popping rectifier diodes

[ganzonimx]

Hello there and Mary Christmas to everybody on the board., it has been a loooong time since I showed up here, as i got kind out of amp building and repair but as a I go troubles here I'm back again..   

Now I.v got problems with  my 6xKT88 Bass amp. Last week it popped the 8A power fuse and the 1A fuse on the OT centre tap right in the moment of switching on the amp. It was a friend of mine operating the amp so I have to rely on his telling.

I tested the circuit without tubes and found, after popping the fuse again, shorted rectifier diodes. I use 1N4007 in a classic full wave centre tap circuit. I use 2 parallel diodes on the PT is 590-0-590 VAC and all 4 diodes were shorted. I installed now 3 paralleled 1N4007 on each side and powered up again.. everything looked cosy so I installed the tubes. I found normal voltages on al pins and tried with a signal but got no output.

I went to sleep that night and checked again today. All voltages with range, normal cathode voltages on all pre amp and power tubes, normal plate voltages but not even static noise on the speaker. I kept on checking and than after about 5 min  the main fuse popped again, just the main fuse. I found again burnt rectifier diodes on both sides.

I thought to check tomorrow with a current limiter again, but I cant recall any reason for this behaviour and thought to ask fist if a I miss something obvious...

so thanks in advance for your help as always!!

Chris

[PRR]

> the PT is 590-0-590 VAC

Do you really have 800 Volts B+ ??

IAC, this amp is far too big to be using 1N4007. You need well over 1,600V 1A rating per diode. IMHO you should be using 2,500V 3A, a large rare part.

[birt]

and you might want to check the power rail and power tubes. if it worked for a long time diodes rarely go bad all at the same instant.

[kagliostro]

You need well over 1,600V 1A rating per diode. IMHO you should be using 2,500V 3A, a large rare part.

Can 2 of this in series be a good alternative with low price ?

Fairchild FFPF10F150STU

1.5KV 10A 2-TO-220F



http://docs-europe.electrocomponents.com/webdocs/0d1d/0900766b80d1dd40.pdf

http://it.rs-online.com/web/p/rettificatori/6708902/

€ 0.90 each

Kagliostro

[Tiny_Daddy]

Series a couple of Doug's 3 Amp 1000V diodes. You will need balancing resistors and capacitors, for example 1 meg and .01 ceramic disk across each diode. We used to do this in "the old days" for that 3KV transmitter supply. Reference ARRL handbook, 1965 or so.

[ganzonimx]

PRR: yes there is about 760-780 VDC on the plates and the amp worked without any trouble for over  400 - 500 hours...

I know the design is pushed over the limits and I got criticized before for it ... if the Pt would die one day I would get a replacement with lover voltages...

TD : thank you for the input, I'll write that down as a possible solution ... but I feel I should get clear about the reason why the start popping now after all these hours of working..

[Tiny_Daddy]

I have found that the usual cause is an output tube that shorts momentarily when it is hot. Also could be the output transformer arcing.
It doesn't work, paralleling diodes for more current. One of the diodes will always hog all (or >90%) of the current.

[sluckey]

I use 1N4007 in a classic full wave centre tap circuit. I use 2 parallel diodes on the PT is 590-0-590 VAC and all 4 diodes were shorted. I installed now 3 paralleled 1N4007 on each side

A 1N4007 is rated at 1000v PIV and 1A forward current. Using two in parallel 'may' double the overall current rating but the voltage rating remains the same. Using three diodes in parallel 'may' triple the overall current rating but the voltage rating still remains 1000v PIV. You must increase the voltage rating.

yes there is about 760-780 VDC on the plates and the amp worked without any trouble for over  400 - 500 hours...

You have been very lucky. You must increase the voltage rating.

but I feel I should get clear about the reason why the start popping now after all these hours of working..

OK, here's why. Your plate voltage is 780VDC. Before the tubes start conducting the B+ is 1.414 x 590 = 834VDC. You would think a 1000v diode could handle 834VDC. Well it can, but that's not the whole picture.

There is 834VDC on the cathode of the diode. But what's on the other side? There's 590VAC _RMS on the other side. That equates to 834VAC _PEAK. This voltage is positive 834v for one half cycle of the AC and it is negative 834v for the other half cycle. See where I'm headed?

During the positive half cycle, there is positive 834v on the anode of the diode and there is 834v (B+) on the cathode of the diode. So there's zero volts across the diode. Actually the forward drop is about .7v. This sure ain't popping the diodes.

But, during the negative half cycle, there is negative 834v on the anode of the diode and there is 834v (B+) on the cathode of the diode. So there's 1668 volts across the diode! POW!!! And when one pops, the others will likely follow quickly. Hopefully the fuse will prevent the filter caps and PT from smoking as well.

This is why the voltage rating of a diode needs to be twice the rating as the peak (not RMS) rating of the applied AC voltage in this simple conventional full wave rectifier. This is bare minimum. For safety, you need to increase this voltage rating even more!

Get some better diodes. You don't need those skinny 1N4007s paired with a 6 pack of 6550s!

[Tiny_Daddy]

One other thought: I was informed by Motorola that they do not "make" 1000 volt diodes. A wafer is cut into diodes and the various voltage ranges are selected out. The diodes with the sharpest turn-on knee are sold as 1000V.

[ernie_jr]

sounds like you would need more than 1000 volts if your P/T is 400-400 volt as well
Ernie

[ganzonimx]

thank you sluckey, at least I learned something substantial about rating of rectifier diodes.. je je je ...

I will try to make the amp work with the 1000 A rated diodes installed and a light bulb in serie. If I can't find any other error I buy some of the suggested higher rated diodes and have a try with these. That will take some time to get them, but I post back once I have tested with the light bulb.

thanks so far... Chris

[Colas LeGrippa]

what I do now, is use an already done diode bridge, 10A. Fast installation, reliable, CHEAP. If you go with this, dont use the B- but the Pt center tap or you will have a big surprise.
just hook up the 2 HV from the PT, and use ONLY the B+ . 

[HotBluePlates]

what I do now, is use an already done diode bridge, 10A. Fast installation, reliable, CHEAP. If you go with this, dont use the B- but the Pt center tap or you will have a big surprise.
just hook up the 2 HV from the PT, and use ONLY the B+ . 

Hoffman doesn't seem to stock a 10A bridge. The ones I found at Mouser were only rated for 1000v, which leaves us with the same problem.

But six 1N5408's are only $3; you get a composite rectifier that's good for 3A, 3kV.

I think we see from this example that under-rated parts will still work, maybe for a considerable time, before dying suddenly.

[PRR]

> I feel I should get clear about the reason why the start popping now after all these hours of working..

You need 1,668V rating. That's the rule for the 2-diode CT capacitor power supply. On the first half-cycle the winding swings up to 834V peak, on the next half-cycle it swings 834V the other way while still hanging on the cap's 834V... 1,668V across the diode. (If I am repeating Sluckey, it's cuz the point bears repeating.)

The "1,000V" diodes are not sure to break-over at 1,001V. Long ago I tested a bag of "60V" diodes. Some were soft at 80V and most were really 97-115V breakdown.

A 1N4007 is sure to be good for at least 1,000V. It may be good for more; maybe 1,800V or maybe 1,100V.

We got the 1,668V number from nominal PT voltage. Power lines have spikes. Very short spikes may not be instant death. They may sustain micro-damage. If this happens many times, the diode is weakened enough that it may fail "for no reason", but really worn-out by repeated stress.

It is good policy to pick diodes for TWICE the calculated voltage. So you want 3,400V rating. This is difficult.

There's also current. A six-pack KT88 is liable to pull 0.8A of DC at full roar. That's 0.4A per diode which seems OK. But the peak currents in cap-input rectifiers flow in narrow spikes of very high current. As a rough rule of thumb you should double or triple the current rating of the _PT_. This suggests 2A or more.

It was VERY common in the early years of Silicon rectifiers for rectifiers to blow after many dozen starts or some hundreds of hours. I made a mini-career of fixing gear, and got tired of exact-replacements failing over and over.

Oh, and Sluckey sure is right. Once one diode goes, they all go. Domino effect.

> Series a couple of Doug's 3 Amp 1000V diodes

That's perhaps a bare minimum to be pretty-sure it won't come back dead in another 100 starts or 1000 hours. I would really seek a 3,000V rating.

> use an already done diode bridge

As you say, the FWB needs a different connection using half the winding, which is lossy and wasteful. Probably fine for smaller amps. Six-KT88 is a very heavy load. And 1,668V is well beyond common FWB ratings.

> they do not "make" 1000 volt diodes

Well, they try for over-1000V but historically they get a lot of "rejects" which have to be sold as 100V or 400V parts. The old sort was with a very low-current near-breakdown, non-destructive. Maybe now they know their stuff so well they can by tests far below the knee.

[kagliostro]

Source for free high voltage diodes

broken microwave oven

Kagliostro

[ganzonimx]

ok I tested with a light bulb in series and I get sound, so it was just the popped diodes that made the problem. I'll go with some higher rated diodes and have to make a decision:

the ones Kaliostro proposed:

Fairchild FFPF10F150STU,Rectifiers 10A/1500V Damper for $ 0.83 each

ore some

IXIS DSA1-18D, Diodes (General Purpose, Power, Switching) 1800V 2.3A for $ 4.46 each

After Sluckeys and PRR's reading I guess I have to go with the IXIS even so I'm not fare enough into save territory. 

What about the microwave oven diodes? are the higher rated? 

 

[HotBluePlates]

I say again...

I'd buy six 1N5408's. Put three in a series string; do the same with the other three. Use each string as you would a single diode. The result is rated for 3A, 3kV. Hoffman sells them for $0.50 each; that's still cheaper for three (for a string) than two of your cheaper diode opitions (also to create a string).

[ganzonimx]

thank you HBP!

I dind't understand that first.. so diodes in series add up there voltage rating, but in diodes parallel do not add up there amp rating? Am I correct?

[eleventeen]

Forget diodes in parallel. It's just something that you don't do. One conducts first and gets the vast majority of the load current, and if that is beyond its rating, it blows. The idea that they would equally "share" the voltage and current is just not correct. It's not a democracy! It's a diodocracy!

[Willabe]

It's a diodocracy!

Ahhhh!   Dog Gone Diodocracy!         

                  Brad                

[HotBluePlates]

thank you HBP!

I dind't understand that first.. so diodes in series add up there voltage rating, but in diodes parallel do not add up there amp rating? Am I correct?

Right. That's what Eleventeen and Tiny Daddy are pointing out.

And we've over-rated the reverse voltage rating of the three series diodes compared to what your amp requires, just in case the voltage doesn't divide equally among each diode.

[ganzonimx]

  to all of you, I  ordered the 1000V/3A and installed for the moment tree en series 1000V/1A diodes  as we need the amp still this year ..


happy 2012 to all of you and 'tll next time!!