Modification of the Harris Amplifier is not the reason for this article, there are many documents on the web covering this [Site 1, Site 2 part1 & Site 2 part 2], I followed Mark GM4ISM’s as Mark was the source of my Harris amplifier bits.
I actually purchased a Harris amplifier ‘pallet’ from Mark, this contains 4 combined amplifiers each good for 400w, and a control board all in in a tray which can be hot swapped in a commercial environment, all 1500watts of it (the difference is the combiner losses). This image shows the 4 amplifiers in the pallet. W7GJ shows all this better still (although his page is for 50MHz).
My intention was only to use 2 of the amplifiers, the CUWS got the other 2. This setup gives each of us 750watts of power, not quite my usual target of a 3dB overhead for an amplifier, in this case 400w which is the UK maximum power on 144MHz, but acceptable to ensure the hardware (components) are not being driven to their design maximums. Don’t forget to get a good iron machine.
Each amplifier block (4 of these make up the pallet) looks like this before modification (click to enlarge);
This is a pair of amplifiers (750w) in a frame with the input and output combiners in place.
John G4BAO actually done the required modifications to get the CUWS and my own amplifiers onto 144MHz, this is basically changing the tuning and combiner lengths to 144MHz.
My Harris amplifier was built initially to be used on the Camb-Hams Mull 2013 dx’pedition, however we decided to use a Beko HLV-1000 amp instead, this was decided due to the lack of testing and protection in my Harris amplifier, if things went wrong on Mull we had no spares. The Beko is a very capable amp with nice protection. The Harris amplifiers are robust, but without time to test and get confidence in it we decided it was poor planning to use it so far from home for the first time. As such it was never used before I started the project!
This article covers my work in boxing the Harris amplifier into a single box with all the bits needed to make it self contained. But wait, hadn’t John G4BAO already done that?
This is no criticism of Johns work, his remit from me, and the CUWS, was to modify the amplifiers to get them working, I only supplied John with the amplifier. coax relay and a basic 2U box which was just enough to fit the amp and RF switching into. The amplifier worked great thanks to Johns work and he supplied me with some very insightful details about the amplifier including a datasheet of its performance and efficiency graph, without Johns work I would not have been able to start or complete this project.
Also it goes without saying that my boxing/construction was built with what I had to hand so I don’t expect anyone to be able to copy it exactly, however the hope is it might provide some ideas or inspiration to others wishing to do the same.
This is the current look of the front and rear of the re-boxed amplifier after I fitted the meters, quite a change from the G4BAO build.
And the rear with the additional 4 fans fitted.
On the rear I added some additional fans.
Part of the reason for re-boxing the amplifier was to add protection and add some flexibility in the overall design. This included;
- fitting in a 48v PSU
- adding an input power detector
- adding an output power detector to read
- forward (FWD) power
- reflected (REF) power
- integrating a control system with failsafe shutdown
- add an analogue status meter
- fully boxed
One thing I like in my projects is to have something self contained and neat. In the case of an amplifier this means a box which includes the amplifier (obviously), the PSU (for all voltages), filters, control/protection circuit and display. One of the reasons I have this design philosophy is that my projects get used by others. mainly within the Camb-Hams group, some are not technical, so I like to make it simple to use and harden it against failure, damage or improper use.
Basically I would like 240v in, PTT in and RF in with an RF out; simple.
My RF path design is basically;
I might integrate a sequencer into the box also for external device sequenced switching. I have a few ready build BXF Sequencer modules built.
I knew this project would take time, I purposefully didn’t rush it like I usually do. It has taken about 4 weeks so far, many hours of work, and it’s not quite complete – there is probably another 2 weeks of work before I can declare it ready.
I was lucky in that I was given a few surplus boxes by Adrian M0HKW some time back. Some were too small and the one I ended up using, pictured below (without the covers on), sat on my shelve for a year before I really investigated if it could work for this project. The box is a 19” sub-rack type by Schroff. The beauty of this type of box for a new project is the design specifically allows many configurations and positioning of ‘struts’ due the regular mounting holes along the panels. I recommend them if you can get your hands on one, this one has 84HP wide front and rear panels, also vent holes at the top of the front panel surround.
I stripped out the original bits from the box and after removing the Harris amplifier from the original 2U box I checked how it would fit. It actually fitted well. I now just needed to work out how to fit the other bits, this required a bit of thinking…
When I bought the Harris amplifier pallet I also bought an Eltek SMPS 1800 48v PSU, this can provide about 37.5A but was much to big to fit into the box with the amplifier. I could handle this, as the amplifier needs about 25A at 48v for 750w output, so I appreciate the size of the PSU could be an issue., i.e. high powered PSUs don’t come small.
Well not really. When in Friedrichshafen 2013 I picked up an Eltek Flatpack-1500 for EUR25, I wish I bought more. This PSU is small, 214 x 41.5 x 243mm (W x H x D) and can supply 31.25 Amps at 48 VDC!! This is about 1/4 the size for the Eltek SMPS 1800.
Something else good about this PSU is it has a current meter, in the form of an LED bargraph, on the front and has built-in cooling. My original plan was to make a hole in the front panel where the LED bargraph was but instead I decided to take a tap off the input pin of the LM3914 LED Bargraph Driver chip to feed to an analogue meter.
The first thing I wanted to try was to get the Harris amplifier and PSU fitted into the box. The Eltek Flatpack-1500 PSU fitted perfectly into the box, but only with no power connectors (240v and DC) fitted. To make it fit these connections on the PSU needed to be removed and instead cables had to fitted directly to the PCB and then brought out the top of the casing.
Mains on the left and +48v on the right. The work was simple, removing the connectors on the PCB and then soldering wires direct to the PCB, mains left and DC right.
The case then had two 10mm holes drilled above where the connectors originally where with rubber grommets fitted to protect the wires passing through the case.
This overcame the issue if fitting it in the box, so now it would fit front to back without the connectors on the back. However left to right the amplifier and PSU were 45mm to long.
Mounting the Amplifier and PSU
The next thing was some metalwork to work out how the PSU and amplifier combined could be reduced by 45mm to fit inside the box. Out comes the hacksaw.
Cutting a notch out of part of the amplifier mounting allowed for the PSU and amplifier to fit side by side.
You can also see the amplifier is on pillars, this was for a couple of reasons; first to increase the gap between bottom of the box and the amplifier to make space for the RF T/R switching relays and also to decrease the gap between the heatsink fins and the top of the box so the airflow is directed and contained along the fins, see the Amplifier Cooling section.
The image below show the amplifier after I fitted a plate to shield the amplifier and also make a mounting surface for the switching relays which would now mount on to. This is the space between the amplifier block and the bottom case of the amplifier, one of the reasons for mounting it on the pillars.
Talking of airflow I also had a bunch of 25mm fans. These are 12v types so each bunch of 3 are wires in series such that it needs 36v or so to run them, on test 48v was also ok with no increase in current.
The final arrangement was done that in RX both blocks of 3 are in in series (all 6 fans in series over the 48v) and in TX each block of 3 are put in parallel (2 sets of 3 fans over 48v) so the fan speed is increased (and airflow). In time this will be integrated with a temperature controller – in actual fact the amplifier controllers have a temp sensor output, I just need to see how best to use that but the schematic shows there is a temp sensor (MJE170) on the heatsink which is used in the temperature alarm circuit.
This though is probably best left and used as the alarm signal and not as a ‘instantaneous temperature’ sensor, so some additional work will be need to see how to make an ‘instantaneous temperature sensor’ as the MJE170 is not a bonafide temperature sensor but is an NPN transistor with a very linear temperature profile thus voltage gain, so I guess that’s why it’s used.
This is the switching schematic which basically switches the fans from series to parallel. I combined the amplifier and PSU fan switching (see PSU Cooling) relays on a single PCB.
Along with the box, which had custom front and rear panels for the previous purpose it was used for when I got it, I was given a spare unused panel which I used on the rear as shown below, the reason was it has nice air vents in it, as the front panel frame has the same then it meant the front panel could be solid – which would also look better.
This is the rear panel and is the air intake end for the amplifier block cooling. This direction is reversed for the PSU as will be seen later.
One last bit was to help direct the airflow at the front of the amplifier out the exhaust holes in the frame. Otherwise it is directed at the panel which would cause turbulence and is inefficient.
I had some thin off-cuts of copper which can be cut easily with good scissors, it was then bent so it angled the airflow above the front panel and out the air holes as hoped.
Notes after first use on 12/1/14
After I done all this work above it was time to give the amp a a spin. On the first Tuesday of each month there is a 144MHz contest in the UK. I normally operate as G3PYE/P (with a few others) and we enter the restricted section, this limits us to 100w and a single antenna. The 100w limit was useful as it’s a low level in which to start trying the amp out at before trying anything more demanding. As it was at 100w I learnt a few things from the night.
- It worked fine, flawless even for the 2.5 hours of use at 100w.
- It seemed to run a bit hot, the PA’s do draw about 10A (at 48v with no RF that’s 480w heat) when the BIAS is applied, about 50% of the current as when at full power, so even with no RF the cooling requirements are quite high.
- There were numerous comments about how narrow this signal was, even when I tested with a little more power post contest with 2 stations 10Km away.
- Transporting the amplifier needed some work, it’s not heavy but has no handles to easily carry it. (see the Carry Handle section).
As for cooling I thought a bit about this, there are air intakes slots on the rear and exhaust holes at the front. The airflow was there, perhaps just not enough?
If that were the case it should be anymore. I concluded that forcing more cooler air into the rear was the best option, so with that in mind I’ve fitted an additional 4x40mm fans, 2 for each amp, on the rear panel which forces cooler air into the smaller 25mm fans on the heatsinks. This should mean more cooler airflow into the heatsink cooling.
The result is better cooling though the PA system. If this is not enough cooling then I have the option to add more fans on the rear panel, as shown below, then “seal” the path between them and the PA heatsinks, such that any additional fans and their airflow are forced through the PA heatsinks improving cooling more, for now I’ll try this new arrangement.
The space removed from the rear panel for the new fans and the existing 25mm fans on the PA heatsinks.
While at it I took the chance to add an LM35DT Precision Centigrade Temperature Sensor to measure the PA heatsink temperature. I was hoping to use the embedded temperature sensor but on further looking its [accessible] output is only a high temp warning switch, I’ll use that in the protection circuit.
The LM35DT is a great device and I use it in many projects, it is basically +5v and GND in, and +10 mV/°C out. Easy to calibrate to PIC ADCs etc or build a comparator to activate on high temperature – as the voltage of a given temperature can be calculated in advance.
You can see it here pushed down between the 2 heatsinks.
Currently it’s only wired to the meter on the front, it will later be connected to a temperature control circuit.
The PSU was designed to be rack mounted so the fans on the front were orientated to ‘suck’ air out the PSU as in the rack air is forced in the rear. As this is not the case anymore I turned the fans round so they now force air into the PSU from the front. This meant I needed to make some air intake holes on the front panel and exhaust holes on the rear.
On the front panel 2 holes needed to be made for the 40mm PSU fan intakes, this just needed marks on the panel which align with the fans then using a Step Drill, the largest I have is 36mm which was fine, the holes where drilled. Fan grills where then fitted.
On the rear panel I first marked up a grid for the exhaust holes then drilled them
Once done a warning label was added…
As can be seen in the above image there is also a mains IEC connector. The way I do this is to draw the required hole then punch a marker to drill a hole large enough that it takes as much of the required material away.
I then use a nibbler tool to remove the rest before filing everything smooth and mounting holes drilled.
All the scratch marks are on the inside, which is typical of a nibbler tool. On the outside it all looks good…
RF and PTT Connections
This is possibly the only bit I’m not happy with. Originally I planned to have N-Type chassis connectors with coax tails to the relays so I drilled the holes and then the 4 mounting screw holes then realised I could be smarter buy having the RF input power detector connector extend out the box for the RF input and the common port of the output coax relay for the RF output. This meant on the input the 4 holes were no longer needed and on the output only 2 were used, so I made 6 holes I didn’t need.
+15v, Bias and Relay Voltage
The amplifier can have 48v connected all the time, for PTT +15v is needed to put it into class A/B mode. This required current at 15v is quite low, about 200mA total for the bias. This 15v supply is also used for the coax relay switching which is an additional 150mA.
You won’t believe how hard it is to generate 15v from a 48v supply though.
A 7815 regulator can only handle a maximum of 35v input. The solution I used is to use an LM371HV which can have a voltage differential of 60v, we only need 33v (48v-15v). The HV is a the High Voltage variant but the only thing is, the datasheet indicates, a very quick drop in current vs voltage differential.
With a voltage differential of 33v the maximum current is about 250mA, for that reason we use 2 devices in parallel only gives us a maximum of about 500mA.
However writing this I’ve just had the thought of using an LM317HV to drop 48v to 35v, this drop (voltage differential) of 13v keeps the LM317HV in the 1A range, I can then use an LM78S15 to drop the 35v to 15v.
The LM7815 (1.5A version) datasheet (figure 10) shows that with a voltage differential of 20v the current is limited to about 1A also, this gives me about twice the current, but probably more importantly produces less heat. The other advantage is an LM7805, for any PIC circuit (display, SWR detect, sequencer) can be taken from the 35v also.
I decided to change the DC regulation to a mix between LM317HV and LM78S15 (and 78S05).
Initial tests shows the 15v line is much more stable, the 2x LM317HV design has about 0.7v drop between RX and TX (when there was 350mA drawn). Also the heat being disipated seems much lower, but this is only anecdotal and not been measured by anything.
RF / PTT Switching
In terms of RF switching within the amp, the original 2U build had a Spinner changeover relay fitted which unfortunately doesn’t fit into the new box, but I have lots of other RF relays available. The good thing with the Spinner relay is it has an integrated DPDT switch which was used to switch the bias supply when the PTT switched the relay, this added a mechanical delay to the bias switching ensuring the RF path was made before the bias circuit was made.
The new relays I used were an MD951 relay (middle) for the input switching, usually found in PYE kit and is good to about 150w at 144MHz, which is fine for the input here, for the output I used a CX-600N relay (bottom right) which is good to 1Kw at 144MHz.
The MD951 is useful as coax solders directly to it reducing the space needed as there are no connectors, I also have a box full of them.
The PTT circuit has a relay on the input. This helps isolate the connect equipment (on the PTT In) as it limits the current to 35mA on the relay coil with the actual amp DC switching done on the relay contacts.
The PTT input goes via a switch on the front panel, which can switch the amp between OPERATE and STANDBY. I added a PTT Out circuit which is driven from a TIP31C transistor. This is for connecting to ‘something else’, whatever that may be but I’ve added it now in expectation it will be useful later.
Low Pass Filter
The output from the amplifier first goes to a Low Pass Filter which has a roll off at 253MHz. While not ideal for 144MHz it is low loss and high powered. It has 28.8dB of attenuation at 288MHz, at 70cm (432.2MHz) the attenuation is 69.8dB. The Harris amplifier is very clean anyhow so I’m ok with this.
RF Power Detection
The RF input is via a Directional Power Detector by Coaxial Dynamics which fits between the Harris amplifier cover and box frame, the input N-Type protrudes out of the rear panel. This was picked up at Friedrichshafen for £5 and it provides a DC voltage proportional to the RF power in the forward and reflected directions, 60w forward and 15w reflected, I don’t plan to use the reflected port though. This is what the detector looks like, the one I use covers 144MHz, I just don’t have a picture of it.
This input Directional Power Detector will ensure the amplifier is not over driven.
On the output side after the Low Pass Filter there is a dual port directional coupler for forward and reflected output power detection. After playing about with the layout, the LPF and directional couplers are not small, I decided to bolt the directional coupler onto the LPF then mount the LPF onto the side panel of the box using rivet nuts on the LPF.
On the side of the LPF I managed then to attach a simple diode power detector to rectify the forward and reflected RF signals out the direction coupler as unlike the input power detector the output from the directional coupler is pure RF, albeit at a low level so it needs rectified for a DC voltage.
The detectors are identical and each has 2 outputs; one raw (at the level rectified) which will goto a PIC to compare forward and reflected power detector outputs (essentially SWR) and then an output via a 10 turn variable resistor which will be used to drive the power meter on the front panel.
I decided early on I wanted to have 2 meters on this amplifier. one showing output information and one showing input information. It always helps when you can see/read 2 parameters at once, it gives a much clearer picture of the status, like PA current vs output power, or input power vs output power.
I started with a meter I again picked up in Friedrichshafen, which seems to have been for a Bang and Olufsen audio amplifier – it was terrible. The backlight was useless, the movement unpredictable and in the end I managed to damage it beyond repair trying to get the pointer to return to 0.
At one point the left meter stopped working so I needed to re-solder the joint, after that the right meter started playing up by sticking and not returning to 0. In the end, as can be seen in the right image, the right meter fell apart.
I had enough of this sub-standard part and decided to order some new meters. The issue this presented was more metalwork to the front panel. I don’t mind metalwork but the issue is at this point it’s easy to mark the front panel and the original meter holes needed to be covered.
The panel comes with a protective cover which I removed after fitting the last meters as I thought all the metalwork was complete. The simple fix here was to cover the panel in masking tape to protect it.
After looking for a while for a suitable meter, I was limited by size due to the rotary switches, I settled on a pair of 60mm*60mm MULTICOMP – SD60/0-100UA meters which fitted. I use my PowerPoint technique to make a stencil of 2 meters side by side so I knew where I should position them and to mark the holes I needed for them.
I’m not so proud of this next picture, it’s not upto my usual standards! I started by using a nibbler attachment for my drill but found the panel was about 0.5mm too large for it to work effectively so resorted to my electric jig-saw as the new meters needed quite a bit of metal to be removed.
However the work paid off and the new meters look good. I also added some labelling while the switches where off.
One of the points I found after the first use was it needed a carry handle to make transporting easier. The complete amp is not heavy however there are not parts to hold it for carrying. I decided a carry handle was needed.
Hammond produce a few handle types, I had thought about rack handles on the front, while these might help protect the front panel somewhat they are more designed for inserting and removing an item from a rack, not necessarily carrying. I decided a vinyl handle, like those on the side of radios would be a good option and settled on the Hammon 1427J.
This now makes carrying the amplifier much easier.
Ongoing… Check back…