Problems with carbon composite in a HV string.

Pure curiosity: what kind of power supply is your LHC based on? ;-) To me (an amateur) it seems insanely hard to produce 2MV of rectified current by electronic means. An old school van de Graaf generator or sth. like that? Please tell more! :-)

Best regards, Piotr

how so?

negtive TC could mean the the hottest ones get the least energy.

--
umop apisdn 


--- news://freenews.netfront.net/ - complaints: news@netfront.net ---

So the above was with SF6 gas? Did you say what the pressure was? I'm not really sure, but I'd guess that SF6 has a lower thermal conductivity that air*. And so that a 2 W resistor in air may not be as good in SF6. (and perhaps you've gotten a new batch of R's that are designed a bit closer to the edge... so to speak.)

George H.

*well there will be convective cooling too.. and who know what that is...

But a single resistitor or a string can run away if there's a constant voltage applied. True, they'll run away together. Kinda romantic.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation

Hi, Maynard -

Years ago I attempted to use a string of carbon comp 22M resistors to sample an HV supply. I was totally surprised that, as the voltage went up, the resistance of the string decreased. I asked AB if it was possible that the resistance changed with voltage stress, and they emphatically said yes.

If you are using a FLIR camera to view the heat being dissipated, you can dismiss this reply. If you are not, you might want to consider that the resistance has reduced to the point of overheating due to the impressed voltage.

In any case, this is just a suggestion.

[resistor failures in an accelerator, using SF6 gas fill]

Maddening, that a 2W 22Mm resistor ought not to be expected to tolerate the voltage across its terminals that generates 1.1W. How would it ever tolerate a 2W test?

Sadly, this is believable. The density of SF6 gas, too, might be modulating the power-handling capability: 2W in free air isn't the same as 2W in (compressed?) dense gas. I'd call around to several big van de Graaf operational sites and ask for experienced help (or for the phone contact info of retirees, because there aren't a lot now running).

uses a 275Kwatt tube (head room), as an oscillator at 100Khz into RF pie transformer. On the other side of the PIEs you have the secondary and at full bus voltage on the oscillator (15kV), you can exceed 50kv.

For rectification to DC, a schenkel diode string is used. This string is composed of diode modules. Each module is a complement of 300 hundred or more SI diodes in series. You can think of this a single diode that can handle 50+kv.

Schenkel multiplier

• | Secondary Xformer output | | | ++---------------+-----------------+ | | | --- --- --- --- --- --- + + | + | | | +---+++|

Jamie-

Your "fix" still stresses each resistor to 4KV. What is the manufacturer's voltage rating of your resistors? I think a garden-variety 2 Watt carbon resistor might be rated for 500 Volts.

How confident are you that other similar resistors will withstand 4KV? Unless you develop statistics by testing a large number, you can't be sure an over voltage sensor will solve the problem.

Another solution might be to use a longer string so their rated voltage is not exceeded.

Fred

We only put ~ 95 lbs of gas in the vessel, its not much but it has to stay no less than 75ls and can not be over 105. The gas is for insulation of course but it also acts as a dialectic for the corona rings, which is the capacitors for the rectifier multiplier..

The problem is not thermal in this case, it is a unit that has other issues in the HV voltage metering auto ramp circuit and is causing the rectifier stack to exceed the voltage rating of the resistor string. Doing the math on a 22MEG 2 watt r's of the composite types (RC), you can not apply the required voltage that would bring it close to the 2 WATT operating point, the required voltage is higher than the package can handle for constant use. You can do impulse with very short duty cycles that would equate to 2 Watt current.

I found with the 22MEG 2 Watt R, you will only be able to use it 100% duty at not more than 1 watt on the average. I think this must be one of the reasons the 22MEG@2W is the limit on this class of R. So no more than 5Kv. I did some test and at 5K across the R, it's fine, however, get near 6K and the package will split!

The unit we are having issues with at the moment (small one), normally does not exceed 4k across each R at MAX Voltage.

Jamie

Yup, carbon comp have quite a pronounced voltage coefficient, or nonlinearity rather. Besides the failure mode, I guess. Which I didn't know about, not that it matters as.... who the heck even uses comp anymore?

High voltage means either, to put it in technical terms, a Metric Shitload of parts (required to stay within, say, 80% of rated voltage per resistor), or thick/thin film types that are rated for high voltages.

Film resistors are absolutely the way to go, 10kV in a single part, well within full ratings, quite common. They usually aren't cheap, which is to be expected of special purpose parts. 'Cuz hey, it's not like there's a consumer application for the sensing of 2MV...

Laser trimmed precision ratios are also common. Not that Tektronix is exactly a shining example of economical design, but they've used laser trimmed thick films in their CRT HV supplies for years. 2-10kV on a single part, with precision taps, no problem. Not that you'd get much mileage out of a single precision divider part, but I guess it's worth mentioning. The way that'd work is, you'd use one at the bottom; unlikely anyone makes a single piece ratio high enough, so you'd stack the rest of the divider's resistance above it. So, one ratio part at the bottom, only. The remainder of the chain must be calibrated for best results. So, almost zero advantage over doing it fully discrete.

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com

Yes, it's normal for that type and some others to decrease their values in HV ranges.

If you apply a variable supply with uA meter across a R to graph the response change, you can then create a log curve and this would be the V coefficient of the R.

If you want to get critical, you can build in a LOG amp to over come this issue in the monitoring circuit, and we do have a crude example of that in the old non-computerized machine. The other machines have a coefficient table in software to correct for that.

Jamie

--
Sounds to me like instead of doing the right thing, which would be 
to get rid of the carbon comps and use proper resistors, all you 
people are doing is blindly adding epicycles to band-aid yourselves 
around an inherent design flaw. 

John Fields

They can take 4K all day long and what good would a 22MEG@2W be at

500V? It maybe possible the phenolic package not be suited for near paths to drain off to, but these units are suspended in SF6 gas and also far away from any potential for discharge. Like I said, they handle 4K all day long.

The machine these are in comes from a 1970's design and only recently has any of these R's been getting destroyed, but only due to other reasons. I think that has a lot to say for itself. We are contemplating in adding a few wraps of these resistors and recalibrating it, there is room for more. This has come to interest lately when they discovered that the unit will produce more voltage with little modification.

Jamie

Hmm... sounds a lot like a Dynamitron... :^)

We use a 150 kW 5 MeV version to inject a gajillion electrons into acrylic specimens to make Captured Lightning... :)

Bert

--
Bert Hickman 
Stoneridge Engineering 
http://www.capturedlightning.com*********************************************************************** 
World's source for "Captured Lightning" Lichtenberg Figure sculptures, 
magnetically "shrunken" coins, and scarce/out of print technical books 
***********************************************************************

If you knew what you were talking about, you would realize there is a reason for using carbon composite resistors in these units, and it's beyond you so I won't go there.

Non inductive and high impulse ratings are couple.. Jamie

o

nt

u

hat

The reason is that they are cheap. Even Jamie can understand that. what he can't understand are the slightly more subtle arguments that make them a ba d choice.

Non-inductive is nice, but scarcely relevant in the application.

High impulse rating is a bit short-sighted. Negative temperature coefficien t resistors don't respond well to impulse loads.

They have a tendency to form low resistance "hot channels" which isn't real ly what you want in a resistor with a couple of MV across it.

Since the various resistors in a multi-resistor stack won't form hot channe ls at the same rate, the effect is to concentrate the voltage drop across t he resistors that are forming their hot channels more slowly, increasing th e power dissipation in these resistors, possibly to levels consistent with catastrophic failure, rather than hot channel formation.

--
Bill Sloman, Sydney

Count me with Frank. Absent a data sheet to the contrary, I'd bet money(*) the resistors aren't 5kV-capable.

(*) One Barackian dinar.

My 2nd vote would be Jon Elson's theory. (SF6 depressurization-spalling)

Yes, but it also suggests that possibly the resistors have never been stressed like this before, IIUC.

Cheers, James Arthur

On a sunny day (Mon, 5 May 2014 19:53:46 -0400) it happened "Maynard A. Philbrook Jr." wrote in :

So... what your dv/dt with all them caps? :-)

--- So you _won't_ go there, but then you start to go there???

Whatta schlemiel!

They're not called carbon composite resistors, they're called carbon composition resistors, and the only reason you'd be using them is because they're a cheap way to get a bulk resistance which can handle high momentary peak loads.

However, your entire discussion seems to be based on what happens to a single resistor when you test it with high voltage DC across it which, if that's what it's subjected to in real life, is another problem.

From:

"Advantages and disadvantages

The big advantage of carbon composition resistors is their ability to withstand high energy pulses. When current flows through the resistor, the entire carbon composition body conducts the energy. The wirewound resistor for example, has a much smaller volume of the wire to conduct. So the thermal mass of the carbon composition resistor is much higher, resulting in a higher energy capability. Carbon resistors can be made with a higher resistance than wirewound resistors, and are considerably cheaper. However, the properties are less good in terms of temperature coefficient, noise, voltage dependence and load.

Fifty years ago, carbon composition resistors were widely used in consumer electronics. Because of the low stability of the resistance value, this type of resistor is not suitable for any modern high precision application. For example, the resistance value can change up to 5% over a shelf life of one year. With heavy use the value changes even more: up to 15% for a 2000h test at full rating with

instability is inherent to the design of the resistor. The carbon composition contains materials with different heat expansion properties. When the conducting carbon particles and the nonconducting binder heat up or cool down, stresses arise in the resistor body. The mechanical contact between the conducting particles will change, and this leads to a change in resistance value. Also noise properties are not good due to the mixture of different materials. The noise level increases when current flows. Resistors of 0.25W and 0.5W, have a maximum voltage of respectively

150V and 500V. Insulation resistance is poor with approximately 10^9 O (order of magnitude worse than other types). One more reason for the decrease in the use of this type of resistor, is the high

So, it seems to me that an appropriate approach to doing away with the problem - for good - would be to replace the carbon comps with resistors specifically designed for pulse service instead of happenstance providing that function and, of course, to fix the electronics you seem to think is broken.