Nice Job Tom. The truth is that we need the transformer to provide much more than is normally perceived. Your data may very well have brought out the balance that is needed between VA and Farads. Could you direct me to the decoupling data
Well, It would take very little more time to sweep the capacitances as well as the VA raitings. But it would be a lot of data...
But remember that these last tables are not the same as the earlier ones. These last two tables are only calculating the max output power using the assumption that the peak output voltage can reach right up to the bottom of the ripple voltage waveform, which is definitely not the case, based on all of my other recent simulations, where we could see four or five volts or more between the signal peaks and the ripple troughs and still the signal got clobbered (because at certain points in time the caps didn't have the right conditions to enable them to release the CURRENT needed at that moment to form the signal voltage across the load resistor. And the latest tables also just used a blindly-selected fixed capacitance, just to get a rough estimate of the avg ripple voltage that could be expected.
I made these latest tables, with measurements from simulations for a range of transfrmer VA-rating values, just to try to make it more clear which combinations of VA and VRMS and output power might make sense to even compare, when looking at the capacitance requirements and thinking about a possible "rule of thumb".
In order to find a simple rule of thumb for minimum required capacitance, we would "probably" need to only compare combinations that have the same sort of "safety margin" built into them. Maybe, additionally, we will want to add a couple more variables, besides just current, and devise a simple equation or some set of steps that can be used to predict the minimum necessary capacitance.
Obviously, a transformer with a VRMS output raiting, or a VA rating, that is marhinally too small MIGHT be able to support a particular load impedance and output power combination, but only with the use of larger-than-typical reservoir capacitance.
It would be nice to be able to take everything into account and still have a simple formula to calculate the minimum viable capacitance. But, at a minimum, we should be able to discern some minimum or desirable safety margins for things like the relationships between VA rating, VRMS rating, load impedance, and desired max output power, and then develop rules of thumb that seem applicable within known boundaries of those.
Decoupling capacitance will show almost no additional effects at the frequencies we have been considering. That will basically be a whole separate study. However, if the decoupling capacitance is of significant total value, it could be subtracted from the reservoir capacitance requirement.
I would like to compare a power board with ordinary caps and then swap in some higher spec'd caps and play a game of: "Where did my bass go?"
Daniel,
What are you saying here? That you want to try capacitors with different ESR ratings or self inductance verses another capacitor or are you just talking about some fancy boutique capacitors with fancy wrappers? Somebody please explain what in the .... is supposed to be so different about the boutique capacitors in the actual manufacturing. I know that in the polypropylene capacitor market that this is pure hype, I have actually seen the exact same capacitors with very name brand wrappers that are identical in every was except for the covering in production. No lie, just a pretty wrapper and a multistrand axial lead out wire and nothing more....
What are you saying here? That you want to try capacitors with different ESR ratings or self inductance verses another capacitor or are you just talking about some fancy boutique capacitors with fancy wrappers? Somebody please explain what in the .... is supposed to be so different about the boutique capacitors in the actual manufacturing. I know that in the polypropylene capacitor market that this is pure hype, I have actually seen the exact same capacitors with very name brand wrappers that are identical in every was except for the covering in production. No lie, just a pretty wrapper and a multistrand axial lead out wire and nothing more....
Well Daniel, it really does depend on the type of amp - for a classA amp like an F5, try a typical Mundorf (or F&T - quite similar), Roedestein, Phillips, etc for a quite pronounced heavier bass, and then swap them out for something like a BHC or Rifa, and hear a complete change in the balance of the amp's sound.
A similar thing happens when you use a cheaper Torroidal transformer and block diodes, - the sound of the amp changes quite a lot with everything else the same so the final amp is the same ol' compromise all over.
Steven, I can see you're not a believer in the better sound of the more expensive parts?
A similar thing happens when you use a cheaper Torroidal transformer and block diodes, - the sound of the amp changes quite a lot with everything else the same so the final amp is the same ol' compromise all over.
Steven, I can see you're not a believer in the better sound of the more expensive parts?
JH,
No that is not what I am saying. Just that many of the claims are just untrue and though the price is higher the actual construction is not different, just the price and the packaging. I can't mention names but I have seen very high priced polypro caps that were utter frauds, very nice looking and all but nothing different in the least. If someone wants to just believe the hype that is their business, I want to see that there is some measurable differences in the device. That is what I am saying.
No that is not what I am saying. Just that many of the claims are just untrue and though the price is higher the actual construction is not different, just the price and the packaging. I can't mention names but I have seen very high priced polypro caps that were utter frauds, very nice looking and all but nothing different in the least. If someone wants to just believe the hype that is their business, I want to see that there is some measurable differences in the device. That is what I am saying.
That should be easy, especially if the effect is very easily noticed. You just need some details of the models of the capacitors in question. What is different about the specs? i.e. What does "higher-spec'd" mean, in this case? Also, how do their form factors and lead spacings compare?
Have you played around with soundcard measurment software, maybe with a measurement microphone and mic preamp and software like ARTA?
Even easier, and much more reliable when looking at only the amp's performance, would be a simple two-resistor voltage divider with its input tied into the amp's output and its safe-level output wired directly into the sound card's input, with no microphone and room modes to make things questionable.
There are several good free-download software packages that include spectrum analyzers and oscilloscopes for the audio-frequency range only, that look at the sound card inputs.
A little "actual measured data" could go a long way toward improving everyone's understanding of such things.
I believe that there is similar free software, that also uses the sound card's outputs, that could stimulate, measure, and display all of the characteristics of a capacitor, in the audio frequency range only, to provide the parameters needed for a simulation model.
And if frequency-response plots, etc, from simulations, don't float your boat, we could always use a WAV file for the input signal, in the simulation, and then listen to the resulting output WAV file. It might be a quicker way to test aspects of a design that you want to try, without having to build them, and without any danger of blowing up anything. And sometimes it might be nice to have everything else be ideal and unchanging, except the part we want to tweak and evaluate the effects of, on the sound. It might at least enable you to more-easily and cheaply identify the magnitude and direction of tendencies in change-of-sound-quality, for various configurations you might want to experiment with.
Well, anyway, what are the specs of the capacitors in question? Or what is the main difference between the specs of the caps that lose the bass and those that don't?
gootee,
I will have to think about some of what you are proposing to Daniel in your last post. I have a Cleo test setup that I use for loudspeaker testing and I never thought of using it for testing an amplifiers output. Perhaps if I limited the output to 1 watt or 2.83 volts like I do when testing a speaker I could use that as a baseline for testing an amplifiers output? How that compares to a sound card I am not sure as I guess it also depends on the card and the bit rate and clock that the card is using. My sound card in my desktop is very old and I am not sure what its specifications would be. My laptop has a built in on board sound card with hdmi output but what besides the USB port for an input I can't imagine. That doesn't look to promising to me. What do you think about using the Cleo system, I usually use a measurement microphone and a dedicated mono preamp for testing.
I will have to think about some of what you are proposing to Daniel in your last post. I have a Cleo test setup that I use for loudspeaker testing and I never thought of using it for testing an amplifiers output. Perhaps if I limited the output to 1 watt or 2.83 volts like I do when testing a speaker I could use that as a baseline for testing an amplifiers output? How that compares to a sound card I am not sure as I guess it also depends on the card and the bit rate and clock that the card is using. My sound card in my desktop is very old and I am not sure what its specifications would be. My laptop has a built in on board sound card with hdmi output but what besides the USB port for an input I can't imagine. That doesn't look to promising to me. What do you think about using the Cleo system, I usually use a measurement microphone and a dedicated mono preamp for testing.
I tried a 36va, 36vct (1a, 18,0,18) transformer with my published DC coupled LM1875 design (I don't promote it because I'd rather not omit useful parts; however, that amp is quite direct, level and useful for test).
The non-inverting split rail amplifier was given 330uF caps at the pins.
Two power supplies were tried:
"All in one board" with bridge rectifier, 2200uF, 330uF, chip
Separate boards with bridge rectifier, 2200uF//2200uF, --cable--, 330uF, chip
The smaller capacitance example had deeper, bigger, louder bass.
But not clearer. The quality of the voice band was inferior and quite typical of all-in-one board sound. Credits to member "TheProf" for the repair--the all-in-one board amp was fixed by series schottky (instead of cable) between the 2200uF and 330uF; and then the blur sound/noise was then gone.
The report is:
Capacitance located much closer to the output devices appears to work for decreasing power supply reservoir size requirement. Apparently, capacitance works better when it is located more effectively. I do not know the extent, so the only meaningful thing I have to say about it is: THANK YOU!!!
P.S.
Other parts used and not previously mentioned:
Heatsink, cable, 10K input load, and the gain divider is 27k with 820R
P.P.S.
Perhaps the experiment would have been more revealing had I chosen a TDA7294, an STK or a discrete amplifier for the opportunity of addressing the predrive versus the outputs separately.
For example, without wasting time on PCB development, it is possible to pin-bend the TDA7294 with the v- pins close to the chip face, the v+ pins bent half that far, and lay BY type or MR type diodes horizontally across to easily form the rails and "isloate" the predrive. On the right side is 220u//3.3n capacitors for predrive (and small signal ground), on the left side is 2200u capacitors for output devices (and power cable). A solid wire across the center (from center of the 2200u to center of the 220u) forms the 0v rail. The mute is energized via 10k. The standby is energized by resistor with cap delay per the datasheet. Small signal groundlift resistor is ~2 ohms. The input load is range 10k to 20k. The gain divider is ~60k with 2.7k. It is miller comp amp with comp options sealed inside, so don't increase the gain much--use preamp if necessary. The bootstrap cap is range 47u to 100u. The NFB cap is 220u and low voltage polar may be used. Believe it or not, this will all fit. If the amp doesn't idle dramatically cool, add one 2u polyester from V+ to V- at power cable attachment point for noise blocker and next add picofareds cap from in+ to in- for stability helper. We will need flux. Assembly time ~20 minutes. The output section has no gain of its own. That very real amplifier can demonstrate the idea of big capacitance directly at the output devices. What do you think of it?
Last edited:
I would love to use them. Can you name them? I need a bit of a shove in the right direction.
Here's an on topic, yet craptastic, mistake that I did to a power board.
All of the caps were 2200u but there were different models (oh no!!). Assuming that the first cap at the bridge rectifier really takes a beating, I put a high spec cap there, since it had better datasheet ratings, and then that was directly followed by ordinary caps to complete the rest of the capacitance. Yep, you guessed it--shouty amp with no bass. I replaced the high spec cap with Cornell Dublier's Mallory SEK, both durable and fairly awful for signal (you'd never want it for an output cap, but it worked great as a power cap). The results were beautiful and the bass reappeared. Unfortunately for me, replacing the deluxe (low esr cap) with something ordinary (mallory sek), wasn't my first guess. This took a while.
Question:
Could I have fixed that situation by putting a loss (cable or resistor or diode or longer trace) between different models of caps?
All of the caps were 2200u but there were different models (oh no!!). Assuming that the first cap at the bridge rectifier really takes a beating, I put a high spec cap there, since it had better datasheet ratings, and then that was directly followed by ordinary caps to complete the rest of the capacitance. Yep, you guessed it--shouty amp with no bass. I replaced the high spec cap with Cornell Dublier's Mallory SEK, both durable and fairly awful for signal (you'd never want it for an output cap, but it worked great as a power cap). The results were beautiful and the bass reappeared. Unfortunately for me, replacing the deluxe (low esr cap) with something ordinary (mallory sek), wasn't my first guess. This took a while.
Question:
Could I have fixed that situation by putting a loss (cable or resistor or diode or longer trace) between different models of caps?
Daniel,
Could you do me a favor. At this point I do not care about the name on the cap. List the actual specification differences that are making the changes you are talking about. Otherwise what it appears to be is you are trying every perturbation to try and use an expensive boutique cap somehow. Is that really the point here, or are we looking at specifications and damn the manufacturer of the components? If standard capacitors are giving you a superior result in your opinion why keep chasing this application. Numbers and actual differences will mean much more to me than calling out a brand name.
Could you do me a favor. At this point I do not care about the name on the cap. List the actual specification differences that are making the changes you are talking about. Otherwise what it appears to be is you are trying every perturbation to try and use an expensive boutique cap somehow. Is that really the point here, or are we looking at specifications and damn the manufacturer of the components? If standard capacitors are giving you a superior result in your opinion why keep chasing this application. Numbers and actual differences will mean much more to me than calling out a brand name.
That was a very early observation in the thread Tsiros, that the sound character may be affected in a greater or lesser way. Re-read some of the comments around post #45 and your own post #50. I believe that power supply hash is entering via the overall NFB loop and because it is not present in the input signal, thus the difference which is power supply related could actually be injected into the signal that we hear.
I also believe this is Tom, Frank (I did not call you Fred this time) DF and Terry's observations with their simulations and would appear that both complex impedance and filtering effect of the power supply has some critical effect on the "apparent sound quality" which is of course distortion of some kind or another - probably related to inter-modulation distortion.
Now one final comment, could this be why some speculate that an amplifier with no overall NFB sounds "better" (or one with a non differential input). Or am I just opening another can of worms here. I do believe that we will come to conclusions that will benefit us all.
Last edited:
Hi,
this is a topic of fire. anyone, should demonstrate that the sound changes even at low power. but those who are familiar with this problem, use a lot of high flux toroidal, with lot of high low ESR capacitors. this is what you find inside an amplifier from € 20,000, if you open it. so..problem exist!
the solution is very complicate, if solve with low hot.
Thanks for that post.
Multistrand cable? In an audio amp? I just said the "f" word. That cable has the copper fibers individually coated slightly for corrosion resist. Why do I want 16 bad copies of my signal distorted like a house of mirrors? Does that help tweeters? Apparently, they are selling tone differences, and they must mark up the price or else call it noise? The durability of the botique polypro caps looks acceptable for speaker crossovers, especially for shunt.
But, I never use expensive tone/noise caps in audio amplifiers.
The sole exception is inexpensive--Cerafine 4.7u or smaller paralleled with a tiny economy polyester cap can be an interesting choice for an input cap, but not in every application.
Except for availability of accurate datasheets, the manufacturer is unimportant. I quoted Cornell Dublier (Mallory) because they provide the most comprehensive data. Compare LP for the high efficiency cap verses SK or SEK for the standard ordinary cap. There is plentiful published data, but I don't really know the cause of the differences. For one thing, LP is good enough to use for an output cap, but SK and SEK are definitely not. The LP works great for any job at the amp board, the SK/SEK works great on the power board. . . or you can turn that example backwards to get some very poor performance.
Because, in comparison to the low ESR cap, I want to know if the greater loss of the Standard cap is providing helpful ballast for more easily charging many in parallel at the power board.
P.S.
If I can get a better/faster charge, that looks just like using a bigger/stronger transformer. Right?
But if I have inadvertently disabled/hindered some of the charging, that looks just like using a smaller/weaker transformer. Right?
At the power board, easier to charge wins, but they don't mention it in capacitor datasheets. Do they?
Last edited:
Assuming the NFB loop is correctly implemented, then this means that hash is already in the output so the NFB should be sending it back to the input for partial cancellation. This is what NFB is supposed to do.Nico Ras said:
It is easy to get the NFB wrong, and pick up ground noise, but that is not the fault of NFB but the fault of the designer. The important thing is to ensure that the input signal and NFB are both referenced to exactly the same ground point.
'Smaller caps mean better bass' might just mean 'smaller caps mean more distortion and more PSU IM, so bass sounds louder so it sounds better even though it is really worse'.
Rubbish can enter the amp from the output terminal. This is then injected into the -IN of the LTP.
Dr Cherry suggested a good way of attenuating this cable rubbish.
Tap off the NFB from between the R & the C of the R+C Zobel.
I have rarely seen it implemented. In fact I cannot recall a single instance on this Forum.
I have used it in a couple of chipamps. But, I could not detect any improvement in attenuation of interference. However you all know I use proper filtering on the input and rigorous use of twisted pairs.
Dr Cherry suggested a good way of attenuating this cable rubbish.
Tap off the NFB from between the R & the C of the R+C Zobel.
I have rarely seen it implemented. In fact I cannot recall a single instance on this Forum.
I have used it in a couple of chipamps. But, I could not detect any improvement in attenuation of interference. However you all know I use proper filtering on the input and rigorous use of twisted pairs.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.