I have build a PSU for a DAC output stage which need 44mA. I used a
http://www.lundahl.se/wp-content/uploads/datasheets/1683.pdf
for this, which works into Mullard GZ34. A switch allows me to switch between FW, so the transformer has 250-0-250 and 200 Ohm total for the secondaries in series and 0-250, where two secondaries are in parallel and 50ohms. the bridge is complemented with two hexfred diodes. a second switch takes these hexfreds in or out, so I have either a bridge or FW.
The rest of the filtering is 30uf-16H(550ohm)-100uf-2k-47uf, all Mundorf tubecaps/polyprop.
When switching between the two, the FW has approx 7V less B+, so approx 275V vs 282V. A bit of a bias change, but not dramatical, so I would not have expected a lot sound difference by a change of 3%.
What I hear:
- The FW is significantly softer. Very musical. You do not hear anything nasty or annoying at all. The only critismn: Lack of dynamic a bit, seems to be "nice" but as well less airy/ transparent and less energy. You can follow female voices at ease, a real pleasure.
- The hybrid-bridge: Super-Turbo-Charged. Like one of these muscle cars where you can inject something into.Very dynamic, seems a tad louder and more transparent. But has by far not the sweetness and the musicality of the FW. More hifi like.
Well...I am surprised. And not sure what exactly causes what. 3% bias change should not be the source of this I guess.
I have read in a forum where some people reported only to use FW and low impedance transformer / chokes and parallel the diodes of an GZ34 etc to get impedance even further down.
So, is it the efficiency of the Bridge design, the impedance of the transformer which is reduced, the stiffness of the ss-diodes in the bridge or what causes this difference in sound ?
Obviously I am trying to find a way where both is combined, the musicality of the FW-Tube-only solution with the dynamics and transparency of the Hybrid-Bridge.
http://www.lundahl.se/wp-content/uploads/datasheets/1683.pdf
for this, which works into Mullard GZ34. A switch allows me to switch between FW, so the transformer has 250-0-250 and 200 Ohm total for the secondaries in series and 0-250, where two secondaries are in parallel and 50ohms. the bridge is complemented with two hexfred diodes. a second switch takes these hexfreds in or out, so I have either a bridge or FW.
The rest of the filtering is 30uf-16H(550ohm)-100uf-2k-47uf, all Mundorf tubecaps/polyprop.
When switching between the two, the FW has approx 7V less B+, so approx 275V vs 282V. A bit of a bias change, but not dramatical, so I would not have expected a lot sound difference by a change of 3%.
What I hear:
- The FW is significantly softer. Very musical. You do not hear anything nasty or annoying at all. The only critismn: Lack of dynamic a bit, seems to be "nice" but as well less airy/ transparent and less energy. You can follow female voices at ease, a real pleasure.
- The hybrid-bridge: Super-Turbo-Charged. Like one of these muscle cars where you can inject something into.Very dynamic, seems a tad louder and more transparent. But has by far not the sweetness and the musicality of the FW. More hifi like.
Well...I am surprised. And not sure what exactly causes what. 3% bias change should not be the source of this I guess.
I have read in a forum where some people reported only to use FW and low impedance transformer / chokes and parallel the diodes of an GZ34 etc to get impedance even further down.
So, is it the efficiency of the Bridge design, the impedance of the transformer which is reduced, the stiffness of the ss-diodes in the bridge or what causes this difference in sound ?
Obviously I am trying to find a way where both is combined, the musicality of the FW-Tube-only solution with the dynamics and transparency of the Hybrid-Bridge.
Last edited:
the FW has better transformer utilization than a center tapped rectifier, why?
because a FW rectifier has current flowing all of the time,
whereas in the center tapped psu, one diode conducts current per 180* phase,
half of the time, the diodes are doing nothing in the other half of the cycle....
magnetizing effect is absent with the FW case than with the CT,
that is why more power is available for the FW than with a CT...
and you want to do that for a DAC?
ever heard of physics? what is the load presented by the DAC to the psu?
knowing this helps a lot to knowing what you need....
but then again, what you need and what you want may not be the same....
all for a DAC?
because a FW rectifier has current flowing all of the time,
whereas in the center tapped psu, one diode conducts current per 180* phase,
half of the time, the diodes are doing nothing in the other half of the cycle....
magnetizing effect is absent with the FW case than with the CT,
that is why more power is available for the FW than with a CT...
and you want to do that for a DAC?
ever heard of physics? what is the load presented by the DAC to the psu?
knowing this helps a lot to knowing what you need....
but then again, what you need and what you want may not be the same....
all for a DAC?
I believe I have defined the load with 44mA...
When you talk FW: You mean bridge, right ? I believe a ct transformer withtwo diodes is as well fullwave, bit as you say only one diode is working at a time...
And you are right: It is very surprising that you can hear the difference...as this is "just" for a dac output...I would have not been surprised not to hear any difference at all....by the way: it is the unbalancer from broskie, so four 12bh7 in total.
When you talk FW: You mean bridge, right ? I believe a ct transformer withtwo diodes is as well fullwave, bit as you say only one diode is working at a time...
And you are right: It is very surprising that you can hear the difference...as this is "just" for a dac output...I would have not been surprised not to hear any difference at all....by the way: it is the unbalancer from broskie, so four 12bh7 in total.
How constructive. There are many people who discuss the difference of the sound of the rectifier tubes. People who pay more thann 1000 bucks for a We274 etc. They do this because they love nonsense ? I think sharing findings and knowledge is the purpose of a forum and not bashing fellows.
You can experience with different SS diodes in the bridge to tweak the sound. For low voltage applications, Schottky SB160 are my favorite. For high voltages, try Ge diodes, the Russian D7J for example. For me, they come very close to the sound of a tube rectifier.
A bridge rectified psu (four diodes, full wave bridge) is lower impedance, and will utilise the windings of the transformer throughout the whole cycle, as the diodes steer the current from each end of the winding throughout the cycle.
A center tapped psu (two diodes, each rectifying half the wave each, which is still a full wave design, btw) is higher impedance, and does not source current equally through the windings at any given time, and basically pulses current through each cycle. The transformer is not working as efficiently in this case.
FWIW, I think a low impedance supply is better in 90% of cases. Shunt regulators for low current, series for high current is my preference.
As for your noted differences, it makes a little sense, as the topologies are a bit different. I personally don't care for center tapped psu designs unless the center is being used as reference, instead of ground. Personally I like using solid state for my rectifiers, and a full wave bridge. Unless you are working with a fully constant current design, the power supply can vary in drop throughout the range of operation, which doesn't sound like what you want in a preamp for something intended to be low distortion and high fidelity. I'm a huge fan of voltage regulators, especially on something like a DAC output stage, where you want everything to stay put and stay clean.
Slap a voltage regulator on it, and be amazed at how open and clean everything will sound, especially considering you felt the hybrid bridge felt "turbocharged" you will notice it feels like swapping your Volkswagen 1200cc single carb engine to a big dual carb stroker.
As for all the difference, in the end none of it is very reliable unless you were measuring or using an AB box in a blind test, many of the Golden Ear crowd has been very embarrassed by the differences they could and couldn't hear in such tests.
I've grown wary of anyone making drastic claims, as to being able to hear the differences between brands of rectifier tubes, resistor brands of the same construction, audiophile parts, etc. In most case they try to sell something or shill for someone else, or are simply full of it.
Please don't take any of this the wrong way, as it's all personal experience, opinion, and results of my measurements and schooling. Just trying to be constructive. When I was doing testing in my former life at college, it was interesting to throw circuits up on the bench, and measure identical circuits that had several expensive parts being the only difference. There were often identical results from identical parts values, and occasionally worse results from using Audiophile grade parts.
Do you have a schematic of your circuit in question? Is it a current-to-voltage conversion stage? I'm getting ready to start on a PCM1794 project, and was going to most likely do a tube output stage as well.
No. What you have described is what happens in a choke input PSU, whether fed by a bridge or not.Lingwendil said:
No. What you have described is a capacitor input PSU, whether fed by a full-wave rectifier or not.
A small change in bias could have a noticeable effect on sound if the original bias point was at the end of the 'good' range and the change pushed it just further into the good range or just out of the good range i.e. it could be a sign of poor bias choice.
Another possibility is a change in ground currents causing some change in hum or intermodulation.
It is very difficult in electronics to isolate one change to a circuit and ensure that this change alone is what is affecting the result. There are almost always knock-on effects too.
My guess, given the OP's description, is that the change in bias did it due to a poor bias choice. What he is hearing is a change in low order distortion, from low ('hi-fi') to slightly higher ('musical').
Well, that's the way my textbooks and professor described it, so I need to write a couple angry letters now 😛
I'm also thinking that some circuit parameters may have changed due to the differences in bias, but it also depends on the circuit in question, the way these switches he used are wired, etc. Hum can inject some really goofy harmonics that can be seen on a scope...
Well, I can build the last r in the rc switchable as well so that we have the exact same voltage. The current through the first tube is set by a current source though.
http://www.tubecad.com/2011/03/blog0203.htm
I realized that Lynn Olson took the path in his karna towards hybrid as well:
http://www.nutshellhifi.com/Karna.gif
...while before he used tube-only...I have send him a mail and asked him for his reasoning. I guess that was intended.
I will build a fullwabe bridge now with four 6d22s and make it switchable between four tubes vs. ss/tube. Let's see if that makes a difference. Maybe I will as well switch additional resistance between transformer and bridge, so that we have 100ohm to be seen by the rectifier in both scenarios instead of 50 vs 100(measured from Ct).
On regulators: The Salashv2 is build and waits to get into the game and Broskie's janus as well...Maida is on its way...superreg from Allen W. ist as well waiting...but I wanted to do it step by step...
http://www.tubecad.com/2011/03/blog0203.htm
I realized that Lynn Olson took the path in his karna towards hybrid as well:
http://www.nutshellhifi.com/Karna.gif
...while before he used tube-only...I have send him a mail and asked him for his reasoning. I guess that was intended.
I will build a fullwabe bridge now with four 6d22s and make it switchable between four tubes vs. ss/tube. Let's see if that makes a difference. Maybe I will as well switch additional resistance between transformer and bridge, so that we have 100ohm to be seen by the rectifier in both scenarios instead of 50 vs 100(measured from Ct).
On regulators: The Salashv2 is build and waits to get into the game and Broskie's janus as well...Maida is on its way...superreg from Allen W. ist as well waiting...but I wanted to do it step by step...
Last edited:
One of two things is true - either your books and professor were wrong, or you didn't correctly understand what they said. DF96 is completely correct in how he has described the circuit operation for the two cases.
If this is for a DAC output stage - then what is the point in feeding it with a crap supply that you can hear causing effects? If the supply was any good it would not be audible, period. Even if the final filter stage was audible, if you can hear a change caused by rectification (as opposed to caused by a shift in biasing) then the power supply filter isn't doing its job, either due to poor a design or a poor layout.
probably a little bit of both 😀 Pothead guitar guy electronics professor, community college, and my (at the time) limited understanding of the subject matter. It's easy to remember it wrong when it was explained wrong 🙄 and over time you get so used to using parts that you never really think about it. I stand humbled after a graceful correction by the mighty DF96.
That's why I like regulators, properly designed they make a great addition to the project. Especially considering that the line voltage these days can swing a good many volts during peak times in some areas (like here in the SF bay area, California) my wall voltage can go from 115 to 135 depending on time of day within an hour or two!
I know what you mean - I had to "unlearn" a few things after I thought I knew how stuff worked but found out that I didn't!probably a little bit of both 😀 Pothead guitar guy electronics professor, community college, and my (at the time) limited understanding of the subject matter. It's easy to remember it wrong when it was explained wrong 🙄 and over time you get so used to using parts that you never really think about it. I stand humbled after a graceful correction by the mighty DF96.
115 to 135...? Wow! That's a hell of a jump, and could easily cause problems. In such a case regulation makes complete sense. Here in the UK at my home is see about 238 on a low day and maybe 240 or so when it's high. Of course, there will always be localised bad spots.
This is a good reference for transformer / rectifier combinations (and it doesn't matter whether it's a valve or solid state rectifier):
Last edited:
This is a good reference for a beginner to potentially make things go wrong, potentially too wrong, if followed blindly without further information.
What are the circumstances needed for the choke input schematic to perform the way shown by the formulas below?
I have the same question towards the capacitor input ones.
I dont have these questions as I simulate mine with PSUD upfront and look at the Q-Factor of each node. Do you as well ?
As much as we may want things to go our way, they often don't.
The Hammond info sheet isn't a manufacturers opinion, it's the law. The diagrams describe most of the common transformer / rectifier arrangements.
The formulas are very basic and aren't open to interpretation; Ohms law rules.
I suspect the differences you claim to hear would not pass a blind test.
There is nothing in that minute supply voltage variation which could cause such a change.
And in the hypotethical case it did (which I don´t think possible) , that would mean that your DAC is horribly designed.
All of which I posted earlier in a way more succint way, but which meant basically the same.
And please don`t "you don´t know what I hear" me, those who make extraordinary claims are those who must offer extraordinary proof.
These formulas alone are barely enough to predict the behavior of a power supply shown in this article.
Are you aware that, for a choke input filter to output Vdc=Vac*0.9, a critical load is needed that depends of the choke value and flowing current? Why it isn't mentioned in your article?
Are you aware that practically, a PSU with a C input filter will only reach Vdc=Vac*1.41 if unloaded? What happens when it's loaded? How does the C value affect the Vdc drop vs. load?
And I'm neglecting Ohmic losses here.
That is what I meant. The Hammond sheet is nice as a rule of thumb estimating sheet. But it does not deliver you the values which depend on many more parameters, starting with the different transformer regulations, intermediate modes as Morgan Jones called them, DCR if the chokes following, current flowing, diodes used when talking tubes etcetc. And then we still do not talk filter characteristics, resonancing of LC nodes, step behavior for current changes etc.
PSUD is helpful or something more comprehensive as spice....
By the way I have changed yesterday the rectifier from Mullard GZ 34 to Psvane 274A replica and made a couple of adjustments, so that bias was the same before and after.
The WE replica blew the Mullard way in a breath. Unbelievable liquid, fine resultion and grainless, like go from blue-ray to 4k. It is now burning in...people reported it needs quite a while before it is up at its full performance. I will build a timer into it, so that the directly heated 274 gets heated a few seconds later than the indirect tubes to safe some tube life as diretly heated rectifiers conduct fully after few seconds...currently each has its own switch.
If someone thinks this all is crazy for a DAC and snake oil...most of us (at least in Europe) live in free countries, right ? So why not try as a Diyer what other try to sell for BIG $$$$ like here : http://www.lampizator.eu/Fikus/DACs.html ...the guy is well known and has made a company out of this....so I guess people are valueing his findings and he ended up with big, directly heated rectifiers...and even DHT in the output stage of his dacs...
Back to the subject:
I studied yesterday again Morgan Jones and there was an interesting graph showing that choke-input PSU generate the lowest internal impedance, cap-input second and intermediate-mode (cap-input with first cap max 10uF) highest impedance....soooooo....trying to get the musicality captured of a FW (non-bridge) with the turbo-effect (lower impedance of a bridge as suggested here?) I will built a switchable PSU next where I can compare the sonics if choke-input vs. cap-input, I have a transformer laying around with the right taps.
PSUD is helpful or something more comprehensive as spice....
By the way I have changed yesterday the rectifier from Mullard GZ 34 to Psvane 274A replica and made a couple of adjustments, so that bias was the same before and after.
The WE replica blew the Mullard way in a breath. Unbelievable liquid, fine resultion and grainless, like go from blue-ray to 4k. It is now burning in...people reported it needs quite a while before it is up at its full performance. I will build a timer into it, so that the directly heated 274 gets heated a few seconds later than the indirect tubes to safe some tube life as diretly heated rectifiers conduct fully after few seconds...currently each has its own switch.
If someone thinks this all is crazy for a DAC and snake oil...most of us (at least in Europe) live in free countries, right ? So why not try as a Diyer what other try to sell for BIG $$$$ like here : http://www.lampizator.eu/Fikus/DACs.html ...the guy is well known and has made a company out of this....so I guess people are valueing his findings and he ended up with big, directly heated rectifiers...and even DHT in the output stage of his dacs...
Back to the subject:
I studied yesterday again Morgan Jones and there was an interesting graph showing that choke-input PSU generate the lowest internal impedance, cap-input second and intermediate-mode (cap-input with first cap max 10uF) highest impedance....soooooo....trying to get the musicality captured of a FW (non-bridge) with the turbo-effect (lower impedance of a bridge as suggested here?) I will built a switchable PSU next where I can compare the sonics if choke-input vs. cap-input, I have a transformer laying around with the right taps.
Last edited:
- Status
- Not open for further replies.