If you use a table-lamp near your audio system, run the lamp through your mains filter.
It may help prevent coupling from the mains into (eg) preamp stages, or MC cartridges.. &c
It may help prevent coupling from the mains into (eg) preamp stages, or MC cartridges.. &c
Hi Rod,
Have you ever been thinking of redesigning your gyrator for high voltage use? I mean a design round high voltage semiconductors with some protection diodes to make it reliable. Ripple filtering of the gyrator is massive and could be used as an E-choke in HV power supplies.
Peter
Have you ever been thinking of redesigning your gyrator for high voltage use? I mean a design round high voltage semiconductors with some protection diodes to make it reliable. Ripple filtering of the gyrator is massive and could be used as an E-choke in HV power supplies.
Peter
Gyrators for HV B+ regulation work OK, but it depends on the application.
Absence of second-breakdown, high dc input impedance, and reasonable capacitance at Vds > 20V make power-FETs a better bet for this duty, than the BJT or Darlington used in the filament regs.
I use them like this for regulating the screen Vg2 in pentode designs: 900V 40W FET + 1,5M gate resistor + 0,47uF/630V MKP + zener + 120R stopper.
But for preamps I prefer the shunt regulator for its better isolation from the input (current source buffer), and for power amplifiers I would not use series regulation at all, unless a very heavy bleed current was applied. The leakage inductance of the OT can provide nasty surprises for a series regulator, when the output slews or loudspeaker back-EMF throws energy back toward the supply.
Carefully configured LCLC + bleed always seems to work (& sound) best, IME, at least until I try out a power-Shunt regulator.
Absence of second-breakdown, high dc input impedance, and reasonable capacitance at Vds > 20V make power-FETs a better bet for this duty, than the BJT or Darlington used in the filament regs.
I use them like this for regulating the screen Vg2 in pentode designs: 900V 40W FET + 1,5M gate resistor + 0,47uF/630V MKP + zener + 120R stopper.
But for preamps I prefer the shunt regulator for its better isolation from the input (current source buffer), and for power amplifiers I would not use series regulation at all, unless a very heavy bleed current was applied. The leakage inductance of the OT can provide nasty surprises for a series regulator, when the output slews or loudspeaker back-EMF throws energy back toward the supply.
Carefully configured LCLC + bleed always seems to work (& sound) best, IME, at least until I try out a power-Shunt regulator.
The gyrator will also have a high output impedance, which is not what you want to feed any amp with. A shunt regulator would be nice, but I haven't seen a robust high voltage power shunt regulator yet. I have a few prototypes but haven't tried them on a real amp yet.
You could build your HV gyrator with the excellent ST STW12NK95 (see st.com). Not only does this have 950V durability, and integrated gate protection, but you should get 3,0S or more of transconductance, even at 100mA. Provided the gate is propped up with a decent cap, the output impedance is only 0,33-ohm at audio frequencies, which should present little problem for 100mA amplifiers.
Ach, I seem to be talking myself into building one. FETs have certainly got a lot better lately, especially for the price.
Paul Hynes is slowly attracting attention with his 100W-capable shunt:
Z100A shunt Voltage Regulator
.
Hi
Paul Hynes also offers the SR5H shunt regulator, which incorporates a constant current source to feed the shunt. As soon as I sort out adequate heating, I'll be trying a pair of these in my 300B SE amplifier- set to 380V and a shunt current of 160 mA (unloaded). The dropout voltage across the reg is only about 10V, and so it can be used easily with existing supplies.
At the moment I use a massive GZ37-rectified, 30H choke input supply for each channel, but Mr Hynes is confident the regulators will improve them!
Paul N
Paul Hynes also offers the SR5H shunt regulator, which incorporates a constant current source to feed the shunt. As soon as I sort out adequate heating, I'll be trying a pair of these in my 300B SE amplifier- set to 380V and a shunt current of 160 mA (unloaded). The dropout voltage across the reg is only about 10V, and so it can be used easily with existing supplies.
At the moment I use a massive GZ37-rectified, 30H choke input supply for each channel, but Mr Hynes is confident the regulators will improve them!
Paul N
Attachments
Then we should have a diy option too. I'll post it in this forum when I'm done. It'll have good line regulation and very low output impedance past 20kHz.
Has anyone considered Rod Elliot’s capacitance multiplier to filter the raw DC before the Coleman regulator?
Capacitance Multiplier Power Supply Filter
I wired up the darlington version on my test bench with a 2N6387 I had lying around. Voltage drop can be as low as 2V and even less. With 130mV ripple at the input I measure only 0.7mV at the output.
Pro: smaller capacitors and improved ripple rejection.
Con: darlington needs to be cooled.
I don’t know if the pre regulator / raw DC supply has any influence on sound quality.
Peter
Capacitance Multiplier Power Supply Filter
I wired up the darlington version on my test bench with a 2N6387 I had lying around. Voltage drop can be as low as 2V and even less. With 130mV ripple at the input I measure only 0.7mV at the output.
Pro: smaller capacitors and improved ripple rejection.
Con: darlington needs to be cooled.
I don’t know if the pre regulator / raw DC supply has any influence on sound quality.
Peter
An externally hosted image should be here but it was not working when we last tested it.
If you use the circuit of the positive half you will get some good ripple rejection, but more importantly you will add some high frequency insertion-loss. With 2V across the darlington, the circuit will look like a series capacitor of 200 - 300pF at high frequency, with some shunt loss, too.
The Coleman regulator is designed to reject ripple and pass very little into the filament: Across the filament, my scope shows no disturbance at all on the mV scale, when operated at the recommended supply voltage, and with the standard capacitor values after the rectifier - which leave 150mV of ripple at the raw supply.
If you try this, please check that the capacitor after the rectifier can handle ripple of 3A (for the 300B model). This is usually a limitation to the choice of capacitor here.
For a 300B, and 2V across the darlington's pass transistor, the dissipation is < 3W, so a very small heatsink should suffice; e.g. 2mm thick Alu: 100mm x 100mm (check it keeps to 60 deg C or so).
As for sound quality, I think it's worth trying, in the raw dc before the regulator - for the HF loss. Comparing its effect to a common-mode choke would be instructive.... I find that isolating trafo is worth the trouble, so it has a good chance. IME, Filaments are incredibly sensitive.
The Coleman regulator is designed to reject ripple and pass very little into the filament: Across the filament, my scope shows no disturbance at all on the mV scale, when operated at the recommended supply voltage, and with the standard capacitor values after the rectifier - which leave 150mV of ripple at the raw supply.
If you try this, please check that the capacitor after the rectifier can handle ripple of 3A (for the 300B model). This is usually a limitation to the choice of capacitor here.
For a 300B, and 2V across the darlington's pass transistor, the dissipation is < 3W, so a very small heatsink should suffice; e.g. 2mm thick Alu: 100mm x 100mm (check it keeps to 60 deg C or so).
As for sound quality, I think it's worth trying, in the raw dc before the regulator - for the HF loss. Comparing its effect to a common-mode choke would be instructive.... I find that isolating trafo is worth the trouble, so it has a good chance. IME, Filaments are incredibly sensitive.
Still haven't built one of these, been off on a tangent researching 40khz AC suuply for a 45. Just curious if the gradient in transductance across the cathode that occurs with DC heat is going to cause more distortion than a 40khz ultasoni AC heater?
I've done some tests with RF heating long ago. You will see the intermod caused by RF, but it is outside the audio band, see page 5 here
http://www.tentlabs.com/Components/Tubeamp/Tubefilament/assets/Heatingmethods.pdf
The RF implementation I listened to sounded very good, but just wasn't as clean and transparaent as my own modules.
By the way, when using current sources, it is interesting to actually listen to the (small) audio voltage across the filament. If the filament is one-side-grounded, you can easilly cap-couple the other side to any other amplifier.
best
Thanks Guido I have no reservations using DC supplies with tubes like the #26 that were originally designed for battery and hence have a low filament resistance (low voltage), but the more powerful DHTs with the bigger filaments you can't help but think the cathode having a defferent GM across it would add distortion vs AC where the GM across the cathode is constant. May amount to nothing, but I am going to use 45's or 46's for their excellent linearity and want to make sure I don't compromise that. This is a headphone application so 60hz AC is out of the question.
My simulations of the "wein" style 40khz oscillator show a ton of noise in the audio band, also see the IM above the audioband. Plus I would have to come of with some sort of mofset driver after the oscillator circuit which will just add more noise.
My simulations of the "wein" style 40khz oscillator show a ton of noise in the audio band, also see the IM above the audioband. Plus I would have to come of with some sort of mofset driver after the oscillator circuit which will just add more noise.
The gm gradient in dc filament heating is not a cause for any distortion - in fact such distortion is only present with ac-heat!
The gradient exists because of the voltage gradient along the filament on account of the filament voltage - it's not a thermal effect.
The same gradient exists in ac, at any snapshot in time. Only with ac, the gradient is shifting up & down at 50 or 60Hz, and this causes the 50-60 and 100-120Hz cross-products the appear in the spectrum of ac heated dhts of all types.
That is the first-order effect. There are other undesirable influences of ac:
- much more difficult to exclude wideband mains noise;
- the filament heating current is directly mixed with the cathode (music) current - they flow in the same internal wiring, and tube socket pins. This calls for a humbucker pot to be used to null 50/60Hz currents of opposite phase. This does not completely work, because of harmonics of the mains waveform generated in the DHT..... Interestingly, this harmonic phenomenon has been examined by Dmitry Nizhegorodov, whose work is well worth a read:
On Correlation Between Residual DHT Filament Hum and AC Frequency. Distortion-induced hum in directly-heated triodes.
The gradient exists because of the voltage gradient along the filament on account of the filament voltage - it's not a thermal effect.
The same gradient exists in ac, at any snapshot in time. Only with ac, the gradient is shifting up & down at 50 or 60Hz, and this causes the 50-60 and 100-120Hz cross-products the appear in the spectrum of ac heated dhts of all types.
That is the first-order effect. There are other undesirable influences of ac:
- much more difficult to exclude wideband mains noise;
- the filament heating current is directly mixed with the cathode (music) current - they flow in the same internal wiring, and tube socket pins. This calls for a humbucker pot to be used to null 50/60Hz currents of opposite phase. This does not completely work, because of harmonics of the mains waveform generated in the DHT..... Interestingly, this harmonic phenomenon has been examined by Dmitry Nizhegorodov, whose work is well worth a read:
On Correlation Between Residual DHT Filament Hum and AC Frequency. Distortion-induced hum in directly-heated triodes.
It is difficult to see any advantage to be gained from RF filament heating.
In the first place, the amplitude noise from the signal generation AND the output buffer would need to be as good as the best dc implementation, as this noise will be directly transferred to the DHT, if present. Proper dc implementations achieve <<1mV of noise injection, so why settle for worse?
In practice, this means the power supply must be much quieter than the equivalent-performing dc solution, because the 40kHz signal source is going to multiply the problem.
Beware also of 32768 or 40k tuning fork crystals - these have very high sensitivity to supply noise, XX-high impedance and show large amounts of phase noise, which may also find its way into your output spectrum.
You also throw away the chance to see what high impedance dc drive can do for the sound, since the drive impedance must be LOW for any kind of ac.
On the practical side, the complexity is huge, and electrical efficiency a least as poor as linear dc.
I don't see any motivation for this scheme. Alright, I may be biased, but anyone show me a motive that stands up to any scrutiny .... let's have some fun.
In the first place, the amplitude noise from the signal generation AND the output buffer would need to be as good as the best dc implementation, as this noise will be directly transferred to the DHT, if present. Proper dc implementations achieve <<1mV of noise injection, so why settle for worse?
In practice, this means the power supply must be much quieter than the equivalent-performing dc solution, because the 40kHz signal source is going to multiply the problem.
Beware also of 32768 or 40k tuning fork crystals - these have very high sensitivity to supply noise, XX-high impedance and show large amounts of phase noise, which may also find its way into your output spectrum.
You also throw away the chance to see what high impedance dc drive can do for the sound, since the drive impedance must be LOW for any kind of ac.
On the practical side, the complexity is huge, and electrical efficiency a least as poor as linear dc.
I don't see any motivation for this scheme. Alright, I may be biased, but anyone show me a motive that stands up to any scrutiny .... let's have some fun.
I think you hit the nail on the head with the noise issue with these ultrasonic AC heaters, thanks for explaining the gm myth.
Very interesting thread. My thanks to all. I wonder why no one talks about floating both sides of the filament (for audio signal) and connecting together with equal-value resistors, signal from the mid point?
A side note: TV transmitters used to have DC on the finals' filaments that was swapped in polarity at any turn-off, turn-on cycle. Something about wear life.
Thanks,
Chris
A side note: TV transmitters used to have DC on the finals' filaments that was swapped in polarity at any turn-off, turn-on cycle. Something about wear life.
Thanks,
Chris
Thanks Chris. Actually, the tapped resistor variant can be seen in my July 2004 post. The two resistors, rather than being equal, have a higher value on the FIL+ side, in order to equalise the cathode current in each side. I don't remember much performance change with it, and did not like the complexity so well. But, things may be different with other DHTs.
Swapping of polarity of dc filaments will aid wear levelling in filaments: the grid-to-cathode voltage is a little higher at the FIL+ side, so it may burn off a bit more emissive agent in high power applications like the TV burner.
It's unlikely to be a lifetime limiter in receiving tubes, but if anyone is worried about such an effect the remedy is simple enough: wire the stereo amplifier with the filament polarity reversed on one channel. Then, on the Harvest Moon, or perhaps the feast of St Jude, Take the DHT from the left channel and swap it with the right.
Swapping of polarity of dc filaments will aid wear levelling in filaments: the grid-to-cathode voltage is a little higher at the FIL+ side, so it may burn off a bit more emissive agent in high power applications like the TV burner.
It's unlikely to be a lifetime limiter in receiving tubes, but if anyone is worried about such an effect the remedy is simple enough: wire the stereo amplifier with the filament polarity reversed on one channel. Then, on the Harvest Moon, or perhaps the feast of St Jude, Take the DHT from the left channel and swap it with the right.
I just ordered a couple of filament supplies from Rod. My understanding is that the supply is comprised of a gyrator followed by a CCS, and that the filament is hooked between the gyrator and the CCS, as follows:
a)
filtered DC positive --> gyrator --> tube filament --> CCS --> filtered DC negative
My question is, why is the tube filament not connected after the CCS, where the line rejection is much higher; like this:
b)
filtered DC positive --> gyrator --> CCS --> tube filament --> filtered DC negative
I have measured an implementation of a simple gyrator/ccs supply hooked up as version a) above and the ripple on the tube filament is much higher than if hooked as in b). We're talking 20mV p-p for a) and below 30uV p-p for b).
So then, why is a) used by everybody?
a)
filtered DC positive --> gyrator --> tube filament --> CCS --> filtered DC negative
My question is, why is the tube filament not connected after the CCS, where the line rejection is much higher; like this:
b)
filtered DC positive --> gyrator --> CCS --> tube filament --> filtered DC negative
I have measured an implementation of a simple gyrator/ccs supply hooked up as version a) above and the ripple on the tube filament is much higher than if hooked as in b). We're talking 20mV p-p for a) and below 30uV p-p for b).
So then, why is a) used by everybody?
think about modulating filament PSU from cathode side
job is to isolate cathode from PSU , same as PSU from cathode
job is to isolate cathode from PSU , same as PSU from cathode