P-Series Valves, what to do with them?

[Murphyv310]

Hi.
ignore the above.
I was using and old Avo 7 and it is faulty.
Here is some more info.
The chain adds up to 165v, its cold resistance is 90 ohms. the dropper less thermistor is 240 ohms, initial switch on current is just over 700ma with 235v mains.
Same test using a 4UF cap and the heaters left for 20 mins to cool, the initial switch on current is 360ma this stabilises to 309ma which is within spec of the valves.
Hence the reason we don't see valves like the EB91 glowing like a bulb for the first few seconds.

[Murphyv310]

Andy, thanks for doing the simulation. Any chance of doing it with a resistor whose resistance is comparable to the reactance of the cap?

I don't see how a capacitor can act as an approximation to a constant current source. Unless its reactance is substantially greater than the load resistance. If the reactance of the cap (or resistance of the dropper) is fairly large compared to the heater chain then it will indeed approach being a current source. But that isn't what's normally happening in a real life heater chain.

Quite right Jeffrey - see the attached screen-grab : It's a simulated circuit with six parallel paths having the same value capacitor but with progressively smaller load resistances.

I must admit that when I was experimenting with a dropper circuit, the capacitive reactance was much greater than the load resistance giving the apparent results I described.

Andy

Hi Andy.
The figures are just about what I get, a 4UF cap with no series resistor and 235v AC into the load that adds up to 165v gives me 309ma.
From my calcs the heater chain would be equivalent to 550 ohms when hot?

[ppppenguin]

Andy, I just crunched the numbers in your simulation from first principles and got small differences from your numbers
203.8
258.2
280.3
286.8
288.5
288.9
Your currents are about 2 or 3mA larger.
Very odd.

Trevor, I can't make your numbers add up. Again from first principles.
X = 1/(2*PI*4e-6*50) = 796R
R(cold) = 90
Z (cold) = SQRT( 792^2 + 90^2) = 801 ohms
I = 235/801 = 293mA

That's the cold current.

The hot resistance of the chain should be 165/.3 = 550R. This gives a current of 243mA.

[Murphyv310]

Yes you are correct Jeffrey. The numbers don't add up and I've just poured a red wine to calm down. I'm using a true RMS meter and only monitoring the heater current. I did the calcs the one on the UKVRRR. I do think it came out at 4.4uf but found the 4uf to give me just over 300ma on hot valves, it is though a 20% motor run cap.

[pwdrive]

I also banged the figures into my calculator and scribbled a bit on my pad using the known formulas, considering my sometimes dodgy calculator, the figures that came out agree as good as with Jeffery's.

Lawrence.

[Murphyv310]

Figures Figures Figures, my head is nipping.
Ok so can we explain why we don't get the switch on surge?

[Murphyv310]

My total resistance of 790 ohms in the dropper circuit is just about spot on.

[ppppenguin]

For a capacitative dropper on 235V mains and a 165V 300mA chain the computed value is just shy of 6uF.

790R (resistive) and 235V gives 297.5mA.

Running through the sums again, with a 6uF cap, gives a surge current of 437mA into 90R load and a running current of 307mA.

A resistive dropper would need to be 240R. Surge current is 712mA into 90R load, running current 297mA.

So the cap certainly gives a much lower surge current. Why? X=530R for a 6uF cap at 50Hz. So it's closer to a current source than a 240R resistor. Certainly the smaller the load resistance the closer the approximation to a current source. I suspect it would be easier to visualise if I drew the phasor diagrams.

In the limit, with a short circuit load, the cap will take 443mA, the 240R dropper will pass 980mA.

[Murphyv310]

Thanks Jeffrey.
Now here's a thing, Up to now I have used a 4uf in this particular set and thought and checked I was around the 300ma mark. I agree that the value should be nearer 5.5uf and even today all looked fine with a current of 309ma with heaters looking correct. not dull or overly bright. For the life of me I must have mistyped in Paul's calculator but with the cap marked 4uf the current was OK. So I disconnected the cap and checked it, and it was wrong as was the other two I have here that are unused. These are Italian caps bought from CPC so this explains the anomaly, no wonder my my head couldn't get round the problem. It is well over a year since I got the caps and I have also passed on at least one as well.
So yes I have made an error on the calcs but I have been unlucky with caps that are most likely wrongly marked from new, what a PITA. At least we agree that the surge with a cap will be less, I've done a good few sets with capacitor droppers and have had total reliability. Oh well you live and learn and now I shall make sure I check these caps, I have to say that I did come across this with my last job with start/run caps not at the value that is stated, I just thought my employer was buying cheap rubbish.

[ppppenguin]

At least we've cleared up the mysteries. It certainly wasn't obvious to me why a cap should give lower surge current than a resistor and it's still not intuitive why this should be so. The calculations prove the point but that's different from instinct. A diode dropper won't reduce the surge current, in fact it makes matter worse. I've just crunched the numbers for Trevor's example. As it happens a diode gives almost exactly the right amount of drop with no extra resistance needed.

With a resistive dropper the surge is 727mA. With a diode it's the efective mains voltage (165V) divided by the cold resistance of 90R. Just over 1.8A. More than double!

That surprised me too. To try and reach an intuitive answer, the greater the fraction of the chain that has a low cold resistance, the worse the surge. I'll think twice before using a diode in the future.