Medium power 70cms PA

[g8gtz]

I'm just doing a Portsdown rebuild and have room for a 70cms PA probably giving around 20 watts.

What is available around that power level from 12v - are any of the surplus or Aliexpress modules suitable?

Thanks
Noel

[g0mjw]

What's wrong with a Mitsubishi brick? You can still get them.

e.g. https://www.aliexpress.com/item/32892822365.html

This search should work https://www.aliexpress.com/w/wholesale- ... ifier.html

Mike

[Pa3fbx]

Ra50 or Ra60 will give you more power than you would expect and have also good gain.

[g0mjw]

Ra50 or Ra60 will give you more power than you would expect and have also good gain.

My experience is you can get 10W out of the RA60/80 modules before the spectral regrowth becomes an issue. You can easily get 80W out CW and with the high bias, around 5A, that they need for linearity, they have very high gain, over 30 dB, so may oscillate.

That 80W of power could be terminal if driving a larger LDMOS PA expecting 10W input. Ditto if using a Lime mini with its calibration power spike - basically don't risk it with a Lime, or be extremely careful with the PTT sequencing and prey it never goes wrong.

To avoid oscillation it is really important to ensure a good input match at all frequencies. That can be done with an attenuator >6 dB on the input. It is also important to avoid any feedback routes. Most of the RA module PCBs have provision input attenuators, but some designs have included TX/RX relays or switching diodes, E.g. the DMR amplifiers I linked above. Often there is insufficient isolation between the input and output when the module is biased up for DATV. They are fine in class C as their designer intended because the gain is so much less. The solution to that is to physically remove that feedback path. I.e. take out the relays and diodes. Don't just ignore them. If you do that they make nice small PAs for not a lot of money.

Also, a low pass filter at the output is mandatory, some boards come with them, but some others don't. A filter is very easy to make at these power levels with a few ceramic capacitors and self wound coils.

Finally, completely different option is to use one of the high power pallets, e.g. from the NEC PAs, but run it off 24V rather than the usual 48V. At low voltage, they will tolerate extreme output miss-match, but as always with LDMOS, don't ever over-drive it. They will only deliver 1/4 of the power, but that's what you wanted.

Mike

[G3GJA]

I've found that they perform much better than Mike describes. Having looked at several examples of amplifiers made using the Mitsubishi modules, most of the problems are caused by construction.

They don't oscillate provided the ground plane of the PCB is bonded with a large solder tag to the module's mounting bolts. Remove the PCB's solder resist if necessary. The PCB ground plane should be screwed to the heatsink using drilled & tapped holes for M3 screws in at least three places.

The input attenuator is not need if the grounding is as above. I bypass the one built into the Enigma boards.

I do not use any switching within the amplifier's housing. Housing kits are available here: https://www.pe1rki.com/amplifiers.html and they need a fan blown heatsink of 150mm x 120mm x 50mm minimum.

Alternatively, mount the module & PCB onto the heatsink and then cut the bottom out of a diecast box and mount it over the module. Use semi-rigid coax to link to the input & output connectors. Ordinary coax with tails of braid is not good enough.

Run the fan off the switched by PTT 12v bias supply. I usually add a three terminal regulator to drop the fan's supply down to 8v. Bias voltage for the module should be set to around 4.5v.

Output at 16W RMS as measured on HP432A with HP487A head; shoulders at -41dB on 333kS/s

b1.jpg (54.82 KiB) Viewed 301916 times

Output at 34W RMS as measured above; shoulders at -32dB. Both driven by the Lime Mini and a SPF5189Z buffer amplifier module from Amazon. Note how the noise floor of the Lime shows as a 10dB hump around the centre frequency.

d1.jpg (56.22 KiB) Viewed 301916 times

Exactly the same conditions as above, but this time using the Pluto instead of the Lime. Note the absence of raised noise floor and the 1.5dB better shoulders.

e1.jpg (68.99 KiB) Viewed 301916 times

Based on this performance, 30W RMS of DATV is easily achieved with the RA604047M1 module. That's why I ditched a Nokia TTRX amplifier that needed a 28v supply and 170W DC in to produce 45W of RF with the same shoulders. No contest!

Clive

[g0mjw]

As I was challenged, I had better explain. The power depends on what level of shoulders are considered acceptable, e.g. -30 dB, -40dB, -50dB. The consensus seemed to be around -40 dB or better on 70cm. Shoulders of -30dB, 20dB and even worse we sometimes see on QO-100 are certainly not acceptable. Hence my 10W limit.

Individual modules may and probably do vary, I could only get 10W before shoulders went above -40 dB. Clive got 16W. My bias was around 5A standing current for best linearity. However, you can tune that a bit. It is not efficient and needs good cooling.

Shoulders are much more critical on 2m where we can't radiate out of the band. What is acceptable also depends on the bandwidth, 333ks vs 1Ms the shoulders of the 333ks will be inside the band required by the 1Ms so from an interference to other users perspective, a narrow transmission even with significant shoulders is going to be better than a relatively clean wide transmission. It will go further too for the same power.

The stability issue is largely due to the feedback paths and/or poor input matching. It won't appear with external switching and some input attenuation. Grounding as Clive indicated, is vital. Maybe it depends on the level of bias but I have seen even well grounded modules oscillate with the Pluto if there is no matching resulting in power instability and worse case full power out.

Mike