this. Jerry, KE7ER On Sun, Apr 4, 2021 at 05:12 PM, John (vk2eta) wrote:
As expected, CW and SSB output power are identical and vary with the bands. Eg 11 watts at 40M for both CW and SSB. They exhibit a small dip at 60M which is seen across all my uBitxs on SSB.? |
Bob Lunsford
to navigate to use esc to dismiss