From: jeroen@nospam.please   
      
   On 1/18/26 00:33, john larkin wrote:   
   > On Sat, 17 Jan 2026 15:58:01 +1100, Bill Sloman    
   > wrote:   
   >   
   >> On 17/01/2026 4:19 am, john larkin wrote:   
   >>> On Sat, 17 Jan 2026 03:59:00 +1100, Bill Sloman    
   >>> wrote:   
   >>>   
   >>>> On 16/01/2026 11:01 am, john larkin wrote:   
   >>>>> On Thu, 15 Jan 2026 23:01:38 +0000, John R Walliker   
   >>>>>  wrote:   
   >>>>>   
   >>>>>> On 15/01/2026 18:15, john larkin wrote:   
   >>>>>>> On Thu, 15 Jan 2026 17:51:59 +0000, liz@poppyrecords.invalid.invalid   
   >>>>>>> (Liz Tuddenham) wrote:   
   >>>>>>>   
   >>>>>>>> john larkin  wrote:   
   >>>>>>>>   
   >>>>>>>>> On Thu, 15 Jan 2026 15:18:31 +0000, liz@poppyrecords.invalid.invalid   
   >>>>>>>>> (Liz Tuddenham) wrote:   
   >>>>>>>>>   
   >>>>>>>>>> john larkin  wrote:   
   >>>>>>>>>>   
   >>>>>>>>>>> I need something like 1.5K resistance across a 750 volt pulse.   
   >>>> Pulse   
   >>>>>>>>>>> widths will be below 1 us.   
   >>>>>>>>>>>   
   >>>>>>>>>>> Three 1206's in series, 499r each, would work. Peak power   
   >>>> dissipation   
   >>>>>>>>>>> per resistor will be 125 watts at 250 volts. I think that's OK   
   >>>> but I   
   >>>>>>>>>>> want to test it.   
   >>>>>>>>>>>   
   >>>>>>>>>>> Here's the tester. The DUT (device under torture) will go   
   >>>> across the   
   >>>>>>>>>>> gap on the left.   
   >>>>>>>>>> f   
   >>>>>>>>>>> I have both regular thickfilm resistors and some thinfilms to   
   >>>> test. I   
   >>>>>>>>>>> theorize that the thinfilms will hold up better.   
   >>>>>>>>>>   
   >>>>>>>>>> Would a non-inductively-wound wirewound resistor work well   
   >>>> enough?  You   
   >>>>>>>>>> would have plenty of mass to average-out the pulse energy.   
   >>>>>>>>>   
   >>>>>>>>> WWs are great for pulse overload, not so great for PCB density. The   
   >>>>>>>>> best would be to use three (or two, or one) surface-mount 1206   
   >>>>>>>>> thickfilm that we have in stock.   
   >>>>>>>>>   
   >>>>>>>>> I could stand a micohenry or so parasitic inductance. The 1.5K   
   >>>> will in   
   >>>>>>>>> fact be in series with a small inductor.   
   >>>>>>>>   
   >>>>>>>> There's your answer; make the resistor and the inductor one and   
   >>>> the same   
   >>>>>>>> component.  For a small investment in suitable machinery this   
   >>>> gives you   
   >>>>>>>> total security of supply, quality control and an edge over any   
   >>>>>>>> competitor who can't make things but just buys them in (or tries   
   >>>> to copy   
   >>>>>>>> your design without realising what that component really does).   
   >>>>>>>>   
   >>>>>>>> Vertical integration was the cornerstone of nearly all the successful   
   >>>>>>>> electronics firms.  (Philips even owned the sand quarries to   
   >>>> supply the   
   >>>>>>>> sand to make the glass to make the valves and light bulbs.)   
   >>>>>>>>   
   >>>>>>>> Experiment with winding a number of turns of resistance wire on a   
   >>>> former   
   >>>>>>>> in one direction, then winding some more in the opposite   
   >>>> direction.  The   
   >>>>>>>> ratio between the two sets of turns can be adjusted to give the   
   >>>> required   
   >>>>>>>> inductance and the total number of turns gives the resistance.  The   
   >>>>>>>> former could be a small piece of heatproof material shaped like a   
   >>>> dog's   
   >>>>>>>> bone to retain the wire, with a notch to catch the wire and prevent it   
   >>>>>>> >from unwinding at the reversal point.   
   >>>>>>>   
   >>>>>>> Yikes. That would be a huge diversion from getting a product done.   
   >>>>>>>   
   >>>>>>> I found one paper that shows that thinfilms are tougher than   
   >>>>>>> thickfilms, but thinfilm MELFs are even better. That makes sense.   
   >>>>>>>   
   >>>>>>   
   >>>>>> When I visited the factory of a smart meter manufacturer I noticed that   
   >>>>>> they used melf surface mount resistors for mains voltage sensing.  There   
   >>>>>> were several in series.   
   >>>>>> John   
   >>>>>   
   >>>>>   
   >>>>> Makes sense. For a given pcb footprint, they have about pi times the   
   >>>>> surface area to work with, for a correspondingly bigger conductor   
   >>>>> area. The cooling might be even better.   
   >>>>   
   >>>> This misses the point. The Vishay resistor data showed that - at least   
   >>>> for their surface mount thin film resistors - the heat doesn't get   
   >>>> beyond the resistive track itself for about 300usec.   
   >>>>   
   >>>> If you get the track too hot for any time shorter than that it can melt   
   >>>> (or at least get hot enough to let the atoms move around). For their   
   >>>> resistors, nothing lower than 10k can take 1kV, which equates to a peak   
   >>>> current of 100mA.   
   >>>>   
   >>>> Once you've work out how much resistive area you need to use to work   
   >>>> with any pulse shorter than 300usec, you then need to work out the duty   
   >>>> cycle of your short pulses and make sure that you can dissipate the   
   >>>> average power to ambient without getting the average temperature too high.   
   >>>   
   >>> I don't expect to have much average power dissipation. The resistor on   
   >>> my prototype is rising about 15c at 180 watts and 1 us/1KHz pulses,   
   >>> according to my thermal imager. Do the math on that.   
   >>   
   >> You'd need to identify the resistor so that I could get the thermal   
   >> resistance of the resistor to ambient before I could do that. If you   
   >> knew what you were talking about you'd know that   
   >   
   > I'll reveal the secret mathematics:   
   >   
   > 180 watts at 0.1% duty cycle is 0.180 watts.   
   >   
   > 15c divided by 0.18 is 83 watts per degree C.   
   >   
   > Please keep this confidential.   
   >   
   >   
   >   
   > John Larkin   
   > Highland Tech Glen Canyon Design Center   
   > Lunatic Fringe Electronics   
      
   Degrees per watt.   
      
   Jeroen Belleman   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)