overloading wirewound resistors

re

e

Wirewound is incredibly tough, especially if it is made with welded connections. Crimped connections are not as reliable, especially if you have lots of temperature cycling. I have seen wirewound parts spend weeks glowing red-hot - and when finally removed from the circuit test within their rated tolerance.

I am surprised that a device rated at 600W dies so easily with a kW transient. Is this rating not realistic, the part is failing on some transient which reduces the power consumption to the mere 1kW you measure/calculate; or you are simply wrong about the power consumption.

So its either a thick and heavy resistor body or a very thin one with an enormous surface. What is the resistance value?

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
                     "If it doesn\'t fit, use a bigger hammer!"
--------------------------------------------------------------

I don't think he said this one (600W) is failing. The existing one may be failing.

MRI is a very simple idea but very difficult to implement. I am sure he is forced into linear power managements due to RF radio silence requirements during relaxation and inversion sequences. However, he can broadcast as much RF as he wanted outside the critical sequences. Sound like he is just an outside contractor trying to solve the critical processes. Ideally, the power managements need to be coupled to the MRI sequencings.

Don't forget that they are pulsing high energy RF in a superconducting magnetic field. I think you guys (the OP and UCSF) need to have better communications.

Above, John Larkin wrote: The power resistors in question is in my "boost" circuit, which recharges/overcharges the power supply caps between pulses, so users have extra voltage to make fast leading edges (to drive lots of L di/dt) for pulse-type waveforms. The higher duty cycle imaging waveforms are stressing the charging resistors past what we designed for.

That's why I had some thought this job might be done with an ultra-shielded switcher--if it's just recharging power supply caps between pulses. What's neater than a high-muscle box that runs stone-cold and saves power?

But John knows the constraints and full functions, and I don't. A switcher / hybrid might not make any sense at all.

Cheers, James Arthur

It is assumed that pulsed power handling (joule-seconds) is demonstrated to be within reason on current hardware, from previously succesful lower rep testing testing and operation. The datasheet seems to indicate capacity of 1000joule-sec at 50% duty cycle. This limit does not seem to decrease significantly at lower duty.

This given, even a replacement device that glows intentionally will have it's limits for continuous power in the final hardware.

RL

I believe they had tried switchers in early developments, but noise problems were unacceptable. Anyways, it's just buffering charges between pulses. I don't know how a switcher would run cooler. They still need to pump the same among of energy into the capacitors.

The Ohmite thickfilm thing is 20 ohms. The boost supply voltage is about 180. It gets bad when the main HV supply is set low, 50 or 75 maybe, and a high-duty cycle waveform keeps the caps from charging much between pulses. Some MRI sequences really never rest, but keep making weird high-current waveforms... not really pulses at all.

I'm probably getting 850 watts or so into a 600 watt resistor. If it were an old-fashioned air-cooled wirewound, it would probably be fine. The thickfilm, bolted to relatively thin chassis metal, doesn't like it.

John

That's what I'd expect. But I'd also wonder whether innovative design could overcome that.

If others tried it, they must've been thinking the same, and seeking the advantages. But they may have missed a few tricks.

By not wasting all that resistor-exploding power, methought. But I don't know what the average power is--the idea might make no sense at all.

Cheers, James Arthur

The big medical imagers use switchers to drive the gradient coils. Copley and others make these. They do tend to be noisy, but a human body is sort of a moving target anyhow, so resolution is adequate. My little (120 amp, 150 volt) linears are used for more researchy stuff, with small-bore magnets, for tissues and small lab animals, and the people who do that want resolution. People are tending to image lab animals more and killing/dissecting them less.

I'd like to do a water-cooled linear or maybe a low-noise switcher, but as noted I don't have a sponsor. This heavy stuff is a pain, too: I can sometimes sell one VME module for as much as a rackmount gradient driver, and the VME doesn't need a forklift to move it around.

John

Do you really mean 18kW or "upto 120A or upto 150V"? I don't think MRI needs 120A.

Why can't you go to old-fashioned then? Those can be had with low-L winding technique.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Many of these types of rheostats are wound on a stove-enamelled steel tube, their inductance is quite high [1]. It would be perfectly possible to pulse them with high voltages and yet retain a low dissipation (until they flashed-over).

[1] I discovered this by accident when using one as a dummy load for a P.A. amplifier. As I took the frequency up, the apparent efficiency of the amplifier increased until it exceeded100%. It was only when I measured the load current that I caught-on to what was happening.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Or multiple resistors. Two should be fine.

John

L doesn't matter, but I'm constrained on size. The thickfilm thing was tiny. What would work is using wirewounds IF I can overload them somewhat (2:1 would be nice) based on having lots of air.

John

From the picture you provided earlier, there seems to be lots of space in the front. Perhaps you need to change the box a little?

--
       O

Oooooooh! Isn't that a bit heavy for Ms Polly Prissypants ?:-)

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

You high-falootin' IC designers and your "ohms per square..." :-)