Modification of 260..280W DVB-T PA modules for 70 cm use

When utilizing four PA modules along with two 1:4 combiners/splitters, the overall cost is €181,-.

This setup delivers an impressive 1+ KW on the 70 cm band for just €181/US$190,-!

ALL INFORMATION ON THIS PAGE IS WITHOUT ANY GURANTEE FOR CORRECTNESS !

Quick links to sections on this page:

Background

EBay frequently offers affordable PA modules used in the 470 to 860 MHz band for DVB-T and analog TV.

These modules, often priced under €40 each, come from disassembled amplifiers once part of Plisch's ULE1021/Tristan models.

Each complete amplifier originally housed eight modules (four on each chassis side) driven by two driver modules (also often available on EBay). The output from both sides combined externally with a large hybrid combiner utilizing 7/16 connectors.

The water-cooled amplifier featured a 3-phase power supply (32V @ 70 Amp).

Modified versions of the amplifier for 70 cm use are predominantly found in Europe.

The amplifiers also came with four 1:4 combiners, also available at reasonable prices on EBay (around €12 each).

Some Hams uses these combiners to combine four modules, you can even split one of the 1:4 combiners so you get two 1:2/2:1 combiners, allowing you to combine two of the PA modules (for 500+ W output).

Many Hams have acquired one or more PA modules from these amplifiers and modified them for use on the 70 cm band, achieving output levels of around 250 to 270W per module.

Some Hams have done modifications to the PA Modules and documented their efforts and results on the internet. I tried some of the modifications, but did not get optimal results, so I experimented some more (more details below).

When looking at these modification pages, it is VERY IMPORTANT to know that there are different versions of the PA module PCB. There is (as far as I know), the VEM1050A, VEM1050A2, VEM1050A3 and the one described here, the VEM1050A4.

A big "thank you" to John DK9JC and Joshua DL3JOP for brainstorming with me during this project.

Also a "thank you" to Verner OZ5TG for help with confirming my measurements were correct.

Below is a list of some of the pages I have found (there are probably more):

DL2OCB (including comments by DK4RC): https://dl2ocb.de/umbau-plisch-ule1021-endstufe/

PA0EHG: https://www.pa0ehg.com/70cmpa.htm

DB6NT: https://db6nt.de/wp-content/uploads/2024/12/Umbau_der_Plisch_ULE1021_Endstufe_fuer_432_MHz.pdf

DG9BFJ, modifying power combiner: https://www.emptron.de/70-cm-leistungsteiler-oder-combiner.html

The design

The PA module uses two BLF861A LDMOS transistors from Ampleon (former NXP). The datasheet can be downloaded here.

Each of the two LDMOS devices are able to work up to 860 MHz, each with an output (CW) power of 150W at 32VDC supply voltage.

The two LDMOS devices are combined at the input and output with splitters/combiners on the PCB. On the input side a 50 Ohm resistor is mounted at the isolated port of the splitter. This resistor is mounted "inside" the module. On the isolated port of the output combiner, a 50 Ohm resistor is also needed. This one is a bit bigger (150W) compared to the one at the input (10W) and is mounted "outside" of the PA module/enclosure.

If you use four modules with the combiners (as shown on the picture at the top of this page), three of the four terminating resistors on the outputs are mounted inside the enclosure for the output combiner. The last of the four terminators are mounted to the heatsink using a spacer (to lift it up so its terminal is flush with the top surface of the PCB of the PA module).

The PA module comes as shown on the pictures, it is enclosed in an aluminium housing with a lid. The lid is secured using five screws. The PCB is mounted (and soldered) on a thick heat spreader. The complete module should be mounted on a suitable heatsink (to get an idea about what is needed, you can have a look at the calculations on this page I made).

Mounting the module to the heatsink is done using eight M3 screws. I drilled and tapped the heatsink for these. Make sure that the heatsink you use is completely flat (this is VERY important!). Further down on this page, you can see the spacing of the holes needed on the heatsink, the dimensions indicated are the best I could measure using a caliper.

The PA module is originally designed to cover a broad frequency range of 470 to 860 MHz. In order to optimize the PA for operation in the 70 cm (432 MHz) band, a few components needs to be removed/changed on the input and output side of the LDMOS devices (for "our use" no need for the PA to be that broad banded).

Supply voltage for the module is 32 VDC. Max current are around 11 to 13 Amp.

The idle current (no RF on input) can be adjusted by two trimmers (accessible thru the lid of the box), at default these are set to approx. 2A for the complete module (1A for each LDMOS) for class A/B operation.

As with all LDMOS amplifiers it is VERY IMPORTANT NOT TO OVERDRIVE them! Overdrive is the number one reason LDMOS stops enjoying life!

Test setup

The test setup consists of the module attached to a heatsink using M3 screws (heatsink is maybe a bit small, but has a 80 x 80 mm 12V fan blowing air to it). I used my R&S ZNLE3 VNA, spectrum analyzer, 800W dummyload, 100W attenuator (30 dB), calibrated 500W attenuator (30 dB), IC-9700, SMU200A RF generator and various "DC instruments" during the tests.

Current was measured over a, calibrated before tests (better than 1%), 0.001 Ohm power resistor (same as used in my PA Overcurrent/Switch project) using a DMM6500 multimeter.

The heatsink used for the test, is a 20 cm long piece cut from a Wakefield 125410 heatsink (1 meter long). Thermal resistance for this 20 cm long piece is 0.45 K/W in natural convection (1.2 K/W for 3 inch). Below the heatsink a 80x80 mm fan was placed, supplied with 10V.

The pictures below is of the module is before I did any modifications to it.

Used also my Dual RF Head - USB together with a directional coupler from Plisch (two couplers in the original amplifier)

The module

The test setup consists of the module attached to a heatsink (a bit small, but has a 80 x 80 mm 12V fan blowing air to it). I used my R&S ZNLE3 VNA, my spectrum analyzer, a 800W dummyload, a 100W attenuator (30 dB), my IC-9700, my SMU200A RF generator and various "DC instruments" during the tests.

Below is a picture of the module as it is delivered. The input is on the left, the output on the right. On the right you can also see the feed-through capacitor (a red clip is attached to it in the picture)

At the lower left bottom, you can spot a 50 ohm terminating resistor (labelled "JMC 50 ohm").

At the top right, you can see a 50 ohm terminating resistor that are located outside the PA enclosure, this one is attached to the heatsink (the picture below the first one shows this one better). The terminating resistor to the right (at the output) needs a small aluminum spacer to raise it high enough so it can be soldered to the trace on the PCB.

These two termination resistors will dissipate power if one of the sides of the PA fails for some reason.

Specifications of the PA module (revision A2/A3, as far as I know, specifications did not change for revision A4).

Measurements of the holes that are needed in order to mount the module to a heatsink (as good as I could measure).

Revisions of PA module

There exists different versions of the VEM1050 boards. As far as I know, there is the A2, A3 and A4 revisions. The A4 is what I have in the modules I have.

Using some detective work, I think I have found (all?) the differences between A2, A3 and A4:

The changes from A2 to A3:
- C62,C63 are changed to 5.1 pF.
- C7,C8,C27 and C28 changed from 5.1 pF to 3.0 pF.
The changes from A3 to A4:
- C64,C65,C67 and C67 have been added.
- C10 and C30 have been changed from 33 pF to 470 pF.

Changes needed

BE VERY CAREFUL! THE LDMOS GATES ARE SENSITIVE, OBSERVE PERFECT ESD PROTECTION WHILE WORKING ON MODUL!!

I looked at some of the other websites that had descriptions of what changes to make. I tried pretty much all of them, but did not quite get the expected results. I think at least some of the reason is that most of the pages shows the A2 revisions of the PA module, the modules I have are the A4 revision.

However, on the DL2OCB page linked above, DK4RC comments what he modified on the A4 revision, I tried to follow this but I did still not get the same results for some unknown (at least to me) reasons.

I have later seen another conversion on A4 modules that is close to what I ended up doing though.

On the input side, I followed most of the work that others have done:

C2/C22 (brown circles) are 3.0 pF in the original design, these are removed and two 18 pF 1206 SMD capacitors (some modules were a bit better off with 22 pF instead of 18 pF!) are mounted instead. I tested both larger and smaller values than 18 pF, but it seems 18 pF fits most modules (except a couple of modules I tested worked better with 22 pF, higher output and gain, so do some experimenting).
For the 18 pF I used is this one, KEMET part number C1206C180FGGACTU.
For the 22 pF used in some modules, I used KEMET part number C1206C220JGGACTU.

The capacitors C5 and C25 (10 pF) that sits right between the two gates of each LDMOS, each have a "piggybacked" capacitor on the top with the designators C62 and C63 (red circle). The two piggy-backed capacitors are on the A4 revisions 5.1 pF. Remove the two piggyback capacitors C62 and C63 from the board (C5/C25 10 pF that sits below C62/C63 are still mounted).

The output side needs a bit more work:

Remove C7, C8 and C27, C28 (green circles). These four capacitors are no longer used and can be put aside. They are all 3.0 pF in the A3/A4 revision (5.1 pF in A2) (high quality ATC-B capacitors!).
The two capacitors C6 and C26 (yellow circles) are also removed (and used in next step) from their original positions (they sit right between the drains of the two LDMOS transistors).
Mount C6 and C26 to the left of C9 and C29 (pink circles), see picture below the schematic.

The above means that you will have six times 3.0 pF ATC capacitors left on the table (C2/C22, C7/C27 and C8/C28).
You should also have two 5.1 pF ATC capacitors left (C62/C63).

BE VERY CAREFUL! THE LDMOS GATES ARE SENSITIVE, OBSERVE PERFECT ESD PROTECTION WHILE WORKING ON MODULE!!

A4 schematic.

Changes done to A2 and A3:

(A2 to A3: C7/C8/C27/C28 changed from 5.1 pF to 3.0 pF)
(A3 to A4: C10/C30 changed from 33 pF to 470 pF, added C65/C65/C66/C67 1 uF and 4.7 nF)

All the changes that are needed in the module (assuming you have a VEM1050A4 version!)

Placement of components (compare to schematic). Please note this is the A2 revision of the board!

Frequency sweep before modifications

I did a number of measurements on the module both before and after the modification shown above.

First I did a sweep of the module with my spectrum analyser (with tracking generator) to see the gain versus frequency. The first sweep is of the original module before any modifications had been done.

Frequency sweep after modifications

After the modifications described above, I did a new sweep of the PA Module. This time the sweep was done from 100 MHz to 2 GHz. The levels shown are only relative, but it is clear that the max gain has been moved very close to 432 MHz (peak is shown at 432.5 MHz).

When injecting a CW carrier at 432 MHz, the gain is measured to be around 17 to 18 dB (more data below)!

Harmonics

Next I did a test of harmonics from the PA module (without any filters). PA was driven with 5.4W, output was 293W, Idq was 1.8 Amp.
There were no harmonics detected apart from those in the measurement below. The 2nd, 3rd and 4th harmonics were all approximately 43 (40) dB below the carrier (due to measurement, the 3rd and 4th harmonics are probably approx. 3 dB higher than shown).

This means that a filter should be used at the output of the PA module(s), depending on the regulations in the country you are in.

Measurements of power, current and gain

I measured the power out and current consumption for two of the modules I modified.

On the graph, the X axis is the drive power (measured right at the PA input). The left Y axis is watt, the right Y axis is current (Amp) and gain (dB).

The blue line is the output power at the PA modules output, the orange line is Idd (in Amp) and the gray line is the calculated gain.

The power measured is done using a Bird 43, it was checked using a attenuator and a calibrated mW meter and is expected to be within reasonable tolerances.

The measurement where done with 32.0 VDC supply.

Both modules shows close to the same performance. Module #002 is with 18 pF at the inputs, module #004 is with 22 pF.

Module #002

Data in a table, including the efficiency(!) which is extremely high!

Data with fixed 3.6W input at different supply voltages.

Module #004

Same type of data as above (minus the Pout for different Vdd voltages).

(The 3 Pout measurements before the last one are not correct, was only estimated, PA is anyway in saturation at that point).

Carsten Groen's (OZ9AAR) personal place