|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
Hi,I've read here couple times that filament (or heater, as I have no DHT) current draw under a lower filament voltage is higher.
For this I really can't understand as it runs counter to Ohm's law. Unless the resistance of heater reduces faster than the drop of voltage, otherwise there is no way that current increases with lower voltage. And I can't see any reasons for any feedback mechanisms to cause heater resistance to drop. Well, maybe it will do so slightly as a cooler filament will have slightly lower resistance.
Maybe I misunderstood something? Maybe it does happen to DHT?
But I'll appreciate if anyone can at least confirm my belief that for tubes like 12AX7 or 6SN7 that's indirectly heated, the filament current drops with lower voltage. Or a correction will be good too.
Follow Ups:
Whoever told you that the current increases was wrong. The current decreases but it doesn't follow Ohm's Law. As a rough rule of thumb, a ten percent drop in voltage will reduce current by 5%. If you think about it the negative resistance characteristic would make nonsense of current regulating a heater.Your supposition that the filament is not a constant resistance is correct. Tungsten like most other metals has a positive temperature coefficient of resistance of about 4.5 x 10 (-3) K(-1) which is a jargon laden way of saying its resistance increases by nearly half a percent for each degree (Celsius or Kelvin) temperature rise.
When you underheat the heater it reduces in temperature therefore the resistance drops.
Mark,Thanks for the thoughts and info. With my own technical judegments I completely agree with you.
From datasheet of 6SN7 I see 0.6mA heater current with 6.3V. The resistance at working temperature thus looks to be around 10.5kohm.
However, anyone has the info of heater resistance of this tube (and/or others) at room temperature, right at the moment it is fired up?
Or if I can find out biased heater temperature, and assuming constant temp. coeff. w.r.t. temp, the value can also be calculated from this 4.5 x 10 (-3) K(-1) coeff.
Knowing this will also give me some idea of surge current of a cold heater.
Cold resistances are about 5 to 7:1 for typical indirectly heated V/Ts (although the Cathode surface is around 1100K, the heater itself is well above this); 10:1 for typical Thoriated Tungsten dull emitters (2100K) and 14:1 for pure Tungsten bright emitters (2700K)(and light bulbs). The surge current through a DC heated 6SN7 at start up is thus about 4.2 Amps if the supply has zero impedance (not possible, but a large cap approximates this).With AC heating it depends where on the AC waveform the heater is switched in - at the peak of the waveform it is around 6 Amps again assuming zero supply impedance. Again, this is not possible to achieve, but a big transformer operating on mains voltage comes close.
I was curious so I made some quick measurements on random samples of 6-6SN7 pulled from my junk box: all used all test good. Not a rigorous meas set up nor were meter accuracy/precision statements consulted:Tungsol #1 1.54 ohm cold = 4.1A calculated: 0.63A @ 6.3VAC measured
Tungsol #2 1.47 ohm cold = 4.3A calculated: 0.66A @ 6.3VAC
Sylvania #1 1.54 ohm cold = 4.1A calculated: 0.62A @ 6.3VAC
Sylvania #2 1.48 ohm cold = 4.3A calculated: 0.62A @ 6.3VAC
Sovtek #1 - 2.34 ohm cold = 2.7A calculated: 0.66A @ 6.3VAC
Sovtek #2 - 2.35 ohm cold = 2.7A calculated: 0.66A @ 6.3VACInteresting how the Sovteks meas about twice the cold resistance of the US brands but they all draw close to the spec 0.6A nominal when hot.
That would be consistent with a lower purity of the filament tungsten in the Sovteks.
I can understand how lower purity might lead to higher cold resistance but in this case it also leads to lower cold inrush current (a good thing?) and somewhat longer warmup time. I would think that lower purity would also lead to higher hot resistance too but this is not the case in the 2 samples I measured. Could this be intentional so that the term "lower purity" is not a negative characteristic but rather just a different filament alloy designed to achieve reduced cold inrush current and longer warm-up characteristics???Actually, my observation is that ALL Russian tubes take a longer time for complete warm-up, esp 6550 types and EL84M. I'm going to make the same typ of measurements on these tubes when I get the time.
I was thinking about the same thing...My C-J came with Sovtek 6SN7GT's. If this is what they designed their amp for, then it explains why the filament fuse tended to blow out; and the filament supply does look more aged than it should after couple years of use. Maybe the lower cold resistance of NOS tubes do put more stress on the filament circuit? Agree? Disagree?
Clearly the Sovteks will be less stressful at start-up than the NOS if the trends we've measured are consistent. HOWEVER...If I were designing tube equip for the consumer market, I sure wouldn't design so close to the edge that tube inrush current variations would be a factor in reliability. One way around the cold filament inrush current issue is to use neg temp coef "inrush limiters" in the primary circuit of the PT/FT. My primary interest is vintage and whenever I service a unit I always add an inrush limiter to the primary circuit. They really work well for me. Must admit that I got the idea from McIntosh: they used them on most of their tube power amps after the MC30/60 series.
I strongly concur with the use of thermistors, especially in vintage equipment. Not only do they drastically reduce inrush current but they also reduce incoming voltage. My experience has shown that using a Keystone CL90 on 1 leg of the incoming primary will reduce voltage by 4VAC, both legs 8VAC. A good solution to today's higher mains voltage. Very inexpensive too!
Good call guys. I did plan on applying inrush limiter to the primary of a digital amp I'm building (20,000uF), but never thought of applying it to the tube amp.You know what, my stock C-J came with a 4A slow blo fuse on heater supply. But it blew up once a while even the total current draw is only 2.1A. The manufacturer now recommends 5A fuse and I'm expecting it to last much longer. Still, I think an inrush limiter is in order.
The amp is a modern design so I would like to change voltage as little as possible. In this case, with steady state total heater current of 2.1A (~14A turn-on surge), is KC006L-ND from Digikey good enough?
It has the specs of:
No load res @ 25C: 10ohm
Max. steady state current: 5A
Approx. res @ 5A: 0.18ohmOr the next one may be good too (KC007L-ND):
No load res @ 25C: 16ohm
Max. steady state current: 4A
Approx. res @ 5A: 0.27ohmMaybe the later one will have close to 0.2ohm at 2.1A steady state. Then it can simply replaces the in-line 0.2ohm voltage dropping resistor I plan to use.
My experience with inrush limiters is that you need to size them as close as possible to their max current rating if you want min resistance/voltage drop when hot because their temp/resis curve is so nonlinear. Based on conditions posted so far (2.1A continuous at filament trans pri) I'd be more inclined to use the KC008 version:Cold resis = 47 ohm
Hot resis = 0.49 ohm
Max current = 3.0AYou might even get away with the KC009 (the one I use most often)
Cold resis = 120 ohm
Hot resis = 1.18 ohm
Max current = 2.0ASince these things are relatively inexpensive, I've purchased a range of them and experiment to find the best match. I've also parallelled them for increased current rating. If this is done, I find they operate best if in physical contact with each other.
Steve,You are right, I should experiment with wider selections. I just changed the order placed through Mouser to include these two items, additional to CL60 (KC006) and CL70 (KC007). But gosh they probably won't arrive until after Thanksgiving.
However, I expect 60 and 70 to have higher than 0.2ohm at 2.1A. Maybe none of these limiters' resistance at 2.1A is low enough for me. Well I'll give them a try anyway. Maybe I can find a low forward voltage Schottky diodes to get somewhere between 5.8V and 6.3V.
Waiting is the toughest part... Pity that the Fry's clerk on the other side of the phone was clueless on the limiter. Otherwise I may drive over to pick them up.
Really?! But then Sovtek does have lower turn-on surge. They don't sound anywhere near those NOS though.Since it's so simple, I just went through few pairs myself:
Sovtek 6SN7GT: 2.4, 2.4ohm
Sylvania VT-231: 1.8, 1.8
Tung-Sol 6SN7GT: 1.6, 1.8
Raytheon 6SN7GT: 1.6, 1.8
Mullard ECC32: 1.2, 1.4ECC32 is not really 6SN7. And its lower cold resistance does reflect the higher heater current.
Now I realize cold start is indeed tough on filaments and their power supplies.
Actually, I did try to measure the surge heater current on my amp that supplies 2 12AX7, 1 12AU7, and 2 6SN7 with 6.3VDC. However a DMM is not fast enough to show the real peak current, I only saw 5A at the peak.Well, by your ratio it's very high. If they all have 7 times higher current at turn on, the total heater current for those 5 tubes is 14.7A! Although I'll guess it lasts for a couple seconds at most (and then gradually backs down to nominal 2.1A).
Judging by the "flash" phenomenon which often occurs with ECC8* / 12A*7 types (esp British ones)the duration of the surge is less than a tenth of a second.
Sorry to intervene ... heater currents are in Ampere range, not 0.6mA. You lost three zeroes somewhere.Surge current: it's simpler! Current through a cold indirect filament is about 3-4 times higher than a fully warmed up.
Around the nominal Uf voltage, If changes +1% for every +3% of Uf voltage. It's not linear regulation, but approximation is OK in a narrow 'legal' Uf range.
Some actual warmup data - see the link. In the good old days, in addition to brief datasheets, factories issued detailed docs with similar warmup charts, glass surface temperature charts, etc...
Klaus,You are absolutely right. The filament current should be 0.6A, not 0.6mA. I knew it but was absent minded in the previous post. The calculated filament resistance should also be corrected by 1/1000 to be 10.5ohm.
Thanks for the cool link to that interesting analysis. I'm still reading it.
This is probably touched in that article. But anyway I've been wondering about the advised practice of waiting for 5min after shutdown before firing up again. Actually with heater still hot right after power off, the surge heater current will be less on turn on. Where is that 5min wait coming from?
Some safety margin won't hurt.
I suspect all you need is to wait until the rectifier tube cools down safely - 2 minutes or more. Turning it back on when rectifiers are still hot is a bad idea, IMO.
What if my amp uses only diodes for rectifying? I don't see them needing any "rest"... Does it still hurts in anyway to turn the power back up say a minute right after power down?
It's your money... you can try :))
You may be right, but still take care, especially if the amp is DC-coupled.BTW, the thread above is an interesting development. You see, all the tubes I ever have are Soviet makes and some 1940-s TFKs, with surge factor under 4x, and the 5-7x surge factor came as a surprise.
I suspect the reason is not in the tungsten itself, but in the ceramic coating on the filament.
The amp is DC coupled between input 12AX7 and driver 6SN7, but AC coupled between driver and output EL34. With still-hot filament and yet-to-discharged filter caps, I really only expect hot power-on to be less stressful than a cold one. At least with my diode rectified tube amp.One question still not answered is whether the higher cold resistance of Sovtek is a better thing than lower ones of NOS.
Maybe there is trade off involved between long filament life and low surge current?
This post is made possible by the generous support of people like you and our sponsors: