|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
128.194.160.53
In Reply to: RE: ALT. 1. posted by drlowmu on January 09, 2015 at 07:30:08
"I was already fully aware of all that you wrote...."
So you agree that the DC resistance of your preferred power supply is greater than that of the one you are comparing it with?
But you didn't address my main points:
1) What do you mean by "resistance to the 300B Finals Filter"? The figures you presented are just the sum of the DC resistances of the chokes. But why would that figure, of itself, be significant? The transformer secondary and the rectifier are also in that circuit.
2) A stepped load is not representative of the power demand from an SE amplifier. The current draw by the amplifier is, on average, constant, with the timescale of the fluctuations being the timescale of the frequencies in the audio signal. So what really matters as far as the amplifier is concerned is the output impedance of the power supply at audio frequencies, and how that impedance varies as a function of the audio frequency.
Chris
Follow Ups:
Yes it was explained in that other thread. I myself self learned something from that thread.
It was interesting when I modeled a typical PSU compared to the low-everything (Farkwin), the low everything was worse in regards to how long it could hold the original desired voltage.
Drlomu's PSU drops voltage like a stone in comparison, specifically in the region where transient response is important.
This reenforces my experiences that "stiffer" power supplies deliver better transients.
While I agree that PSUD cannot replicate the amplifier load accurately, it does however give an idea as to how the PSU will respond to a dynamic request.
Since the desired response is a flat line, then you want a PSU that holds a flatter line, if only for a short time.
Music is transient and random. Liveliness IMO comes from the attack.
If you look at my model below from at the start (4.1 seconds), to say 4.12 seconds you will see that Drlomu's PSU has already dropped more then twice the voltage of my more typical design.
Even though tubes sag more over time then SS diodes, I find that the initial slope is comparable between the two. Therefor one doesn't have to be overly concerned on the choice of rectification from this particular transient viewpoint.
I used a similar concept but with SS diodes in my little RH84 with wimpy OPTs and 604 duplex, it has dynamics that will make you blink.
"I used a similar concept but with SS diodes in my little RH84 with wimpy OPTs and 604 duplex, it has dynamics that will make you blink."
How ironic. I have recently set up a variation of the SE RH EL84.
Exactly what PS are you using and what is the B+ off the rectifier?
Did you get any distortion or bandwidth measurements?
I posted some measurements on Tube DIY of my circuit performance.
"Did you get any distortion or bandwidth measurements?"
NoHere are the details of the RH PSU in the sim below.
Also for comparison with the two examples I submitted earlier, I applied the stepped load and the same 2V - .125 second scale to the chart.
As you can see this RH PSU is a little stiffer again then my earlier example. It's got some snap to it.
Edits: 01/09/15
You have taught yourself how to download and input data into PSUD2. Good job.Now, you need to learn how to properly interpret what it shows AND how it applies to audio design in power supplies. You are not doing that at all, you have several things backwards.
Swenson, a graduate EE, has measured music and he tells us essentially, that the first 50 mS are "free". He tries to get "his" supply to settle within 50 mS. ... but I don't do that any more, I have other design priorities.
Opposite of what you wrote and thought, it is beneficial for the supply to " fall like a rock " and settle quickly ( but smoothly ) in the first 50 mS. After the 50 mS. time, the quicker it settles "overall", the more able it will be - to play back the very next large transient, inherent in the music.
With all that UNusual storage ( 30 HYs and 150 uF ) the supply you simulated has a better dynamic Z than ALT.1, but, ALT.1 settles fully, and smoothly with no overshoot, in 250 mS, ( 4.35 mS. ) whereas yours is still recovering in up to 400 mS. of time ( 4.5 mS. ). When music's next transient comes along, your supply is still dropping, and ALT.1 has possibly fully recovered, and is ready to please.
The worse thing of all is, the human ear / brain will instantly pick up on large slow B+ filter parts to the Finals ( high L and high C ) as being slow and out of time with the music. So it sounds out-of-time, slow, like an audio amp - like 99% of the amps out there, and not like real music.
Where do we find a 30 HY choke at 220 Ohms ? Would 19 Ohms of series resistances in a B+ filter to the FINALS sound better than 270 Ohms ?
Jeff Medwin
Edits: 01/09/15
"Swenson, a graduate EE, has measured music and he tells us essentially, that the first 50 mS are "free"."What event precedes the "free" 50mS that you're referring to? Surely not the onset of a musical demand by the amplifier?
" the human ear / brain will instantly pick up on large slow B+ filter parts to the Finals"
The amplifier can only be properly powered when the supply maintains constant voltage against changing current demands. The opposite of this is a sloppy, quickly moving supply that sags and resonates. That seems to be the sort of thing that you equate with high quality reproduction.
--------------------------
Buy Chinese. Bury freedom.
Edits: 01/09/15
"The amplifier can only be properly powered when the supply maintains constant voltage against changing current demands. The opposite of this is a sloppy, quickly moving supply that sags and resonates."
Absolutely! The one thing we DON'T need is a B+ that squirms about.
My PT is not 65 Ohms to center tap as your simulation shows, there is no hump and not that the hump is even relevant.
I have 30H 220 ohm chokes on the bench right now, shouldn't be hard to find.
As it has been explained to you, the part of the Sim that is relevant is the is the beginning.
The rest of the Sim shows a steady state draw on the PSU, which isn't whats happening with an audio signal.
Try again.
GG,
Ohh my goodness, I now do see you are using 165 Ohms, and I "saw" the "one" in one sixty five as a "parenthesis" on my small monitor. My visual error. Are you suggesting 1/2 your HV secondary is 165 Ohms?
As for this " As it has been explained to you, the part of the Sim that is relevant is the is the beginning."
That is incorrect, you have the "beginning" bass ackwards, the drop in the first 50 mS. is GOOD to be sharp and get "most" of the settling done smoothly and quickly as possible.
I will re-PSUD2 with 165 Ohms. BTW, That 165 Ohms PT DCR is a LOT worse than 65 Ohms. It WILL however, damp the waveform, and make EVERYTHING go in slow, out-of-time motion. Now, I'll go back to PSUD2 with 165 Ohms DCR, yikes !!
Jeff Medwin
"That is incorrect, you have the "beginning" bass ackwards, the drop in the first 50 mS. is GOOD to be sharp and get "most" of the settling done smoothly and quickly as possible."
I disagree with this, you want to hold as much voltage in the beginning as possible.
The sharper the drop, the more voltage you have lost, as in your case.
A shallow drop on the other hand delivers a stronger leading edge to transients because it holds a higher voltage for a longer period of time.
Remember transients are gone as quickly (aside from residual decay) as they arrive, so I fail to see why you are so concerned with what is happening a half second later on a simulated resistor.
A step drop off means you have less power when you most need it.
A fast charge time is nice, but doesn't mean much if you don't have the immediate capacity when needed.
It's the area under the curve that is important.
Have a look below, you tell me which one sounds better.
Your approach:
Or mine:
gargoyle -what happens if you extend out beyond 4 seconds and allow the 150uf cap to discharge?
I *think* the idea is that the low storage supply would recover from overload faster. it's hard to see that in a small window of time.
yes it drops 1 or 2 volts faster than the supply with more henries and capacitance, but also recovers faster. the other thing is the belief that while the 150uf capacitor is sinking current to recharge the output tube is starved for current, whereas the smaller caps won't have so much of an effect.
I think another issue here as the simulations get more elaborate, it's hard to nail all the variables for an accurate simulation.
the document that jeff speaks of, where he put all of the JLH posts he found that reinforces his power supply ideas, is VERY good IMO.
at some point if you guys are really curious, spend the $50-$100 on some chokes, mock up the supply and get your test gear out.
I'd recomend the flywheel stuff over this more recent power supply...
the stuff to use that is proven is the triad c56u and 2 stages of triad c40x.. hammond 159za is a bit better than triad c40x but costs about double.
three stages, .035h, 15uf, .3h, 50uf, .3h, 50uf
very low dcr toroid transformers where used by JLH. he got his "dynamic impedance" down to around 78 ohms with cree diodes in a bridge. not 650 ohms, he'd call that junk. 200-250 ohms with tube rectifiers.
he also did test with a frequency generator and found very little change in b+ across the frequency range, and was satisified with supply impedance at that point.
Jeff's got the document, and I think its worth a read. goes WAY beyond Jeff's interpretation..
I wonder how all the odd TW/DLM power supply configurations affect IMD measured at the output of the amp?
ARTA has a few preset 2 sine tests and a user defined option.
Have you played around with any of those measurements?
"gargoyle -what happens if you extend out beyond 4 seconds and allow the 150uf cap to discharge?"
-For curiosity sake I will see what I can do.
"I *think* the idea is that the low storage supply would recover from overload faster. it's hard to see that in a small window of time."
-I still fail to see the value in a fast charge time if it sounds bad during use.
Don't forget that the B+ voltage will be wildly swinging up and down simultaneously for the higher frequencies. This undulating reference will inter-modulate with the higher frequencies garbling the sound.
"yes it drops 1 or 2 volts faster than the supply with more henries and capacitance, but also recovers faster. the other thing is the belief that while the 150uf capacitor is sinking current to recharge the output tube is starved for current, whereas the smaller caps won't have so much of an effect."
-As I mentioned on the other page, it is the area under the curve that counts.
Capacitance fills in the dip. "starved for current" is kind of a red hearing, the tube will prefer to pull the current from the last cap anyways. It does not compete with the capacitor, it attempts to empty it.
So long as the voltage is higher then then ground, voltage will flow from the capacitor.
Farkwin supplies aren't "fast", they are saggy.
Those step vertical waveforms are there because it dropped like a stone and there is nothing to charge.
"Fast" would imply you had a transformer rectifier combination that could deliver much more current then typical devices used.
Since the Farkwin uses a tube, and usually the cheapest iron this isn't the case.
"I think another issue here as the simulations get more elaborate, it's hard to nail all the variables for an accurate simulation."
-It's OK if the results are somewhat arbitrary. We are comparing topologies.
"the document that jeff speaks of, where he put all of the JLH posts he found that reinforces his power supply ideas, is VERY good IMO."
-I've read too many ridiculous things claimed by Drlomu over the years to go chasing unicorns.
"at some point if you guys are really curious, spend the $50-$100 on some chokes, mock up the supply and get your test gear out."
-I've read too many ridiculous things claimed by Drlomu over the years to go purchasing unicorns.
Don't forget the humble PC is a pretty good piece of test equipment as demonstrated by this thread.
"I'd recomend the flywheel stuff over this more recent power supply... "
-Sounds catchy!
"three stages, .035h, 15uf, .3h, 50uf, .3h, 50uf
very low dcr toroid transformers where used by JLH. he got his "dynamic impedance" down to around 78 ohms with cree diodes in a bridge. not 650 ohms, he'd call that junk. 200-250 ohms with tube rectifiers."
-Seems like a lot of fuss to try and compensate for low capacitance, not to mention noisy.
"he also did test with a frequency generator and found very little change in b+ across the frequency range, and was satisified with supply impedance at that point."
-Good for him.
I don't think Drlomu should drag someone into the mud based on some half assed implementation of something he read somebody else do.
I'm no EE, but I don't think the term "Dynamic Impedance" is being used properly, or is even relevant beyond the application cited.
A quick look at yesterdays sims show the voltage wavers steady after the rectifier. Dynamics seem to be pulled from the caps which seem to be much lower then the "78 ohms" you implied to be a good number.
Everything has trade-offs. My preference for power supplies are slow to charge at turn on.
This is a good compromise because the tubes are warming up anyways.
Slower to charge is also slower to drain which equals a longer amount of time closer to the optimal B+, much stiffer too.
To justify your VERY misguided intellect .
Edits: 01/12/15 01/12/15
Jeff, it is hard at times to figure out quite what it is that you believe. You surely must agree that the output voltage of your low C, low L power supply fluctuates more with the audio signal than a high C, high L supply will do? Is this, at least, an area of common ground between what you are saying and what many of the rest of us are saying?
If you want to argue that despite fluctuating more, or indeed perhaps because it fluctuates more, it "keeps better time with the music," then that could be a separate discussion. (I don't know what "keeps better time with the music" is supposed to mean, but still, that is a discussion one could have.) But the fact that small C and L will cause a greater level of fluctuation at the audio frequencies can surely not be in dispute?
Chris
The stiffer a power supply is--- that is, the less it current-starves the amp stage it is powering, the better it will follow the "groove" of the music. It is IMPOSSIBLE, therefore, to follow the music's groove with a power supply that fluctuates (sags) enough to momentarily current-starve any amp stage that it is powering.Current starvation can come from several sources:
(1) Capacitors that store too much energy and release it too slowly. The cap's re-charge cycle takes current away from what is to be powered, and the release cycle of a large cap simply doesn't even get started (release isn't fast enough) until the musical transient has come and gone.
(2) Current starvation can come from a power transformer that's operated too close to its rating. Run about 30% of the rating, or less.
(3) Current starvation can come from too much stored inductance in chokes-- as in capacitors-- energy charge-up hogs current away from the device to be powered and energy release is delayed for too long.
(4) Current starvation can come from unnecessarily high resistances in the wrong places. These can be high-DCR chokes (anything over 20 ohms in most cases in tube amp power supplies).
(5) Current starvation can come from attaching devices onto the plate or cathode of a tube, in the attempt to correct for a sagging power supply. Examples of this include CCS devices, SRPP, etc., and several more.
(6) If a current-starving (under powered, or slow due to large capacitors, high value chokes, etc.) power supply is being operated, use of these (above) correction devices will usually improve bass performance at the expense of causing some current starvation of High notes that are extended and extremely dynamic, such as repeated cymbal clashes-- something very few amps ever get right-- so no one should feel bad if he's never heard an amplifier do it right. The use of these add-on devices nearly always also cause slight-- or sometimes worse-- degradation of signal purity, similar to a Tetrode or a Pentode VS a true triode-- that is-- slight signal homogenization and slight (or worse) image smearing.
The above effects (defects, actually) are euphonic and pleasing to many listeners because by homogenizing some signal information together, they more fully "flesh-out" the music-- it sounds solider and fuller than it actually is. This problem largely disappears with the use of medium or lower-efficiency speakers (usually under 96 db/watt) simply because those speakers can't reproduce it, but becomes apparent as an unwanted musical distortion when powering High-EFF, speakers that have large radiating surface areas, and have cables/wiring that is clean, efficient, and wideband.
The above are observed results and are not a set of personal opinions.
---Dennis---
Edits: 01/13/15 01/13/15
" The above are observed results and are not a set of personal opinions.
---Dennis---"
Since you observed these results instead of measuring them, they are all personal opinions. Further more, claiming that improved bass and more fleshed out sound is the result of a defect is incorrect. If an amp reproduces an acoustic bass in a natural manner, it can't be said to have exaggerated bass. If the mids and highs have a full sound that the original music has, it is correct.
Using examples like these as an excuse for an amp having a thin sound or lacking in bass is not being true to the source material. You may prefer that type of coloration but others may prefer a more natural sound.
Why bother to answer this. Anyone care to try his hand at it?
---Dennis---
" Current starvation can come from several sources:(1) Capacitors that store too much energy and release it too slowly. The cap's re-charge cycle takes current away from what is to be powered, and the release cycle of a large cap simply doesn't even get started (release isn't fast enough) until the musical transient has come and gone."
No. This is not correct. The larger the capacitor, the more steady the voltage across it will be under a given fluctuating load current. This is just a basic fact of how capacitors work.
Chris
Edits: 01/14/15
Sorry, but you are ignoring the re-charge cycle particularly. When a cap is recharging it is HOGGING CURRENT AWAY FROM the load.
When it is too large for the application, it's discharge/energy release cycle is VERY SLOW----wwwwwwwwwwwwww! YOU CAN HEAR THAT! It is NOT Thrilling!!!!!
This is so blatantly obvious to those builders that have experimented with capacitor and choke values in designing amps, that it's become ridiculous and repetitive to keep on repeating it.
Those old theories taught in those old tube books are not high fidelity-to-signal.
They NEVER WERE. I've used just about every tube amp out there for some sort of commercial app-- not all of them, but most of the best ones.
They DO NOT reproduce music RIGHT! NONE of those tube amps ever did...
Nothing built with those old, obsolete theories does today.
This is SO EASY to test! EASY! Get a Spectral DMA200S amp, or any 200 watt or greater Boulder, or a large, newest Bryston., etc., and see how it handles fast transient attacks in music.
Then, get out your tube amps-- I don't care-- get the best you know of, and watch them fall all over themselves trying to get out of the way of a GOOD amp.
Well, it just so happens that I designed a TUBE amp that actually works. For a while, it was the ONLY one. Now, it's starting to get some decent competition, and that is good-- VERY good. It means that I can give this up, retire, and have a good time.. That's right, we actually reproduce music, something tube amps normally fall down at.
Better look out! That old crap might include your amps if you followed all those old cap and choke theories for old push-pulls and tried to stuff them into a S.E. amp. Yes, it will sound slow-- just like all those other tube amps of yesteryear-- perhaps with today's VASTLY better resistors and caps, it will be a bit better, but will still fall short as far as reproducing today's best recorded music is concerned.
Someday, what's going on here will be common knowledge. Today, you can still get away with throwing darts at people like me who know what's up-- but all you're getting away with is Old Theory-- not anything scientific or provable by actual measured and listened-to musical performance.
Oh yes, it's s good idea to stop obsessing about what those old-fogies were measuring to sell cheaply designed amps with, and start measuring what actually matters-- musical reproduction SPEED, RESOLUTION, and DYNAMICS-- both Micro and Macro.
We're Light Years beyond that old stuff now. Your still popular Theory Train is slowing down and it's railroad tracks are starting to widen and loosen.. Reality could lie around the next Bend....
---Dennis---
You like the sound of a sloppy pose supply. There is mother wrong with that. At least you know what it is, so you can duplicate it in your production.
> Those old theories taught in those old tube books are not high
> fidelity-to-signal.
> They DO NOT reproduce music RIGHT! NONE of those tube amps ever did...
> Nothing built with those old, obsolete theories does today.
Oh, yes, all names and brands from Frank McIntosh to Bob Cordell, from Marantz to Sony must be wiped out from the history of audio in favor of Great Fraker Theory.
Peacockery and laughing-stock.
"We're Light Years beyond that old stuff now. Your still popular Theory Train is slowing down and it's railroad tracks are starting to widen and loosen.. Reality could lie around the next Bend...."
OK, so you like the sound that comes from a floppy power supply that wobbles with the load. Nothing wrong with that, if it's what you enjoy.
Chris
You are only talking about a SMALL portion of the power supply in my amps, and its an incomplete picture, not representative of whats really going on.
My power supply, unlike many / most others, is "all over the amp " and we have said this, going back many years now.
For example, I have a simple one-part shunt regulator, located ONLY 1/8 th of an inch from the point of use, the front end's plate resistor. It shunts over 15 times the audio current, than what the audio stage uses. Ninty percent of what we hear in a two stage 2A3 amp is the front end. WHAT do you think that Shunt Regulator does, to lock-in high performance??
And I am not even discussing the additional use of a Dennis Fraker " Final Filter", L/C, located 1/8th of an inch from the input stage's plate resistor. Who has those TWO items in their amp designs, no one except Dennis and I.....and anyone we have mentored.
So Chris, we can't intelligently discuss my preferred amp designs, unless we look at the whole circuit. My power supply, copying Dennis', is ALL OVER the AMP !! What you are doing is trying to play poker with one third of the deck !!
Enough of this .
Jeff Medwin
"So Chris, we can't intelligently discuss my preferred amp designs, unless we look at the whole circuit."
Maybe you could post a schematic, and then we can discuss it?
Chris
15 times .5ma is 7.5ma. So there is 8ma. total of which only .5ma. is moving.In the old days they call this form of voltage regulation a "clamped down" PS.
It does a good job preventing voltage fluctuations caused by the .5ma that is moving but it does nothing to prevent the voltage from changing because the source voltage is changing do to the large current swing of the output stage and a poorly regulated supply that's feeding the whole thing.
In this case the voltage regulator would need to regulate against both sides, the changing load and the changing source, to be of any real benefit.
I hope I wrote that clearly enough to be understood.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 01/13/15
Not even going to read it lol.Judging by the picture you posted you don't grasp the difference between energy storage device and something that dissipates energy.
Its very simple Drlomu it goes like this, I will even use your attachment for an example.
Power from the voltage source will prioritize the cap because it is the lowest impedance. 36 ohm vs 650 ohm
The tube conversely loads the capacitor because 36 ohm is lower vs whatever the resistance is of the choke etc.
It's a relay.That is why it is called a reservoir capacitor.
Similar to a water reservoir that is designed to collect and store sporadic additions of water, that can later be distributed in a even controlled fashion. Through valves even.-The reservoir needs to hold enough water to get you by between rainfalls.
-You need more rain then the water you consume per year.The bigger the reservoir is, the less it is going to drop in level per dry cycle. This is of particular interest to us because this water line is used as our voltage reference.
If this waterline stays fairly constant, then you can control valves in a predicable, dare I say linear fashion, because the water pressure is nearly always the same.
Edits: 01/12/15 01/12/15
Very simple. A friend's son was granted an interview with Microsoft for a summer internship when asked to explain the concept of "reduction" to a child. His example was Russian Nesting Dolls. Perfectly simple to understand.
my only long winded point being that what Jeff is describing can be made to work and don't dismiss it because of his troll like ways of spreading the information.
there is an EE or two out there that have spent time making it work that like variations on the "low everything supply." people with a lot more education and understanding on electronics than myself. jeff's gorilla marketing can be frustrating, but he means well.
Why is it so hard for people to say they like the way a sloppy supply sounds and measurements don't matter.
GG,
This is really sort of funny !! You have it very wrong. Please send me your private email address, and I will email you, as attachments, all the SWENSON and HASQUIN posts, in chronological order, for you to read and digest some.
If you don't want me to contact you directly, no problem, find a Forum Member we both know, and It could be forwarded, them first, to you second.
I understand your line of thought, its just that it is not correct. The first 50 mS. are "free" according to SWENSON's measurements of music.
Your call on proceeding, I have the posts in Word Files. I am just being fair and open with you, certainly you recognize that.
Jeff Medwin
Edits: 01/10/15
I can't see GEO's last two posts.
Hi again Jeff,
> > > the drop in the first 50 mS. is GOOD to be sharp and get "most" of the settling done smoothly and quickly as possible. < < <
I dunno, isn't the speed of the drop that only part of what matters? I would have thought the *magnitude* of the drop against time would also be an important aspect (wanting to keep the voltage up as transients hit), as would the time to recover and (lack of) overshoot and ringing.
Non-technical conceptualisation:
When a big transient hits, the PS voltage should fall fast enough to support sufficiently quick recovery, but should not fall far so as to maintain voltage for the transient. Recovery should be smooth and fast enough that voltage has recovered and the PS stabilised before the next big transient hits.
The caveat for me is, does the 15% stepped current in PSUD2 reflect anything like real-world conditions and can we really extrapolate audible effects from it?
Of course, this ignores PS impedance and maximum accepted ripple etc.
Regards,
91
"When a big transient hits, the PS voltage should fall fast enough to support sufficiently quick recovery, but should not fall far so as to maintain voltage for the transient. Recovery should be smooth and fast enough that voltage has recovered and the PS stabilised before the next big transient hits."In a class A circuit why would a big transient cause the current draw to increase longer than 1/4 of one wave length before the current draw starts down?
And 1/2 wave length later wouldn't the current draw be decreased from idle and the worry would be increased voltage?
1/4 wave length of the slowest frequency of interest (20Hz) is 12.5ms
If the supply is slow to react to current increases (and current decreases) and the voltage holds relatively steady for the first 12.5ms, until the current starts to decrease heading back to the idle current point, isn't that what we want?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 01/16/15
I think you have it nicely said on a conceptual basis. But I am not a EE Guru PSUD2 expert.
The 15% step was something EEs Swenson and Hasquin used. I just copied their examples Raymond.
Jeff
GG,
We "match" more now, wondered why we differed !! Now I know, GIGO.
You still have this LONG fully-settle time - 400 mS. It stays the same.
Now however, this cheap-wimpy-soggy 165 Ohm Power Transformer has increased the Z of your supply, to where its worse than ALT.1. Your B+ filter drops 8.83 VDC peak to trough settled, and ALT.1 drops about 7.83 VDC peak to trough. On a similar change of 12.25 mA. ( the current step ) your B+ Filter NOW has a calculated Z of 720 Ohms, and ALT.1 has a Z of 631 Ohms.
Amps with 165 Ohm Power Transformers and 270 Ohms chokes sound like you are listening to the music on your 604s, with a big woolen blanket thrown over the coaxial driver. What I may need to do is bring one of my DC amps up to Chicago area, next time I visit there, and let you hear it for yourself.
Jeff Medwin
.
Have Fun and Enjoy the Music
"Still Working the Problem"
HINT: The current draw in a Class A amplifier is NEVER, EVER "constant" when the amplifier is reproducing musical dynamics.The THEORY that current draw is constant in ANY kind of amplifier is an old-wive's tale that needs to be put to rest.
Constant current draw DOES NOT EXIST in any circuit that is processing musical dynamics.
---Dennis---
Edits: 01/09/15
"Constant current draw DOES NOT EXIST in any circuit that is processing musical dynamics."
Assuming perfectly linear tubes, a push pull Class A output stage will draw constant current on a moment by moment basis.
Assuming reasonably linear tubes, the current draw of a push pull Class A output stage will not change by very much at all.
So while you statement is in fact true, it's misleading.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Assuming reasonably linear tubes, the current draw of a push pull Class A output stage will not change by very much at all.
This comment always bothered me. Chris covered it when he included the "over a period of time" but when we look at the actual current through the output the picture is quite different.
this is a few watts out of a 2A3 with a DC bias of about 58.5ma
dave
That looks like the current draw for one tube playing music.Add to that another tube with an input signal out of phase but otherwise identical and the total current draw change on the power supply should go down.
If both tubes were 100% linear and if the drive signals were 100% out of phase with each other but otherwise 100% identical then the current draw change would be zero.
There's no such thing as a 100% linear tube so Dennis' statement "Constant current draw DOES NOT EXIST in any circuit that is processing musical dynamics." is true but in the context that he made it, IMO, it is misleading.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 01/09/15
What is all this PP talk in a SE forum.... what next those ideal transistors?
dave
And furthermore Dave, isn't this discussion about a B+ filter to a SINGLE ENDED A1 amp, and not a PUSH PULL A1 output stage ?? I am certainly no electronics whiz, but might that be significant ? If so, why would one anyone inject P-P examples in a SE amp design discussion ?
Jeff Medwin
" "Assuming reasonably linear tubes, the current draw of a push pull Class A output stage will not change by very much at all."
This comment always bothered me. Chris covered it when he included the "over a period of time" but when we look at the actual current through the output the picture is quite different."
That statement you quoted was for a push-pull class A amplifier. I think the plot that you displayed is for single-ended, right? The current as a function of time looks just right for single ended; fluctuating roughly equally up and down around the quiescent level.
For push-pull class A, the total current should be very nearly constant, since it should be like the sum of the plot you showed plus an inverted version of that plot.
Chris
"The current draw in a Class A amplifier is NEVER, EVER "constant" when the amplifier is reproducing musical dynamics..."
Indeed, of course it isn't. As has been said many times, for an SE amplifier the current draw fluctuates in step with the audio signal. But the average current draw, averaged over the timescales of the audio frequencies in the signal, is approximately constant. (The current oscillates up and down around the quiescent current. Assuming there is no gross distortion, the fluctuations above and below the quiescent level, on the positive and negative halves of the audio cycles, are essentially equal in magnitude, and hence the average current is essentially constant.)
Chris
Please don't try to hijack my thread. Just honestly answer the key, germane question I posed, IF you seek to participate . We should think in terms of instantaneous pulsed power delivery, rather than constant sinusoidal.Question : Which amplifier B+ filter do YOU think would sound best? Why ?
Edits: 01/09/15 01/09/15 01/09/15
"Don't hijack my thread. Just answer the question I posed, IF you seek to participate ."
First of all, it's not your thread. The OP was xcortes. And all I am doing is asking you to explain some points relating to the posts that you made on this thread.
As I said before, "what really matters as far as the amplifier is concerned is the output impedance of the power supply at audio frequencies, and how that impedance varies as a function of the audio frequency." Without knowing that information for the two power supplies, I don't know how to answer your question about which would sound better.
Chris
The amp is held constant, ALL we are discussing is the B+ Filter to the Finals, three or four parts choices, and what will perform best, on the basis of instantaneous pulsed power delivery.I am amazed that you don't know, but that is all you really need to say.
Lets see if others will answer and tell us why.
How "deep" does this SET AA Forum go ?? Who responds, and what they say as to "why" will be telling.
Paul Joppa has always been interested in us respectfully commenting on designs he has posted, as a way of possibly improving a circuit's audio performance.
Jeff Medwin
Edits: 01/09/15
"The amp is held constant, ALL we are discussing is the B+ Filter to the Finals, three or four parts choices, and what will perform best, on the basis of instantaneous pulsed power delivery."
I'm not sure what you are driving at when you say "The amp is held constant."
As I have tried to point out several times, the nature of the power demands of the amplifier, as it plays the music, governs the criteria that one needs to consider when designing the power supply. For a class AB amplifier, the net current draw by the amplifier can grow a lot larger during the course of a loud, and maybe and prolonged, crescendo in the music. By contrast, in a single-ended amplifier, which is class A, the net current draw, averaged over the timescale of the audio frequencies it is reproducing, is essentially constant and independent of whether the music is loud or quiet.
Thus, the design criteria for the power supply can be very different in the two cases. For a class AB amplifier, your criterion of looking at how the power supply responds to a stepped load would be quite relevant. For a class A amplifier like an SET, on the other hand, the power supply will not encounter any such prolonged increases in the loading. The important thing for a power supply for an SE amplifier is that it should allow the audio signal to pass through it with minimal let or hindrance. Thus, the important point is that the output impedance of the power supply in the audio frequency range should be small in comparison to to the impedances associated with the output stage of the amplifier itself, so that there is no significant frequency roll-off.
One can fairly easily make some estimates of the impedance at the output of the power supply, but one would probably want to use SPICE in order to do a better job.
Armed with such information, one would be in a better position to say which of the two power supplies you presented would be better. Neither of them looked very good to me.
Chris
and increasing the values of the caps in the CLC version
Dave, since you are doing those simulations in SPICE, I wonder if it is fairly straightforward to include the transformer and rectifier in the sim also? One could then look at the behaviour when the power supply is loaded with a constant quiescent current and a superimposed sinewave current (so I = I0 + I1 Sin(w t), simulating a typical load of an SE amplifier reproducing an audio frequency f=w/(2 \pi)). It would be interesting to see how the magnitude of the audio "ripple" on the power supply output voltage behaved for various frequencies f in the vicinity of the resonance in the low audio spectrum in the drlowmu supply.The reason I am wondering about this is that the approximation of putting that 100 ohm resistor to "represent" the power transformer and rectifier is a bit of a rough and ready one. The behaviour of the impedance curves changes significantly if one alters that 100 ohm value. I think the qualitative feature of the resonance is always going to be there, however.
I'm not familiar with using SPICE, and although I can handle LCR networks fairly easily (when I'm not making stupid typos!) using Mathematica, it's a rather tougher proposition to try to deal with rectifiers without the aid of a dedicated program like SPICE.
So it might be interesting to see what SPICE would reveal, along the lines I was suggesting. Oscillatory loads are much more relevant than stepped loads for probing the relevant features of a power supply for an SE amplifier.
Chris
Edits: 01/09/15
Here is my attempt to simulate the drlowmu power supply, using the same set-up as I think you have in your SPICE simulation. I took the two chokes, 1 H and 1.1 H, to have DC resistance 9.5 ohms each, and I copied your 100 ohms as a model for the power transformer plus rectifier. (I'm really not sure how good a model that is, which is why I've been a bit hesitant about making strong claims about how the PS would perform.) Anyway, with 30 uF and 50 uF for the two capacitors, my calculation in Mathematica then gives this result for the impedance as a function of frequency.
Chris
(This is a corrected plot. Had a typo before.)
Edits: 01/09/15
Guys,
ALL I did was change the Finals filter in Paul Joppa's schematic. This is hardly a supply I would use, as an overall supply. Nor is it an amp I would build.
I was only changing out the high HY high DCR choke, and asking everyone, "which will sound better and why? " !!!
Have fun, hardly anyone has answered my question above.
Jeff Medwin
I'm going to watch the results of these sims be hashed out amongst the group here... and enjoy Dave and Chris' bringing simulations to the table...
I'd just like to point out that it'd be next to impossible to speck a choke at 1.1 henry, vs 1 henry... that has to be an error on jeff's part in the sim.. I can't imagine that .1 henry difference making a significant difference, just know that the chokes will measure differently under different current levels and that he couldn't speck two difference chokes, one at 1h and one at 1.1h.
Jeff we've had debate over different chokes meausuring differenctly...using a LCR or a real bridge, etc.
that is the main probelm with tuning in PSUD2 to these extremes...its difficult to get an accurate indutance on the chokes, it is all guess work.
on another note, I'd still stick with the two 1h chokes and believe Dave's simulation shows lower impedance across the board, albeit with the resonance around 50-60hz... which might help dynamics :)
Rage,
I wonder,..... what if one 1 HY choke is wound with 10% more turns than the other, and maintains a similar gap?
Jeff
Out of Depth.
from my limited experience inductance doesn't necessarily work that way.
is this extra .1 henry essential to your tuning?
"from my limited experience inductance doesn't necessarily work that way."
Indeed. The inductance goes like the square of the number of turns, so an increase of 10% in inductance would be achieved with roughly a 5% increase in the number of turns, not 10%.
Chris
Well, we do have the resonances in the same spots :-)
what concerns me in your sim is that it drops below the series resistance values at low frequencies and mine levels out at the series resistance numbers.
dave
"what concerns me in your sim is that it drops below the series resistance values at low frequencies and mine levels out at the series resistance numbers."Sorry! You are quite right. I had a stupid and inexcusable typo in my Mathematica routines. I've corrected the two plots now.
Chris
Edits: 01/09/15
Thanks Dave,Not too many of us amateurs are comfortable with Spice. Could you please tell us ( me ) what that means?
Also, could I ask you what "you" think might sounds better between the two filters, and why ?
Thanks, you are a benefit to us all.
Jeff Medwin
Edits: 01/09/15
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: