>
>
>>>IOW, what is the optimum value for the glitch resistor all other
>>>things (surge capacity) being equal? I think Ian recommends (as a rule
>>>of thumb) something like 20 ohms per/KV on his website.
>
>Not my thumb, but Eimac's. Based on the 40A limit that Phil summarizes
>below, it would actually be 25 ohms per kV.
> However, many people tend to
>use less than this (in consideration of the voltage drop in normal
>operation, especially if there is significant ESR in the capacitor
>stack) and it obviously protects the tubes OK.
// Typical ESR in a modern 200uF, 450v electrolytic filter cap is c.
0.5-ohms. Eight such caps would be c, 4-ohms. At 3200v, peak discharge
current would limit at c. 700A. In my opinion, this is not good
engineering practice. And yet there are amplifiers that are currently
being manufactured that use no glitch resistor whatsoever.
>
>>
>>Eimac's bulletin says that they want the peak energy limited
>>to 50 joules. They say that a 50 ohm resistor will limit current
>>to 40 amps. If the value of the resistor is larger than necessary,
>>it will delay the action of the primary breakers/fuses. All these
>>things have to be considered in designing a HV glitch protection
>>circuit.
>
>As Phil says, it's actually based on the *energy* that might otherwise
>be dumped into the tube. For oxide cathode tubes, the limit they quite
>is just 4 joules - yes, four! For tubes up to 1500W, the 40A peak
>prospective current (= Vmax/R) is Eimac's rule of thumb to achieve this.
>
>The limits for tubes with thoriated-tungsten filaments are much higher,
// good point. With a solid molybdneum wire grid, there's no gold to
evaporate.
>but for smaller triodes such as the 3-500Z they quote a limit of 50J,
>based on protecting the grid structure.
>
// Chortles. I have never autopsied a 3-500Z that had a damaged grid.
The most common cause of death is a bent filament helix that contacts the
grid. (photo on p.15 of Sept., 1990, *QST*, "Parasitics Revisited").
>The energy stored in the capacitor bank is typically 100-200 joules (eg
>30uF @ 3kV = 135J) so you have to do *something* to protect the tube...
>not to mention the other components through which the surge current
>flows (like your grid current meter).
// Amen, Ian, unless one has a freebee supply of valves/tubes, or one
has more money than brains. / In a TL-922, the grid meter is protected
with a glitch diode, However, the anode current meter is not, so it
sometimes gets ditched by a parasitic glitch. Been there, done that.
The bottom-line is that smallish diodes are not suitable for glitch
service. After that chapter, I started using 3A-avg./200a-peak diodes.
>
>>The Eimac bulletin is a valuable tool in designing and
>>testing such a device.
>
>It's high time we had a comprehensive "Flashovers FAQ" page!
>
// It's a flashover all right that makes the big-bang and turns off the
ON light. However, after investigating the matter, neither I, nor any
other amp, repair guy that I spoke with ever found a dead mouse or
insect or whatever that could have caused such a clatter. My guess is
that the initiator of is an intermittent regeneration that lies ouside
the tank's passband. Since such energy can not escape, it runs amok.
>As well as Eimac Bulletin #17, my file of photocopies contains similar
>information from Siemens and Burle (RCA), and others going all the way
>back to the 1930s and earlier. It also contains information about
>several kinds of electronic trip circuits.
>
// It takes stong medicine to interrupt several kV at possibly a kA..
>I will put this all together so that it's all publicly available in one
>web page. Then other contributions can be added in (all sources
>acknowledged, of course) too pool everyone's knowledge.
>
// From a practical experience standpoint, 10-ohms seems to protect
3-500Zs and 20-ohms (2, 10-ohm, 10w in series) seems o protect 8877s.
>Can't say exactly when this will be done, because there's a lot of
>material to scan/OCR/edit/write/format (which sounds a lot like my day
>job... which is waiting, and it's Monday) but I *will* do it.
>
cheerio, Ian
- R. L. Measures, 805.386.3734, AG6K, www.vcnet.com/measures.
end
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|