Topics

uBitx V3 low power continues

Evan Hand
 

I used 10x probes, direct on the TP.

David Forrest
 

Is that 1ms/div? It looks like your dial will go down to 0.5us/div, which could show a couple cycles of a 7.2MHz wave in a division.

But voltage-wise with 1mV/div and 10x you’re at about 20mVpp and  50mVpp  versus the reference traces at 672mVpp and 1.96Vpp.  

I’d look upstream.

Dave


On Nov 6, 2019, at 9:55 AM, Evan Hand <elhandjr@...> wrote:

I used 10x probes, direct on the TP.

Evan Hand
 

The scope traces are dim, so not sure if I am reading correctly.  To me it looks like the TP3 signal is stronger than the TP4 signal.  As David suggested, increase the scan rate of the scope (reduce the horizontal trace time / div) to get a better picture of those two points.  I would also check the bias voltages on Q911 and Q912.  That could be your issue.

73
Evan
AC9TU

 

Hi Ted,

Scope triggering should be set to "Auto".  Since your signal is in the MHz range your sweeptime needs to be a lot shorter.  If you had a 1MHz signal, the time of one full sinewave would be the inverse of that -- in other words 1uS.  So if you set your sweep time to 1uS, then a 1MHz cycle would show one full cycle per division of the scope screen, and with a 7MHz signal you would have 7 full cycles per division.  This would be a bit hard to see, so you want an even shorter sweeptime.  The 0.5uS as suggested.  For measurements, make sure the red "SWP Var" and "Var Volts/Div" knobs are all the way clockwise.

Where did you measure the Q90 signals?  The input signal would be on the base (not emitter) and I would take the output signal from RV1's wiper side.

Hand-help scopes are crappy compared with what you have now unless you spend megabucks on one (and then you might as buy a better used analog scope).

73,


Mark

Ted
 

OK.  Now that you fellows have reminded dad me of what I've forgotten over time, let's revisit some of those points.  Scope is set to .5ųs/div. The probe is 1x, as I'm getting nothing while using the 10x option:




Here is TP 3 (RV1 wiper) @ 20ųv/div:





TP 4 (C90) @ 20ųv/div:




TP 5 (R95) @ .2V/div:




TP 6 (R92) (ibid):




TP 7 (T11-5) @ 2V/div:



Please note that I have a pot on the CW level and have that down by about half. Here's TP 7 at full output (same settings):





Aaannnd, here's an approximate 600-Hz whistle into the mic (same settings):




This will hopefully give a better impression of where things are.


With much appreciation,

Ted
K3RTA

Ted
 

Here's TP 2 (C80 at Q90) @ 10ųv/div:


Power supply hum acknowledged. Battery power was used for TP 3-7 measurements.

And here's the collector of Q90 at T8, with lots of jitter (which disappears by the time the signal goes through RV1):

Ted
 

Problem located.

The scope pictures posted above showed no gain out of Q90 and I missed that obvious failure.  It seems that R83 measured out at some 1.8K when it's supposed to be 10 ohms.  I did wonder why the emitter of Q90 had close to the same scope reading as the base.  Now, with a proper R83 in place the waveform on the emitter is present, but low; meanwhile, the output of Q90 is now quantum levels above where it was.  RF out of the radio on 80m, 40m, 30m, and 20m are (about) 13, 12, 10, and 8 watts right where they ought to be.

Incidentally, I shoved a can-type 2N2222 into the Q90 spot some time ago to see if it would add the same benefit that changing out the 6 driver transistors gave. If anyone has a better suggestion for the application, please do tell.  Whether that swap "caused" R83 to open up is anyone's guess.

Thanks again to each and all. 


73,

Ted
K3RTA

 

Glad you found it!

"Whether that swap "caused" R83 to open up is anyone's guess."

I'll take that bet hihi.  I always look at the last thing I did when I have an unexpected problem.  Sometimes this works with the XYL too.

Ted, for getting a better view at close-spaced waveforms you can use the x10 Mag control to cut the sweep time by 1/10th.

73,


Mark.

Michel Dupuy
 

Thanks Ted, a few days ago my output power had gone from 12W to 1W and reading your message this morning I checked and indeed the resistance R83 of 10 Ohms were cut off, I replaced it and I retrovated the nominal power. Indeed transistor Q90 instead of amplifying attenuated the signal. Thank you from F1GTX

Ted
 



Michel,

I'm rather pleased that an obscure find such as this, helped someone else.  At least one or two transistors bravely gave their lives in the search for this fault, as weak output led to an assumption that Q90 was the issue.  Best wishes always.




Mark, 

"Denial" is a sneaky adversary in these pursuits; I mean, I'm not entirely sure what the purpose of that capacitor + resistor is, in parallel with the bias resistor on that leg of Q90. There's no denying that its important!

Having a strip of deer-tick sized 22-ohm resistors from the 2N2222 driver stages mod, I may just pair a few up and replace the R83 with a heavier "resistor" as a precaution. 



With appreciation,

Ted
K3RTA

 

Looks like I lost the bet!

It sounds more like a run of defective resistors maybe?  I can't imagine there is so much current that it exceeds the rating of the resistor. You could do a check for heating after a 30s TX or something. Modeling could show the current through it. 

Cheers!

Jerry Gaffke
 

The R82 collector resistor to ground (along with bias resistors R80 and R81) determines the quiescent DC emitter current through Q90.
Resistor R83 increases the gain at Q90 for AC signals with no affect on that quiescent current, since DC cannot go through C81.

Each stage of that power amp is increasing power by something like 10 or 15 dB,
so RF power generated by Q90 is down around 5mW.
And the RF power dissipated by the 2.2 ohms at R83 is an order of magnitude down from that 5mW.
R83 should not be getting hot.

Perhaps this was a stress fracture (if internal you would not see it even with a microscope), or a cold solder joint?

Jerry, KE7ER



On Thu, Nov 7, 2019 at 02:49 AM, Ted wrote:
I mean, I'm not entirely sure what the purpose of that capacitor + resistor is, in parallel with the bias resistor on that leg of Q90.

Robert Freeman
 

Well done Ted!  This was helpful to me as well. 

73s de NQ0T, Robert