DHT has some eccentricities. Indirectly heated may have some long term interface changes to Rhk, and possibly dendrites, but I highly doubt that could be made noticeable in any objective way.
I prefer AC only for simplicity.
I have AC problems (hum) only with 6C4C triode. May be due its special filament struccture.I´m not sure.
A humpot is not quite here.I use now DC not regulated with them... 4x Cree Sic diodes and 20.000 uF.
The diferent sound may be is the IMD teh sound with 50/60 Hz and harmonics.Is a personal taste for everyone.
Cheers
I have AC problems (hum) only with 6C4C triode. May be due its special filament struccture.I´m not sure.
A humpot is not quite here.I use now DC not regulated with them... 4x Cree Sic diodes and 20.000 uF.
The diferent sound may be is the IMD teh sound with 50/60 Hz and harmonics.Is a personal taste for everyone.
Cheers
When you use a indirect heater tube w/little cathode/filement space your AC heater will cause hum, so people will guilty the tube <this tube have hum>
However builders really love AC heaters as it need a few caps and resistors less than a decent DC heater.
Last edited:
No. There is no hum unless the DC is poorly filtered/regulated. The amp I'm listening to right now uses AC for the output tubes, and regulated DC for the gain/splitter/drivers (16VDC -> Cap -> MC29300 -> Cap -> LM7806 per 6SN7 heater with cap). One of my others uses 12V ATX to power the entire amp using boost converters for B+, and all the heaters run series-parallel.
Dunno…
The only time I consider DC for filaments is for extra-high gain preamplifiers (think “microphone input”, or RIAA moving coil phono sections). Indeed — even in a single chassis — I might only reserve DC for the very first valve. The rest of 'em can sing from the AC filament score!
But when running DC, I also am a hard-nose about selecting a filament regulator that provides the actual 6.3 VDC (or 12.6 VDC). Not some arrangement of series diodes and big capacitors to hopefully create the right DC source. Not at all. Use a (low drop out) LDO regulator like NJM2396F63 NJR | Mouser … and Shotkey rectifiers from a 6.3 VAC winding, full wave bridge, decent pre-regulation capacitors.
FYI: for those of you who have expressed misgivings about twisting pairs effectively, consider constructing a small twisted-wire jig. Mine is a pair of ½" (ID) PVC tubes, each about 5 feet long. Duct tape the tubes together.
When you make twisted pair, just cut a piece of straight wire off, about 2× the length of the tube set. Bend in half; fish each half down a separate tube. Connect the bent portion to a drill. At modest speed, twist away!
The TRICK here is that the wires can independely spin on the inside of their individual PVC tubes, while being twisted by the drill bit. And let me tell you, it makes a huge difference. The wires, thus allowed to independently twist, become “twist-locked”. They don't unravel! I try to twist tight enough to have 3–5 twists per inch. (Pretty tight!)
Once you “get into making twisted pair” effectively, you can noodle around and make several dozen lengths of it. I like "brown" insulation — because I have mentally reserved that color for all filaments. Except if I've got a mixed-voltage chassis (6.3 and 12.6 VAC). Then I use brown for 6.3 and grey for 12.6. And the 5 volt rectifier winding? RED baby!!! Its going to become the epitome of HOT voltage! LOL.
GoatGuy
The only time I consider DC for filaments is for extra-high gain preamplifiers (think “microphone input”, or RIAA moving coil phono sections). Indeed — even in a single chassis — I might only reserve DC for the very first valve. The rest of 'em can sing from the AC filament score!
But when running DC, I also am a hard-nose about selecting a filament regulator that provides the actual 6.3 VDC (or 12.6 VDC). Not some arrangement of series diodes and big capacitors to hopefully create the right DC source. Not at all. Use a (low drop out) LDO regulator like NJM2396F63 NJR | Mouser … and Shotkey rectifiers from a 6.3 VAC winding, full wave bridge, decent pre-regulation capacitors.
FYI: for those of you who have expressed misgivings about twisting pairs effectively, consider constructing a small twisted-wire jig. Mine is a pair of ½" (ID) PVC tubes, each about 5 feet long. Duct tape the tubes together.
When you make twisted pair, just cut a piece of straight wire off, about 2× the length of the tube set. Bend in half; fish each half down a separate tube. Connect the bent portion to a drill. At modest speed, twist away!
The TRICK here is that the wires can independely spin on the inside of their individual PVC tubes, while being twisted by the drill bit. And let me tell you, it makes a huge difference. The wires, thus allowed to independently twist, become “twist-locked”. They don't unravel! I try to twist tight enough to have 3–5 twists per inch. (Pretty tight!)
Once you “get into making twisted pair” effectively, you can noodle around and make several dozen lengths of it. I like "brown" insulation — because I have mentally reserved that color for all filaments. Except if I've got a mixed-voltage chassis (6.3 and 12.6 VAC). Then I use brown for 6.3 and grey for 12.6. And the 5 volt rectifier winding? RED baby!!! Its going to become the epitome of HOT voltage! LOL.
GoatGuy
I always run 6V or 12V, not 6.3V or 12.6V respectively. It's within spec and in theory the heater will last longer, although I've never seen a heater failure personally.
@Goatguy that is a good idea, I'm just too lazy to bother. DC doesn't need twisting, doesn't need special placement, doesn't radiate an AC field. I basically only use AC when I'm being cheap.
@Goatguy that is a good idea, I'm just too lazy to bother. DC doesn't need twisting, doesn't need special placement, doesn't radiate an AC field. I basically only use AC when I'm being cheap.
Been doing that for simplicity, however didn't know the wires could twist inside the insulation for staying twisted.
Good to know so many diyer are using the AC heaters as they are supposed to be made.
Two tricks I may add:
Look for phase of the AC amplification chain.
1. gain stage (AB heater polarity)
2. Splitter stage (AB if the phase change, BA for the other section / or if the phase don't change)
3. Class AB push pull : AB one tube and BA the other.
Trick is to null the push pull output noise of the heaters with phase.
Also use the opposite polarity from one stage to other when the phase change.
When the phase don't change use the same heater polarity.
Many tubes have separate section heaters which enable you to make it almost 100% cancelation.
Ex for one concertina simple amp :
Input gain stage AB
a) splitter anode AB, push pull section AB
b) splitter cathode BA, push pull section BA
in such an amplifier all the noise is cancelled along the amplification chain.
Further, if you use 12xx tubes you can place 2 in series and use a 25.2V transformer reducing the noise 4x , current is decreased 4x.
Further, to diminish capacitance with heater winding in the transformer and power tubes you can use a separate transformer for AC heaters of the pre-section and a separate transformer for the power section. This really improves the sound.
DC heaters exacerbate the capacitance and inter-windings problems within the power transformer. This is a serious issue for 100W+ amps.
DHT (Direct Heated Triodes) and AC Versus DC Filaments:
AC Filaments, 50Hz and 60Hz:
Each end of a DHT filament becomes negative 50 or 60 times a second. The result is one or the other end of the DHT filament becomes negative 100 or 120 times a second.
The grid voltage comes nearer to one end of the filament. Then the grid voltage comes nearer to the other end of the filament. That happens 100 or 120 times a second (not 50 and not 60 times a second). And there are 100 or 120 times a second when the grid voltage is the same amount to both ends of the filament (filament AC is at zero crossing).
You may think this cancels out. Take a signal generator, and a spectrum analyzer and see:
Apply a 1,000Hz test tone to a DHT amplifier, you can see 100Hz or 120Hz sidebands, even long before the grid goes near 0V. The result is a spectrum of 900Hz, 1,000Hz, and 1,100Hz. That is what you get with AC filaments on a DHT. Sometimes I use AC filaments on DHT. I will not argue how that sounds. Sounds good, sounds bad, who knows?
DC Filaments:
One end of the filament is negative, and the other end is positive. Lets use 2 resistors in series to create a grounded Pseudo-Center. Connect one resistor to one end of the filament, and connect one end of the other resistor to the other end of the filament. Join the unconnected ends of the two resistors together to Ground. Then use fixed bias on the grid.
Try this on a 300B, 300V plate, -61V fixed bias, 60mA plate current, and 5V DC across the filament, with the 2 resistor pseudo-center to ground. The 300B filament has the shape of a W across the face of the two sides of the plate structure (sorry, I could not turn the W upside down in the word processor, like it is for a rightsize up 300B). A. Grid voltage to one end of the filament = -61V +2.5V = - 58.5V grid to filament. B. Grid voltage to the other end of the filament = -61V -2.5V = - 63.5V grid to filament.
(And C. Grid voltage the the center of the filament = -61V +/- 0V = -61V).
Hmmm, that means there will be More plate current from the filament end A. (-58.5V bias) Hmmm, that means there will be Less plate current from the filament end B. (-63.5V bias) There are uneven currents along the sides of the plates; one end of the plate sides, versus the other end of the plate sides. That is what you get with DC filaments on a DHT. Sometimes I use DC filaments on DHT. I will not argue how that sounds. Sounds good, sounds bad, who knows?
AC Filaments, 50Hz and 60Hz:
Each end of a DHT filament becomes negative 50 or 60 times a second. The result is one or the other end of the DHT filament becomes negative 100 or 120 times a second.
The grid voltage comes nearer to one end of the filament. Then the grid voltage comes nearer to the other end of the filament. That happens 100 or 120 times a second (not 50 and not 60 times a second). And there are 100 or 120 times a second when the grid voltage is the same amount to both ends of the filament (filament AC is at zero crossing).
You may think this cancels out. Take a signal generator, and a spectrum analyzer and see:
Apply a 1,000Hz test tone to a DHT amplifier, you can see 100Hz or 120Hz sidebands, even long before the grid goes near 0V. The result is a spectrum of 900Hz, 1,000Hz, and 1,100Hz. That is what you get with AC filaments on a DHT. Sometimes I use AC filaments on DHT. I will not argue how that sounds. Sounds good, sounds bad, who knows?
DC Filaments:
One end of the filament is negative, and the other end is positive. Lets use 2 resistors in series to create a grounded Pseudo-Center. Connect one resistor to one end of the filament, and connect one end of the other resistor to the other end of the filament. Join the unconnected ends of the two resistors together to Ground. Then use fixed bias on the grid.
Try this on a 300B, 300V plate, -61V fixed bias, 60mA plate current, and 5V DC across the filament, with the 2 resistor pseudo-center to ground. The 300B filament has the shape of a W across the face of the two sides of the plate structure (sorry, I could not turn the W upside down in the word processor, like it is for a rightsize up 300B). A. Grid voltage to one end of the filament = -61V +2.5V = - 58.5V grid to filament. B. Grid voltage to the other end of the filament = -61V -2.5V = - 63.5V grid to filament.
(And C. Grid voltage the the center of the filament = -61V +/- 0V = -61V).
Hmmm, that means there will be More plate current from the filament end A. (-58.5V bias) Hmmm, that means there will be Less plate current from the filament end B. (-63.5V bias) There are uneven currents along the sides of the plates; one end of the plate sides, versus the other end of the plate sides. That is what you get with DC filaments on a DHT. Sometimes I use DC filaments on DHT. I will not argue how that sounds. Sounds good, sounds bad, who knows?
Well, like in any other case design is about optimization, and means have to lead to end results. A first, rectify 6.3V or 12.6V by Shottky diodes. Second, select filter capacitor according to ripples on the current that you draw, that you can allow, taking in account tolerances of a voltage in outlet. Third, use a voltage regulator with almost zero voltage drop capabilities.
Here is an example attached.
Attachments
I have slight hum coming from the speakers in a breadboard test. I decided to add another 6922 to perform pre-volume pot duty. heating is AC from 7.5v winding. Sound is awesome and hum will be below hiss level (-110db) when done. Hum not coming from power supply. I am going to shorten signal lines and test again.
5V on 300V is about a 1.66% difference in electric potential/electric field, and 5V on 60V is about an 8.33% difference. Electrons emitted by the cathode follows the Fermi-Dirac distribution, and on a coated cathode, emission itself is not uniform, you can expect huge differences along its surface, current density can vary in a factor 30 between different points...
Even more, the cathode has an M shape or something like, what about the spaces not covered by the M?
Fortunately, between cathode and grid there is an electron cloud (hate the name space charge) and electrons arrive at the anode in a more or less decently uniform form.
You have explained very clearly the ugly effect of AC, do not worry for a minor detail using DC.
DC definitely sounds better, even on indirect heated cathodes, the only issue IMO is the possibility of magnetization due to the DC magnetic field, changing the polarity on each start-up is not *that* difficult though.
Last edited:
Even when your posts are too long for my Tarzan-English, I read them carefully and so far did learn a lot; unfortunately this is wrong, let me explain
The time varying magnetic field produced by the filament depends on voltage, not current.
Neglecting the AC magnetic field around the wire, supposedly well twisted, for the filament
Bac = (Uac x 10⁸) / [√2 π f S N]
Some specially designed valves have twisted heaters, the 7025 could be one of them, for the other cases the filament should be like a single spire (N=1) and the enclosed area should be very small and hence the AC magnetic field could be very large altough it is shielded in part by the cathode.
In conclusion, if you want to reduce the AC magnetic field use the lowest possible voltage, for 12Axx valves the heaters must be connected in parallel.
Last edited:
Ok I see your point, inside the tube the heaters are a spiral, likely an inductor which radiates a field.
I was referring to the wires inside the amplifier and not to DHT in particular or any other inside tube noise. Only to the mutual inductance -M * (Delta I / Delta t) I was referring to in regard to the other wires in the amps and the other components.
Bac = (Uac x 10⁸) / [√2 π f S N] Its been many years since I studied physics... where is the Voltage in the equation? I see only permeability constant uo times I in amperes. If Iac is lower the Bac is lower.
This also brings another issue with DC, self inductance of heaters....
I was referring to the wires inside the amplifier and not to DHT in particular or any other inside tube noise. Only to the mutual inductance -M * (Delta I / Delta t) I was referring to in regard to the other wires in the amps and the other components.
Bac = (Uac x 10⁸) / [√2 π f S N] Its been many years since I studied physics... where is the Voltage in the equation? I see only permeability constant uo times I in amperes. If Iac is lower the Bac is lower.
This also brings another issue with DC, self inductance of heaters....
Uac is the RMS AC voltage.
No, current is responsible for another magnetic field, H
Iac flows in opposite senses, so Hac is a lot lower than on your alleged equation, badly spoken BTW.
Popilin,
Who changes the polarity of the DC to the filament each time the amp powers up?
Some indirectly heated tubes have Spiral Filaments, once up, and once down. The magnetic fields created are vertical, because the spiral filament is like a vertical solenoid.
If you wire a 12A_7 for 12.6V, 4 to 5, you get the same magnetic field from the tube when you wire the 12A_7 for 6.3V, 9 to 4&5. That is because in each case you have 0.15A for filament 1 and 0.15A for filament 2. Perhaps you meant the filament wires leading to the tube, with 2x the current when it is wired for 6.3V, and that magnetic field will be bigger, than when wired for 12.6V.
If the wires to the filament are not twisted, then the e field is different for the 6.3V versus the 12.6V case (especially if one end of the filament is grounded).
Who changes the polarity of the DC to the filament each time the amp powers up?
Some indirectly heated tubes have Spiral Filaments, once up, and once down. The magnetic fields created are vertical, because the spiral filament is like a vertical solenoid.
If you wire a 12A_7 for 12.6V, 4 to 5, you get the same magnetic field from the tube when you wire the 12A_7 for 6.3V, 9 to 4&5. That is because in each case you have 0.15A for filament 1 and 0.15A for filament 2. Perhaps you meant the filament wires leading to the tube, with 2x the current when it is wired for 6.3V, and that magnetic field will be bigger, than when wired for 12.6V.
If the wires to the filament are not twisted, then the e field is different for the 6.3V versus the 12.6V case (especially if one end of the filament is grounded).
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.