Gates 5 blowing FETs

[ChuckG]

Just happened again.
Gates 5. Perhaps 1-2X per year a random power amp FET shorts, never the same quad or the same amp board. I replace the pair, inspect the board, make sure everything is clean and tight and switch it back online. No further drama until the next episode.

I've replaced the big blue caps, the load (tower) is stable, there is a decent surge suppressor on the incoming AC. The xmitter readings remain very close to the factory checkout sheet. IPA drive is good measured at the quads, waveform looks fine.
I'm almost suspecting some sort of spike on the incoming audio, but can't prove it. It is fed fairly dense, asymmetrical audio, the mod meter mostly sits parked at -98 +110. Scope confirms the mod meter isn't too far off.

I did have a FET problem a few years back that traced to a cracked toroid, but that was solved long ago. Since then it's random and infrequent, but IMO it shouldn't happen at all... but again this isn't the only Harris AM I've run into that eats FET's on occasion either.
Ideas on where to look?

[Radio Ranger]

Chuck, If the RF and PDM sections of the Gates 5 is similiar to the SX-5 and I think it is, then there's a procedure in the book that outlines adjusting the phase of the oscillator (?) outputs so that the zero crossing points are the same. How's that for a nebulous statement??!!
Anyway, it's been a while ago and I don't have the transmitter manual here to refer to, but I had a similar problem with blown FETS on the PA modules and blown fuses on the PDM pull up boards, so I went thru the procedure and the problems went away.
As I recall, the text describes exactly how to measure the degree of phase shift, but what's important is to get the time alignment to be something like 90 or 180 degrees apart. In my case,things weren't too far off, but making the adjustment made the transmitter happy. Someone else can probably chime in to make this paragraph read a bit more eloquently.

Steve Brown
Radio Rangers
Minneapolis,MN

[BroadcastDoc]

Actually, I was going to say something similar (and probably not eloquent either!). In addition to what you mentioned, there is a process that I believe they call "balancing" the PDM's. According to the Big H, there should be no more than +5 degree drive phase spacing difference on the PA modules. If it's on the jagged edge, it could cause random failures. I just checked and they mention it in the manual...though this is also an SX-5 I'm talking about. I'm thinking they're probably similar in design.

[Dale H. Cook]

I'm thinking they're probably similar in design.

They are, except that, IIRC, the Gates boxes don't have the PITA metering that the SX series does. Mind you, it has been some years since I have dealt with either.

[DonIam]

Yesterday, I had a GATES1 blow several FETS ..... I've been replacing FETS like popcorn thinking it was lightning related.

Yesterday, relative humidity was below 25% and a strong wind from the northwest. Blue Sky and NO lightning.
Guy wire johnny balls were "snapping" once every three seconds .

I'm now suspecting that the arcing johnny balls changes the load impedance of the transmitter and blows the FETS.

I have a ball gap ( one credit card gap ) / Horn gap and static drain at the ATU output / gap at top of transmitter + rf choke to ground ( factory ). Silver soldered six inch strap everywhere with a single point ground at the ATU output
( the ATU is inside the transmitter building and the tower is 20 feet away )

I could hear the flash-overs from 100 or more yards away, so they are pretty intense.

Don


ps Hello to the group from Huntsville, Alabama

[Deep Thought]

Hello Don, and welcome.

Yeah, that's the problem with guyed towers...the tower can be perfect in a storm but the guy wire segments can charge up and cause even worse problems. There are static shunts (basically resistors) you can clamp across the johnny balls but that is spendy. Better to just deal with it when it comes up.

I haven't run into a situation where this would cause a transmitter malfunction though. Is this a high impedance tower?

[K4WRF]

Years ago, had this same issue with static discharges across the johnny balls. Sounded like rifle shots going off and it drove the magnaphase on the Conti crazy! very annoying but there were no MOSFET's in that thing. As part of much needed rusty guy wire replacements, we added 6' fiberglass insulators where the guys attached to the tower's and about 3' (IIRC) right at the turn-buckles along with the 18kV johnny balls in between. That stopped it! Now that ain't no cheap fix and if you compare the cost, the bleeder resistors that might be the least expensive. Like Mr Deep said, I don't know if that would directly cause the MOSFET failure trouble but that might be serial with some other unknown problem.

Also makes one wonder what the electrical length is between the LTU output arc gap and the PA output bus. LTU+line+Transmitter TEE network+Third harmonic trap + "L" network? I once built a system that totaled exactly 180 degrees; after finally discovering and changing that condition it resolved the issue, that WAS MOSFET transmitter.... The GORILLA balls firing at the tower would place short right on the PA. Possibly, it could be opposite or 90 degrees causing a high-voltage condition at the same junction.
Just a few thoughts.

Best regards Don and welcome to the group....

Let us know what you find.

K4WRF

[DonIam]

Also makes one wonder what the electrical length is between the LTU output arc gap and the PA output bus. LTU+line+Transmitter TEE network+Third harmonic trap + "L" network?
K4WRF

How is that measured ?

All of my other GATES ONE transmitters are fed through 200-300 feet of underground 7/8 heliax.
This one is twenty feet from the transmitter ATU which is inside the TX building.

I did put a 1/4 wave of RG-8 ( still on the spool ) between the TX and ATU... didn't seem to help.

Don W4DNR

[Dale H. Cook]

How is that measured ?

There are a number of components in that electrical length. One is the coax - you need to know the line length. I would determine that with a TDR because I have one. I would check the manufacturer's spec for the velocity factor of the coax, disconnect the far end at the ATU, disconnect the near end from the transmitter, set the TDR to the velocity factor of the line, and measure its length.

An alternative, if you do not have a TDR but have an accurate digital capacitance meter, is to get the capacitance/meter spec for the coax, disconnect both ends, measure the capacitance, and divide by the capacitance/meter spec to get the line length in meters. (Illustrating that there are multiple methods for flaying felines).

As for the other factors, if you know the design of the transmitter output network and ATU network you can calculate their electrical length.

[DonIam]

Ten feet of RG-8 to a Kintronics Main/Alt relay then ten feet to the ATU. Typical LCL ATU ( assume 90deg )......
I know I know ......

I have put .01s on both input and output legs just to DC isolate the TX from the ATU. Didn't see improvement.

Of course, you know that the GATES ONE doesn't have a direct DC path to the fets ..... only the torroids that couple to the output networks. The torroids aren't damaged and there is no pattern of which fet quad may fail.

After last weekend, I'm fairly confident that it's the johnny balls arcing over and changing the impedance faster than the PDM kill can act.

[boiseengineer]

I've dropped a direct short (angle bracket) on the output of a Gates-5 with no failures.
After hours & shut down short the tower base and see what the impedence at the transmitter is.

Side note. When Harris developed the DX series they discovered the the main reason the FETs were blowing with VSWR events. When VSWR would mute the RF amps, the decaying RF envelope would drop in frequency causing the gate of the FET to be out of phase with the RF drive and the voltage would excede the limits of the device. When the DX mutes the RF it replaces the RF exciter drive to the FETs with a sample of the output network so the gate's RF drive stays in phase with any RF feeding back into it.

[DonIam]

Don't have the manual here, but I'm pretty sure the ONE keeps RF on the FETs
( depending on good high value capacitors in the osc and drive amp )

The PDM Kill turns off the modulators.

[boiseengineer]

Yep. The RF drive stays the same and the PA voltage from the PDM goes away.

[Deep Thought]

After last weekend, I'm fairly confident that it's the johnny balls arcing over and changing the impedance faster than the PDM kill can act.

Again, I don't think I've ever seen this cause a transmitter failure. It shouldn't be nearly as bad as an arc gap firing, and shouldn't cause enough of a tower impedance change that the transmitter would see it anyway.

Do you have effective surge suppression on the AC line?

[DonIam]

Do you have effective surge suppression on the AC line?

I am running a 30 amp 220vac ZERO SURGE ( series inductance ) filter and between that and the utilities, I have a 5KW isolation transformer.

Don W4DNR