Opinion needed for testing HV diode arrays.

I dunno if I'm grasping the problem.. Something avalanched?? In a CDCDCDCDCDCDC... capacitor diode chain? (voltage multiplier)

Perhaps simulate the failure modes and to develop a test technique.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

"Jamie" wrote in message news:oLKTk.8315$ snipped-for-privacy@newsfe08.iad... | | At work, we have irradiation units that comprise of a vessel | that contains a rectifier stack along with it's chorona cap rings | etc... | Basically, this unit is driven by a single triode (250Kwatts) to | create 100Khz into a transformer which is inside. This transformer | simply steps up, to supply a HV ac of 100Khz, that connects to a | voltage multiplier rectifier etc... | | The official book, states that when testing the diode array modules, | simply apply 35 volts in the forward direction across the connections | and you should see ~ 50ma's or on the amp meter of your supply. | | Question: | Since these units are subjected to much higher voltage, like around | the 30's,50's or more volts for each one, wouldn't you think it would be | more sensible to fault test these units closer to their operational | voltage for break down test? | | Monday, we'll be taking a unit a part "once again", I've made | suggestions this time, to vear away a little from the trouble shooting | bible of these units and start inventing some new techniques, like, | using a HI-POT unit and subject these units to their actual operating | voltage.. | | Apparently, the low voltage and practical guide test suggested by the | manufacturer of these units are not sufficient for the problem we have | at the moment. | | These diode array's are composed of 2.5 amp 1kv generic SI type | diodes, each having a snubber across each. I don't know the exact | count of each, But if I had to take a guess, 50 in each sounds about | right. | The manufacture does not suggest to test the single units inside, | they suggest using 35 volts on the over all that are in series. | | This to me only tells me if one have blown open, how about those | that may have avalanched and there by, has inhibited it's voltage | handling ? |

Forward/reverse resistance is a good indicator of HV diode health. Also, low voltage operation should catch some problems.

Cheers

es,

around

Well, no; the break down is under REVERSE bias, at inconveniently high voltages.

Probably a single unit short (of the dozens of rectifiers in the stack) will not disturb normal operation; if several short, the forward current will show a significant increase, so any important fault DOES show up in your test. Capture the unit conduction readings in a log book, and look for changes with time.

A failure by opening up of one rectifier (less likely) will obviously show up clearly on the forward conduction test.

So, the forward conduction test should do the job for you. If there's something subtle happening at extreme voltage, you can see the corona at the fault point.

We will better know tomorrow.

Trying to resolve this issue also has the techs at the main head quarters complexed at the moment.

Currently, we can bring the bus voltage on the triode up to ~ 5K where we just start to see ~20-30Kv from the end of the main stack coming back via the HVD (high voltage divider) string. If the bus voltage to the triode is increase from this point, we either get an arc at the Spark Gap on the plate (anode) xformer in the RF junction box or, the HVD drops down to 0 volts and you see the grid voltage drop down. If you lower the bus voltage back a little, it will magically output it's 30Kv.

At this point, the anode current isn't even generating enough to show on the display..

I wasn't there when they did the initial test on all the components in the vessel, i'm only going by what was told me.. I kind of told them to bite the bullet and open it..

I was thinking that maybe some caps in side a diode module have generated some low break down voltages which will not show on that simple forward diode test laid out via the bible. TOmorrow, we will be HI-POT testing the modules along with the filament RF xformer for the beam tube.

ules,

ns

e around

The 35V relates to the forward voltage drop. In a typical diode, the forward drop is roughly 0.7V. High voltage diodes are made by putting many diodes in series. Thus, the forward voltage drop of a high voltage diode can easily be 10's of volts. You can find information on multiplier circuits like yours at

Today, 3 units were found to not comform with the operating designed conditions. Using the method laid out by the bible yields these units to be ok how ever, applying 15k DC which is only a fraction of what they are suppose to handle caused 3 units to fail.

So, we took it a step future and found not only do they break down below the expected operating voltage, the capacitance of the failed units has greatly increased over a working unit.. This, in it self is most likely causing the LC tank which composes the rectifier stack of D's to not resonate correctly..

So we summized that maybe some of the capacitors in the HV diode modules are shorted. I was told today that none of these units have ever been replaced because they'ev never seen one fail.. This may explain why we have a couple of other lines in house with poor efficiency even with a new triode which runs ~$12k bucks.

It would be interesting to hear from others that work in this field on this matter.

Always use the manufacturer's specs for testing those parts. At low voltage, a faulty HV cap can appear as being perfectly fine. To HV, however, they appear as a dead short from one plate to the other.

In other words, the individual diode element of a multiplier array do not "see" the final voltage of the multiplier. The most they see will be the multiplier drive signal voltage times 2.

The diodes of a multiplier rarely can handle the final voltage of the output node. At least on the small circuit board scale stuff. We only use the diode needed because they get very expensive, very fast.

Also, if capacitors are "built in" to your multiplier diodes, then they are "multiplier stages". That means that it could be the caps that are what has gone south.

Each diode array module is rated to handle 50KDC.

The over all stack is a Cockroft & Walton multiplier using chorona ring dynodes for the capacitors in a SF6 gas chamber.

Each diode module is an equal of each diode you see in the schematic.

the final net voltage can go up to 1.8 Mv.. The system can actually do more but it's limited for safety reasons.