Interesting. I wonder what his thinking was behind that.
I have had a C3m as a pentode into an output pentode with Schade feedback floating around my head for some time. Thankfully I would post my schematic here first to get a critique! 🙂
Bill
I have had a C3m as a pentode into an output pentode with Schade feedback floating around my head for some time. Thankfully I would post my schematic here first to get a critique! 🙂
Bill
I think thats exactly the point. Feed in the ripple on G2 and cancel it with a capacitive divider.
The same is done in tube regulated power supplies sometimes
The same is done in tube regulated power supplies sometimes
Last edited:
Bill Brown,
I added (edited my post # 260) just as you were posting # 261.
It has a little more information for you.
I added (edited my post # 260) just as you were posting # 261.
It has a little more information for you.
v4lve lover,
Generally if you eliminate the B+ ripple in the first place, it is not necessary to cancel it out.
Generally, cancellations are a narrow deep null. If the cancellation is less than perfect, the deep null is not deep anymore.
If Either the Screen G2 to Plate gain changes (from his tube to the exact tube you use), Or if the exact 300B tube you use has different rp than the 300B tube he used (the 300B is powered by the exact same B+ ripple voltage), there will be little or no ripple cancellation.
And, the frequency response to the fundamental ripple frequency to its harmonics is basically flat for the SV83 circuit,
but the frequency response of the 300B circuit is not flat, it is dependent on the output transformer inductance, and also on the varying impedance versus frequency of the loudspeaker.
Complex indeed (I am not joking here, but there is a Pun in that).
I am sorry for my poor description of how this works (no two SV83 tubes are exactly the same, and no two 300B tubes are exactly the same). The right channel might need 10uF and 23uF, and the left channel might need 11uF and 22uF capacitive dividers to work with the exact SV83 and 300B tubes you have. After that, do not switch them from one channel to the other.
So, you have to change the either the 10uF or the 22uF capacitance to get the right ripple division, so that it will cancel the total ripple of the 2 stages.
If you like to do "tube rolling",then be prepared.
Each time you change either the SV83, or the 300B, you have to re-adjust the 10uF and 22 uF cap ratio.
A very complex solution to a simple problem.
How many of you are going to put capacitors in parallel with either the 10 or the 22uF to get the ripple of the particular SV83 and and the particular 300B you are using, in order to to get ripple cancelation?
Personally, I find a little ripple sound from my speakers of 180, 240, 300, or 360Hz to be more offensive than a little ripple at 120Hz.
Generally if you eliminate the B+ ripple in the first place, it is not necessary to cancel it out.
Generally, cancellations are a narrow deep null. If the cancellation is less than perfect, the deep null is not deep anymore.
If Either the Screen G2 to Plate gain changes (from his tube to the exact tube you use), Or if the exact 300B tube you use has different rp than the 300B tube he used (the 300B is powered by the exact same B+ ripple voltage), there will be little or no ripple cancellation.
And, the frequency response to the fundamental ripple frequency to its harmonics is basically flat for the SV83 circuit,
but the frequency response of the 300B circuit is not flat, it is dependent on the output transformer inductance, and also on the varying impedance versus frequency of the loudspeaker.
Complex indeed (I am not joking here, but there is a Pun in that).
I am sorry for my poor description of how this works (no two SV83 tubes are exactly the same, and no two 300B tubes are exactly the same). The right channel might need 10uF and 23uF, and the left channel might need 11uF and 22uF capacitive dividers to work with the exact SV83 and 300B tubes you have. After that, do not switch them from one channel to the other.
So, you have to change the either the 10uF or the 22uF capacitance to get the right ripple division, so that it will cancel the total ripple of the 2 stages.
If you like to do "tube rolling",then be prepared.
Each time you change either the SV83, or the 300B, you have to re-adjust the 10uF and 22 uF cap ratio.
A very complex solution to a simple problem.
How many of you are going to put capacitors in parallel with either the 10 or the 22uF to get the ripple of the particular SV83 and and the particular 300B you are using, in order to to get ripple cancelation?
Personally, I find a little ripple sound from my speakers of 180, 240, 300, or 360Hz to be more offensive than a little ripple at 120Hz.
Last edited:
Hmm i took a glance at the schematic. Should have seen the 6AS7 power supply. I am now quite excited. As i have another tube regulated power supply amplifier to add to my collection.
I will have to research what deep null means. But we are splitting hairs here. The inner mu and overal mu are related from what i gather. in any case 0.3mv ripple max on G2 is gonna contribute only slightly to noise and IMD i think. Once you count in cancelation. assume 75% is cancelled. that leaves what?
But your reason is correct, if you build it like described, it will work either way
I will have to research what deep null means. But we are splitting hairs here. The inner mu and overal mu are related from what i gather. in any case 0.3mv ripple max on G2 is gonna contribute only slightly to noise and IMD i think. Once you count in cancelation. assume 75% is cancelled. that leaves what?
But your reason is correct, if you build it like described, it will work either way
Heres some work i did on regulated power supplies some years back but that flew under everyones radar. Tube regulated power supplies
Ive used that after a CRCRC supply and have gotten under 300uV noise+hum at about 4VPP in.
Edit that was at 500 in 325 50 mA out with a different pass tube.
Ive used that after a CRCRC supply and have gotten under 300uV noise+hum at about 4VPP in.
Edit that was at 500 in 325 50 mA out with a different pass tube.
Last edited:
Andy assigned 12mA plate current in his design the Japanese design used 2ma, so here we see how different designers do it differently...
I understand that your position is: you have to cancel the ripple completely, otherwise you suffer intermodulation products of the fourier components of the power supply noise? That is perhaps true, Heptodes were used for mixers in tuners after all. My educated guess is that the designer weighed the options or had more complete data on the tube in question than we do.
There is a line stage amplifier on the Tubes / Valves threads that has B+ ripple cancellation of 2 stages. Seems to be very popular.
I just choose to solve the ripple problem another way.
I let the power supply be a power supply.
I let the amplifier be an amplifier.
I just choose to solve the ripple problem another way.
I let the power supply be a power supply.
I let the amplifier be an amplifier.
Yes. A power supply that has low ripple to start with is a good plan. I remember researching a what was considered acceptable ripple for power tubes and thinking why not get rid of as much as possible? The goal of low series resistance in a power supply can conflict with the goal of low ripple, as can space considerations. Some builders don’t want to use solid state components in their builds. I figure do what sounds best, but I must admit being annoyed with building the adjunct solid state circuits that are a part of many of the DHT designs today. Thorsten's Legacy is pretty much all tube with the exception of one component in his filament supply. I am still building and not in a position to listen properly yet. When I do listen, I will report back.
It’s great to have a forum with members who have already tried different things and who can report with their results. It’s like being given a partial map with which to find a treasure. That said, whether or not what I find is a treasure is going to be a personal choice, and entirely my call. I have heard too many crappy sounding systems that others love to believe that anyone else can provide the entire map for me.
I first determine the amplifier voltage needs, and current needs.
If I am stuck with working with a particular power transformer (I often am), then I have to determine if I can use a real choke input power supply, or a real capacitor input supply, or a modified 'choke input supply' (has a small cap before the choke, 0.5 to 4uF).
After the choke, i never use less than 50uF; then a resistor, usually 100 to 180 Ohms, and then a capacitor of at least 100uF, sometimes up to 500uF, that powers the output stage; then another resistor, perhaps 1k or more, and the last capacitor to power the input/driver or splitter/driver stage depending for single ended, or push pull respectively.
This is a starting place.
Since I use solid state diodes, the rectifier drop is almost a constant, and it is very small compared to a tube rectifier, a tube rectifier voltage drop changes more with variable current needs.
Then I begin to calculate the ripple all the way along, at each point.
I estimate the power supply rejection of each amplifier stage, and all the gain after that stage to predict what the ripple (hum) will be at the 8 Ohm output, from that particular stage and its power supply rejection of the ripple that powers it . . . after that has gone thorough the rest of the amplification stage(s), and the output transformer voltage reduction.
I can not think of a better way to describe this.
I calculate the ripple manually using my very old HP11 calculator.
I retired my Post VersaLog slide rule decades ago.
I estimate power supply rejection using stage rp, RL, and whether it is single ended, push pull, differential, etc.
I think it is not nice to allow lots of ripple on the B+ for a stage, just because it is either a push pull stage, or is a differential stage.
I try and treat those stages just as if it was a single ended stage (can't hurt to do that to a stage that already has a good power supply ripple rejection).
Software and I get along like a fish in air, instead of a fish in water.
Ditto for simulation software.
If I am stuck with working with a particular power transformer (I often am), then I have to determine if I can use a real choke input power supply, or a real capacitor input supply, or a modified 'choke input supply' (has a small cap before the choke, 0.5 to 4uF).
After the choke, i never use less than 50uF; then a resistor, usually 100 to 180 Ohms, and then a capacitor of at least 100uF, sometimes up to 500uF, that powers the output stage; then another resistor, perhaps 1k or more, and the last capacitor to power the input/driver or splitter/driver stage depending for single ended, or push pull respectively.
This is a starting place.
Since I use solid state diodes, the rectifier drop is almost a constant, and it is very small compared to a tube rectifier, a tube rectifier voltage drop changes more with variable current needs.
Then I begin to calculate the ripple all the way along, at each point.
I estimate the power supply rejection of each amplifier stage, and all the gain after that stage to predict what the ripple (hum) will be at the 8 Ohm output, from that particular stage and its power supply rejection of the ripple that powers it . . . after that has gone thorough the rest of the amplification stage(s), and the output transformer voltage reduction.
I can not think of a better way to describe this.
I calculate the ripple manually using my very old HP11 calculator.
I retired my Post VersaLog slide rule decades ago.
I estimate power supply rejection using stage rp, RL, and whether it is single ended, push pull, differential, etc.
I think it is not nice to allow lots of ripple on the B+ for a stage, just because it is either a push pull stage, or is a differential stage.
I try and treat those stages just as if it was a single ended stage (can't hurt to do that to a stage that already has a good power supply ripple rejection).
Software and I get along like a fish in air, instead of a fish in water.
Ditto for simulation software.
Last edited:
With enough practice and experience, the rules of thumb are working properly.
Calculations? Nearly nothing. Its all in the practice of knowing how to do it.
I never calculate hum or ripple, and I think all good amps of the past were so easily designed with great experience, that no one does this.
Just apply what you've learned from studying great designs. They share all the same rules of designing, because thats what a great design is about. It couldn't be fooled, just by amateurs who think they have to change values by factor 10 for better measures. But they are wrong. I see this often with amateur designs.
Then your venerable HP 11 can lay to rest.
Its just that easy.
P.S. I once met a wise man with experience in tube radios. All he needed for locating of faults was an old screwdriver and his finger. No Oh-silly scope, no voltage meter, no calculator. He repaired every radio with this method. Just an old trained professional.
Calculations? Nearly nothing. Its all in the practice of knowing how to do it.
I never calculate hum or ripple, and I think all good amps of the past were so easily designed with great experience, that no one does this.
Just apply what you've learned from studying great designs. They share all the same rules of designing, because thats what a great design is about. It couldn't be fooled, just by amateurs who think they have to change values by factor 10 for better measures. But they are wrong. I see this often with amateur designs.
Then your venerable HP 11 can lay to rest.
Its just that easy.
P.S. I once met a wise man with experience in tube radios. All he needed for locating of faults was an old screwdriver and his finger. No Oh-silly scope, no voltage meter, no calculator. He repaired every radio with this method. Just an old trained professional.
Last edited:
Thank you, too, for providing your approach to PS design. I take the same approach, and have read how to do the calculations manually, but must admit to using PSUD II 🙂
Bill
Bill
Almost everything of importance was known hundred years ago. The design improvements have aimed at improving efficiency not performance. Efficiency and performance stand in sharp contrast to each other.
Last edited:
I know many inventions that improved performance that are not very old (Allen Wright's cathode follower, David Berning's Zotl, John Strickland's Transnova and HV OTL's just to name a few). Many here like the Mu Follower and that is not so old either. PPP amps and OTL's like the Tim Mellows version or Atmasphere. I can go on and on....
Try John Broskie's website or Steve Bench...
Try John Broskie's website or Steve Bench...