Friday, 28 July 2023

Honey, I blew up the linear amplifier

Well, the linear amplifier blew up, but it seems it was not necessarily my fault. Nor was it the fault of the linear.

This took months to resolve so this is a long posting.

Of course, I have a long history of blowing up linear amplifiers, going back almost 50 years. I trust linears less than any piece of equipment I have. They are almost always about to let me down. Having said that, this posting is NOT about a linear letting me down. It is about me trusting a radio to transmit a low power output, and me taking too many risks by assuming that it would behave itself.

I have tried to write this piece several times, but it looks as if I am blaming the linear, which I am not. I will keep going on about that, just as I go on about lots of things in this blog.

Icom IC-7100 "It was him what done it, Constable".

There is a risk using modern high gain solid state amplifier devices in amateur service. Due to the enormous gain available the danger is that it will prove impossible to reliably restrict the RF power supplied to the device. The attenuator built in to the linear has to be set at a level compatible with the radio, and thought has to be given to the possibility of a sudden RF spike over-driving the device and its supporting circuitry.

So the spike in this case was generated by the Icom IC-7100 (or possibly the computer too) and that caused damage to the linear. Multiple spikes over a long period, or so it seems. How could I let this happen? Well one factor is that the IC-7100 has been operating into other linears with higher value attenuators for years. These larger attenuators make the chances of overdrive are much less. As a result I had become lulled into thinking this was not a big issue.

I have been thinking about this type of thing for a long time. The same problem arises with transverters as I explained here back in 2019. In fact I was extremely careful with transverters and even then I managed to over-drive the attenuator in one of them, though in that case without any calamitous results. It seems as if I failed to learn that lesson.

When it comes to VHF and above the amplifier has the RF power on hand to destroy any masthead preamp or other device further along the chain, and that also has to be factored into the equation.

Despite all the knowledge I had, I still managed to over-drive my linear despite having set the "RF Power" control to the correct level. The simple cause was that I had relied on the ALC operated "RF Power" control to keep the drive power low at all times.

If the RF driver relies on using the ALC circuitry to limit the output power, the chances of a random power spike are quite high. Connecting this transmitter to a computer which has its audio turned down and expecting that level to be uniform is another risk in the chain of potential for over driving the later stages.

At this point I can hear a voice over my shoulder saying that valve linears are much more reliable. You know who you are. Well, maybe. Not necessarily if you take care and if your linear is carefully designed, as this one was. I have three solid state linears which have given more than five years trouble-free service. And some that didn't.

So here I am talking about my Tajfun 1000 500W 2m/70cm linear. I do not think that it was in any way at fault in what happened. Neither was the excellent SHF-Electronik MMV 432-VOX masthead preamplifier which went arrrggghh at the same time. Nope, it looks like RF spikes from the rig that caused the problem.

I have to say that both the suppliers, VH Electronics for the linear and SHF-Electronik for the preamp, could not have been more helpful in fixing the consequences. Aside from the postage, the charge was €70 for the linear and €7 for the preamp. From this you can conclude that it was not the main RF device which failed in the linear but simply one diode.

Tajfun 1000, the victim of the problem, not a problem itself
This is not a review of the Tajfun, a linear amplifier which I think pretty highly of. It is very difficult to photograph as it is glossy black on the front and the display panel is pretty bright so it presents me with a technical problem when it comes to making an image of it.

Anyway, moving on to fixing things, ...

The fault showed up when suddenly the Tajfun lost output power and the preamp suddenly had more gain when it was out of circuit than when it was in circuit. Something had happened.

At this stage Vlado at VH Electronics, who had supplied the amplifier, went to great lengths to try to diagnose the problem and fix it. Thanks to the nature of the design, a lot of things could be fixed and indeed updated over the internet. The software for the display was sent to me and I was able to re-load it on a new SD card, and then the same was done for the firmware for the control board. These things helped but did not solve the underlying problem. After eliminating the relays and cabling, the overdrive had caused a fault which had to be in the amplifier RF stage.

Another clever thing that Vlado was able to do was to interrogate the records of overdrive and power output warnings which are stored in the control system. From this we discovered that the Tajfun had (successfully) coped with a series of huge overdrive situations. The output produced by the linear was at least 10dB above the power settings I had used and 3dB more than the full peak power the linear was rated at. Running any linear at twice its rated power is asking for trouble.

In each case the Tajfun protection circuit had cut out as it should have done. However, it was clear to me that this was not me deliberately operating at these ridiculously high power levels but probably something which was momentary. I was setting things up for 200W output so it was not that which was the root cause of the trouble.

The control circuits cannot generate more power than the transmitter can produce so the basic fault had to be in the transmitter. I will deal with that later.

For now I need to say that once he had eliminated all the other possible causes Vlado asked me to send the Tajfun back, which I did. The Freescale device had survived running at twice rated power, and the power supply had coped with supplying the necessary extra amps. Obviously, this overdrive was indeed  momentary. Vlado quickly identified a diode in the bias circuitry and the linear issue was fixed.

ITB RF board inside the Tajfun 1000.

The Freescale MRFE6VP5600H device is in the centre of the photo under the alloy plate, firmly screwed down. At the time of this photo I had tagged on a direct co-ax feed at the bottom of the board, bypassing the relays to check that they were not at fault. This also bypassed the attenuator which can just be seen below the red wire, bolted to the bottom of the RF enclosure.

Early in our search for the fault Vlado had considered that the attenuator could be at the root of the issue. While he had the machine with him he wisely changed the attenuator for the most recent version, increasing the power reduction from 5dB to 6dB. We could find no evidence that the attenuator was at fault, and anyway it seems to me that it would be unlikely that it would fail momentarily, and then recover, several times.

I have nothing but praise for the Tajfun. Having been all round the inside of it I can say that it is carefully designed and well made.

So the Tajfun returned and quickly resumed producing the power I was looking for.

You cannot overdrive any linear, not even a good one.

Siggi at SHF-Electronik quickly repaired the preamp which had been driven with 6dB more than it was rated for. Once again, not the preamp's fault. The service offered by Siggi remains exemplary. It took me a while to get around to dealing with the preamp, and in the meantime I had replaced it. He fixed it and returned it very quickly, and now I have two 432MHz preamps.

Even I cannot use two 70cm preamps. 

I need to thank Richard GI4DOH for helping with the logistics of getting the linear repaired by Vlado. I will not go into the details, but Richard was key to resolving the whole thing.

The fundamental problem was my Icom IC-7100. Well, that conclusion is inescapable. This was the source of the RF drive which blew both devices. I have had spike problems before. I had grown complacent.

I had connected the IC-7100, capable of delivering 35W on 432MHz, and set the output power to 4W as that was what the linear needed. The possibility of the rig delivering a sudden +10dB spike may sound improbable, but actually the full rated input of the linear at 500W peak would have required just 10W to achieve. Anything over 10W would have pushed it too far. The evidence would suggest that it reached 20W at least.

So what to do next? I do not want to have fixed the Tajfun and preamp just to overdrive them again. 

I could have replaced the internal attenuator in the Tajfun with one of a higher value. That is the solution adopted in my Gemini linears which are set for 25W drive. However, I was reluctant to raise the value of the attenuator too much as the heat dissipated has to go somewhere and the linear generates as much heat as we might expect already. Also, what if that single attenuator fails?

In the end I decided to add a second attenuator in the tx line. The idea was that I could change the value if necessary, and move it outside the linear for cooling if that was needed too. Doing a few sums, it looked as if a 5dB attenuator would involve the drive increasing from about 4W to about 12W for 200W output, and leave 8W to be added to the heat inside the Tajfun. The Tajfun has efficient temperature controlled turbo type fans. This means that any spike would be reduced by 5dB, and the result would not over-drive the linear or the preamp.

Additional attenuator which is now fitted inside the Tajfun 1000

Then I have considered other ways of avoiding the spike in the first place. The big problem here seems to have been me relying on the ALC to limit the power into the amplifier. One of the points made at the launch of the IC-9700 suggested that it has a "true power limiter". This appears to be the "TX PWR LIMIT" control. It is in addition to the "RF POWER" control which seems to be a standard ALC-type limiter. I say "appears" as this is not clear in the manual which states "The Transmit Power Limit function limits the output power to the preset level for each band.".

Anyway, the worry I have with this is that if anything did go wrong with this (for example if I messed up the settings) the IC-9700 can deliver 75W into the linear as opposed to the 7100s max of 35W. And, yes, I can get settings wrong.

I am guided by the idea that if something can go wrong, it will go wrong. Years of being steeped in statistics and probability have taught me the validity of that old saying. However, I can also get too carried away. It is not very likely to go wrong. Adding 5dB attenuation to the linear still gives some protection. If I remain worried I can increase the second attenuator and help cooling by placing it outside the amplifier casing.

In any case, I am losing confidence in the IC-7100. The years roll on and new better designs appear. It now looks pretty dated as a superhet radio in a shack full of SDRs.



No comments:

Post a Comment