Page 1
http://www.tubezone.net/pdf/12ax7ecc83.pdf
But Philips = Mullard. Page 7.
http://www.r-type.org/pdfs/ecc83.pdf
Last edited:
No, nothing to do with noise. Everything to do with reasonably well-defined potentials. Note that the effective impedance seen at the cathodes of an LTP is likely to be much smaller than the actual cathode resistor because of cathode follower action.tapehead ted said:
Heater-cathode leakage is an entirely different issue, because only in a very bad valve would the leakage be bad enough to provide the required conductive path between cathode and heater to avoid potential problems. An ideal valve would have zero leakage i.e. infinite intrinsic Rh-k.MarcelvdG said:
Thanks Merlin for the datasheet and RDH4 links.
RDH4 does point nicely to the hum ingress issue for the typical phase inverter of that era before an output stage. RDH4 would normally put a reference in for that topic if there was one, but sadly no.
I guess there was in-house assessment - such as assuming a grounded heater, and calculating the hum voltage across the phase inverter cathode resistance (with a maximum Rh-k datasheet value, eg. 20k), when the intrinsic Rhk was at some benchmark low level, and how that would relate to then getting amplified in to a 'just acceptable' amp output hum level.
For Mullard, they would have had a lower acceptable limit for intrinsic Rhk. I've seen acceptance levels down to 2Megohm (100Vdc with up to 50uA leakage) in the 5687WA datasheet.
Post #17 makes a reference to (I think) intrinsic Rhk and some comment by Morgan Jones about a hot heater. I recently did some testing and recall observing no difference in Rhk leakage current for a hot or cold heater. I'll have to re-check that - may be a good use of my new cheapy Aneng AN8009.
RDH4 does point nicely to the hum ingress issue for the typical phase inverter of that era before an output stage. RDH4 would normally put a reference in for that topic if there was one, but sadly no.
I guess there was in-house assessment - such as assuming a grounded heater, and calculating the hum voltage across the phase inverter cathode resistance (with a maximum Rh-k datasheet value, eg. 20k), when the intrinsic Rhk was at some benchmark low level, and how that would relate to then getting amplified in to a 'just acceptable' amp output hum level.
For Mullard, they would have had a lower acceptable limit for intrinsic Rhk. I've seen acceptance levels down to 2Megohm (100Vdc with up to 50uA leakage) in the 5687WA datasheet.
Post #17 makes a reference to (I think) intrinsic Rhk and some comment by Morgan Jones about a hot heater. I recently did some testing and recall observing no difference in Rhk leakage current for a hot or cold heater. I'll have to re-check that - may be a good use of my new cheapy Aneng AN8009.
Surely if hum was the issue then there would be a minimum Rh-k spec, not a maximum? A truly floating AC heater supply cannot inject hum into the cathode, as there is no complete circuit for the hum current to flow in. As soon as you add a conductor (as the datasheets insists you do) then you can get hum. Note that the spec is not maximum resistance from cathode to circuit ground (which is relevant to hum) but maximum resistance from heater circuit to cathode circuit.
How many pF is the filament to cathode capacitance? And . . . How well shielded is the filament secondary from the B+ secondary? How many pF is it from the filament secondary to the B+ secondary? Yes, the very same B+ secondary that may have solid state diodes driving a large capacitor, and that creates Lots of high frequency transient noise! Think about that one when you let the filament secondary float. You will not have to think, just listen to it.
Brimar datasheet/report shows a 4pF spec for Chk - I haven't seen that reported in other datasheets. I sort of measured it down around 3pF in one test jig that was looking for other info.
I reckon the datasheets are implicitly referring to applications where heater is grounded (either by one leg connected to chassis which would be the worst case common situation of that era, or its CT, or humdinger, or bypassed elevated DC voltage) - so the generic path for hum in a phase inverter stage would be via the unbypassed common cathode resistance to ground of the inverter. So the hum loop would be the common cathode to ground unbypassed resistor of the inverter stage, through to the heater winding, and back to the common cathode via what could be down to a 2Meg Rhk interface resistance in parallel with 4pF Chk interface capacitance.
I reckon the datasheets are implicitly referring to applications where heater is grounded (either by one leg connected to chassis which would be the worst case common situation of that era, or its CT, or humdinger, or bypassed elevated DC voltage) - so the generic path for hum in a phase inverter stage would be via the unbypassed common cathode resistance to ground of the inverter. So the hum loop would be the common cathode to ground unbypassed resistor of the inverter stage, through to the heater winding, and back to the common cathode via what could be down to a 2Meg Rhk interface resistance in parallel with 4pF Chk interface capacitance.
Last edited:
This applies to most valve circuits; basically any circuit with a DC path straight from the grid to ground and a resistance from anode to supply that is not >> Ri has a low-frequency impedance of about 1/gm at the source.
If the maximum external Rh-k spec is just a way to ensure that the internal heater-to-cathode leakage is swamped, like Merlin wrote, then it is silly that they don't just specify a conservative value for that leakage.
In any case, it has to be related to some leakage path or other, because for a valve with zero leakage you could use any impedance you like to prevent the electrodes from floating, so there would be no need to specify a rather low maximum.
Last edited:
I don't agree that that the aim is to swamp the internal leakage; if that is what Merlin wrote then I believe he is mistaken. The aim in valve manufacture is to minimise leakage by maximising the intrinsic Rh-k. The ideal is infinite Rh-k. As this is a quality issue they won't want to specify a value, except perhaps for 'special quality' valves.MarcelvdG said:
Not so, because in a working valve there are electrons flyng around. They all end up somewhere. The aim is that those which somehow end up on the heater (which will be a tiny minority, I assume) don't push it to a high potential. Note that this need not necessarily be a high negative potential, as secondary emission could mean that the floating electrode goes very positive if every arriving electron on average causes more than one secondary electron to be emitted. An added complication is that the heater itself will be doing some thermionic emission - this could dominate other reasons for the heater potential to change.
So we have an electrode which is emitting or gaining electrons. We don't want it to change potential too much because that could disrupt valve operation, so we specify a maximum resistance between it and the cathode. This is being overcautious, because in most circuits the heater and cathode are fairly well fixed in potential even with a greater resistance value between them, so we say that for LTP you can exceed the limit.
My take is that the manufacturers are covering three bases at once, all of which are 'fixed' (directly or indirectly) by swamping the h-k insulation resistance:
1) Hum (several mechanisms);
2) Potential drift;
3) The effect of unknown h-k insulation resistance on tuned circuits.
Last edited:
Relevant articles (I'm sure Tim has them already but others may like to see):
http://www.clarisonus.com/Archives/TubeTheory/Heater.pdf
Page 216 onwards:
http://www.4tubes.com/LITERATURE/ENGLISH/Other/RCA 1962 Electron Tube Design_text.pdf
Last edited:
Maybe this is a semantic discussion, because what you are describing here is exactly what I would call leakage paths; there are relatively small currents flowing to places where you don't want them to flow.
The manufacturer is going to assume both k and h are referenced to a common point in any ordinary (1960s) circuit. So they know in advance that there will be a resistance between h and k, typically formed by a cathode bias resistor (or phase inverter tail resistor etc), and a heater reference resistor, respectively. This total resistance is in parallel with the internal h-k insulation. Not because you need a resistance in parallel with the insulation, of course; it's there as a result of normal circuit design.
If one side of the heater supply is grounded, then making Rk (or a tail resistance etc) very large will cause a larger hum voltage across it, due to h-k leakage and h-k capacitance.
If the heater reference resistance is very large or infinite then you get a large hum voltage across Rk due to mains-to-earth leakage (mostly via h-k capacitance).
If the heater reference resistance is infinite then the heater might float to an unknown potential and affect linearity (that's all I meant by 2).
If the heater supply is grounded (one side or centre, it doesn't matter), h-k insulation is in parallel with any tunued circuit in the cathode and hence will spoil the designer's calculations.
So the manufacturer can cover all these situations by speccing an Rhk limit and washing his hands of the whole malange.
Last edited:
Agreed.Merlinb said:
Are you saying that there will be leakage (via stray capacitance) from mains to the heater circuit, then from the heater circuit to the cathode? If so, it is the cathode to circuit ground resistance which matters more than the heater reference resistor.
OK. Agreed. In my view this is the main issue.
Possibly, but 'swamping' it with a resistor will make things worse. If Rh-k is a problem then the designer has to ensure that it does not appear across a tuned circuit e.g. by grounding the cathode. This is only likely to be a problem in an oscillator, and you certainly don't want to add more loss resistance to an oscillator.
I'm not sure you can make such a generalization covering all RF and video circuits ever! It must have been an anticipated circuit kludge or it wouldn't get a special mention in RDH4, no?
Possibly. It just seems odd that this isn't cited in the literature as being the actual reason for the rating. Not even once, that I can find. What little is mentioned is all about hum and tuned circuits instead *shrug*
Last edited:
The hum leakage is driven by the heater AC voltage acting across the intrinsic heater-cathode insulation barrier (Rhk and Chk), and looping through the typical phase inverter cathode resistance to ground (given the heater is connected to ground). Effectively the same way that hum is induced in an input stage when the cathode resistor is not bypassed. So yes, in that circuit situation, the typically large unbypassed value of cathode to ground resistance is the concern for audio applications.
That Electron Tube Design book is an amazing reference. I've only had time to glance at it, but I recognize a couple graphs on p.233 showing the heater bias curve- I've seen in a recent web article on heaters and hum where the hum current is much smaller biased well away from 0 volts. That article- sorry I don't have a reference for it now- says that the cathode follower and LTP benefit from the heater and cathode voltage differential in a way like elevated heaters can reduce hum currents, by being up on the flat part of this curve.
Perhaps that's part of what Mullard was making an exception about.
I still note that their exception was specifically for a phase inverter driving the power tubes, so possibly it's partially the heater to cathode voltage differential acting like elevated heaters and partially a less sensitive location.
Perhaps that's part of what Mullard was making an exception about.
I still note that their exception was specifically for a phase inverter driving the power tubes, so possibly it's partially the heater to cathode voltage differential acting like elevated heaters and partially a less sensitive location.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.