Professors sometimes get things wrong - they can misunderstand textbooks or their own professors. Textbooks rarely get things wrong, although they sometimes oversimplify. Students often get things wrong, and such misconceptions can stay with them through their professional career. Some students end up as professors, and so the cycle is complete.Lingwendil said:
There is just enough information there to confuse a newbie.zelgall said:
For example, full-wave choke input has Idc = 1.54 Iac: what does that mean? I am surprised to see Hammond putting out such low quality information.
Lampizator's ideas are popular. I don't know why.
> This is a good reference for transformer / rectifier
It is actually very flawed. (The half-wave cap is obvious.)
I challenged Hammond. They said they got it off the Web, and basically denied responsibility.
What is "Vavg" on a cap-input DC supply? If ripple is small, the most-usual audio-supply need, it is very nearly Vpk adjusted for source/load resistances. That chart took the no-filter result and wrote in on the filter case.
> I am surprised to see Hammond putting out such low quality information.
I was speechless.
Here is _my_ mark-up of the points I am very sure of.
It is actually very flawed. (The half-wave cap is obvious.)
I challenged Hammond. They said they got it off the Web, and basically denied responsibility.
What is "Vavg" on a cap-input DC supply? If ripple is small, the most-usual audio-supply need, it is very nearly Vpk adjusted for source/load resistances. That chart took the no-filter result and wrote in on the filter case.
> I am surprised to see Hammond putting out such low quality information.
I was speechless.
Here is _my_ mark-up of the points I am very sure of.
Attachments
The Sowter page is also potentially very confusing, because it appears to use the convention that HT secondary windings are rated according to DC current draw from a particular PSU circuit rather than the AC RMS rating used more commonly. Thus the page may be fine for someone specifying a Sowter transformer, but could be quite wrong for another manufacturer.
Clearly you know the reason for this. So why not just say it? Some form of political correctness perhaps? 😎
It's the bloody hexfreds of course. Can you seriously expect them to be an inaudible addition?
But then, what about the bass and dynamics? Musicality is fun for a day or two but gets old quickly 😀
Surprisingly you can have your cake and eat it too. Just need to forget about vacuum rectifiers and any form of solid state and get yourself a nice selection of mercury: a simple 83 is an excellent start for a preamp or dac.
Yes, i know, every day you spend listening to mercury a kitten dies, but it's still worth it.
Don't you know that the internet forums are also used by shills to promote their products? And the list doesn't end there. Sharing audiophile fantasies, bragging about their fancy looking components,... etc.
Time to wake up and notice the stench of the reality.
Patrick Turner wrote a nice page about tube rectifiers.....powersupplies
if only they knew the physics of things....personal anecdotes are, well, anecdotes...
never matters to me....i am never swayed...
if only they knew the physics of things....personal anecdotes are, well, anecdotes...
never matters to me....i am never swayed...
> Sowter page is also potentially very confusing, because it appears to use the convention that HT secondary windings are rated according to DC current draw ...rather than the AC RMS rating
I agree it is confusing. It is easy to get yout Vac and Vdc flipped. It is easy to be confused about CT, whole winding or half-winding.
However take the Full Wave Bridge formula: Iac = Idc x 1.61
This is in the correct form. I would quibble that "1.61" may apply to heat-limited designs/sizes and "1.8" is better when sag is the limiting factor. However this is mooted because we can't buy "exact" current and are likely "forced to buy" over-size by more than a 1.6:1.8 difference. And in my sloppy work I just use "2".
The CT 2-diode connection has each winding utilized only half the time. This seems to equate to my Vac=2*Vdc over-guess and may be conservative, but not real wrong.
The choke input connections "look right". The larger effective factor (allowing for half-winding) for the 2-diode CT form may be right.
I am however suspicious that the choke-in and cap-in forms work out to the "same" VA/Watts ratios. CT 1.41 or 1.44 (same), FWB 1.14 or 1.18 (same). Transformer heat will go with RMS current and the spike-wave of cap-input leads to much higher RMS heating.
If it really mattered, mattered as in "big money", we would send a test-mule and specify the PT to do "with 117VAC input, deliver 350V DC across onboard dummy load with less than 50 deg surface rise and 5,000 hour expected life". MIL-spec gear does this. For 1 or 1,000 it is cheaper to buy stock parts.
> Sowter page is also potentially very confusing
Then again, that page confuses me too. I doubt folks who have not spent years smoking windings will be comfortable.
I agree it is confusing. It is easy to get yout Vac and Vdc flipped. It is easy to be confused about CT, whole winding or half-winding.
However take the Full Wave Bridge formula: Iac = Idc x 1.61
This is in the correct form. I would quibble that "1.61" may apply to heat-limited designs/sizes and "1.8" is better when sag is the limiting factor. However this is mooted because we can't buy "exact" current and are likely "forced to buy" over-size by more than a 1.6:1.8 difference. And in my sloppy work I just use "2".
The CT 2-diode connection has each winding utilized only half the time. This seems to equate to my Vac=2*Vdc over-guess and may be conservative, but not real wrong.
The choke input connections "look right". The larger effective factor (allowing for half-winding) for the 2-diode CT form may be right.
I am however suspicious that the choke-in and cap-in forms work out to the "same" VA/Watts ratios. CT 1.41 or 1.44 (same), FWB 1.14 or 1.18 (same). Transformer heat will go with RMS current and the spike-wave of cap-input leads to much higher RMS heating.
If it really mattered, mattered as in "big money", we would send a test-mule and specify the PT to do "with 117VAC input, deliver 350V DC across onboard dummy load with less than 50 deg surface rise and 5,000 hour expected life". MIL-spec gear does this. For 1 or 1,000 it is cheaper to buy stock parts.
> Sowter page is also potentially very confusing
Then again, that page confuses me too. I doubt folks who have not spent years smoking windings will be comfortable.
I'm not sure they did it in reverse. The voltage equations are correct, once you realise that they specify a CT winding by the total voltage. The current equations, amd hence the power equations, are suspect. They imply that a cap input supply uses the transformer almost as efficiently as a choke input supply.
Ok, I have swapped the Lundahl-Transformer against a normal EI which happens to have 340-250-0-250-340 Taps at 80mA. The secondaries have now 210 Ohm per 0-350 Legs and 170 per 0-250 leg....so a bit more than Lundahl.
I built the switch into the PSU and when I take the PIO 4uF (Sangamo i believe) out and change from 250 to 340, I have +- 1 Volt the exact same HV.
It is clear that ripple goes up, so I have experimented with an additional 500Ohm-47uf in the chain, so four filter settings tried:
- 4uF(pio,Sangamo)-21H(90ohm, amorphous core)-100uF(Mundorf tubecap, polyprop)-two legs:2000Ohm-47UfMundorf tube cap 250-0-250
- same, but without 4uF and 340-0-340
- 4uF-21h-100uf-500ohm-47uf-1000ohm-47uf(two legs) 250-0-250
- same, but without 4uf and 350-0-350
What do I hear:
- The additional RC in 3 and 4 reduces the live charakter quiet a bit, it kills microdynamics and does not bring an advantages, the higher ripple of choke input from PSUD simulation is not audible anyhow. It reminds me a bit on the effect when you go in a loudspeaker crossover from 12db to 18 or 24 db filters.
- When switching between 1 and 2: The choke input sounds very stiff and extra-dry. Maybe very correct, but as well the naturalness of the tone and its vibrations seems to be choked away. I have read in a guitar forum that to high inductance can cause this.
I am a bit surprised again. I expected a clear winner, being the Choke-input with additional RC-filtering and in fact it was the worst sounding combination. In fact I like the cap-input with one simple RC behind best.
I thought about it and maybe the low current I need (44mA) is an explanation. Basically the stress on the rectifier in cap inpit is as low as if I would have used it in LC at its max specs...peak current on the diodes at 22mA if I look into PSUD simulations...
Kevin Carter from Kandkaudio did once a comment which was like: At a Metalevel, an amp is a machine which transforms AC from the power plug into music and as such the psu is very audible. It seems to me that this is dramatically true, it is even more like building a really good cross over where time domains and frequency amplitudes and that stuff comes together.
Next steps:
-I will experiment with size of first cap, will exchange more caps from polyprop without oil to ASC/GE oil filled
- Will experiment with more chokes, lower value and CLCLC instead CLCRC
- Will revisit FW vs. Bridge with TV damper diodes like 6ax5 etc. ( I believe someone had the same suspicion than I have about being able to hear thehexfreds in the hybrid bridge)...
- Will experiment if the high-cost parts I used are really necessary (amorphous core vs. normal, Psvane 274 replica vs. Damper diodes)
- Will test the above mentioned 83, 866 (which looks extremly cool) and some xenon filled tubes
I built the switch into the PSU and when I take the PIO 4uF (Sangamo i believe) out and change from 250 to 340, I have +- 1 Volt the exact same HV.
It is clear that ripple goes up, so I have experimented with an additional 500Ohm-47uf in the chain, so four filter settings tried:
- 4uF(pio,Sangamo)-21H(90ohm, amorphous core)-100uF(Mundorf tubecap, polyprop)-two legs:2000Ohm-47UfMundorf tube cap 250-0-250
- same, but without 4uF and 340-0-340
- 4uF-21h-100uf-500ohm-47uf-1000ohm-47uf(two legs) 250-0-250
- same, but without 4uf and 350-0-350
What do I hear:
- The additional RC in 3 and 4 reduces the live charakter quiet a bit, it kills microdynamics and does not bring an advantages, the higher ripple of choke input from PSUD simulation is not audible anyhow. It reminds me a bit on the effect when you go in a loudspeaker crossover from 12db to 18 or 24 db filters.
- When switching between 1 and 2: The choke input sounds very stiff and extra-dry. Maybe very correct, but as well the naturalness of the tone and its vibrations seems to be choked away. I have read in a guitar forum that to high inductance can cause this.
I am a bit surprised again. I expected a clear winner, being the Choke-input with additional RC-filtering and in fact it was the worst sounding combination. In fact I like the cap-input with one simple RC behind best.
I thought about it and maybe the low current I need (44mA) is an explanation. Basically the stress on the rectifier in cap inpit is as low as if I would have used it in LC at its max specs...peak current on the diodes at 22mA if I look into PSUD simulations...
Kevin Carter from Kandkaudio did once a comment which was like: At a Metalevel, an amp is a machine which transforms AC from the power plug into music and as such the psu is very audible. It seems to me that this is dramatically true, it is even more like building a really good cross over where time domains and frequency amplitudes and that stuff comes together.
Next steps:
-I will experiment with size of first cap, will exchange more caps from polyprop without oil to ASC/GE oil filled
- Will experiment with more chokes, lower value and CLCLC instead CLCRC
- Will revisit FW vs. Bridge with TV damper diodes like 6ax5 etc. ( I believe someone had the same suspicion than I have about being able to hear thehexfreds in the hybrid bridge)...
- Will experiment if the high-cost parts I used are really necessary (amorphous core vs. normal, Psvane 274 replica vs. Damper diodes)
- Will test the above mentioned 83, 866 (which looks extremly cool) and some xenon filled tubes
Last edited:
The aim with amp and PSU design is to ensure that both the amp and PSU are inaudible. Only the music should be audible.Blitz said:
So, you have an amp which seems sensitive to PSU - so the amp may have poor PSRR or poor bias choice or insufficient feedback.
You also have a variety of PSUs which seem to have varying levels of output voltage, hum, buzz and output impedance across various frequency ranges.
Basically, you seem to be thinking and acting like a medieval alchemist instead of a 19th century chemist. "A bit more ground toad will improve the elixir, or should I try finely-chopped bat?"
Df96, you make me smile a bit ...as the circuit and its original PSU design is coming from one of the Spice-maniacs, John Broskie:
http://www.tubecad.com/2011/03/blog0203.html
...so I guess that he put a little bit of thinking into thoses things...
...so, instead of becoming personal, let's conclude: There is a difference between spice and sound and not everything can be simulated. And that makes the differenct between good and great.
http://www.tubecad.com/2011/03/blog0203.html
...so I guess that he put a little bit of thinking into thoses things...
...so, instead of becoming personal, let's conclude: There is a difference between spice and sound and not everything can be simulated. And that makes the differenct between good and great.
- Status
- Not open for further replies.