HV diode

This forum is for specialized infomation important to the construction and safe operation of the high voltage electrical supplies and related circuitry needed for fusor operation.
Post Reply
Posts: 51
Joined: Wed Nov 19, 2008 3:31 am
Real name:

HV diode

Post by Eldarion » Fri Feb 20, 2009 1:48 pm

I have some diodes rated for 1kv 1a, and more fore 400v and 5av, both rated at 1 amp, surge of 30 amps, can these be used in a series as a HV diode for a 5.6kv transformer?

Posts: 250
Joined: Sat Jun 30, 2007 8:07 am
Real name:

Re: HV diode

Post by MarkS » Fri Feb 20, 2009 2:13 pm

They sure can. You're going to want to make sure that you have enough distance between the diodes so that there won't be any arcing. So for 5.6 kV put 6 or 7 diodes in series, and if you want submerge in oil.

Posts: 1482
Joined: Wed Oct 24, 2001 6:14 pm
Real name:

Re: HV diode

Post by Starfire » Fri Feb 20, 2009 11:02 pm

If the diodes are rated for 1kv it would be wise to de-rate them to about 66% voltage if you are going to connect them in series. For a 6kv application I would use nine 1kv diodes.

Dan Tibbets
Posts: 578
Joined: Thu Apr 17, 2008 5:29 am
Real name:

Re: HV diode

Post by Dan Tibbets » Sat Feb 21, 2009 11:18 am

Mmm...derating the diodes in series makes sense. I had strung 4 x 1 kv diodes to handle the MOT output (max of ~3000 V peak to peak) in the first stage of a voltage multiplier. It quickly failed-shorted. I wondered if one or more of the diodes was initially defective. I had not considered that there could be that much variation in the voltage tolorance of 'normal' 1 kV diodes. Using 66% of the rated voltage tolorance would give only 2400 V tolorance for the string, so once the weakest diode failed the others would quickly follow suit. In fact, it makes sense to use the 66% rating guideline, then add one or two additional diodes to make the system more fault tolorant.

Question is- how fast does the current drop, diodes heat up in a chain, if each diode is only ~ 90 % efficient? Or, does that only refer to the voltage drop across each diode?*
Also, if each diode switches at ~ 100 ns, how many can be stacked in series befor there are problems? With a 60 Htz AC input, I assume there is alot of leeway.

* I understand that at least some high voltage diodes are actually made up of a series of lower voltage diodes stacked together in one package. I have seen refrences of voltage multipliers only increasing voltages by 87% per stage (vs 100%) due to diode inefficiencies (two per stage). Was this based on their specific example (5 kV input). If high voltage diodes are stacks of smaller diodes, does the inefficiency scale directly with the voltage tolorance? If so, then I guessing there is no disadvantage to stacking seperate lower voltage diodes, as compared to expensive larger diodes.

Dan Tibbets

Posts: 2346
Joined: Sat Jun 30, 2001 5:13 am
Real name:

Re: HV diode

Post by DaveC » Sun Feb 22, 2009 6:10 am

HV diodes (the fast kind) are indeed made of stacked chips... typically about 2 kV per chip.... The PRV is usually about 40% of the breakdown voltage, to provide some headroom.

The derating issue is more related to reverse conduction resistance variations when the diodes get warm. The reverse bias resistance creates a voltage divider of sorts. If the diodes do not all have reasonably similar reverse conduction resistances, then the ones that have the highest values get the largest portion of the total voltage across them.

SInce reverse conduction increases with temperature, paradoxically, the poorest diodes, cause the best to fail. Once one goes... the rest can follow rather quickly.

A string of high resistance resistors, each one shunting one diode, can help keep voltage division linear, at the expense of some losses in output..

Dave Cooper

User avatar
Chris Bradley
Posts: 2930
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: HV diode

Post by Chris Bradley » Sun Feb 22, 2009 6:31 am

If you *need* to link diodes together to get the voltage rating you want, then the above all applies. Under-rate your diodes, use a resistor divider chain, possibly use a capacitor divider chain as well to mitigate big switching differences between them, and use identical diodes.

But for this voltage, why bother? Always best to use electrical components rated for your purpose rather than 'make' them yourself if you can, and for 5.6kV you can easily find the ratings you want as a price you'll hardly notice.

Cliff S
Posts: 48
Joined: Fri Nov 21, 2008 6:25 pm
Real name:

Re: HV diode

Post by Cliff S » Mon Feb 23, 2009 7:09 pm

What are the part numbers of the diodes you would like to series?
Generally speaking, putting diodes in series is not a problem. But there are a few things that can bite you. Assuming the diodes are all of the same type, and from the same, or close production batches, the critical parameters affecting voltage balancing will be closely matched. In general, if we are talking high frequency power supplies, the Trr is the critical thing to watch. If you are using a high frequency drive, (20kHz or over), you probably already picked a diode with Trr under 100nsec. Most ultra-fast recovery diodes in the 400V to 1200V range will have Trr in the 45-100nSec range. If you are running near or over 100kHz, you really need to watch Trr balancing. I have not found reverse leakage to be a concern if one is not running the junction temp above 100C. Above 100C, things can runaway real fast. (Trr starts to increase, aggravating the thermal runaway situation).

The most difficulty I have ever encountered with putting single junction diodes in series is the effect of parasitic capacitance on the voltage sharing. If you look at the physical geometry of the series-diode array, you can envision that each diode may have an equal parasitic C to ground, (for example), or to another electrode. (Oil and potting materials can amplify this effect due to their higher dielectric constant). Now also visualize the top two diodes in this array. The shared node of these two diodes has a capacitance to ground, equal to each node in the array. Now depending on the reverse voltage dv/dt, the top diode will see a disproportionate share of the total array voltage due to this parasitic C. It goes down the array, with the lower diode seeing the least voltage. This voltage sharing issue is present during reveres recovery transients, and externally induced transients (like HV arcs). If you have this issue, unlikely you will be willing to change your overall geometry, and then capacitive snubbers need to be added. (Shunting across each diode to swamp the parasitic C). In these real world conditions, the parameter that may make one type of diode work and another not will be its avalanche energy rating. (The amount of energy the diode can handle during a reverse breakdown condition).

The “diode stacks”, (multi-junction diodes), commercially available are good at low currents. At higher currents, they suffer from one major problem: getting the heat out of the inner junctions. The main way a diode cools itself is the thermally conductive path through the metal leads. In the stacks, the inner junctions will get significantly hotter than the outer junctions. This aggravates the conditions mentioned above, (Trr and reverse leakage), and are the primary reason why you won’t see these types of diodes achieve current rating greater than 10’s of milliamps, or a few hundred milliamps at best. (Even then, they are generally spec’d in oil and at 25 C).

Even with this, the suggested derating of 66% is a good one. You can probably go to 75% if you have some experience to back it up.

Best Regards,
Cliff Scapellati
Executive Vice President of Engineering
Spellman High Voltage Electronics

Post Reply