Hi everyone,
Someone asked me how to estimate the anode-to-cathode capacitance of a valve rectifier near Vak=0, and the truth is that I have no clue. N.O.C.L.U.E.
So I turn to you folks, domain experts, to ask if there exist reasonable approximate values that apply to two general classes of valve circuitry,
Thank you from someone whose valve-virginity is intact,
-- Mark Johnson
(link to original query)
Someone asked me how to estimate the anode-to-cathode capacitance of a valve rectifier near Vak=0, and the truth is that I have no clue. N.O.C.L.U.E.
So I turn to you folks, domain experts, to ask if there exist reasonable approximate values that apply to two general classes of valve circuitry,
- Class #1: rectifier valves that supply power to valve preamps (phono + linestage)
- Case #2: rectifier valves that supply power to valve power amps (?? 50 WPC RMS ??)
Thank you from someone whose valve-virginity is intact,
-- Mark Johnson
(link to original query)
Once again thanks for your efforts Mark
I was considering starting working with tube rectifiers for my projects, but I need some more know-how about them, since they seem to be quite different animals than SS diodes, with much higher voltage drops and a rather resistive character
I have found that as a start, but I think I have to build something and just check it.
http://frank.pocnet.net/sheets/049/6/6CL3.pdf
I was considering starting working with tube rectifiers for my projects, but I need some more know-how about them, since they seem to be quite different animals than SS diodes, with much higher voltage drops and a rather resistive character
I have found that as a start, but I think I have to build something and just check it.
http://frank.pocnet.net/sheets/049/6/6CL3.pdf
Valve rectifiers have desirable characteristics regarding capacitance. It is lower than many semiconductor alternatives, and it does not increase with decreasing reverse-voltage.
Valve data sheets for rectifiers do not often give a value, but smaller Damper diodes (efficiency diodes) have about 9pF, larger dampers about 14pF.
Many of us are turning to these damper diodes now, as they give good performance and are still plentifully available.
Valve data sheets for rectifiers do not often give a value, but smaller Damper diodes (efficiency diodes) have about 9pF, larger dampers about 14pF.
Many of us are turning to these damper diodes now, as they give good performance and are still plentifully available.
When used for their intended purpose, rectification of AC power mains, directly of off a power transformer secondary, the anode-cathode capacitance of power rectifiers is entirely negligible.
Perhaps if you can tell us why you need to know the capacitance, as you'll only need to know it if you have some non-standard application in mind, we can give better advice. The capacitance of any vacuum tube diode has two parts:-
a) the static capacitance, which can be calculated from the tube structure dimensions using standard formulae;
b) the additional capacitance due to teh sapce charge. This capacitance depends on the applied anode voltage.
Of course, you can, with care, measure the static A-K capacitance. The Q-meter method is the best way. If neccessary this can be done with nothing more than an oscillator and a radio reciever (used as a detector) and a hand made coil.
Measuring the dynamic capacitance under working conditions takes a bit more ingenuity.
Perhaps if you can tell us why you need to know the capacitance, as you'll only need to know it if you have some non-standard application in mind, we can give better advice. The capacitance of any vacuum tube diode has two parts:-
a) the static capacitance, which can be calculated from the tube structure dimensions using standard formulae;
b) the additional capacitance due to teh sapce charge. This capacitance depends on the applied anode voltage.
Of course, you can, with care, measure the static A-K capacitance. The Q-meter method is the best way. If neccessary this can be done with nothing more than an oscillator and a radio reciever (used as a detector) and a hand made coil.
Measuring the dynamic capacitance under working conditions takes a bit more ingenuity.
Thank you Keit! The original requestor is diyAutio member dimkasta, who has joined this thread. He is considering whether or not to connect a dissipative snubber across the secondary of his power transformer, which feeds valve rectifier(s). Why? I don't really know but I suspect the answer might be, why the hell NOT arrange the secondary to be critically-damped (Zeta=1.0); it might help and it cannot possibly hurt. Even if valve rectifiers exhibit no reverse recovery spikes whatsoever, certainly it's harmless to add a couple capacitors and a resistor that set the damping ratio to a number of the designer's choosing? (rather than letting random parasitics + Murphy's Law (evil, vindictive, mean-spirited Murphy's Law) set it for you).
To choose a starting point for a critical-damping snubber, it helps to have a guess of the rectifier capacitance and to have a guess of the transformer self-capacitance. That's why dimkasta has asked the odd-sounding question. In fact, for the "CRC" snubber he contemplates, all he really needs is an upper bound on rectifier capacitance. It sounds like Crect <= (Ntubes x 30pF) will take him where he wants to go.
Last edited:
Snubbers are usefull with semiconductor diodes because the diode charge storage time is appreciable - microseconds. The diode keeps pasing current in the reverse direction for an appreciable time.
Charge storage occurs in vacuum tube rectifiers as well (it's the space charge) but its nanoseconds. In rectification at power mains frequency it is entirely negligible.
As I have said, you can completely ignore the tube capacitance. It is of the order 10 pF or so, whereas the self capacitance of a typical tube power transformer secondary is of the order of nanofarads.
The diode capacitance (picofards) is essentially in parallel with the transformer self capacitance (nanafarads) (the return path is via the first filter cap - effectively a faction of an ohm ESR). So, the tube capacitance adds something like 0.1% and anybody who thinks it is a starting point in any calculation lives in fantasy land.
Snubbers are routine with semiconductor rectifiers in major part because low ripple capacitor input filters are used, making the diode conduction time very short, so the peak diode currents are considerable. With vacuum tube rectifiers, tube ratings mandate that the first filter capacitor must be of low value such that the ripple and diode conduction time is a substantial fraction of the total half cycle time. So the diode peak currents are relatively low.
I cannot see the logic in adding components "because it cannot do any harm". Any good engineer soon learns that the best way to excellence is to find the simplest way to meet the specification.
The more parts you add, the more there is to go wrong. And any such snubber capacitors (& resistors if used) are going to be seriously stressed. Take a typical example: Secodary 350V per side; DC voltage on fisrt filter cap 410 V. The rectifer, and any snubber capacitor must withstand a peak voltage of 760 V. And there should be an allowance for mains voltage variation and mains switching surges. This is NOT a standard capacitor. Any series resistor used in snubbing must also withstand a peak voltage of 760 V. This is most certainly not a standard resistor from the point of view of voltage rating. And, to comply with Ul and house contents insurance requirments it would have to be an intrinsicly fire-proof part, installed so as to guarantee the fire proofing.
Don't do it, unless you know what you are doing. An if you knew what you were doing, you woudn't do it.
I could not access the Morgan Jones article. But going on Morgan Jones' other writings, it is likely to be a mix of fact, complete nonsense, and misunderstood principles.
Charge storage occurs in vacuum tube rectifiers as well (it's the space charge) but its nanoseconds. In rectification at power mains frequency it is entirely negligible.
As I have said, you can completely ignore the tube capacitance. It is of the order 10 pF or so, whereas the self capacitance of a typical tube power transformer secondary is of the order of nanofarads.
The diode capacitance (picofards) is essentially in parallel with the transformer self capacitance (nanafarads) (the return path is via the first filter cap - effectively a faction of an ohm ESR). So, the tube capacitance adds something like 0.1% and anybody who thinks it is a starting point in any calculation lives in fantasy land.
Snubbers are routine with semiconductor rectifiers in major part because low ripple capacitor input filters are used, making the diode conduction time very short, so the peak diode currents are considerable. With vacuum tube rectifiers, tube ratings mandate that the first filter capacitor must be of low value such that the ripple and diode conduction time is a substantial fraction of the total half cycle time. So the diode peak currents are relatively low.
I cannot see the logic in adding components "because it cannot do any harm". Any good engineer soon learns that the best way to excellence is to find the simplest way to meet the specification.
The more parts you add, the more there is to go wrong. And any such snubber capacitors (& resistors if used) are going to be seriously stressed. Take a typical example: Secodary 350V per side; DC voltage on fisrt filter cap 410 V. The rectifer, and any snubber capacitor must withstand a peak voltage of 760 V. And there should be an allowance for mains voltage variation and mains switching surges. This is NOT a standard capacitor. Any series resistor used in snubbing must also withstand a peak voltage of 760 V. This is most certainly not a standard resistor from the point of view of voltage rating. And, to comply with Ul and house contents insurance requirments it would have to be an intrinsicly fire-proof part, installed so as to guarantee the fire proofing.
Don't do it, unless you know what you are doing. An if you knew what you were doing, you woudn't do it.
I could not access the Morgan Jones article. But going on Morgan Jones' other writings, it is likely to be a mix of fact, complete nonsense, and misunderstood principles.
Last edited:
Thanks, Keit! I am sure diyAudio member dimkasta appreciates your comments.
The we-like-snubbers crowd tend to select 650VAC rated (1600VDC rated) film capacitors like the MKP10 from WIMA (here's one at Mouser). You are right, they ain't cheap. At 50Hz the magnitude of impedance of this 150nF cap is about 20Kohms. This limits 50Hz current through the snubber, a series RC circuit, to something less than (350V/20KR) = 18 mA. For the snubber resistances typically encountered {5R <= Rsnub <= 500R}, snubber dissipation is less than 200 milliwatts: IsquaredR. So the inexpensive Stackpole CF1 1-watt resistors, rated for 500VAC, are plenty adequate. Thanks to the voltage divider effect of the snubber capacitor in series, the snubber resistor sees less than 10% of the secondary voltage. Conservative people could of course put three resistors in series if they wanted an even greater margin-of-safety.
The snubber design starting point is (Ctrafo + Crectifier) or an upper bound thereof. In cases where Crectifier totally dominates Ctrafo {often seen in solid state poweramps that use 35 ampere bridge rectifier assemblies with toroidal transformers}, Ctrafo is so small it is negligible. Irrelevant. In other cases, such as the valve ones you describe, where Ctrafo totally dominates Crectifier, Crectifier is irrelevant. It is good that member dimkasta asked the question, and even better that you kindly provided an answer!
The we-like-snubbers crowd tend to select 650VAC rated (1600VDC rated) film capacitors like the MKP10 from WIMA (here's one at Mouser). You are right, they ain't cheap. At 50Hz the magnitude of impedance of this 150nF cap is about 20Kohms. This limits 50Hz current through the snubber, a series RC circuit, to something less than (350V/20KR) = 18 mA. For the snubber resistances typically encountered {5R <= Rsnub <= 500R}, snubber dissipation is less than 200 milliwatts: IsquaredR. So the inexpensive Stackpole CF1 1-watt resistors, rated for 500VAC, are plenty adequate. Thanks to the voltage divider effect of the snubber capacitor in series, the snubber resistor sees less than 10% of the secondary voltage. Conservative people could of course put three resistors in series if they wanted an even greater margin-of-safety.
The snubber design starting point is (Ctrafo + Crectifier) or an upper bound thereof. In cases where Crectifier totally dominates Ctrafo {often seen in solid state poweramps that use 35 ampere bridge rectifier assemblies with toroidal transformers}, Ctrafo is so small it is negligible. Irrelevant. In other cases, such as the valve ones you describe, where Ctrafo totally dominates Crectifier, Crectifier is irrelevant. It is good that member dimkasta asked the question, and even better that you kindly provided an answer!
There is also substantial bulk capacitance between PT windings typically used in a valve rectified amp. For a 240V mains to 385-0-385 secondary, I measured 3.5nF between each of the 385V half secondaries (when CT tap was separated, and with windings shorted), and from one 385V half secondary to primary was 0.4nF, and 0.2-0.4nF to heaters, and 1.3nF to earth screen.
Just in case the OP is still temped to install an RC snubber, I must point out an error in this thinking.
Even though the dissipation is less than 200 mW, and ordinary 1 watt resistor is NOT adequate. A Stackpole CF1 is certainly not adequate. The flameproof version, CFF1 would be MUCH better, but still not adequate.
There are two things that must be considered:-
a) the impact of mains borne switching transients. These can pass straight through the power trasformer and having a high dv/dt will not be fully attenuated by the capacitor. So the voltage stress on the resistor must be considered to be the sum of (Vsec x 1.4 + Vdc + allowance for transients). Experince shows that allowing for a transient of 150% of AC mains gives acceptable reliability. So the resistor voltage rating for a 350V as side secondary must be (350 x 1.4 + 410 + 175) = 1075 volts.
A CF1 or CFF1 is varely adequate on a voltage basis, and not adequate at all if the body is in contact with the chassis or another part.
b) Failure mode analysis.
The capacitor may fail short circuit. If it does, the full voltage (RMS 760 V) is now across the resistor. If an ordinary type it will burst into flame. If you are reasonably lucky it will quickly burn out, go open, and that is the end of it. (Expect you now have no snubber and may not be aware of it). But you may not be lucky.
For house safety, if a stereo manufacturer were to install an RC snubber, UL (in USA) or the insurance industry or govt body (elsewhere) would require the resistor to be fireproof. And installed so as to preserve the fireproof rating. A home constructor should think the same way.
An alternative is to use a cpacitor type that is intrinsically safe, ie cannot fail short circuit.
It is no fun if the capacitor fails, the resistor bursts into flame, sets adjacent wiring and parts alight, and burns your house down. If the insurance investigators manage to discover your amp was the cause you built it yourself, they will invoke some clause that lets them deny payout.
If you look at the failure modes relavent to resistors in tube amps (and in SS amps), the fire risk does not apply to almost all resistors. It usually does apply to some parts in TV sets. In such cases fireproof resistors are used.
Last edited:
Depends how one measures the capacitance. If you just put a cap meter between the windings, the screen does nothing. If the screen is really grounded in operation, it will have about twice the meter read capacitance to each winding. But any signal on one winding will get shunted to ground through that doubled capacitance. Ie, screens don't remove capacitance, they make it worse. But they will block signal transfer.
Not necessarily.
Intruments for measuring capacitance frequently have a guard or earth terminal. If you connect the screen to the guard terminal, the meter will read correctly the capacitance between primary and secondary arising from electric flux going around or bypassing the screen, which will be a very low value.
It's much the same as the screen in a tetrode or pentode tube. The scren grid does not affect the capacitance beween grid and plate on its own. But when it is earthed, the effective capacitance for circuit operation between grid and plate becomes very low.
Or in other words, you'll get the correct answer in measurements if you understand how to do the measurements. If you don't, you may not....
The guard ring is the common ground in the meter between the oscillator and receiver sections, so connecting it to the shield will cause all the osc. capacitive coupled current to return via the guard terminal. And the receiver will not see the signal (except for what leaks around the shield). But the osc. section will be sending current to the shield at around 2X the non shielded capacitance. Ie, if you have a parallel plate capacitor and insert a shield plate half way in between, you end up with 2 capacitor sections in series, each with 2X the capacitance without the shield. The guard ring terminal feature simply shunts the osc. capacitive current off through the shield.
But most hand held portable C meters don't have the guard ring feature, so they will measure the same capacitance, shield or no shield (as long as the meter has no other connection to ground).
For a real circuit, this means that the feedthrough capacitance just gets replaced by common mode capacitance to ground of twice the original feedthrough capacitance. So the idea of putting shield layers into an OT is quite damaging to freq. response. But fine for a power xfmr. that wants common mode line noise blocked.
But most hand held portable C meters don't have the guard ring feature, so they will measure the same capacitance, shield or no shield (as long as the meter has no other connection to ground).
For a real circuit, this means that the feedthrough capacitance just gets replaced by common mode capacitance to ground of twice the original feedthrough capacitance. So the idea of putting shield layers into an OT is quite damaging to freq. response. But fine for a power xfmr. that wants common mode line noise blocked.
Last edited:
Don't y'all vacuum tube folks put MOV clamps across the primary (mains)?? If not, why not? I read the "Afterglow" valve poweramp article by John "Buddha" Camille, and he strenuously advocates MOVs in the primary. Is he dead wrong? Is he in the tiny minority of valve designers and modern thinkers? Why?
Suppose that a conservative DIYer placed 3 Stackpole CF1 resistors in series, as previously mentioned, to implement their snubber. Each resistor is rated 1000 volts, end-to-end, max overload voltage. Are you still worried?
Putting MOVs on the primary side helps reliability on solid state stuff. But unlike solid state stuff, vacuum tube equipment is inherently tolerant of switching spikes & surges - that is, if you don't go doing stupid things like adding snubbers comprising of inadequate parts.
MOV's can fail open circuit. In such cases there will be no amplifier fault symptoms - the music will still sound fine. So you must consider the consequences of MOV failure - in this case snubber failure.
Yes. If the capacitor shorts, you've got the full transformer secondary voltage in series with the rectified DC across those three resistors. Instant enormously massive dissipation overload, something like 1000 watts dissipated in resistors designed to cope with 1 watt. So three resistors on fire. So the resistors MUST be fireproof, which the CF-1 is not.
A commercial amplifier manufacturer would not do this. Neither should you.
It is not sufficient to ensure compoents are within ratings during normal operation. You must consider reasonably posible failure modes, and ensure NO such failure mode results in a safety hazard.
Last edited:
I add a primary side MOV to reduce stress on the mains switch contact, and the mains side winding insulation - especially for older valve amps.
The bulk winding capacitance measurements were just simply made, as a two terminal measurement, with all other windings and core floating on the PT. As such, a capacitance measurement may be influenced a bit by the other windings depending on the physical configuration.
To a first order, one could use a simple 'multiple parallel plates' model, where some plates are physically larger than other plates, and some plates are relatively thicker than other plates. So a capacitance measurement from say plate position 2 to plate position 5 would depend on the physical separation of the two plates, and the alignment of the areas, and that any intervening plates just acts to reduce the dielectric thickness in that area.
A secondary HT winding will benefit from some parasitic capacitance to shields on either side of it for directing the stray rectifier dI/dt induced energy back to the winding CT. The ES is typically on one 'side' of the HT windings, and a heater (with eg. CT to to ground) would be a reasonable shield on the other 'side'.
Yes if you use valve diodes, or eg. UF4007, instead of 1N4007, then the dI/dt at diode turn off is effectively devoid of the reverse recovery discontinuity at peak reverse current. And so the remnant dI/dt is when the diode current hits zero.
The bulk winding capacitance measurements were just simply made, as a two terminal measurement, with all other windings and core floating on the PT. As such, a capacitance measurement may be influenced a bit by the other windings depending on the physical configuration.
To a first order, one could use a simple 'multiple parallel plates' model, where some plates are physically larger than other plates, and some plates are relatively thicker than other plates. So a capacitance measurement from say plate position 2 to plate position 5 would depend on the physical separation of the two plates, and the alignment of the areas, and that any intervening plates just acts to reduce the dielectric thickness in that area.
A secondary HT winding will benefit from some parasitic capacitance to shields on either side of it for directing the stray rectifier dI/dt induced energy back to the winding CT. The ES is typically on one 'side' of the HT windings, and a heater (with eg. CT to to ground) would be a reasonable shield on the other 'side'.
Yes if you use valve diodes, or eg. UF4007, instead of 1N4007, then the dI/dt at diode turn off is effectively devoid of the reverse recovery discontinuity at peak reverse current. And so the remnant dI/dt is when the diode current hits zero.
Thank you everybody! It seems to me that your suggestions and recommendations to diyAudio member dimkasta, can be summed up roughly as follows
1. Vacuum tube designers have learned through bitter experience, that it imperative to build circuits which can tolerate much higher voltages than their supposedly normal operating point. If the power transformer secondary outputs "V" volts RMS, be sure all circuits and components can tolerate AT LEAST sqrt(2) x 3 x V volts. For example: if the secondary is 350 volts, make sure all circuits and components can tolerate AT LEAST 1500 volts.
2. For 350 volt secondaries, we are comfortable with the WIMA MKP10 film capacitor's ratings (650VAC, 1600VDC). Be sure to analyze what happens if this capacitor (a) fails as an open circuit; (b) fails as a short circuit, and arrange the design so that neither condition starts a fire or initiates some other kind of catastrophe.
3. In a snubber for 350V secondaries, we are not comfortable with any resistor in series with the WIMA MKP10 capacitor, unless that resistor can tolerate 1500 volts AND is flameproof rated. Fortunately, inexpensive 1W flameproof resistors, rated 600V, are available. A series string of three of these resistors (end-to-end rating 3x600 = 1800V), is acceptable. Stackpole RSMF1 ($0.35 qty=1), and/or TE Connectivity ROX1 ($0.30 qty=1), and/or Panasonic ERG1S ($0.33 qty=1) are some possibilities.
4. If you follow the above guidelines you can make a snubber that meets the requirements vacuum tube designers have learned through bitter experience. But a snubber seems like a very dumb thing to build; why get your knickers in a twist about critical damping of a resonant secondary circuit, when that resonant circuit cannot possibly be stimulated into oscillatory ringing? Unlike solid state diodes, vacuum tube rectifiers do not exhibit snapback or other reverse recovery misbehaviors.
2. For 350 volt secondaries, we are comfortable with the WIMA MKP10 film capacitor's ratings (650VAC, 1600VDC). Be sure to analyze what happens if this capacitor (a) fails as an open circuit; (b) fails as a short circuit, and arrange the design so that neither condition starts a fire or initiates some other kind of catastrophe.
3. In a snubber for 350V secondaries, we are not comfortable with any resistor in series with the WIMA MKP10 capacitor, unless that resistor can tolerate 1500 volts AND is flameproof rated. Fortunately, inexpensive 1W flameproof resistors, rated 600V, are available. A series string of three of these resistors (end-to-end rating 3x600 = 1800V), is acceptable. Stackpole RSMF1 ($0.35 qty=1), and/or TE Connectivity ROX1 ($0.30 qty=1), and/or Panasonic ERG1S ($0.33 qty=1) are some possibilities.
4. If you follow the above guidelines you can make a snubber that meets the requirements vacuum tube designers have learned through bitter experience. But a snubber seems like a very dumb thing to build; why get your knickers in a twist about critical damping of a resonant secondary circuit, when that resonant circuit cannot possibly be stimulated into oscillatory ringing? Unlike solid state diodes, vacuum tube rectifiers do not exhibit snapback or other reverse recovery misbehaviors.
These resistors would not necessarily be suitable, and would not be a good choice in a home-constructed amplifier, even though they are flameproof.
The maximum dielectric rating in both cases is only 350 V. This rating has nothing to do with, and is not increased by, putting multiple resistors in series. To use these resistors, you would need to use specific installation/mounting arrangments to ensure there is clearance around each resistor, so that the rating is not relavent. And do it so that future servicing by the constructor or someone else he sells/swap/gives the amplifier to does not upset things.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.