My A600 PA project.
Photos updated 20200918.
Copyright pa0nhc.

This PA is still under construction.
More updates follow.
This project demands knowledge and skills. Not for a beginner.

Sokkie : My place.

As the background noise level is becoming worse in many locations, more power than 100W is sometimes welcome for many listeners. I saved some (holiday) cash, and wanted a nice building project.

In June 2020 i spotted a nice and (at the first sight) well by Razvan Vatu designed 600W amplifier kit for 1.8 MHz to 70 MHz.    All info at www.qrpblog.com.
Also available are a "Back panel Kit", containing a TX / RX relays circuit and a +5V supply, and a "Low Pass Filter Kit", switch able with TTL level  DC signals. All SMD components are already soldered. A central command unit with a text screen is under development.

I ordered the amplifier kit, which arrived about 10 days later. Razvan soldered all SMD components before shipment. A nice service. which i appreciated, as my hands are a bit shaky due to an "Essential Tremor", and my eyesight is not optimal due to a bad sector in my left eye. I cannot see depth while soldering small components.

As i do not like to read small characters on a small LCD or O-led screens, i decided to use small, cheap 85C1 moving coil meters for monitoring the amplifiers performance. These six meters cosseted less than $30 in total, and are monitoring (left to right) :

- +48Vdc, 
- consumed DC current,
- cooling block temperature,
- input power, 
- output power, and 
- reflected power. 

Drilled cabinet.

PCB for one 85C1 moving coil meter.

These monitored DC signals come from the J5 output bus on the amplifier PCB, and are accurately representing all levels at the PA PCB itself.
The individual meter PCBs are powered from the 12Vdc supply at the back panel, or (with small component changes) can be fed from the +48Vdc supply.
The metering PCBs contain each their own 5Vdc reference.

The used J5 signal output lines at  the PA unit have different output levels and source impedances. Active buffer stages are thus needed to drive the lowZ 1mA up to 20 mA 85C1 moving coil meters.
I drawled new dedicated meter scales, each adapted for the particular signal. They were glued onto the meter scale plates. 
I used different meters with different coil currents, but i suggest to use 1mA instruments only. You have to draw and install your own meter scales anyhow. It makes calculating and installing some PCB components simpler. 

Warning : some current meters are AC types, and contain a dedicated current transformer and rectifier, which must be removed. I am not sure that these meters have a linear scale. Just order DC voltage meters for all meters. They are 1mA instruments. You can choose ones with a correct FSD indication  for you needs. Then copy the scales, and change the indication.

A versatile metering PCB was developed for it, which are directly screwed onto the 85C1 M3 meter connections. The input has an RF filter. For different meters, some components are different, defining the meter FSD output voltage, and an alarm threshold voltage. If this alarm threshold is passed by the input signal, a warning LED burns above that particular moving coil meter, and a for all meters common buzzer sounds for a short while. Each PCB contains a trim potmeter for adjustment of the meter FSD. Meter calibration to the by Razvan specified PA signal output levels, and threshold testing can be done, by supplying a (for each meter different) DC voltage to that PCB input. 

LPF switching PCB.

After wiring i concluded, that one long PCB, screwed onto the bottom of the cabinet, containing the 6 metering circuits, plus the FAN regulator circuit, should be far neater and handier. 

Such a PCB will be costlier, but the interfacing with the PA module could be much handier and neater using only one flat cable. 
For every meter, only simple wire connections to the meter +/-, and its accompanying "Alarm  LED" must then be made. 
The "Alarm  LEDs" are fitting into 5mm holes into the front, above their respective moving coil meters, and are secured in place with contact glue.

The LPF is by Razvan delivered as a pre-assembled kit. This LPF will simply be installed on top of the PA unit. The filter is switched by my band switch PCB, installed at the front panel, and powered by the 12Vdc from the T/R PCB. It supplies lowZ and RF bypassed TTL compatible DC signals to the LPF PCB. Seven LEDs indicate which filter is activated. 

The fan is regulated according to the cooling block temperature. An automatic air temperature sensing FAN is therefore not useable. 
The FAN is installed at the center of the back of the cabinet, and creates a little under pressure inside the cabinet. Thereby forcing cooling air to be drawn from the outside directly through the ALU cooling block, which is mounted onto the cabinet bottom. 

I installed the PA unit shifted from the center of the bottom plate towards the front, to keep enough room between PA and back panel FAN, to be able to install one or two extra 9cm fans on the top cover if necessary. 

But an extra 12cm FAN under the bottom of the cabinet, blowing into the air intake opening under the cooling block, is another option. 
Fitting cabinet feet are difficult to obtain. My cabinet now stands on  2.5 cm high door buffers. If needed they can be changed for higher ones. 

On the front of the cabinet, the mains power to the internal 12V supply, and to the external 48Vdc supply, can be switched ON/OFF with a heavy duty power switch. When the PA mains power is "ON", the cooling FAN runs constantly at about half supply voltage. When a cooling bock temperature of abt. 35C is reached, the FAN voltage is switched to the full 12Vdc. This cooling block temperature is adjustable.

The accompanying FAN regulating PCB is installed onto the cabinet bottom. Its temperature input is the same J5 output as used for the temperature moving coil meter. 
Cool air intake is in the bottom of the cabinet, under the center of the PA cooling block. Forcing the air stream to run from the center to both cooling block ends, for effective cooling. 

Oversized ALU cooling profile.
l x w x h = 24x12x8 cm3. It has a 1.2cm thick sole.

Warning : testing with open top lid results in NO air stream through the cooling block.

Remember : PA efficiency is about 75%, 25% input power is generating heath. All harmonics are by the LPF reflected back to the PA, also converted into heath. VSWR should be low, as reflected power is again converted into heath. With 600W output, heath generation is at least 200W. 

A safe cooling block temperature is calculated to 50C. This asks for a cooling block thermal resistance of less than 0.2 K/W. In order to prevent buying, drilling and tapping of a 10cm x 20cm x12mm thick and expensive copper heath spreader, I preferred to use a big and thick ALU cooling block. A copper heath spreader is of cause safer for digital modes and contesting, but costs 12mm extra height and more than 100 Euro extra. The use of certain high quality cooling paste is then not possible.

I found at RScomponents a very nice ALU cooling block including a fan, but RScomponents will only supply it to companies. I suggest to use this block, as it is very handy and performs very well.

So i needed to find an equally good solution. At Alibaba i found a 12cm wide x 24cm long x 8cm high "700W LED Heath Sink" with a 12mm thick sole. No thermal resistance was specified, but it seemed to be very useable. The to me delivered one showed some -not straight- cooling fins, which do not appear nice, but do not degrade performance. 

        Cabinet, drilling and tapping.
As this cooling block is 8cm high, and the PA PCB and the LPF on top of the PA  are 4cm and 3cm high, they demanded for an at least 16cm high cabinet. 
On Alibaba i found a 4 units high 19" black cabinet kit. Consisting of six 3mm thck die cast, fraised and drilled ALU panels, which must be connected to each other with small 3mm black flat head screws, and corner connections. All fitted perfectly. 

A lot of holes must be drilled. Use new, sharp drills, cooling the drills with alcohol or lubricate with oil. Use slow rotation. 
When drilling the holes blind into the ALU heath sink, drill deep enough. 

Tapping the holes can best be done using single machine taps, while lubricating with oil. Regularly stop. Also tap all grounding holes in the cabinet, as the electrical contact is then more secure.
After a half to two turns, turn 1/2 turn back to break the curl loose. If to much force must be used, remove the tap and its debris. After finishing a hole, CLEAN the tap fully from debris.

The 50mm holes for the 28C1 type 1mA moving coil meters, MUST be drilled using a drill stand, and a multi-tooth 50mm high speed steel drill. Use a piece of flat wood to support the ALU plate, and screw them together. Pre-drill the holes first using a 6mm drill, to be sure of the correct hole position. Be accurate. Drill with slowest speed, and the correct pressure to prevent blocking of the drill.. 

While drilling, regularly remove aluminum curls and debris, and keep lubricating to reduce drag and hearth. Do not drill indoors, to keep the XYL friendly. Or the house will be infested with aluminum debris.

As the cabinet panels are nicely black anodized, they are INSULTED from each other. I therefore interconnected the cabinet sides using short wires, and then connected one of them to a central grounding point at the bottom plate. The mains safety ground, and the PA unit cooling profile (with PA PCB connected) are connected to this central grounding point too. Tap all grounding holes with M3, to be sure of good electrical contact between ALU part and M3 steel screw.

When installing PCBs, you must remember this. Scratch some black eloxated layer away, and use tooth rings under screws and nuts, which penetrate the insulating layer. MEASURE groundings !

The 12V power supply at the T/R "back panel PCB" is connected to the cabinet back panel. 

The SMA and N connectors on this T/R PCB are electrically FLOATING, and only connected to the PA PCB via their SMA cables. Preventing RF stray currents through the bottom plate. 

        Back panel.
I decided to design my own "BackPanel", as i do not need an Elecraft KPA1500 interfacing, nor a second antenna connection. 

My back panel PCB performs T/R switching, and is connected to two LEDS in the cabinet front, which indicate "Receive / Transmit" mode. During receiving mode, the PA bias current is reduced to zero by the T/R PCB "Inhibit" output. Thereby avoiding unnecessary heath generation, obtaing better cooling during standby, and preventing generation of wide band noise, which could de-sense the connected receiver.

A sequenced relay output from the driving TRX must be connected to the back panel BNC "Key" bus, to switch the PA fast between standby and amplified RF power. A toggle switch at the front of the PA can cut this input, to force "PA standby". The FAN keeps running, but the PA cannot be activated to amplify incoming RF. The TRX is directly connected to the antenna.

My back panel also contains a 12V/10W DC power supply, feeding the metering circuits, the band switch circuit, the FAN regulator and the FAN. The output is protected with a 1A 5x20 mm fuse. It is a simple, not stabilized circuit, reducing heath generation. The metering- and FAN switching circuits are designed for 10-20Vdc variations in supply voltage, making a stabilized supply unnecessary. REM : On the FAN regulator PCB, a larger value 12Vdc buffer capacitor showed to be necessary to enable positive FAN relay switching.

At the back panel, the 240 Vac mains input is fused by a 5mm x 20mm 5A slow blow fuse. This fused 240 Vac output is connected to a 40-50Vdc 1100W surplus power supply, by means of an Euro power cord with female plug.

The 50 Vdc power cable is fused by a 25A automotive fuse. It is connected to the power supply output with a high current DC connector pair.
As the rack mount type power supply needs some convection cooling, i placed it on two wooden feet, creating breathing space below the bottom.

The cooling block is fixed into place to the bottom lid of the cabinet,
by means of four 8mm rods and 3mm screws into the sole of the cooling block. The rods are at the underside of the bottom lightly tightened by M8 nuts, and keep the amplifier unit firmly in place.
10mm rods and M4 screws will have more mechanical strength.

T/R switch + 12V/10W power supply PCB

FAN regulator board.

LPF switching board.

Back-panel board.


            Status 20200918.

     My drive and protection solutions :
According to NXP absolute maximum limits, the allowable gate voltages are -6 / +10V in respect to the sources, or a symmetrical +8 / -8 Vpp in respect to the about 2V bias voltage. Max drive voltage is therefore 16Vpp. With the 2:1 input transformer, the max. drive voltage at the PA PCB input is thus 2 x 16 = 32Vpp or 11.43 Vrms.

Max. drive power into the 50 Ohms PA-PCB input = E2 / R = 11.432 / 50 = 2.61 Wrms input
With 24 dB PA-gain this could lead to 650Wrms max. output. Power attenuation between a 100W TRX and this PA should therefore be at least 15 dB.

I shunted my 20dB attenuator with a 470 Ohm 8W metal film SMD resistor network ((2x 470 ohm 2W in series) two of these strings in parallel).
Theoretically this enables about 483W output. This is enough for me, as the legal power limit in The Netherlands is 400W, and some over-drive safety margin exists. 

For extra protection against damage to the FET gates, i installed two fast acting, single polarity clamping circuits in parallel at each FET. They limit the gate-to-source voltages to about +9.2 / -5.7 Vpeak. REM : this are the voltage limits for the gates directly in respect to the grounded  sources. These clamping diodes also protect against RF overdrive.

It is an attempt to prevent that, during over-diving the PA and / or bad VSWR, high drain peak voltages will be coupled into the gate circuits via the negative feed back resistors R31/32. And accumulate with the drive voltage.


Back-panel inside.
At the left side : 240V~/5A mains power fuse and 50V/25Adc fuse.
At the right : The T/R switching PCB with 12V / 0.8 Adc power supply for the meters, FAN and LPF.

Wiring overview. Arduino cables are elongated with thin insulated wires. At the left on the bottom the FAN regulator PCB.

        REM :
The metering circuit wiring is cluttered, and connections could be mechanically sturdier. A better solution should be :
One long and narrow PCB containing all metering circuits, mounted under the meters onto the bottom of the cabinet.
This metering PCB is then connected to the PA-PCB using a 2x7 flat cable.
Then all meters and LEDs connected to this metering PCB with 1x2 IDC connections.
This solution costs more, but is neater, easier to (dis-)connect and more reliable.

        The back panel. 
At the left the N busses for antenna and TRX, with below them the BNC bus for connection of the transmit signal from the ACC bus at the TRX.
Holes are drilled big enough to enable plugs to pass the panel. The N-busses are INSULATED from the panel, and via the internal cabling connected to internal ground to prevent stray RF currents.

All bigger holes  are drilled using a step-tapered drill. A drilling template is available. 
M3 grounding holes should be drilled 2.5mm and tapped M3 for good ground contact.

Small size door buffers are used as feet.

        At the top-right :
A 50V / 25A automotive fuse for the 50Vdc input, and a 5A fuse for the 240V~ mains in/output.
Below are the 50Vdc cable, the 240V~output to the power supply, and the 240V~ input from the mains power.
The 3 chokes use 29mm #31 cores. FairRite "SnapIt" #31 cores with a 19mm or wider hole are also useable.

At the top two of the meter PCBs, mounted onto the meter connections.
Under it the central ground point at the bottom of the cabinet.
At the bottom of the cabinet the FAN regulator PCB..

All side panels are interconnected, and grounded via the yellow/green wire.
The brown wire is the grounding for all meter circuits.

WARNING : when starting to complete the LPF-PCB, first solder all coils
Remove wire insulation first and securely pre-tin all coil ends (using a HOT 420C iron and fresh multi core solder tin). 
REM : The silver plated wire is ALSO INSULATED !
Then solder all relays.
These are very sensitive to deformation due to a solder iron nearby, and will not switch anymore when its plastic housing is deformed. 

A 10-pole flat cable connects the bandswitch at the front panel (not shown) with the LPF-PCB on top of the PA unit.
3x4 small #31 ferrite cores at 3 places shifted over the flat cable should block induced RF currents.
Over the 50Vdc cable about 20 small cores ferrite #77 (with a bit wider hole to fit over the power cord) are also installed..
The amplifier PCB and MRF300 FETs are screwed directly in contact with the ALU heath sink.

On top of the PA-PCB the LPF PCB. 
REM : one extra wire is connected to +12Vdc, a grounding wire connects it to PA-PCB "ground" and thus to -12 Vdc.

As the heath sink is 8cm high, with on top of it, the PA circuit and the LPF circuit, they demand for a 4-unit high (19"rack) cabinet.