Tuesday, 29 September 2020

Fixing my Gemini 2 linear amplifier

EDIT. I have rewritten parts of this as it is now 6 months since I repaired it and it is working fine. I have calmed down a bit and I can see that the Gemini design is very good, but the SWR protection circuit did not work for me and several others who have e-mailed me. 


OK, I am writing this four weeks after I restored my Gemini 2 linear amplifier. By this stage, after 141 QSOs on those four weeks, I think it is working again.

Gemini 2 at GM4FVM (plus 1296 linear, PTT relay, and bent paperclip for resetting phone)

"Fixing" is the word I used in the title. Not exactly repaired, because I did not put it back the way it was, but "restored". I am using it again even though it is not as it once was.

I have had the Gemini 2 "300 watt" linear amplifier for two and half years now. For the first year it worked fine. Then it developed a fault which caused it to break down four times during the next year. And again after that.

The fault always came on the same way. The amplifier would trip with an SWR fault, even though no fault existed. I could simply restart it and then it ran for a week or two, and then tripped again. And then again after a week or so. And then every day, and then every few minutes, and then it just tripped out all the time. 

It had a two year guarantee and during that time I did not think it was right for me to try to solve the problem myself. Any meddling would just invalidate the guarantee. Various suggestions were made as to the cause such as it might be the centre pins of my N-type plugs, or my driver rig. It wasn't any of these things.

To prove that it was not something at my end I connected the linear to an entirely different set up, a hand held FM driver rig in and a 300-watt dummy load out, with different RF cables. I even changed the mains cables. It still tripped. This was clearly not a fault at my end. 

One reason I knew that it was not a fault at my end is that I was bequeathed (by David GM4JJJ) a very nice WaveNode WN-2 digital power and SWR meter which provides an almost instantaneous reading of power and SWR and it showed nothing amiss. I even took video and analysed it frame by frame. The SWR LED could be seen first glowing dimly and then getting brighter at different stages during a transmission, with the WaveNode in the same shot showing no SWR at all. I tried five different SWR meters, and none showed any problem.

WaveNode WN-2 - superb SWR/power meter (but very expensive!).

The WaveNode uses outboard sensors, and after it arrived here I added several more to cover all the bands I use. I take it to be accurate.

I returned the Gemini to the maker several times under guarantee. Sometimes it came back with no fault found but it still did not work here. 

Other times it came back with work having been done and an explanation. This was the solution that eventually got it working again for the first time, explained thus when it came with this email:- 

"I had a hunch what it was and it proved to be correct. 

The directional coupler pick-up bridge circuit feeds through two 75R 0803 SMD resistors - one for fwd and one for rev. 

The rev one was showing big swings in resistance with temperature and the thermal camera was picking up a bigger variation in temp (as it heated up the resistance was lowering to the current was increasing causing more heat and so on). 

I have replaced both these 75R resistors and the amp has been on soak test now for an hour at 150W out with no issues." 

So this was my clue as to what was wrong too. Well, it was actually exactly what was wrong. I had this clue to go on, and he sent me a photo of the resistors:-

Photo of one of the SMD resistors sent by DXShop (looks rather small, as they do ...)

Replacing the overheating resistors with more of the same solved the problem for a while ...until the Gemini started tripping again after another few months.

After another trip back under guarantee (the fifth trip back I think) I got this message:-  

"As a precaution I have replaced the directional coupler components. If it still trips then there must be some issue external to the amplifier that is failing under rf load."

That was a strange comment. If changing the components fixes the fault that suggests to me that the components are at fault, not some issue external to the amplifier. I suspect that if you change a component and if it fails again then you know where the fault lies. Suggesting something external seemed to me to be unlikely to be causing a resistor in the directional coupler to overheat.

Well anyway, I know now that there wasn't some issue external to the amplifier causing a problem. 

I felt sure that keeping on changing the resistors which keep on altering in value was just putting the problem off for a while. However, by repetition this had occupied the entire 2 year guarantee time and so it is my problem now. 

It ran for a further three months before the same fault happened yet again. As nothing had changed, just keeping on doing the same thing (changing the resistors for more of the same) will not bring a different outcome.

Another way I knew the fault was not at my end was that during the periods the Gemini was away I was using the excellent Microset SR200 linear amplifier. No SWR fault there. When the Gemini came back I sold the Microset (mistake). When the Gemini failed again I acquired an RM Italy VL-250. The Microset SR200 was good, the RM Italy VL-250 was excellent. The VL-250 has a very fine protection suite including SWR protection - and no fault showed up there either. There was no fault at my end to find.

Three months after it came back for the last time from repair under guarantee, it failed again. I just put the Gemini 2 on the shelf and plugged in the RM VL-250 and tried to ignore it. 

There, sitting on a shelf, was a nice piece of kit which I could not use. Obviously I was going to do something eventually, but just then I was really fed up with it. 

A couple of months later I calmed down and started to think that the Gemini needed to be restored. 

I read back over everything that had happened. I decided to assume that it was true that the two resistors in the directional coupler were overheating and varying in value. One thing that seemed to be supporting that idea was that the output meter on the Gemini had started to work erratically - and it is powered by the directional coupler too, albeit by the forward power circuit. If changing the resistors worked for a while each time then perhaps that is the only fault, and the power meter is affected in the same way.

I poured over the Gemini circuit diagrams and read the manual backwards and forwards. I opened it up and examined all the things I thought I could do. It seemed that I had four options:- 

1) Turn off the protection by turning down the potentiometer on the control board (which would stop over-drive and over-temperature protection too) 

2) Replace the SMD resistors in the directional coupler with devices which could withstand the heat/current 

3) Simply disconnect the directional coupler and run without any SWR protection, leaving the other protections in place 

4) Do (3), but add some outboard SWR protection of my own. 

 I rejected option (1) as being too risky. 

Option (2) appealed. I may do this later and restore the Gemini to more-or-less original condition but for now I will not being doing that. Too many imponderable possibilities.

Option (3) was possibly an immediate solution to see if that solved the problem. Although I have used lots of linears over the years with no SWR protection, I felt sure that Murphy's law would step in and blow this one up if I ran it permanently without any protection. I had no strong reason not to do it, and I think it should be safe enough. 

BUT, option (4) looked the easiest way of getting the Gemini working again and having some SWR peace of mind. 

I should perhaps say that the Gemini 2 has NEVER tripped due to high SWR when there really was an SWR fault. It has tripped dozens and dozens of times, but only as a result of this false trip issue. Despite this I though I should do SOMETHING to protect against high SWR, simply to put my mind at rest. 

When David JJJ became silent key he left me his WaveNode, as mentioned above. David and I had talked about my false SWR-trip fault and he was as worried about it as I was (he had two Gemini linears himself). Now, whether he had thought about the possibility of neutralising the SWR circuit and using the WaveNode for protection I do not know. After all, when we spoke about this fault it was still under guarantee. Anyway, the WaveNode he left me has an SWR protection circuit. There is a trend here - everything David left me has been useful, even if I could not work out why he left them to me when they arrived.

So we had two fairly simple tasks:- 

First complete Option 3) - disconnect the connection between the directional coupler and the control board (then test everything works) 

The complete Option 4) - rig up SWR protection from the WaveNode. 

First: cutting off the on-board SWR protection 

This is pretty simple, or it should be. The wire from the directional coupler uses the multi-pin header plug on the W6PQL control board. Despite this being fairly simple to disconnect I managed to disconnect the wrong wire three times. 

Getting them back in once removed was tricky. Use a commonly available crimp tool "under $30 on Amazon" says W6PQL. Anyway, none of the crimping tools I had worked and eventually a home brew method using a screwdriver and extreme force did the job. 

Did I ever mention that I am colour blind? I am not the right man to send on a job to a 20-pin plug to find any particular wire. I am bound to pick the wrong one. Not that I actually needed to do that, as I could have just counted the pins (bet you would have started counting at the wrong end Jim).

The 20-pin header plug on the W6PQL control board (colourful, eh? I would not know)

I did not disconnect the forward power line to the power meter on the front panel, which perhaps I could have done at the same time. 

I found it easier to take the W6PQL control board out each time while working on it. It is screwed to the right hand side of the Gemini case and hemmed in by other components.

Having carefully insulated everything which had been disconnected, I put the linear back in its case and tested everything with no SWR protection. Unlike before, when it tripped immediately, it all worked perfectly. It worked fine for three days before I set up the SWR protection. 

Then setting up the outboard SWR protection. 

Pretty simple too. The WaveNode has an on-board relay which trips at a pre-determined SWR level which you can set in software. 

The WaveNode monitors four power and SWR levels at once (in my case, 6m, 4m, 2m and 23cms). However, it can only do SWR protection on one band at a time as it only has one physical relay. For the sake of simplicity and to avoid any confusion later (hopefully) I moved 2m to monitor One, and set the SWR trip level at 2:1. This is in fact a tighter trip level than the Gemini had initially.

WaveNode monitoring 2m (band 1), set for SWR2:1 (actually it is 1.132:1 in this setting).

 The physical wiring looks complex but is pretty basic really. 

Using the WaveNode to monitor the Gemini SWR

For the sake of clarity the diagram leaves out the 24 volt relay which switches the IC-9700 PTT line between the 2m and 23cms linears. That is the relay you can see in the first photo. Well, not just for clarity: it is wired in the wrong place. It too cuts off the 23cm linear should the 2m SWR trip work. Far too much wiring to get that right so I left it the way it is. This place is like a spider's web already without adding more wires.

The WaveNode has a three pin PC board socket on the back of the main module. Being the sort of person who never throws anything out I had a matching plug in my "computer parts" box. It was simply a matter of attaching phono plugs to that so that it matched my PTT line standards. 

If the SWR on 2m exceeds 2:1 the relay will trip and de-activate the PTT from the Gemini. That leaves the rig facing the high SWR, but it is well protected and that was always the case with the on-board protection anyway. 

I suppose that there is some risk of a cable failure between the Gemini and the WaveNode sensor, but then there is a risk of cable failures inside the Gemini too. And in any case such a fault would probably trip the SWR circuit anyway (I am not about to test this to find out).

I tested it with a high SWR at the antenna and the trip worked immediately and also gave a loud audio warning "HIGH SWR TRIP ACTIVATED" over my PC speakers.  

You could do this with much simpler and cheaper SWR meters than the WaveNode, but that was the one I had.

So finally ... 

The Gemini is restored. 

Maybe one day I will delve in and replace the two resistors in the directional coupler with high wattage and high stability alternatives. On the other hand, maybe not. 

For the first few days there was a "hot electrical" smell which suggested that the resistors were still overheating. Gradually the output power meter got more and more erratic. Then the smell stopped, the output power meter stopped working, and normality was resumed. 

I can read the output power and SWR from the WaveNode, which displays this information for all bands conveniently on my PC screen. It also shows one band directly on the readout on the WaveNode module itself.

I am left with various issues to ponder over. One reason I did not replace the resistors (again) was that I am not sure why they were so hot in the first place. This Gemini 2 is very quiet, compared with my 4m and 6m Gemini linears. Could it be that the fans are running too slow?

I have tinkered with the pot on the control board and poked about at the secondary fan control board, but the airflow seems OK. There is plenty of hot air coming out over the PA heatsink, and the fans do accelerate as you might expect. So is it just the SMD resistors cannot handle the current? Why not? Or is it heat leaking from the PA?

Never mind it works, for four weeks so far anyway (EDIT six months now). 

I have now sold the splendid RM VLA-250, a fine 200W linear. 

I have since bought a new 70cm linear - a Tajfun 1000. The Gemini design is good and the price not unreasonable. It is a pity that those two resistors could not stick the pace in my amplifier..

Thanks again David.

I do not know how you knew, but you knew.





  1. Thanks Gavin, it has generated a few interesting emails too.