Upgrading the PA output transformer

JerryW0PWE notes that in the posts on boosting and/or leveling output power he saw pictures of a binocular core being used as the output transformer in place of the toroid that comes with the uBitx board.  He asks “What size is that binocular core?”

Allison plans to use the longer version of the BN43-6802 (or two BN43-302 cores end to end).

John VK2ETA says he went back to IRF510s in the finals and has used a BN43-3312 as the single 302 was getting warm on the low frequencies and he could not put more than 2/3 turns in the core.  He tried both 1/2T and 2/4T in the 3312 and settled on 2/4T due to better efficiency. The 1/2T drew too much current at 80M.

There is plenty of gain and power at 14MHz and below on 16.8V for the PA but has had to resort to a gain control for flattening the power curve despite 330pF caps in all 6 drivers emitter resistors, 22uH inductor in the base feedback of  the pre-driver,  330 pF across the primary of the finals transformer and 820ohms in the finals feedback resistors.

Nick VK4PLN uses the BN43-202 with a winding ratio of 2:4 using 0.64mm enameled CU.  His rig is outputing 20W+ on 80/40/20 using RD16HHF1.  Nick doesn’t detect any heating… Maybe it will at that power level on 10m?

His plan is to now try using:

– the MPSH10 in the pre-driver
– a single 10T bifilar wound FT50-43 for feeding the power to the RD16HHF1s
– removing the PA feedback loop
– and other pre-driver mods as per Farhans suggestions….

This appears to be the common approach on most of the RD16 amp designs out there…and Nick hopes he will get 20w+ output from 80-10m….

Bill K9HZ tried a BN61-002 and it seems to be a winner.  It doesn’t heat up at full power and it seems to be extremely wide band.


Fake PA FETs

Watch out for fake RD16HHF1 devices.   Buying on Alibaba and Ebay is a risk some are willing to take, but there are plenty of fake devices for sale through these channels.


AE7TO experience with evening up power output

Mark Cantrell (AE7TO) has been having fun getting his uBITx to have more even power out.

All this experimentation was to allow Mark to get enough power out of his µBITx to make up for low efficiency portable antennas.  He didn’t want to carry a separate power amp.  He just needs to build a 24 VDC battery pack now!

His rig can be seen in the photo above.   At the left hand side (with controls incorporated on the rear panel) is an Antenna Tuning Unit and in the middle is a new daughter board.

Step 1

The new daughter board implements the relay-based band output power leveling mod designed by Bill (K9HZ). 

This mod helped quite a bit at 10 and 14 MHz.  Here is a summary of power levels, as observed by Mark, on each band after the mod:

  • 7 MHz and lower frequencies = 11 watts out (no change from before the mod
  • 10 MHz = 8 watts out (twice was I was seeing before the mod)
  • 14 MHz = 8 watts out (also twice what I was getting before the mod)
  • 18 MHz = 6 watts out (never checked that band before the mod)
  • 21-28 MHz = 3-4 watts out (didn’t test that band before the mod)

Mark didn’t bother adding a 4th relay for tuning as he usually uses an antenna analyzer and manual tuner anyway.

A 3D model file is available for the plastic bridge built to hold the daughter-card above the main uBITX circuit board.  Someone else may find that file useful.    

Step 2

Mark split off the PA power line and fed it 25 VDC (two small lead acid batteries in series).  A buck converter was used to reduce the 25 volt main supply to 13.5 VDC for the rest of the uBITX main board.
That gave 25W at 3.5 to 7 MHz, 12W at 10 MHz, 14W at 14 MHz, 5W at 21 MHz, and 4.5W at 28 MHz.

Then he added a 33 uH inductor in series with R86, and 220 pF caps in parallel with R87 and R88, as suggested by Howard WB2VXW.

Here is a summary of power levels after the above changes (plus replacement of RV1 with the 3-relay/4-pot mod discussed below):

Freq / Power Out / Current Draw
  • 3.5 MHz / 28 W / 2.3 A
  • 7 MHz / 28 W / 2.2 A
  • 10 MHz / 16 W / 1.4 A
  • 14 MHz / 20 W / 1.5 A
  • 18 MHz / 11 W / 1.1 A
  • 21 MHz / 7 W / 0.7 A
  • 28 MHz / 7 W / 0.7 A

These numbers are good enough that Mark no longer feels the need to fiddle any more.  In particular, he is not going to change the bias settings on the finals as they seem well adjusted from the factory.   Mark has turned back the power output on the low bands to give no more than 20w out to protect his finals.

VU2ESE power mod for 10m operation

Ashhar Farhan VU2ESE notes that 28 Mhz has, unusually, been open for the last week.  As a result, he realised that the ubitx output was woefully low on 10m.  Hence the experiment below:

Ashhar suggests the following final fix, and asks for others to try it out and see if it is replicable:

Replace C81 from 0.1uf to 470pf
Replace R83 from 10 ohms to 2.2 ohm
(you can short R83 as well)
Replace 97, R98 from 47 ohms to 220 ohms
Remove C261, C262.
So, what happens is that removing the C261 and C262 increases the gain of the finals. They are run open. Hence greater gain at 28 mhz.
However, the gain is very high at lower frequencies. So, in order to reduce the gain at the lower frequencies, the 0.1 uf cap is replaced by the 470 pf. As the frequency of the signal drops, less and less RF flows through the 470 pf, decreasing the gain of the predriver. 470 pf is not a magical value, 220 pf works almost as well.

Step #1 Increase the predriver gain towards the higher frequencies

The predriver Q90 has a emitter degeneration capacitor C81 (0.1uf) and and R83 (10 ohms). Replace C81 with a 470 pf and R83 with 2.2 ohms. Altenatively, short R83.

With this, the emitter reactance decreases with increasing frequency, yielding higher gain beyond 14 MHz.

Step #2 Take off the feedback from the IRF510s.

a) Replace the existing  R97 and R98 with a value of 47 ohms  with 220 ohm resistors.
b) Remove C261, C262.
Reference #1
Reference #2

Photo of mod for evening up power output

Nigel G4ZAL did Howard WB2VXW ‘s mod by adding:

  • 27uH inductor in series with R86 (NB calculated value)
  • Adding parallel caps of 220pF on top of R87 and R88.

The mods to the main board are shown in the photo below.

The power output changes that resulted were as follows:

  • 7MHz from 12.5w to 14.5w
  • 14MHz from 7w to 11.5w
  • 21MHz from 4w to 8w
  • 28MHz from 2w to 4w

Nigel says it is a “very worthwhile mod for just 3 components”.  This mod is something that all constructors should consider.


Evening up power output on high bands

Howard WB2VXW received his µBitx, assembled it, and started testing.  He performed a number of modifications to the µBITx in his lab before moving it to its permanent home in the shack.

In particular, he modified the transmitter to improve output power at the higher frequencies. The power out in the original state is about 3 watts at 28 MHz.  His modification brings the power at 28 MHz up to 7 watts on both the 15 and 10 meter bands. On his board, there was not enough range in VR1 to increase the power to anywhere near this level without modification. The other bands remain unchanged in output, at around 11 watts.  The output stage could be modified as others have done by changing the FETs and transformer to improve this further, but Howard was happy with the improvement for now.  He plans on adding a power amplifier later, something between 50 and 100 watts.

Howard’s fix involves adding 3 passive parts:

  • 33 uH inductor in series with R86. I lifted the resistor mounting it on one pad standing up, and teepeeing the inductor between the top of the resistor and the other pad.
  • Adding a 270 pF capacitor across both R87 and R88.

The values are not very critical, I calculated that I needed 27 uH and 220 pF, but the values I tried were in the lab.

The theory is very simple. The closed loop gain of Q911 and Q912 is set by the ratio of R86/R85, or a gain of 10. Adding an inductor into the path in the feedback loop increases the effective impedance at higher frequencies, adding “peaking” to the circuit, thus making the closed loop gain higher.

I think the poor frequency response is in the transformer as well as the transistors Hfe falling to under 25 typically at 30 MHz.  The open loop gain is also increased by bypassing the emitter resistors for higher frequencies.

The transformer is a 2 to 1 step down, (it looks like a trifilar winding) so the theoretical gain of the stage is 5.

Offer of parts

Howard also made a generous offer to  anyone wanting to try the fix and needs the parts.  He has a full reel of each of the parts (2x 270pF capacitor and 33µH inductor), so just Howard a stamped self addressed envelope, and he will send it to you by return mail. The offer is good until he runs out of parts.

Others try it on

Joel N6ALT completed the mod not long after it appeared on the BITX20 IO Groups list with spectacular results. Before the mod he had only 300mw at 13V on 28.500.  After the mod he had 3.8 watts, and after increasing the drive slightly, he now has 4.5 watts. All of the other bands also benefited from the mod.

Tim AB0WR also tried the mod, winding an FT37-43 ferrite to get 33uh. Not as neat as an smd inductor but he didn’t have any in the shack!


Flatten the power curve and set your power out in software

John VK2ETA  has some pretty good ideas. A few days ago he worked on an AGC system using the first IF stage (at 45MHz) to control gain at the front end of the system.   He has now got some pretty good results using a limited amount of memory in the firmware to flatten out the power curve on transmit.  At this stage it only works in voice and digital modes (LSB and USB). Because CW is achieved by unbalancing the first mixer after the IF filter, this approach will not work on CW.

By shifting the IF frequency on the filter, John is able to produce a fixed attenuation that keeps the output power within the required limits.

The process is two fold:

  • a one off calibration exercise.
  •  a menu item that selects Low, High or Max power.

John has used 5W and 10W as targets for the Low and High values.

His results (remembering he has modified the final stage with RD16s and a number of other changes have been made to his transceiver):

  • 80m to 10m in low power settings have variations between 4 and 6 watts.
  • 80m to 10m In the high power setting has a range between 9 and 11 watts.
  • Only 16 bytes of data points are required (could be stored in EEPROM if desired).

His ATU uses a 2nd Arduino.   He has used a spare digital output to send a 1500Hz tone, low-pass filtered, to the microphone input for calibration purposes.  The audio filtering plus the 2nd IF filter take care of the audio harmonics and the signal is clean. This could be done in the Raduino if digital lines are freed up, such as by using an I2C display.

The tone also provides a tune up facility in low power.

He can transmit in digital modes in low power mode without having to adjust the drive every time he changes bands.  Perfect!

John will update the code in the file section so that others can incorporate this new feature into their firmware if they desire.

John acknowledges Jerry KE7ER’s idea of using fixed tables to perform the High/Low calibration function.

Firmware uploaded to files area

John has uploaded an update of the KD8CEC based software (V1.04 based) for Raduino and ATU Arduino.

Key changes in Version 20180411:

– Made the menu system dynamic so that items can be inserted at compile time or can be context-dependant at run time.
– Added output power attenuation for 80 to 10M using first IF shift.
– Added new menu item “Power Level”: low/high/max for SSB modes.
– Made CW menu items display dependant on having CW modes selected.
– Reduced the level at which the software AGC comes in. Adjusted the correction of S-Meter when software AGC comes into play.
– Changed tone generation for ATU tuning sequence from CW to using SSB with the 2nd Arduino generated tone.
– Change tune sequence to

a)Select the “Low” power setting before performing a tune and
b)Change tune frequency to tune on carrier (offset by audio tone frequency).

Assumes the following hardware change (only if using an ATU): a digital output on the ATU Arduino is used to generate a tone. An audio filter identical to the Raduino CW audio tone filter is used and it’s output is connected to the mic input of the SSM2167 module. If digital outputs are freed-up on the Raduino, for example by using an I2C display, the same can be implemented with only the Raduino.

This allows a low power tune function.


Drive control – 2 methods

Bill K9HZ who we have already reported had developed a relay control system driven by the Low Pass Filter I/O lines.  In addition, Glenn VK3PE has also come up with a circuit that could be adapted using a digital potentiometer module to be driven by the Raduino using a digital I/O line.

Relay drive control

Bill K9HZ has now drawn up the circuit he used to allow for relay switching of drive levels by band grouping, driven by the LPF band grouping I/O lines of the Raduino.  No firmware mod is required to deliver even drive power across all of the bands.

RV1 in the yellow circle is the existing drive pot in the uBITx.  Remove it and connect the wires from the relay as shown.  Q17C, Q18C, and Q19C references connect to the transistor Q17, Q18, and Q19 collectors.  Then set RV1A, RV1B,  RV1C  and RV1D on the basis of the appropriate band grouping selected to keep the power out flat.


Digital drive control

Glenn VK3PE has come up with an old circuit that could be used to work with a digital potentiometer module to use as adrive control set by the NANO  firmware on a band by band basis.

The original drive control, RV1, would need to be removed and a few parts added, along with a digital pot (I2C control) in place of R3 below to form an attenuator in the RF path.

With some careful work it should be possible to design a small PCB that fits into the holes vacated by RV1, either vertically or horizontally.

The schematic is from page 62 of the book by Randy L. Henerson on designing a Transveiver. Its a very old book (1997) ISBN 0-07-028263-3


Power curve flattening in the driver and pre-driver stage

Glenn VK3PE is attempting to flatten the power curve from the µBITx by playing with the driver and pre-driver stages. 

He has completed his mockup of the PA driver stages and

  • used MPSH10’s in all stages.
  • Used FT37-43 toroids all wound with 10 turns 0.25mm wire either tri- or bi-filar according to original Sch.
  • Bypassed all emitter resistors with a 220pF cap.
  • swept from C80 input to T10 secondary (as wound)

The result is a very flat response: better than 1db from 3Mhz to 54MHz.

Glenn plans to mock up the finals now to see what the net impact is on the power response curve.  If it is nice and flat he will replace the parts in the µBITx and we will have a solution that everybody who wants to flatten the power curve can use.