I am using Rod Coleman's regulators in my 6J5 input 45 output parafeed headphone amplifier. The 6J5 will have a cascode MOSFET CCS load, the 45 will have an EL83 pentode CCS load (ri of the EL83 ~260K).
I am trying to decide if I should use dedicated transformers for the 45 filaments, i.e., the Hammond 266L12, or include the two 6.3V / 5A windings on the custom mains transformer I am having made by Sowter UK.
Talking with Rod, if the windings are on the Sowter transformer, I do run the risk of having rectifier noise from the raw DC supplies (also Rod's design) coupled to the B+ supply and having it appear on the plates of the tubes.
Given both input and output tubes have active plate loads, PS noise rejection should be good (especially the 6J5), this is my preferred route. So...
Option A: include two 6.3V / 5A windings on my Sowter mains transformer and trust that my EL83 pentode CCS will reject the raw DC rectifier noise on the 45 (the MOSFET CCS on the 6J5 should take care of it no problemo)
Option B: mount the Hammond transformers under the chassis, possibly subjecting my internal components to their sizable magnetic fields - would try to minimize by optimizing the internal layout
I don't have the experience to say which of these two options is best, would love to hear from those who do, thank you.
I am trying to decide if I should use dedicated transformers for the 45 filaments, i.e., the Hammond 266L12, or include the two 6.3V / 5A windings on the custom mains transformer I am having made by Sowter UK.
Talking with Rod, if the windings are on the Sowter transformer, I do run the risk of having rectifier noise from the raw DC supplies (also Rod's design) coupled to the B+ supply and having it appear on the plates of the tubes.
Given both input and output tubes have active plate loads, PS noise rejection should be good (especially the 6J5), this is my preferred route. So...
Option A: include two 6.3V / 5A windings on my Sowter mains transformer and trust that my EL83 pentode CCS will reject the raw DC rectifier noise on the 45 (the MOSFET CCS on the 6J5 should take care of it no problemo)
Option B: mount the Hammond transformers under the chassis, possibly subjecting my internal components to their sizable magnetic fields - would try to minimize by optimizing the internal layout
I don't have the experience to say which of these two options is best, would love to hear from those who do, thank you.
I would recommend quantifying this. You're going to filter your B+ power supply after all...
With the noise that a #45 makes on its own, this kind of contribution will be like dropping a pebble in the ocean during a hurricane.
Okay, let me give it a shot. That is a pretty strong qualitative analogy you have there audiowize. BTW, I am using a 24:1 turns ratio OPT per your recommendation, which will help my case, thanks.
Here is what I came up with as far as PSRR and ripple goes on the supply, not including any potential coupled 120Hz noise from the filament raw DC supply.
First, here is the EL83 pentode CCS loading the 45:
So looking at PSRR, the AC impedance of the EL83 CCS is ~57.8k at 120Hz.
The ra of the 45 tube is 1650ohm, so:
PSRR = (RL + ra )/ ra = (57.8k+1650ohm) / 1650ohm = 36 = 31dB
With my power supply in PSUD2, I am expecting ~2mV rms of ripple at the top of the EL83. So, the ripple I would expect on the plate of the tube would be:
2mV rms / 36 = 55uV rms
This will be further stepped down by my 24:1 OPT, resulting in:
55uV / 24 = 2.3uV rms headphone output
In my supply, ripple off the resevoir cap is about 2V rms, then down to 2mV post CLC filter (60dB attenuation), then 31dB down across the EL83, then another 27dB down across the OPT.
With all that being said, not sure what kind of voltage I can expect from 120Hz coupling from the 6.3V / 5A winding, but I think it would have to be pretty significant to overcome 118dB of attenuation.
Does that make sense, audiowize? Thanks again for your input.
First, here is the EL83 pentode CCS loading the 45:
So looking at PSRR, the AC impedance of the EL83 CCS is ~57.8k at 120Hz.
The ra of the 45 tube is 1650ohm, so:
PSRR = (RL + ra )/ ra = (57.8k+1650ohm) / 1650ohm = 36 = 31dB
With my power supply in PSUD2, I am expecting ~2mV rms of ripple at the top of the EL83. So, the ripple I would expect on the plate of the tube would be:
2mV rms / 36 = 55uV rms
This will be further stepped down by my 24:1 OPT, resulting in:
55uV / 24 = 2.3uV rms headphone output
In my supply, ripple off the resevoir cap is about 2V rms, then down to 2mV post CLC filter (60dB attenuation), then 31dB down across the EL83, then another 27dB down across the OPT.
With all that being said, not sure what kind of voltage I can expect from 120Hz coupling from the 6.3V / 5A winding, but I think it would have to be pretty significant to overcome 118dB of attenuation.
Does that make sense, audiowize? Thanks again for your input.
Last edited:
With a cap input B+ supply, the danger is not controlling the local loop of the capacitive currents. In that case, you have a noisy ground loop that might get into the sensitive input circuit of the amp.
Attention to controlling ground loops fixes this.
With a choke input B+ supply, the danger is the magnetic field from the choke getting into the output transformer. Designing with proper angular orientation of the choke versus the output transformer; giving enough space between the choke and output transformer; and using a non magnetic chassis (steel conducts that hum from the choke to the output xfmr.
The same goes for the power transformer, orientation, spacing, and no steel chassis (or you get what I have got on some amplifiers).
Attention to magnetic coupling details fixes this.
The rest is simply a matter of filtering the B+ well enough (that is the easy part; noisy ground loops and magnetic couplings are the difficult parts).
I shoot for no more than 100uV noise at the 8 ohm tap measured with a non inductive 8 Ohm load resistor.
That is good enough for speakers, but I admit it is not good enough for some headphones.
Attention to controlling ground loops fixes this.
With a choke input B+ supply, the danger is the magnetic field from the choke getting into the output transformer. Designing with proper angular orientation of the choke versus the output transformer; giving enough space between the choke and output transformer; and using a non magnetic chassis (steel conducts that hum from the choke to the output xfmr.
The same goes for the power transformer, orientation, spacing, and no steel chassis (or you get what I have got on some amplifiers).
Attention to magnetic coupling details fixes this.
The rest is simply a matter of filtering the B+ well enough (that is the easy part; noisy ground loops and magnetic couplings are the difficult parts).
I shoot for no more than 100uV noise at the 8 ohm tap measured with a non inductive 8 Ohm load resistor.
That is good enough for speakers, but I admit it is not good enough for some headphones.
Last edited:
6A3 - thanks, right now I am looking at 2.3uV rms ripple on the headphone output according to PSUD2.
What I did above was to see what kind of 120Hz noise attenuation I can expect across the power supply filtering, output tube CCS, and OPT, the reason being I might include my 6.3V / 5A filament windings on my mains transformer (as opposed to dedicated transformers), and there is some concern the rectifier noise from the raw DC supplies might couple to the B+.
What I did above was to see what kind of 120Hz noise attenuation I can expect across the power supply filtering, output tube CCS, and OPT, the reason being I might include my 6.3V / 5A filament windings on my mains transformer (as opposed to dedicated transformers), and there is some concern the rectifier noise from the raw DC supplies might couple to the B+.
LOrdGwyn,
I use solid state rectifiers in the B+. Solid state diodes are noisier than tube rectifiers.
I use choke input whenever I can (there are sometimes restrictions like not enough B+ versus cap input; and sometimes restriction of space on the chassis).
I have found that the lowest noise in the B+ power supply that might couple to the filament windings on the same transformer is with choke input supplies.
Cap input causes transients with high frequency components.
Choke input is more average current, and low frequency components (with lots of Henrys, beyond the critical inductance number).
I have a pair of power transformers of the same make and model. They have similar filament loads, and similar B+ currents.
The choke input B+ amp has a power factor of 0.94.
The cap input B+ amp has a power factor of 0.76.
I do not have instruments to measure below 100uV; so when the answer is 0 uV, I do not know how much below 100uV the hum is.
I use solid state rectifiers in the B+. Solid state diodes are noisier than tube rectifiers.
I use choke input whenever I can (there are sometimes restrictions like not enough B+ versus cap input; and sometimes restriction of space on the chassis).
I have found that the lowest noise in the B+ power supply that might couple to the filament windings on the same transformer is with choke input supplies.
Cap input causes transients with high frequency components.
Choke input is more average current, and low frequency components (with lots of Henrys, beyond the critical inductance number).
I have a pair of power transformers of the same make and model. They have similar filament loads, and similar B+ currents.
The choke input B+ amp has a power factor of 0.94.
The cap input B+ amp has a power factor of 0.76.
I do not have instruments to measure below 100uV; so when the answer is 0 uV, I do not know how much below 100uV the hum is.