Thanks Allison for this information I have noticed similar heating of the driver transistors and also T11. I see a reduction of output somewhat on the order of 10 to 20 percent with increasing heat.?
I¡¯m going to wind a new T11 using a binocular core the single core toroid is inadequate for the power levels we are seeing. I¡¯m running my UBITX at 13.5 volts but want to build another one in a grab and go case with batteries like you had mentioned.?
Just to let you know what I¡¯ve done is bypass all driver emitters with 330pf caps and added 30uh inductors inline with the 1k bias resistors. I also added a 20pf across the T11 primary this increased output and flattened out the response some at the higher frequencies. I¡¯m seeing a minimum of 4 watts on 10 meters with RV1 adjusted to 14 watts on 40 meters.?
toggle quoted message
Show quoted text
On May 19, 2018, at 11:03, ajparent1/KB1GMX <
kb1gmx@...> wrote:
First all changes are on a production uBitx not an older brassboard model.
The first findings are not about power out, I haven't fully tested.
What I did find is at 13.8V all on the 2n3904s in the power chain run very hot
with no signal.? How hot 125-130 degrees.? Measurements had the first Q90
running at 38ma and for a low level sage thats very high and for a 3904 its
near excessive.
Why is that problem?? THe bata sheet says that as we increase the device
current the HFE goes down and leads to reduced gain and, also lots of heat.
Current solution was to increase the bias resistor R81 from 1K to 2.5K but as two parts
?Its 1K +1.5K and the 1.5K gets a .1uf capacitor across that so the stage gain is maintained.
This lowers the current to about 20MA (19.6 actual).? FYI that means the device is now
dissipating 230mW of power where before it was 390mW.? The device max is 500mw
and I"ve found when you get close to that they fail.
One to the drivers.? Another noted he tried 2n2219s and they get warm.? Likely?
it was the bias level alone causing that.? ?The current?devices in mine are 3904
and they are hot at 130-135F.? Again the bias is wrong?(does not allow for
devices with higher than average HFE) and the device current is much too
high In a class AB push pull the current need not be this high.? The same
technique? of splitting the bias value so the 1K is unbypassed is used and
the same values 1k+1.5K (1.5k bypassed) are more suitable and result in a
16ma standing current per device.??
Much of this is the result of transistors being anything but uniform.? HFE
variations? for 2n3904 is around 50 to 300? and varies with device current.
This is typical.? But that also is variable across vendors.
Those changes alone have had no testing for power out at 20M and 10M yet.
That will be? the next step.? Right now I was trying to get the devices in the
range where they should behave as good as they can.
Allison
?
Skip Davis