...
Sansui big capacitor replacement - disappointment
- Thread starter mplsbob
- Start date
- Status
...
...
...
...
...
...
...Posted by: targeteye
On: Yesterday 08:56 PM
Steve, the transformer in the US would be delivering current 60 times per second not twice! (In Australia, it would be 50) Most amplifiers use a centre tapped secondary and a full wave bridge which means each rail (the + and the -) are receiving 120 'top ups' per second as the centre tap gives out of phase current delivery. The fact that the capacitors can be charged within a few cycles, means the transformer has an incredibly low source resistance and a quick analysis of the unsmoothed vs full load filtered psu rail waveform will also show you the caps only deliver a fraction of what the transformer does under load. I don't say they are not important, but if they are in series, how come your amplifier still works with them removed???? (yes a stack of hum) They are a filter, a reserve plain and simple.
I admit it has been many years since my basic electronics training. However, I don't think i've forgotten so much that I am remembering wrong that a Full wave bridge rectifier coupled with a center tap tranny will send a rectified signal of twice the tranny input signal such as this.
http://www.play-hookey.com/ac_theory/images/rectifier_dual_full_wave.gif
I also seem remember that the Tranny supplies the current to the load and the filter caps only when rectifier Vin voltage is = Vc+1.4. When Vin in < Vc+1.4 the rectifiers switch the tranny connectivity off and the output current it totally by the main filter caps. Exactly how many degrees of a total of 360 that the caps are supplying the output current is dependant on the RC constant of the load and the caps. When I put my amp back together I'll get some pics on the scope to show what i'm talking about.
As far as pulling the main filter caps out? To truly test your theory you'll also need to pull out the decoupling caps that are usually found on the amp boards. If the amp will still function after pulling all the filtering and electrolytic decoupling caps out and you send a 1khz singal into it. The ouput will be a 1kz signal 100 percent modulated by a 120 (or 100 in your case) signal. I'll let you risk it with your amp to test the theory.
That is all I have of the posts in question.
Enjoy,
Rich P
Enjoy,
Rich P
I will post a few thoughts, then I will go away and let you folks hash this out however you wish:
If you remove the main caps from your amp and power it up, you will let the magic smoke out, pretty much immediately. You can ignore me and try it, but, you have been warned.
The current that is going to the load is NOT coming out of the transformer directly. The only time that the transformer is sourcing current is when the transformer output voltage exceeds the charge on the caps + the voltage drop across the rectifier(s). The main supply caps are the reservoir supplying the current to the load (well over 95% of the time), hence, are in series with it. If you look at a plot of the current coming out of the transformer, once the main caps are charged, what you will see is large, short spikes, with very high ramp rates, that are limited by the transformer inductance and series resistance, as well as by the ESR of the caps. These spikes occur only at the peaks of the transformer half-cycles. At any other time the only place the outputs get current from is from the main caps.
If the caps were supplying a completely passive circuit (no active components: diodes, transistors, FETs, etc.), then the capacitive reactance of the caps would play major roles, as would their parasitics (ESR, inductance, shunt capacitance, etc.). Capacitors are probably one of the most complex and least ideal components in the electronics world. They are very hard to model accurately, from cap design to cap design. They are also quite unstable (operating characteristics moving all over the place with time, temperature, applied voltage, etc.). However, the caps are not supplying a passive circuit, but a complex (mathematically), active circuit.
Most commercial power amps (since this is what is under discussion here) of the day (not modern switchers, but analog/linears) utilize feedback from the output back to the input. They are essentially closed-loop servos, where the input voltage dictates the output signal (current or voltage, dependant upon the design). This results in the truth of EW's original statement: "The quality of the output signal is primarily dependant upon the amp topology, not upon the capacitors used" (sorry if I failed to get the essence). Whatever non-linearities, caused by various components of the servo (including and especially, coupling and filter caps), regardless of the source of those non-linearities, will be compensated for by the servo, up to the point where the servo's response hits a limit (frequency response, phase linearity, transient response, etc.). With a specific, instantantaneous, input voltage, a specific, instantaneous, current or voltage is to be present at the output. If the output does not match the input dictation (point in time by point in time), the servo pulls or pushes to make it so. Part of the magic in servo design is to make the phase linearity flat across the required frequency range of the servo, so that the delay from dictation to output is flat, for us, across the audio spectrum (plus a bit, for spec derating). If your servo can't handle a little bit of variation in capacitor response, you need to get out of the business, or you will destroy the reputation of your company. The better the "design" (not the parts), the better the response. The "magic" is in the ability to design a servo that does not need expensive parts to produce an expensive sounding result.
If you analyze any of the significant amps of the day, you will see that they did "not" use esoteric, magic, TOTL parts (anywhere). They did use parts that were optimized for audio, but they were not magic or expensive (relative to what might have been available then, or even today). As a matter of fact, modern power transistors are so much better (frequency response-wise) that you actually have to slow them down a bit to use them (in some cases) in the old power amps, or the servos will go unstable and oscillate. That is with "better" parts, in almost every respect, to the originals.
Keep this in consideration: The amps were designed for the parts they used, and the designs were balanced to maximize performance and profit, while minimizing cost and maufacturing complexity and difficulty. The balance is not an electronically delicate balance, but a financially delicate balance. Audio band servos are not hard to design so that they are not teetering on the edge of response degradation. To maintain the name, you must derate your designs a bit, so that as things age, the device still works up to spec, for a calculated life time of the product (anticipated end of life and predicted parameter drift of capacitors included in the calculation). What this means is that, you want the design to have enough performance headroom to be able to compensate for component aging and other time-related changes and other dynamics, related to operating conditions, without loss of peformance.
If someone replaces old filter caps with new, and the sound suffers, there is something wrong with the servo. What should have happened is, the performance headroom of the servo should have been been restored, or at least increased, back in the original factory direction, and the individual non-linear contributions of the aged parts should have retreated back under the cover of servo compensation. You simply have more troubleshooting work to do, that is all. Test me in this: Give the Sansui amp in question, with the new and old caps, to EW and turn him loose. If the amp does not sound sweeter with the new caps, when EW is done with it, I'll eat my hat. He knows power amp servo circuits probably better than anyone on this board, and most audio techs in the country for that matter.
Sansui (Pioneer, Harmon Kardon, Phase Linear, et. al) did not listen to main filter caps and pick the most appealing. I promise you that what they listened to were designs, and topologies. If you can hear the difference between main filter caps with a particular power amp design, the designer would have been sent back to the drawing board. I am a design engineer with over 25 years of experience in design for manufacturing. I know a little bit about what I am taking about.
Folks, what has been stated before is the fact. Argue with physics, or with each other, if you wish. However, it will not change the physics: "A properly designed audio power amp should be as close to a wire with gain as possible" If you want color in your music, that is a completely different topic, as to what color appeals best to the ear. In my opinion, color, if any, is the business of the audio transducers (speakers or headphones). You can't polish a horse apple. If the design stinks, you won't improve it much with esoteric caps. A good design can tolerate inexpensive caps, if not, it is "not" a good design.
I am frequently complemented in how much better I have made a piece of equipment sound. I am not in the business of mod'ing gear. I will do it, if requested, but I will resist, and sometimes refer the work to others. I use better quality parts in my recapping and restoring, but not in anticipation of improved sound, but of increased life for the gear. Essentially, I don't improve the sound over what was originally there. I simply "restore" it, to what it was originally. If it wasn't very good from the factory, it won't be very good when I'm done with it. Now, sometimes you can improve the performance of a piece of gear, but that requires modification of the "design." Basic cap substitution, except in certain critical points, is not a design mod. Simple mods, like better op amps in the input current mirrors, etc, can be done, and will improve performance over that from the factory. But, what I do normally is to put in, for example, better caps, with longer life, better ability to tolerate heat, etc. This doesn't improve the sonics all that much, but does make the piece of vintage gear more reliable, long term. "If it ain't broke, don't fix it" is a nice platitude, however, I say, "If you want to significantly increase the likelihood that your amp will be singing for many more years, you better recap it." Otherwise, "You pays your money, and you takes your chances." I just lost my bench amp (a beautiful, Pioneer SPEC-4) due to bad main power supply caps. If I had gone in and recapped the unit, in the first place, it would have cost me the $125 for the 4 big caps, + whatever the rest of the caps cost, and labor, rather than now having to replace almost every semiconductor in the box.
There are 3 ways to get wisdom:
1) straight from the source (mfg data sheets and manuals, etc.)
2) you can get it from someone who got it way number 3
3) the hard way.
The bottom line is this: Aluminum electrolytics depend upon a "moist" electrolyte paste to function. The aluminum electrolytic is then greatly dependent upon the seal of the cap can to retain the electrolyte. The electrolyte actually ages, and changes as well as dries. No seal is perfect. Some are better than others, even in a particular production run of a single cap model. Hence, a large percentage of vintage gear is still working. Heck, there is still an original Edison incandescent light bulb burning in a firehouse somewhere back east. But that means that the individual part has greatly exceeded its anticipated EOL (end of life). EOL is not a prediction of when all the parts will fail, but a prediction of when you can anticipate them to begin to drop out (some sooner, some later, some much later, hopefully most, much later). You know what? I bet that no-one would be a bit surprised if that bulb burned out this afternoon. It has had a real good run. Neither should you be surprised if your prized piece of vintage gear releases some of its magic smoke, right in the middle of showing it off to your sceptical young nephew. Whoops
Cest la vie
I now give up. You folks go ahead and go at it, if you wish.
Enjoy,
Rich P
If you remove the main caps from your amp and power it up, you will let the magic smoke out, pretty much immediately. You can ignore me and try it, but, you have been warned.
The current that is going to the load is NOT coming out of the transformer directly. The only time that the transformer is sourcing current is when the transformer output voltage exceeds the charge on the caps + the voltage drop across the rectifier(s). The main supply caps are the reservoir supplying the current to the load (well over 95% of the time), hence, are in series with it. If you look at a plot of the current coming out of the transformer, once the main caps are charged, what you will see is large, short spikes, with very high ramp rates, that are limited by the transformer inductance and series resistance, as well as by the ESR of the caps. These spikes occur only at the peaks of the transformer half-cycles. At any other time the only place the outputs get current from is from the main caps.
If the caps were supplying a completely passive circuit (no active components: diodes, transistors, FETs, etc.), then the capacitive reactance of the caps would play major roles, as would their parasitics (ESR, inductance, shunt capacitance, etc.). Capacitors are probably one of the most complex and least ideal components in the electronics world. They are very hard to model accurately, from cap design to cap design. They are also quite unstable (operating characteristics moving all over the place with time, temperature, applied voltage, etc.). However, the caps are not supplying a passive circuit, but a complex (mathematically), active circuit.
Most commercial power amps (since this is what is under discussion here) of the day (not modern switchers, but analog/linears) utilize feedback from the output back to the input. They are essentially closed-loop servos, where the input voltage dictates the output signal (current or voltage, dependant upon the design). This results in the truth of EW's original statement: "The quality of the output signal is primarily dependant upon the amp topology, not upon the capacitors used" (sorry if I failed to get the essence). Whatever non-linearities, caused by various components of the servo (including and especially, coupling and filter caps), regardless of the source of those non-linearities, will be compensated for by the servo, up to the point where the servo's response hits a limit (frequency response, phase linearity, transient response, etc.). With a specific, instantantaneous, input voltage, a specific, instantaneous, current or voltage is to be present at the output. If the output does not match the input dictation (point in time by point in time), the servo pulls or pushes to make it so. Part of the magic in servo design is to make the phase linearity flat across the required frequency range of the servo, so that the delay from dictation to output is flat, for us, across the audio spectrum (plus a bit, for spec derating). If your servo can't handle a little bit of variation in capacitor response, you need to get out of the business, or you will destroy the reputation of your company. The better the "design" (not the parts), the better the response. The "magic" is in the ability to design a servo that does not need expensive parts to produce an expensive sounding result.
If you analyze any of the significant amps of the day, you will see that they did "not" use esoteric, magic, TOTL parts (anywhere). They did use parts that were optimized for audio, but they were not magic or expensive (relative to what might have been available then, or even today). As a matter of fact, modern power transistors are so much better (frequency response-wise) that you actually have to slow them down a bit to use them (in some cases) in the old power amps, or the servos will go unstable and oscillate. That is with "better" parts, in almost every respect, to the originals.
Keep this in consideration: The amps were designed for the parts they used, and the designs were balanced to maximize performance and profit, while minimizing cost and maufacturing complexity and difficulty. The balance is not an electronically delicate balance, but a financially delicate balance. Audio band servos are not hard to design so that they are not teetering on the edge of response degradation. To maintain the name, you must derate your designs a bit, so that as things age, the device still works up to spec, for a calculated life time of the product (anticipated end of life and predicted parameter drift of capacitors included in the calculation). What this means is that, you want the design to have enough performance headroom to be able to compensate for component aging and other time-related changes and other dynamics, related to operating conditions, without loss of peformance.
If someone replaces old filter caps with new, and the sound suffers, there is something wrong with the servo. What should have happened is, the performance headroom of the servo should have been been restored, or at least increased, back in the original factory direction, and the individual non-linear contributions of the aged parts should have retreated back under the cover of servo compensation. You simply have more troubleshooting work to do, that is all. Test me in this: Give the Sansui amp in question, with the new and old caps, to EW and turn him loose. If the amp does not sound sweeter with the new caps, when EW is done with it, I'll eat my hat. He knows power amp servo circuits probably better than anyone on this board, and most audio techs in the country for that matter.
Sansui (Pioneer, Harmon Kardon, Phase Linear, et. al) did not listen to main filter caps and pick the most appealing. I promise you that what they listened to were designs, and topologies. If you can hear the difference between main filter caps with a particular power amp design, the designer would have been sent back to the drawing board. I am a design engineer with over 25 years of experience in design for manufacturing. I know a little bit about what I am taking about.
Folks, what has been stated before is the fact. Argue with physics, or with each other, if you wish. However, it will not change the physics: "A properly designed audio power amp should be as close to a wire with gain as possible" If you want color in your music, that is a completely different topic, as to what color appeals best to the ear. In my opinion, color, if any, is the business of the audio transducers (speakers or headphones). You can't polish a horse apple. If the design stinks, you won't improve it much with esoteric caps. A good design can tolerate inexpensive caps, if not, it is "not" a good design.
I am frequently complemented in how much better I have made a piece of equipment sound. I am not in the business of mod'ing gear. I will do it, if requested, but I will resist, and sometimes refer the work to others. I use better quality parts in my recapping and restoring, but not in anticipation of improved sound, but of increased life for the gear. Essentially, I don't improve the sound over what was originally there. I simply "restore" it, to what it was originally. If it wasn't very good from the factory, it won't be very good when I'm done with it. Now, sometimes you can improve the performance of a piece of gear, but that requires modification of the "design." Basic cap substitution, except in certain critical points, is not a design mod. Simple mods, like better op amps in the input current mirrors, etc, can be done, and will improve performance over that from the factory. But, what I do normally is to put in, for example, better caps, with longer life, better ability to tolerate heat, etc. This doesn't improve the sonics all that much, but does make the piece of vintage gear more reliable, long term. "If it ain't broke, don't fix it" is a nice platitude, however, I say, "If you want to significantly increase the likelihood that your amp will be singing for many more years, you better recap it." Otherwise, "You pays your money, and you takes your chances." I just lost my bench amp (a beautiful, Pioneer SPEC-4) due to bad main power supply caps. If I had gone in and recapped the unit, in the first place, it would have cost me the $125 for the 4 big caps, + whatever the rest of the caps cost, and labor, rather than now having to replace almost every semiconductor in the box.
There are 3 ways to get wisdom:
1) straight from the source (mfg data sheets and manuals, etc.)
2) you can get it from someone who got it way number 3
3) the hard way.
The bottom line is this: Aluminum electrolytics depend upon a "moist" electrolyte paste to function. The aluminum electrolytic is then greatly dependent upon the seal of the cap can to retain the electrolyte. The electrolyte actually ages, and changes as well as dries. No seal is perfect. Some are better than others, even in a particular production run of a single cap model. Hence, a large percentage of vintage gear is still working. Heck, there is still an original Edison incandescent light bulb burning in a firehouse somewhere back east. But that means that the individual part has greatly exceeded its anticipated EOL (end of life). EOL is not a prediction of when all the parts will fail, but a prediction of when you can anticipate them to begin to drop out (some sooner, some later, some much later, hopefully most, much later). You know what? I bet that no-one would be a bit surprised if that bulb burned out this afternoon. It has had a real good run. Neither should you be surprised if your prized piece of vintage gear releases some of its magic smoke, right in the middle of showing it off to your sceptical young nephew. Whoops
Cest la vieI now give up. You folks go ahead and go at it, if you wish.
Enjoy,
Rich P
TheAnalogman
Tell me about it.
Thank you Rich.
grateful
Super Member
In a fully complimentary amplifier the filter caps are directly connected to the collector of the output transistors.This current flows through the output transistor to the load. So in effect the caps are in series. ideally the voltage should remain constant, the current varies. From this perspective the caps are a power supply and should remain stiff in voltage. The sound does not pass through the filter caps, In fact the caps are the current source that the audio signal modulates. So ideally the cap should have no internal resistance and unlimited source current. Seems to me that a new non leaky cap would be better at this role. Anything less than this will distort the signal.
spaceman
Registered Lunatic
Thanx much, pustelniakr, we appreciate your time. :yes: Just to clarify MY position, I know next to nothing about electronics, so the suggestion of recapping a unit makes a lot of sense to me, since many people have pointed out that they fail with age, and can take other parts with them. I have no intention of trying to "improve" the sound, I'm not that smart, but if it's a side-benefit of recapping, I'll gladly take it. I'll admit, I've played with the idea of spending a little more for "better" caps, but since most of you have pointed out that that's a little foolish, I'll take your advice. I guess my original question to EW should have been, what modern caps are closest to the originals, in quality & function, and he's already answered that, kinda. I've learned a FEW things from this thread (I thought the transformer ran all the time, lol), & I'm sure others have too, so it's not really a waste of time. I can see that what I really need is more understanding about what each type of cap is & where they go in the unit, but that's probly been covered, & just hasn't sunk in for me yet. :scratch2:
Morden2004 said:
I'll rumage thorugh the thread and send you the important post numbers.
If you want.
On the topic of caps. I think, now that I think of it, this sansui 9090 I reacantly got had some caps replaced in it when it was serviced some years back. Before I got it. It has a sticker, and a story LOL
Anyways, there are some ornge caps in key locations that do not look stock.
If I'm right.
Nice theory, but capacitors are indeed not perfect and do produce a sound of their own. Which of course is part of the original design where it was used; i.e. too low ESR or a different internal construction yields another sound than was the intention by the original designer.
And as with all passive components, the added sound character should be one easily acceptable. So I am sorry, but no over-simplified theory can remove the important consideration of the added sound (and distortion) from passive components.
And second, most vintage circuits are single-ended and non-symmetrical.
Just another world.
spaceman said:
Using a bit better cap (105 degC, as opposed to 85 degC, a bit lower ESR, lower leakage) do contribute, but to the added headroom for the servo compensation, but once the component characteristics hide under the servo's compensation response, you will not "hear" any difference. But, you will increase the performance headroom. In other words, things can change inside more, before you begin to hear it again.
As a side note, cap can color is a cap mfg. choice, not an indication of cap quality or type, if you are looking from the outside in. Once you are familiar with what parts a mfg. tended to use, then colors begin to flag differences in part types or characteristics. For example, the orange 'lytics in a Pioneer are typically caps with designs that focused on very low DC leakage. These are frequently used in timing circuits and as coupling caps between amp stages, where DC leakage, even tiny amounts, are significant factors.
Enjoy,
Rich P
grateful
Super Member
The fact is the output circuit is a simple circuit to anyone with a knowledge of basic electronics. The designers of these units did not listen to "magic parts" they bought parts that fit a design specification and a cost to keep the unit competitive and profitable and affordable. If they could have used the better performing parts of today they would have.
I had something long and boring as a reply to Kjell's post, but grateful said it best. This stuff (95% of the gear we own...Japanese stuff built from '73 to '80) was built to a price point, and selection of individual components was done by association with with other manufacturers, contracts with other manufacturers, default selection due to parentage by a large corporation, or lobbying by various manufacturers ('Please use our parts, Mr. Pioneer-san! We rok!'
).
You're kidding yourself if you think some engineer pored over capacitor catalogs, comparing construction and specs for different electrolytic caps for each location on the amp in order to pick 'The Perfect Capacitor' for his design. The design engineers often had minimal input to the selection process...'cause that's how big corporations work. Been there, done that.
). You're kidding yourself if you think some engineer pored over capacitor catalogs, comparing construction and specs for different electrolytic caps for each location on the amp in order to pick 'The Perfect Capacitor' for his design. The design engineers often had minimal input to the selection process...'cause that's how big corporations work. Been there, done that.
I want to see your long boring post. After all. I spent a good 2-3 hours generating mine.EchoWars said:I had something long and boring as a reply to Kjell's post, but grateful said it best. This stuff (95% of the gear we own...Japanese stuff built from '73 to '80) was built to a price point, and selection of individual components was done by association with with other manufacturers, contracts with other manufacturers, default selection due to parentage by a large corporation, or lobbying by various manufacturers ('Please use our parts, Mr. Pioneer-san! We rok!'
You're kidding yourself if you think some engineer pored over capacitor catalogs, comparing construction and specs for different electrolytic caps for each location on the amp in order to pick 'The Perfect Capacitor' for his design. The design engineers often had minimal input to the selection process...'cause that's how big corporations work. Been there, done that.
(I write slow)Enjoy,
Rich P
- Status