Look at V4 and latter and the mods incorporated along the way to V6.


Allison
------------------
Please use the forum, offline and private will go to bit bucket.

Wonder why they died?? ?

First bet is a impulse more than 20V subjected to the gates.
If you look those devices are not subject to much current.

Allison
------------------
Please use the forum, offline and private will go to bit bucket.

Mine and other have reported the corner blocks are anodized aluminum.

It was one of the first things I did.? If they had been plastic I'd have
made up brass strips to connect across the corners.


Allison
------------------
Please use the forum, offline and private will go to bit bucket.

I recall, and someone should confirm, that those screw blocks are made of plastic and that the 2 side panels are consequently not electrically connected to the rest of the box.

Hi Folks,
??????????? Yesterday I had some time to work on my sBitx, mounted a fuse holder, removed the DC power connector, ran heavy duty figure eight cable through fuse to main PCB, used countersink bit to remove powder coating on outer case, removed oxide from the small mounting blocks to improve earth of the radio. Next removed flash drive, did the git update on a spare Rpi , cut track and jumper tx control as per Ashhar's modification. Now working without the peak burst in current I was seeing at switching on/off, will test more today. Thanks to Ashhar & HF Signals for a fun radio.

73 Alf vk2yac

Allison this is for you and anyone else who wants to chime in. The changes to the transmitter driver section have been discussed and from what I gather the Q11.12.13 chain should be like the ubitx V3-6 series. Q15 and Q16 changed to a single 2N2219 with a 4:1 transformer in place of L8. Allison, you said about changing R7 as it cooked things. Do you mean R57?
Thanks
K3bo

开云体育

Inter-process communications (IPC). ?There are several ways to do it in Unix. ?I don’t have experience with it, but time to learn! ?:-)

Allison,

I pulled Q10, the 2n7000, and the SMD 2n7002.? Both tested bad as 200ohm resistors SD.? Q9 tested good.

Now to try to work out what could have caused them to fail.

73
Evan
AC9TU

Allison
?? I wasn't commenting on speed. If you spawn a thread to handle an event you may still have to communicate the result from that thread back to the main process. You could, of course, spawn another thread to watch for that thread's message etc. (turtles all the way down).
--
73
??? Bob? KD8CGH

uBitx band filtering is bi-directional and the same in both directions (RX/TX).

The usual noise blanker (Lamb) samples RF/noise from the antenna and if?
Pulse in nature it operates a FAST RF switch to interrupt the signal for a
short time usually befor ethe primary selectivity..? Hence the name blanker.? ?
There are designs on the net. Generally the biggest difference between
them is the devices used to switch?usually diodes or can be transistor
or fet.? IT is useful for pulse noise but not hash from things like
switching power supplies or other electronic crap.


Allison
------------------
Please use the forum, offline and private will go to bit bucket.

The process chain is in the very core of the kernel and as such highly tuned for speed.
Its not all turtles as if priotitized (see NICE)? but you have finite CPU (granted not
at all a slug) and many things going on.

Its as much a discusstion of good software design and really the SBITX is emerging
design so potential is there but some bugs as well.? Compared to some SDRS
it has a lot of CPU and in many cases far more than others.? An example is the
IC7300 which uses a FPGA with custom coding to do the high speed lifting
with a second cpu to do the slower stuff.? The SBITX that RPi brings a lot of
flexibility including enough power to build he code on the box itself.
That and the code going in is all open to alter/improve.

Think of the active things while receiving:
You have an active data stream (serial baseband audio from the RX) not much
delay permitted there as it 96kbits/s sample rate.
Then you have Wired for Wifi networking
and a list of general housekeeping as well but their urgency is lower.

The whole CW discussion is you close the key, 1mS later its acknowledged
and the TXsoftware switches to TX and then generates a tone for as long as
the key is down.? There is delay there and faster cpu or less fat OS does not
change that.? ?This is common to ALL SDR radios.

Oddly for SSB and data modes none of that is much a problem.

The Rpi compares well to the average winders box running a firebreathing
cpu and all but burdened with Winders11.? I use a Rpi3b+ as a HB laptop
and a Rpi4B as a desktop, to me its linux and plenty fast.

Elecraft took a page from many to create some good radios.

Allison
------------------
Please use the forum, offline and private will go to bit bucket.

Evan,

Mine has that 2n7000 on the bottom, I believe it bridges the SMT part,

The power sequencing is not greatly helped by the software mod as
that is only useful when the radio software is running.? For straight
power up that has to be a delayed or forced off.? ?With the digital?
board in that should be off.? THere is a 10K on the gate of Q10
to ground so off, should be off.

I've added a mod to delay the drive to Q10 using RC and a
additional transistor (2n7000) to force the DC 12V switched bus
to ground when Q10 is OFF.? Reason for that is even if Q10 is
momentarily "on" a bunch of caps reated to that bus ar charged
and the amps then get bias if only briefly.


Allison
------------------
Please use the forum, offline and private will go to bit bucket.

Charudatt,

Thanks for the suggestions! Will make give that a try. The rotary code in KD8CEC is not as isolated as I would like. So I have been hesitant to tackle it until I have plenty of quiet time. But I can at least make sure the GPIO pins can be used as both Analog and Digital so on the next spin it is only a software change.??

It is interesting that the "real" raspberry pi pico" only has 3 ADC where as the RP Connect Arduino version has the full complement of A0-A7...

The new processors do open the prospect of web remote control since they often have a second processor for the wifi/web. Would just have to figure out how to get the CAT commands to the onboard webserver and then loop back to a serial port.?

73
Mark
AJ6Cu

David,

The filters switched in the uBitx are low-pass filters (LPF).? They reduce any signal above the desired frequency, like unwanted harmonics generated by the transmitter.? Most AM interference is below the HF bands.? There are exceptions overseas.? You will still need a broadcast interference (BCI) filter if you are close to a commercial AM station.? The same would be true for the sBitx.

Note that there are more filters in the design for both radios.? These include a 30MHz low pass filter used by both the receiver and the transmitter and crystal filters used to reduce the bandwidth that must be decoded.

73
Evan
AC9TU

I removed the finals. One tests OK the other shows a dead short amongst all the pins.
?
With the final removed I measure 0.45 volts on the gate of Q10 - I presume this is a low.

The input current measures 1.0 amp with the digital board in and 480 ma with the digital board out.

I do not plan on installing new finals until this problem is resolved.?

Lou

The TX button and \t command do not appear to put the sBitx in transmit when in LSB mode. They do seem to work in DIGITAL mode. I was trying to use them with ?mic that did not have PTT.

开云体育

Thank you, Allison.? I think I understood that feedback oscillations could be causing the board failures.? I did not have an issue before making the board mods.? I did afterward.? I chalked it up to not having done the SW upgrade before applying power.? I am not sure that was the cause.? I removed and checked the IRFZ24Ns, and they both check out good.

I have noticed that my board has another MOSFET (2n7000) through hole soldered under the board in what looks parallel to Q10.? See attached pictures.

I did some voltage checks, and it looks like Q9 and Q10 are on without the RPi installed; at least, I am getting the 12volts to the bias circuits.? I adjusted the bias controls, and the current went down, so I guess I will need to replace Q9 and Q10.? I will need to order replacements as I do not have either one in my supply.

73
Evan
AC9TU

Does pin 36 from the PI not extend down to the main board?

In other words, the trace for pin 36 does not get connected to the main board already so we have to tac the wire directly to were pin 36 from the Pi is on the digital board?

Evan,
The sequencing mod is for control output of the Rpi so if the pi is not up
it has no effect.

What I believe is going on is the power amp chain on first power experiences
momentary feedback, high power oscillation due to not enough isolation
from out put to input.? The State of K9 and U3 are unknown at power up
and likely the wrong state if only momentarily.? Further worries is that
just having the input of the power amp near the output opens an
opportunity for oscillation.? There is a lot of gain there (more than 60db
mid band) so even chance coupling can meet the criteria for oscilation.

Farhan, tested by removing C14 that would break that potential path and?
the startup high current pulse I've seen goes away as well as TX.? Of
course with C14 out RX is deaf.

The side effect of all this is the finals get pushed way beyond their
capability to survive and Both Farhan and I believe the ability of the
drivers in this case to provide more than enough power to break
down the gate oxide of the finals, instant dead.

The sequencing I went with a hardware solution that seems to behave
better than software as its not reliant on the RPi.? I still feel its less than
ideal (may impact CW keying)but when we fix the oscillation problem
I may modify it (shorter time constants).? For now it does not blow up
at power up.? Notes and schematics are earlier in this thread.

What I'm doing and I'd encourage others to inquire and ponder it.
Layout based issues if any and RX/TX RF path switching.?
We sort this out and make it reliable we have a really powerful
radio.


Allison
------------------
Please use the forum, offline and private will go to bit bucket.