A Power Supply Troubleshooting Mystery - can you solve it?

[K1JJ]

I just love troubleshooting rigs or helping out others to do the same. It's even more fun when done remotely using "verbal troubleshooting."  After a while we learn a lot of tricks.

Can you solve this one?  I told it on the air today and the guys on 3885 quickly solved it. If you heard the solution, hold off and let's see if others can figure it out...

Today I turned on my homebrew 2KV HV supply used for the 3-500Z linear. The power supply went "POW!" with a flash and a bit of smoke and smell - just like a dead short.  I figured it meant a lot of bad parts and work.  The supply was now completely dead when switched on.

The HV supply is cap input using a string of six beer-can-size electrolytic caps, a Variac, a bridge rectifier string, 240 AC input and no step start.  The only KEY clue I will give is that both 3AG 240AC fuses were still good.  (they were within normal fast-blow ratings for the PS)  How can this be?  The fuses were still good?!!   It was not a bad wiring/solder connection.

The actual reason for the crap-out was the last thing you might expect after a big bang like that.  It took me about 30 minutes to find because I had a "mental bias" and went down the wrong path at first. Afterwards it all made perfect sense.

What was it and why?

I'll wait for a day to post the answer unless someone gets it sooner.

T

[ka1tdq]

My guess is maybe no filament voltage on the 3-500's? I've Googled to see what applying B+ to a tube would be if that were to happen but can't find anything. I've never done it either so I can only guess.

Other than that I can't think of anything.

Jon

[w9jsw]

Glitch R blew?

[w3jn]

The bridge opened up?

[W1RKW]

The strap snapped

[WD8BIL]

Rodent or bug chewin' on things?

[K1JJ]

Most of you are on the right path, but no cigar yet.

Another clue contained in the first post: big capacitor bank, not using a step start...  think about the cause and effects of that plus the good 3AG fuses.  The amplifier itself is OK.... just a bad power supply.

To win the prize, identify the bad component(s) and why the symptoms/clues were relevant.

The ~75 MPH windstorm is getting intense here, so may lose power soon.

T

[WD5JKO]

Fried contacts on the AC primary keying relay....

Jim
Wd5JKO

[W7TFO]

Maybe a line surge, and you have SB fuses in the holders. The peak set up a very low impedance in the Variac and opened a connection in the internal mains wiring.

73DG

[AG5CK]

Did the power switch fail because of the inrush current?

[KB2WIG]

Ghosts.

[w8khk]

I am going to go out on a limb here and suggest that the root cause of the failure is the attempt to charge a bank of six 450 volt capacitors in series on a voltage peak of the input power with no step start.  The equalizing resistors across each capacitor serve no purpose when charging at high current into fully-discharged capacitors.  The total voltage rating of the bank is estimated at 2700 volts, probably sufficient with a step-start circuit.  But because the capacitors are not identical in capacity or ESR, they did not charge equally, thus placing a higher-than-rated voltage across one or more of the capacitors, causing a shorted capacitor.  The slow-blow fuses did not open, but the main breaker for the 240 volt source opened before more damage occurred.  If this initial analysis seems quite foolish, chalk it up to the fact that it is still April, and I am sleep deprived due to thunderstorms and hail all night.

[w4bfs]

variac moving contact arced between windings due to uncharged secondary reflecting a low impedance to the variac.   no step start means turn variac back to zero each time before starting a high power supply. 

yup been there done that 

[K1JJ]

Jim/WD5JKO:  "Fried contacts on the AC primary keying relay."

Travis/AG5CK: "Did the power switch fail because of the inrush current?"


BINGO!

Who would ever suspect an innocent open switch?

Cap input, huge inrush current on turn on...No AC primary relay, but close enough. The AC primary switch contacts blew themselves apart internally. This was a 2-pole DPDT regular 10A AC toggle switch that worked for about three years. But like fuses, contacts can fatigue after a life of surges above their ratings and then self-destruct.  Thus, we had a big bang from an internal sustained arc, the fuses were still good and there was no power when the switch was turned on.  There was also black soot on the switch, but I couldn't tell you or it would give it away.

If I had used a rated primary AC relay like Jim / WD5JKO suggested, it would have never happened in the first place.  

The troubleshooting problem started when I assumed it was a dead short because of the fireworks.  I quickly found the bad switch contacts and the good fuses, but assumed if I put in another switch it would fry like the first one.  I checked every component in the supply and found it all good with no shorts.  I couldn't believe what I had just measured. So, surprise, surprise...I finally just replaced the switch with a 20A version and it worked perfectly.

I usually use step starts on all my HV supplies, but thought I could simpify things and get away this time.

So bottom line is if surge currents are involved, use a step start or other method of softening the surge. (and a primary relay if needed)   And when there are good fuses when they should be blown, the rig is talking to us and we need to listen to the symptoms.

Good job, guys!

T


** Feel free to add your own Troubleshooting Mystery Theater episode below ****

[K1ETP]

Shame on you Tom Vu for not using at step-start on a cap input supply. What is the total capacitance? Huge I'm sure.

Lich

[K1JJ]

Shame on you Tom Vu for not using at step-start on a cap input supply. What is the total capacitance? Huge I'm sure.
Lich

Yep, I'll take the heat for the sake of teaching a lesson.  The six caps are each 600 uFd at 400V  =  100 uFd at 2400 V total.    Now that you mention it, I will add an input filter choke and set the Variac  for 1500V as planned. The choke inductance at 60 Hz will dull the inrush and let me use the 20A switch alone as is. Better regulation too for feeding a linear amp.

The real story is that I intended to use it at 1500V and figured the lower voltage and Variac resistance would add enuff buffer to run with a switch.  It did FB for 3 years, but like a piss beat tube, they always fail eventually from fatigue.

T

[Chuck...K1KW]

A choke will increase the voltage sag with current draw due to the increased series resistance.  A loser if you have enough output capacity.

The caveat here is that the capacitance is high enough to handle the peak power demand.

A little history:

Back many years ago capacitance was more expensive than iron and copper so choke based power supplies were the standard

Today it's easy to get enough capacitance to use capacitor output on a rectified supply and use the 1.4 X RMS rule for the expected output voltage.

Just make sure there is enough energy storage to handle the peaks!  Here is an easy way to do this.  Take the voltage divided by the peak current to establish a
peak minimum resistance.  Make the output capacitor reactance 1/200 that value at 60 Hz if half wave or 1/100 at 120 Hz if full wave.  You will have a very solid supply and don't forget the soft start!

No chokes needed these daze!

Been doing this for years,

[K1JJ]

A choke will increase the voltage sag with current draw due to the increased series resistance.  A loser if you have enough output capacity.
The caveat here is that the capacitance is high enough to handle the peak power demand.

Vely interesting.  Seems I always end up with a capacitor input power supply in the end after running tests. But I believed the old idea that a choke is better for regulation. My big HV supply has a choke, but I always run it bypassed as capacitor input.   (140 uFd at 10KV)   Very stiff.

I did add a choke on this 1650 VAC RMS bridge supply today and find the resting voltage under full load is lower than I want it to be. It slipped to 1200 V now and I wanted 1500V when cap input.  Considering there is 100 uFd of filter cap in there now, we will probably see the choke is a loser as you said. I'll replace the choke with a manual step start instead. I'll use two switches, the first with a 25 ohm resistor to turn it on the primary.  

Thanks for the heads up.

The 70 MPH wind took out the AC power tonight, so who knows how long I will be off the air to run some tests....  On generator power now. See ya when it's back on.

Yep, too bad on the 4X1.  Any builder seeing that ad probably bought it sight unseen over the phone....

T

[K1JJ]

Chuck,

I removed the choke and added in a 2-switch  soft start using a 25 ohm 20 watt resistor in the 240 primary.  It works really well.

Switch #1 on and the voltage slowly climbs to 1200V in two seconds.  Switch #2 and it hits 1500V, ready for action.   The regulation is good and I can now go well above 1500V with the Variac for more flexibility.

All set!  Tnx, OM.

T
 
To the group -  a little secret:  I sometimes wonder where I get my motivation to demand performance perfection from my homebrew rigs. It has a lot to due with Chuck's influence over many years.  He has spent hundreds of hours with me on the phone and thru email discussing rig performance. Antennas included. I couldn't ask for a better mentor.  He has mastered many hobbies and approaches everything with an attitude of excellence and doesn't accept less. A visit to his QTH is a technical mind blower.   I hope he stays around and posts often on this BB. There's many times he could save some of us a lot of time pointing out the right road when we are making design decisions or when we simply get stuck. There's many times I didn't take his advice and regretted it later -  only to redo things the way he first suggested.   

Still on generator power and waiting to get back on the air... [sigh]
 

[KD6VXI]

I cannot stand having a switch to flip to short our Rstepstart.  This is what I do to make an automatic step start

Use a 120v contactor that shorts out the Rstepstart.

One side of the contactor to neutral or ground (whichever you have), the other side to the transformer side of the step start R.

No more switches.

As the filter C charges the transformer voltage will rise on the xformer.  Then the step start contactor will fire and short out the contactor applying full voltage.

Stopped buying timers years ago using this circuit.

--Shane
KD6VXI

[K1JJ]

Shane,

That's a nice, simple way to do it.  Next time I have it apart I may drill some holes and mount a relay for the job.

BTW, there's no free lunches... there's relay coil power consumed to keep the relay on...  

Tnx.

T

[KD6VXI]

Shane,

That's a nice, simple way to do it.  Next time I have it apart I may drill some holes and mount a relay for the job.

BTW, there's no free lunches... there's relay coil power consumed to keep the relay on...  

Tnx.

T

Some people also add a fuse in line with the step start resistor.  If for some reason the contactor doesn't close, it will prevent burning the resistor up if you fail to notice and hit the tx switch.

Power consumed by the contactor minimal....  BUT make sure you don't have a noisy contactor....  One that hums when it closes.  That would drive me up the wall!

--Shane
KD6VXI

[ka1tdq]

I like Shane's step start idea.

So how about this. Instead of a resistor inline with the incoming AC, how about using a massive inductor choke. I have one that is 30mH at 20 amps. Inductive reactance at 60 Hz is 11.3 ohms. It's not exactly 25 ohms but still a significant amount of resistance. The time delay would be less until the 120 volt relay kicks on.

I'll be charging a bank of capacitors that has a capacitance of 92uF (12 @ 1100uF) in series at roughly 3000 volts. The nice computer grade ones. Someone was practically giving them away new out of the box on eBay.

Jon

[ka1tdq]

Nope. Just thought about it. That wouldn't work in an AC circuit. Damn physics.

Jon