Hello all, my objective is to lose as much RFI and other common mode noise as possible in as simple a way as possible for a heater supply.
I'm running balanced heaters with the center tap of the transformer connected to a filtered elevated supply (+40v or so).
I've been scouring Valve Amplifiers on the subject. I've found a suitable split bobbin transformer to lose as much common mode interference as possible on the mains side, and I see that split bobbin transformers have unusually high leakage inductance.
It looks to me like inductance will be playing a role and that I can't just put a capacitor from each filament pin of the input tube to ground to make an RC filter- it will inevitably be an LC filter due to the transformer inductance.
Is it the leakage inductance of the transformer that I would use to calculate this LC filter?
I have contacted a helpful fellow at Signal Transformer who has measured several of their split bobbin transformers for me. I asked for leakage inductance measurements, but I wonder if he misunderstood- he reports 120 mH for a 115/36v .22A transformer, and this seems really high to me. I'm clarifying this point with him, but I realize I'm not certain what aspect of transformer inductance is the one I need for my LC filter calculations.
Anyone?
I'm running balanced heaters with the center tap of the transformer connected to a filtered elevated supply (+40v or so).
I've been scouring Valve Amplifiers on the subject. I've found a suitable split bobbin transformer to lose as much common mode interference as possible on the mains side, and I see that split bobbin transformers have unusually high leakage inductance.
It looks to me like inductance will be playing a role and that I can't just put a capacitor from each filament pin of the input tube to ground to make an RC filter- it will inevitably be an LC filter due to the transformer inductance.
Is it the leakage inductance of the transformer that I would use to calculate this LC filter?
I have contacted a helpful fellow at Signal Transformer who has measured several of their split bobbin transformers for me. I asked for leakage inductance measurements, but I wonder if he misunderstood- he reports 120 mH for a 115/36v .22A transformer, and this seems really high to me. I'm clarifying this point with him, but I realize I'm not certain what aspect of transformer inductance is the one I need for my LC filter calculations.
Anyone?
What about layer on layer wound transformer with electrostatic shielding between coils, if you're so worried about it?
IMHO, people are too worried about outside RFI compared to internal noise, such as rectifiers switching noise. To ease the later, I would be using a power transformer with the lesser leakage as possible, because the higher its value, the higher the Q and the lower the Fr between Crec, Ls and Rsec.
This of course, if you intend the rectify the secondary voltage into DC.
For a transformer used in AC heating duties, a split bobbin design make sense to me. The higher leakage also give a "softer" switch on characteristic, which can be a bonus for cold heater inrush currents.
IMHO, people are too worried about outside RFI compared to internal noise, such as rectifiers switching noise. To ease the later, I would be using a power transformer with the lesser leakage as possible, because the higher its value, the higher the Q and the lower the Fr between Crec, Ls and Rsec.
This of course, if you intend the rectify the secondary voltage into DC.
For a transformer used in AC heating duties, a split bobbin design make sense to me. The higher leakage also give a "softer" switch on characteristic, which can be a bonus for cold heater inrush currents.
1. Ask the Signal Transformer fellow: Is the leakage inductance referred to the primary, or to the secondary. In any case, 120uH at 60 Hz = 45 milli Ohms. By the way, that 120uH is: 'in series' with the inductance of the winding, not across the winding.
2. For RF common mode signals, I would be more worried about the capacitance from the primary winding to the secondary winding. A. The split bobbin should be quite low capacitance. Part of that capacitance is from bobbin to bobbin, and part of it is bobbin 1 to laminations, and laminations to bobbin 2. B. 50AE mentioned a more standard transformer, but with the less common special feature of having an electrostatic shield between the primary, and the secondary. Grounding that shield will also help get rid of RF, and reduce the capacitance from primary to secondary.
In any case, putting capacitors from the secondary connections to ground forms a capacitive divider. You can not completely get rid of that. And any lead wire from an electrostatic shield to ground does have inductance. And the lead wires of a capacitor does also have inductance. Inductance in either case can reduce the effectiveness of grounding, except at the one frequency where the series of the capacitance and inductance is resonant.
2. For RF common mode signals, I would be more worried about the capacitance from the primary winding to the secondary winding. A. The split bobbin should be quite low capacitance. Part of that capacitance is from bobbin to bobbin, and part of it is bobbin 1 to laminations, and laminations to bobbin 2. B. 50AE mentioned a more standard transformer, but with the less common special feature of having an electrostatic shield between the primary, and the secondary. Grounding that shield will also help get rid of RF, and reduce the capacitance from primary to secondary.
In any case, putting capacitors from the secondary connections to ground forms a capacitive divider. You can not completely get rid of that. And any lead wire from an electrostatic shield to ground does have inductance. And the lead wires of a capacitor does also have inductance. Inductance in either case can reduce the effectiveness of grounding, except at the one frequency where the series of the capacitance and inductance is resonant.
tapehead ted,
Is this for a problem you Are having?
Is this for an expected problem?
What are the details?
Is this for a phono preamp?
What kind of interference do you have?
There are many causes of interference, not just filament problems.
Do you have 2 wire power source?
Do you have 3 wire grounded power source?
Is this for a problem you Are having?
Is this for an expected problem?
What are the details?
Is this for a phono preamp?
What kind of interference do you have?
There are many causes of interference, not just filament problems.
Do you have 2 wire power source?
Do you have 3 wire grounded power source?
You need to consider leakage inductance.
Low leakage inductance = extended frequency response. So high leakage is good for power transformer. Of course the leakage inductance has to be low enough to pass the fundamental supply frequency (that's easy if 50-60 Hz) but high enough to stop high order harmonics and other RF garbage. Only need to be far enough with other stuff to avoid leakage flux or screen it.
You want low capacitance (i.e. substantial interleaving). The same is true for filter chokes.
Last edited:
Thanks for all the replies!
This is something I'm designing- I'm addressing anticipated issues. I read in Valve Amplifiers that what RFI does to audio "cannot be overemphasized". That's pretty strong language for him. It's not clear to me how exactly it interferes with the audio band- perhaps if I understood that better I would be less superstitious about it. Meanwhile I'm trying to err on the side of caution and take basic precautions.
I may be overcompensating a bit in trying to design a very clean AC heater supply- I can see the advantages of a DC supply but simplicity has it's charms as well.
My pre (not a phono pre) is remote from the power amp and the power supply and the HT is well filtered, so that rectifier hash and magnetic fields generated in the amp itself shouldn't be a concern at the input stage.
On the other hand, the wire runs are longer (more inductance) and the ground is further away.
I'd like to RF filter the heater supply right at the input valve heater pins.
I've seen recommendations of just putting a small pp cap across the input valve heater pins, and even adding inductance for uncalculated further filtering, but I'm concerned I could end up with a big old resonance somewhere.
The supply output resistance (mainly from the elevated supply at the center tap, which puts some resistance between that tap and ground) in parallel with the resistance of the filaments would help bring the Q down and damp a resonance, but it still would require a very large capacitance to bring the Q down with 60 mH (one half of the 120mH, if that turns out to be the right figure). This could enable filtering well into the audio band however, and the lower the resonant frequency the more filtering at higher frequencies.
I have located a suitable toroidal transformer with an electrostatic shield, but it's quite a bit larger than necessary and I've yet to see if I can get leakage inductance figures for it. Jones says the geometry of toroidal winding is "poor" and so I can probably expect high leakage inductance there as well.
I'm off grid so the power source would be an inverter from batteries. Exceltech makes some really clean inverters with very minimal distortion, so I'd be using one of those. Of course it's tempting to run the heaters right off the batteries, and I've thought about it quite a bit but it would also be nice to be able to run the amp off any standard ac power outlet.
This is something I'm designing- I'm addressing anticipated issues. I read in Valve Amplifiers that what RFI does to audio "cannot be overemphasized". That's pretty strong language for him. It's not clear to me how exactly it interferes with the audio band- perhaps if I understood that better I would be less superstitious about it. Meanwhile I'm trying to err on the side of caution and take basic precautions.
I may be overcompensating a bit in trying to design a very clean AC heater supply- I can see the advantages of a DC supply but simplicity has it's charms as well.
My pre (not a phono pre) is remote from the power amp and the power supply and the HT is well filtered, so that rectifier hash and magnetic fields generated in the amp itself shouldn't be a concern at the input stage.
On the other hand, the wire runs are longer (more inductance) and the ground is further away.
I'd like to RF filter the heater supply right at the input valve heater pins.
I've seen recommendations of just putting a small pp cap across the input valve heater pins, and even adding inductance for uncalculated further filtering, but I'm concerned I could end up with a big old resonance somewhere.
The supply output resistance (mainly from the elevated supply at the center tap, which puts some resistance between that tap and ground) in parallel with the resistance of the filaments would help bring the Q down and damp a resonance, but it still would require a very large capacitance to bring the Q down with 60 mH (one half of the 120mH, if that turns out to be the right figure). This could enable filtering well into the audio band however, and the lower the resonant frequency the more filtering at higher frequencies.
I have located a suitable toroidal transformer with an electrostatic shield, but it's quite a bit larger than necessary and I've yet to see if I can get leakage inductance figures for it. Jones says the geometry of toroidal winding is "poor" and so I can probably expect high leakage inductance there as well.
I'm off grid so the power source would be an inverter from batteries. Exceltech makes some really clean inverters with very minimal distortion, so I'd be using one of those. Of course it's tempting to run the heaters right off the batteries, and I've thought about it quite a bit but it would also be nice to be able to run the amp off any standard ac power outlet.
Dunno… I think your best bet is just to design a solid DC supply for your heaters, and be done with it. Remember, if you need “6.3 VACRMS”, then you need between 6.0 and 6.5 VDC for the same heaters. Double it if you're working with 12 volt heater tubes. Etc.
Turns out that Mother Nature and good old fashioned solid state devices deliver quite a refined DC if you need them to. A split 12 VAC-CT (center tapped) filament winding can quite easily be full-wave-bridge rectified with discrete Schottky diodes, fed to 22,000 µF, 16 VDC caps, (EPCOS/TDK: B41231A4229M000, $2.75 ea), you get a starting point of quite low-ripple DC. Using just an extra Schottky, its easy to get the (6.3 × 1.414 - (2 × 0.8) × 0.97 → … - 0.8 → 6.3 VDC) DC voltage needed for the filaments. Regulated? No. With a CLC filtering scheme (and a small L), you get RFI free heater juice.
Just keep all those transformer leads short.
ALTERNATELY, and again quite successfully, one can purchase some very modest sized chokes and AC mains rated capacitors and just put a low-pass-filter utilizing LCLC→X (to transformer) configuration to cut RFI to close-to-unmeasurable levels. And still do “the same thing” with DC filament excitation. Because I like DC filaments. Takes money, delivers results.
GoatGuy
I'd suggest checking all your household loads to see if any are causing your inverter to have a distorted waveform. Your amplifier's HT power supply may be the worst culprit.
I'd also suggest you divert your technical learning curve to other pursuits. Chasing a ghost when you don't have the technical nous to either know if the ghost is present, or how to exorcise it, is I suggest a waste of your time.
I'd also suggest you divert your technical learning curve to other pursuits. Chasing a ghost when you don't have the technical nous to either know if the ghost is present, or how to exorcise it, is I suggest a waste of your time.
I have designed a few valve amps and not had a problem with heater transformers causing noise. I elevate to about 45VDC and leave it there. I take care routing heater wires as they radiate.
I did in one case resort to a DC heater supply but that turned out not to be the cause of the hum. It was caused by rectifier diodes switching. I switched to schottky didoes and the problem went away. The input signal needs to be as short as possible and any wire connections screened. Route HVAC away from signal wires.
I did in one case resort to a DC heater supply but that turned out not to be the cause of the hum. It was caused by rectifier diodes switching. I switched to schottky didoes and the problem went away. The input signal needs to be as short as possible and any wire connections screened. Route HVAC away from signal wires.
RF on the heater wiring can be a problem when designing a sensitive radio receiver, but unlikely to affect an audio amp. You are worrying about a non-problem; the time you are taking on this could be better spend thinking about real issues. RF usually gets in at inputs and outputs and then goes straight to the audio circuit.
So did I, but I couldn't imagine transformer leakage inductance helping and instead I installed ferrite pairs before the capacitors on the pins.
Who says getting a clean DC supply is any easier? I disagree. A clean AC supply is fine.
Thanks again everyone.
I'm actually learning a lot on my wild goose chase here. Some of the hidden elements and how significant are they really?
FWIW, 160 mH really is the leakage inductance of that split bobbin transformer- only it's 115/36v .35 A, not .22 as I posted. Referred to the primary.
And my electrostatically screened toroid (Antek) only has a wire screen, not a foil one, so that won't help.
Does anyone have an idea of the efficacy (especially at RF) of a single cap across the heater pins of the input valve vs. a cap from each pin to chassis?
I's tempting to think the single cap across the pins of a balanced AC heater supply would give you the equivalent of two (more) perfectly matched caps to ground potential without the complications of inductance and resistance to ground.
I'm actually learning a lot on my wild goose chase here. Some of the hidden elements and how significant are they really?
FWIW, 160 mH really is the leakage inductance of that split bobbin transformer- only it's 115/36v .35 A, not .22 as I posted. Referred to the primary.
And my electrostatically screened toroid (Antek) only has a wire screen, not a foil one, so that won't help.
Does anyone have an idea of the efficacy (especially at RF) of a single cap across the heater pins of the input valve vs. a cap from each pin to chassis?
I's tempting to think the single cap across the pins of a balanced AC heater supply would give you the equivalent of two (more) perfectly matched caps to ground potential without the complications of inductance and resistance to ground.
tapehead ted,
Just exactly what RF are you trying to get rid of?
What is the frequency?
Differential Mode RF:
A capacitor mounted right at the tube across the filament socket lugs will reduce the Differential Mode RF voltage across the filament. But how much reduction?
A 6.3V filament that draws 300mA is 21 Ohms.
A 0.5uF capacitor is 21 Ohms at 15kHz.
Why do I use 15kHz as an example? What if you are located down the road from the Navy's 1.5 Megawatt 15kHz CW transmitter (details changed to protect the innocent, lol).
And any of those transmitters have carriers that are in the audio frequency range.
But most 0.5uf capacitors are not a very good bypass to a 238 kiloWatt 600MHz Digital TV transmitter that is just down the road from you. There is too much lead inductance at 500MHz, even if the leads are short. How many tubes can "detect" (rectify) 600 MHz . . . quite a few, it does not have to very efficient at that frequency, just enough to be a problem.
When one of those digital TV transmitters first went on the air, it brought a hospital's patient monitor system to its knees. The TV transmitter had to be shut down until the problem with the hospital system could be fixed.
Common Mode RF:
A capacitor from one filament lead to ground, and a second capacitor from the other filament lead to ground will reduce Common Mode RF. But again, by how much, and at what frequency?
Differential Mode and Common Mode RF have to have a way to get into the amplifier.
As was earlier stated in this thread, the shielded signal to the input jack, power lead to the power input connector, and the loudspeaker cable to the speaker output connector are the most likely sources of RF getting into the amplifier circuitry. Once the RF gets inside the shielded enclosure, it "Launches" into the whole inside area of the enclosure.
There are various filter boxes, double shielded cables, etc. But no one thing prevents all RF getting into the amplifier. Sometimes, shielded cables and coax are tested for "Reverse Transfer Impedance" to see how good or bad they are at rejecting RF signals in the air.
Lighting ballasts, computer power supplies, switcher supplies of all kinds put interference on power lines, launch (transmit over the air) etc.
And do not forget the magnetic interference modes. Just put a phono preamp directly over some power amp power transformer or its filter choke.
Usually and hopefully, things are not that bad. Don't worry, just be prepared IF you experience a problem.
Just be aware when driving around a curve, but not worried. The tire that is in good condition usually does not blow.
But If you find you are in a location that has extremely high level RF, you will have to try and eliminate the problem, or even may have to move.
Just exactly what RF are you trying to get rid of?
What is the frequency?
Differential Mode RF:
A capacitor mounted right at the tube across the filament socket lugs will reduce the Differential Mode RF voltage across the filament. But how much reduction?
A 6.3V filament that draws 300mA is 21 Ohms.
A 0.5uF capacitor is 21 Ohms at 15kHz.
Why do I use 15kHz as an example? What if you are located down the road from the Navy's 1.5 Megawatt 15kHz CW transmitter (details changed to protect the innocent, lol).
And any of those transmitters have carriers that are in the audio frequency range.
But most 0.5uf capacitors are not a very good bypass to a 238 kiloWatt 600MHz Digital TV transmitter that is just down the road from you. There is too much lead inductance at 500MHz, even if the leads are short. How many tubes can "detect" (rectify) 600 MHz . . . quite a few, it does not have to very efficient at that frequency, just enough to be a problem.
When one of those digital TV transmitters first went on the air, it brought a hospital's patient monitor system to its knees. The TV transmitter had to be shut down until the problem with the hospital system could be fixed.
Common Mode RF:
A capacitor from one filament lead to ground, and a second capacitor from the other filament lead to ground will reduce Common Mode RF. But again, by how much, and at what frequency?
Differential Mode and Common Mode RF have to have a way to get into the amplifier.
As was earlier stated in this thread, the shielded signal to the input jack, power lead to the power input connector, and the loudspeaker cable to the speaker output connector are the most likely sources of RF getting into the amplifier circuitry. Once the RF gets inside the shielded enclosure, it "Launches" into the whole inside area of the enclosure.
There are various filter boxes, double shielded cables, etc. But no one thing prevents all RF getting into the amplifier. Sometimes, shielded cables and coax are tested for "Reverse Transfer Impedance" to see how good or bad they are at rejecting RF signals in the air.
Lighting ballasts, computer power supplies, switcher supplies of all kinds put interference on power lines, launch (transmit over the air) etc.
And do not forget the magnetic interference modes. Just put a phono preamp directly over some power amp power transformer or its filter choke.
Usually and hopefully, things are not that bad. Don't worry, just be prepared IF you experience a problem.
Just be aware when driving around a curve, but not worried. The tire that is in good condition usually does not blow.
But If you find you are in a location that has extremely high level RF, you will have to try and eliminate the problem, or even may have to move.
Last edited:
36V? Are you wiring heaters in series? That raises other thoughts.
160mH reflected across 115V:36V is 15mH.
36V 0.35A is 103 Ohms of heater load. Heaters are fairly non-inductive.
15mH feeding 100r is a 1KHz low-pass. (We could have suspected this, knowing that power windings often fall in the KHz range due to leakage L.)
So how do you define "RFI"? 50KHz bias oscillator? WNBC-AM 770KHz? Cellphone at 2GHz?
50KHz coming in the power line is down by 50:1. (50KHz from an on-board bias osc is a different problem.) WNBC is below notice. However GHz waves are so short that most any lumped-filter will fail to ensure suppression. In between, study AM and FM tuners schematics and see what they did. Sometimes a few-turn choke and 0.047mFd.
Thank you very much. This is much clearer to me.
I'm blessed to have already moved! There is a local internet broadband tower a few miles away on top of the next mountain over and the end of the power lines is less than half a mile away across the street. I read all kinds of RF radiates off the end of the power lines, but I am really blessed in this regard. Some satellite is beaming XM radio right in my south facing window...
(moment of gratitude)
I saw the LR filter there, but I suspected the filtered frequencies were shunted right through the filament, so I wasn't sure that would help.
I see there's a thread on series heaters- I intend to look at that.
I have three 12.6v .15 amp heaters in the pre. I can envision some pretty long and variable cable runs (lots of potential listening setups out here) so I'm looking at keeping the current low for minimal voltage drop.
I'm seeing that cable inductance is playing a role as well.
Inductance has been a kind of out of sight out of mind factor for me, working on accounting for it.
I need to get clear on where this leakage inductance (as the largest L) persists- I wasn't aware of the change reflected across the transformer. I need to get wise on the output L of various filters- I'm thinking it must become frequency dependent.
I'm blessed to have already moved! There is a local internet broadband tower a few miles away on top of the next mountain over and the end of the power lines is less than half a mile away across the street. I read all kinds of RF radiates off the end of the power lines, but I am really blessed in this regard. Some satellite is beaming XM radio right in my south facing window...
(moment of gratitude)
I saw the LR filter there, but I suspected the filtered frequencies were shunted right through the filament, so I wasn't sure that would help.
I see there's a thread on series heaters- I intend to look at that.
I have three 12.6v .15 amp heaters in the pre. I can envision some pretty long and variable cable runs (lots of potential listening setups out here) so I'm looking at keeping the current low for minimal voltage drop.
I'm seeing that cable inductance is playing a role as well.
Inductance has been a kind of out of sight out of mind factor for me, working on accounting for it.
I need to get clear on where this leakage inductance (as the largest L) persists- I wasn't aware of the change reflected across the transformer. I need to get wise on the output L of various filters- I'm thinking it must become frequency dependent.
XM radio signals are quite weak, unless they are targeting your particular house to keep you from having kids (lol).
Parts have parasitics.
Resistors have L and C
Capacitors have L and R
Inductors have C and R
Designing a filter to cover 10 decades of frequencies is not typically needed.
Find out what frequencies and signals are actually causing problems, and how they are getting into your amplifier, and then attack them.
Parts have parasitics.
Resistors have L and C
Capacitors have L and R
Inductors have C and R
Designing a filter to cover 10 decades of frequencies is not typically needed.
Find out what frequencies and signals are actually causing problems, and how they are getting into your amplifier, and then attack them.
OK, so now I'm looking at rectifying the output of a split-bobbin transformer and how the inductance works there.
I plan to use one transformer (6.3-0-6.30) to run the power tube heaters and then to rectify and feed DC to the preamp tubes.
So the leakage inductance, referred to the primary but reflected through to the secondary, first forms LR filters with the power tube heaters. The output impedance of this encounter leaves me a complex impedance composed of the series inductance and the parallel resistance. At higher frequencies, increasingly more inductance than resistance, and vice versa.
This feeds the rectifier diodes, and I can't see why that would do anything to eliminate the effect of the leakage inductance.
So is there something about the reservoir capacitor that prevents it from forming an LC filter with the leakage inductance? I'm thinking that CLC smoothing may be redundant- above some frequency the output is already smoothed, and most of that nasty rectifier hash is considerably mellowed by this LC filter.
At this point, the output impedance comprises the leakage inductance and the reservoir capacitance, which cancel each other out to some extent at some range of frequencies, and the resistance.
Would this complex impedance have any effect on a voltage regulator IC?
Thanks again for helping me think this through.
I
I plan to use one transformer (6.3-0-6.30) to run the power tube heaters and then to rectify and feed DC to the preamp tubes.
So the leakage inductance, referred to the primary but reflected through to the secondary, first forms LR filters with the power tube heaters. The output impedance of this encounter leaves me a complex impedance composed of the series inductance and the parallel resistance. At higher frequencies, increasingly more inductance than resistance, and vice versa.
This feeds the rectifier diodes, and I can't see why that would do anything to eliminate the effect of the leakage inductance.
So is there something about the reservoir capacitor that prevents it from forming an LC filter with the leakage inductance? I'm thinking that CLC smoothing may be redundant- above some frequency the output is already smoothed, and most of that nasty rectifier hash is considerably mellowed by this LC filter.
At this point, the output impedance comprises the leakage inductance and the reservoir capacitance, which cancel each other out to some extent at some range of frequencies, and the resistance.
Would this complex impedance have any effect on a voltage regulator IC?
Thanks again for helping me think this through.
I
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.