Yes if you take take a normal amp or an opamp, if you draw the feedback loop out it is in parallel with the load.
Now the important bit... it's between those two points only (the feedback take off point and the feedback return/input ground) where the output is a copy of the input signal.
So there is a tapping off that feedback network to the amp.
The amp compares the signal at that tapping with the signal at the input, and the output of the amp is whatever voltage is required to make that difference between the two zero. That's basic opamp theory, and the way most discrete power amps work too.
So we can say that,
for an amplifier with feedback, "The amplifier output will do what is required to maintain the difference between the two inputs (inverting and non inverting) at zero.
Now consider where the load is connected too. It's connected to our reference point, our ground, and that's the same place the feedback return is referenced to as well.
So with respect to that point the amp output is a perfect copy of the input (hopefully). The load current flows via the output of the amp into that star.
What happens when it leaves the star doesn't matter, the amp output (the current or signal we want), is with respect to our star point and the amp output.
What leaves the star is a "dirty" current that flows back to the PSU, whether that PSU has good caps, poor caps, odd value caps, batteries in one half etc etc, it doesn't matter in the way you keep thinking. Those caps are not in series with the load.
Do you see
The output capacitor is only one of many other influence factors.
If there are two identical circuit amp topologies, one with symmetrical PS with DC coupled loudspeaker and Relais contact for DC protect for loudspeaker and an other with unsymmetrical PS and AC coupled loudspeaker (i. e. electrolytic cap in series without relais switch contact), both devices with top class devices in all respects and without design errors, the last delivers objective and subjective the best sonic results.
Additional advantages I have by the last AC coupling solution:
1) the NFB cap (for DC-gain = unity gain) can be also polarized type
2) no exact 0 V offset adjust (resp. perfect DC symmetric adjust) nessecary
3) no "0 volts" at both connectons by the NFB network
4) no DC-servo necessary
5) choise of the right capacity value (input cap, NFB cap and output cap - actually a crossover adjust for the smooth frequency response modelling) by the use of small, medium and large vented and closed loudspeaker cabinets (rule for lowest necessary acoustical measured f3: not lower than 20-22 Hz for the the largest enclosures).
6) No relais contact in series to the load (main advantage).
The disadvantages are only follow: twice of the DC voltage for the power supply cap and that one for the output and thus more expensive by the use of the same quality class.
By loads below 4 ohms (like Infinity Kappa or full range ribbon like Apogee) the cap value must be very very large arround 0,5-1F.
BTW - most people think, by symmetric PS and complete dc coupling (perhaps also the use of DC servo) there must be no caps in series to the signal. This isn't true, because the influence of the power supply caps is still very very large, independend of the circuit topology.
The reason therefore is the aera between the -3 db high pass frequency causes the whole capacity value of the power supply caps and the clean DC response condition (created through the connection of the mid-connection from the secundary windings).
By an other Thread here I have this descripted more exactly.
But I know, the most developers don't have my view of this because they think, through the DC coupling the influence of the ps caps are negible (the evidence therefore is simple; look at the quality standart to the caps in currently commercial home hifi amplifier products and compare it with vintage versions).
My opinion for the wrong thinking of the most engineers is the kind of schematic drawing. If one remove the GND/Earth symboles all and create normal connections, you will get complete other understanding and view.
Last edited:
One drawback of single supply + output cap is the shape of the current provided by the psu.
On the negative alternance, the output cap is disconnected from the psu and have to provide the output current. Then, the current demand from the PSU is half wave.
There is nevertheless one way to cure this: split the psu using two identical (double value) caps parallaled with some resistors (in order to keep the middle point at half the total voltage) You obtain a Vcc/2 point where to connect the load, getting rid of the output cap.
On the negative alternance, the output cap is disconnected from the psu and have to provide the output current. Then, the current demand from the PSU is half wave.
There is nevertheless one way to cure this: split the psu using two identical (double value) caps parallaled with some resistors (in order to keep the middle point at half the total voltage) You obtain a Vcc/2 point where to connect the load, getting rid of the output cap.
Last edited:
Mooly,
considering the nfb loop parallel connected to the load i am tempted to say the pSU caps are serial connected to (load //nfb loop). Though, the nfb loop has a high resistance value compared to the load and so, derivates a very small part of the current.
I am tempted to say that the assertion: "the caps are serial connected to the load" is true at more than 999/1000.
I am sorry to insist. That is because i do not really understand what you explain.
I need to think more about what you wrote an i need to read about opamp theory ...
considering the nfb loop parallel connected to the load i am tempted to say the pSU caps are serial connected to (load //nfb loop). Though, the nfb loop has a high resistance value compared to the load and so, derivates a very small part of the current.
I am tempted to say that the assertion: "the caps are serial connected to the load" is true at more than 999/1000.
I am sorry to insist. That is because i do not really understand what you explain.
I need to think more about what you wrote an i need to read about opamp theory ...
Last edited:
for an amplifier with feedback, "The amplifier output will do what is required to maintain the difference between the two inputs (inverting and non inverting) at zero.
Now consider where the load is connected too. It's connected to our reference point, our ground, and that's the same place the feedback return is referenced to as well.
So with respect to that point the amp output is a perfect copy of the input (hopefully). The load current flows via the output of the amp into that star.
What happens when it leaves the star doesn't matter, the amp output (the current or signal we want), is with respect to our star point and the amp output.
What leaves the star is a "dirty" current that flows back to the PSU, whether that PSU has good caps, poor caps, odd value caps, batteries in one half etc etc, it doesn't matter in the way you keep thinking. Those caps are not in series with the load.
Do you see
And for an amp without global feedback?
Both GNFB and relais protection can be avoided (put into the psu, a la JLH) in the design. In this case only the DC-servo issue remains.
As with everything is a design choice. Your analysis is very good, though
Both GNFB and relais protection can be avoided (put into the psu, a la JLH) in the design. In this case only the DC-servo issue remains.
As with everything is a design choice. Your analysis is very good, though
The Linn Klout topology avoid also the relais despite a DC connection of the speakers. But if happens a CE shorting at regulator transistor and power buffer transistor at the same time (e. g. through mains voltage burst while operating), the voice coil overloading is present nevertheless.
Please post a related schematic link to the JLH PSU you mentioned.
Here an "off topic" question: do you know this audio system about
http://www.diyaudio.com/forums/soli...grafo-stereofonico-1965-schematic-wanted.html
from your country? If Yes, please send your reply there.
I want a schematic from it.
Last edited:
And for an amp without global feedback?
I think we already covered that point... if it's any kind of amp other than a buffer it will use feedback.
Can you show me a design you think has no global feedback, other than a buffer ?
A buffer has a voltage gain of slightly less than 1, but has a high current gain to drive low impedance loads. An emitter follower uses for want of a better way of describing "100% feedback".
Try and follow why
Can you see that a DC coupled amp can supply a perfect squarewave all the way down to DC if required ?
You reduce the input frequency 10hz, 9 hz and so on all the way down to DC and the DC coupled amp produces a perfect output across the load.
The load current flows from the amp output to the star. Yes
If we put a 1000uf cap in series with the amplifier output and connect the load to the other end of the cap the amp now does not reproduce a perfect squarewave across the load as frequency goes down. Do you follow that ? That is because that cap is in series with the load.
If the PSU caps were similarly "in series" with the load a similar effect would be seen. The output would not go down to DC.
If you don't agree say so
What about the transformer as a path for DC. Well you can run the amp from the caps for a few moments by disconnecting the transformer... so it's not flowing there.
Hi Mooly, I think you may have missed his point! His was not an argumant about what is, or is not, feedback. Instead his meaning is an amplifier with an output capacitor but without global feedback (i.e. the capacitor is not enclosed and 'corrected' by an output-to-input feedback loop).
So for that type of amp (such as John Linsley-Hood’s Class-A power amp, or a power source / emitter follower) what influence does the output capacitor have? Can it be measured and correlated to sound quality? Apart from capacitance, voltage and current what parameters are important when selecting the component? Any other questions that are important?
What do you think?
While I do look for opinions on various issues on the subject, my own tests finally decide what I like and what type of components or circuitry I would use. Having looked at posts over several years on various forums on the Net it becomes very clear that there is NO perfect way to do things. There are too many permutations and combinations in audio. NO ONE as yet seems to know the magic combination that produces ALL the desired features of an ideal ( or highly desired) system in one package.
MOST of the time I find 'electrolytic' capacitors do detract from the sound if they are in the signal path. Dc bias on them sometimes alters the level of degradation. No dc bias being the worst situation. Film caps are better but vary depending on who makes them. Price and exotic types do not always guarantee a 'better' cap ! You need to try out what you can afford to use.
It's inexplicable but sometimes cap coupling sounds better than a direct connetion ( with hardly any dc off set from the previous stage).
The bottom line is to try out all the variations you can and pick on what sounds best 'to you '.
MOST of the time I find 'electrolytic' capacitors do detract from the sound if they are in the signal path. Dc bias on them sometimes alters the level of degradation. No dc bias being the worst situation. Film caps are better but vary depending on who makes them. Price and exotic types do not always guarantee a 'better' cap ! You need to try out what you can afford to use.
It's inexplicable but sometimes cap coupling sounds better than a direct connetion ( with hardly any dc off set from the previous stage).
The bottom line is to try out all the variations you can and pick on what sounds best 'to you '.
You know, we worked for many decades, in fact 40 years to REMOVE coupling caps from the inputs and output of audio stages. First we were able to do this with +/- power supplies, BUT we left a significant sized cap in the feedback loop that usually had 0V on it. For awhile we thought we were doing fine, BUT in actuality, 0V across an electrolytic cap is problematic, and actually can cause distortion. I published in an IEEE article in 1978, called 'Omitted Factors in audio design' by showing measurements with Tantalum caps (the best at the time) at low frequencies. Adding DC bias helped reduce the non-linear low frequency distortion, BUT not entirely. Later, starting about 1978, Dick Marsh realized that Dielectric Absorption in caps could be very big, especially in Tantalum and Aluminum caps. DC biasing only slightly helped this problem if at all. That's when SERVO correction was added to audio circuits in order to REMOVE all input, output and feedback caps. This is where we have been with high end solid state design for the last 30 years. No coupling cap is the best cap, in audio, except when you MUST roll off the lows, highs, or equalize the frequency response, and then you had better use a GOOD cap, polypropylene, polystyrene or Teflon, if you want no real added problems. Everything else is a compromise, either in non-linear distortion, linear distortion (DA), or reliability.
Actually a well designed subsonic filter can make the sound of a vinyl setup better. Also caps are used in servos and PSUs and there is a never ending debate lurking what Opamp sounds best when used as a servo and where to implement it. Fortunately there are very good low value caps availlabe. I know designers that even avoid caps in RIAA stages and hand trimm their circuits to avoid servos. I personally have no sound problems with properly designed filters and servos but other may disagree.
Hi Gordy,
The thread has got a bit wandery hasn't it... all my long replies started with the questions in post #9 and #10 and the PSU caps being thought of as being "in series" with the load or speaker.
So what influence does an output cap have ?
Well measurement wise it has shortcomings, it raises the output impedance slightly at mid and HF, raises it a lot at LF, it adds distortion at LF, mainly linear as it's reactance increases with decreasing frequency (LF roll off), and a bit of non linear distortion too.
So imo, for choosing a cap go for a reputable brand, check things such as ripple current rating, going for as large a value as possible.
Sound quality... well why not do a simple test and (if you are using as DC coupled amp) make a non polarised cap from two back to back caps of say at least 4700uf and see what you think by adding in series to the speakers.
The JLH is more than just a Class A amp... it has a very specific distortion spectrum that sounds (to many) very musical... that's it's real secret.
output cap vs transformer vs room
I hate to argue with John Curl, but then I have never had one of his high end amps in my living room. Currently my best all time sound is a DJoffe modified ST120 driving two SP2's (15" plus piezo horn) through 10 ga cable in my 14'x30'x12"h living room. The ST120 has output caps and a single supply. As tissuebasr pointed out, no output transistor bias setting is required, for the ST120 was designed to be assembled by novices with no instruments.
I tried out the same SP2's with a Peavey CS800S amp (purchased 98) which has a split supply and direct coupled outputs (through relays) in the seller's trailer studio, a 14x14x8 room with a diagonal placement of the speakers. The bass was superior to my previous 10' driver speakers on ZZ Top, but the treble on Serkin Beethoven Sonatas was just okay. Whether the room placement was wrong or the fact that he was driving through a 100' 24 gauge snake, the SP2's sounded better in the store- and now in my living room, with coupling caps and a reviled amp design.
One will note, the ST120 has no feedback from the output of the cap back to the signal parts. This ST120 had suffered a one channel fire. I purchased it in 1985, and replaced the big caps (output, power supply) the output transistors (NTE60 one side, NTE181 other side) and on the damaged side, the predriver transistors (NTE 49,SK3440) and the phase splitter (SK3275). The other side has the original 1970 RCA TO5 predriver transistors and splitter. Due to no schematic and fire damaged components, I put a resistor and regulator TO3 on the power board that limited the ST120 output to 2 amps or 16 watts. I was drawn to DIYaudio by DJoffe's ST120 improvement article, which allowed me to tune up the power supply to 6.5 amps per spec, and install his bias boards, which make the highs sound better at very low volume. I originally had found this ST120 to have an annoying low frying crackle sound after the 1985 repair, both with my PAS2 preamp on LP's, and with an Ampeg mixer as a keyboard amp. After I replaced all the other electrolytic caps in 2009, the amp is now dead quiet. It still has unmatched transistors on the sides, but scope traces show the power output to be equal on a quasi-sine wave from a phase shift oscillator.
I'm still wresting with the weird connectors of the CS800S to make a full test of cap amp vs direct coupled through relay amp. I can't even get a 16 ga zip cord to fit the phone plugs that I bought for the CS800s to SP2 cables, whereas the old phone plugs easily take 10 ga 4SO cord. And the input requires stereo RCA to XLR cables, and it is taking UPS 7 days to get RCA plugs from Phoenix to Indiana.
And the transformer? I was driven to upgrade my speakers because my Hammond H182 with a push pull 7591 tube output through transformers (35 watt?) puts out way better bass through it's 15" speaker than my amp could put out on the same piece from a pipe-organ LP (E Power Biggs Bach Favorites #1, Colombia) on the 10" woofers of Peavey T300 Hi-Freq Projector speakers.
I hate to argue with John Curl, but then I have never had one of his high end amps in my living room. Currently my best all time sound is a DJoffe modified ST120 driving two SP2's (15" plus piezo horn) through 10 ga cable in my 14'x30'x12"h living room. The ST120 has output caps and a single supply. As tissuebasr pointed out, no output transistor bias setting is required, for the ST120 was designed to be assembled by novices with no instruments.
I tried out the same SP2's with a Peavey CS800S amp (purchased 98) which has a split supply and direct coupled outputs (through relays) in the seller's trailer studio, a 14x14x8 room with a diagonal placement of the speakers. The bass was superior to my previous 10' driver speakers on ZZ Top, but the treble on Serkin Beethoven Sonatas was just okay. Whether the room placement was wrong or the fact that he was driving through a 100' 24 gauge snake, the SP2's sounded better in the store- and now in my living room, with coupling caps and a reviled amp design.
One will note, the ST120 has no feedback from the output of the cap back to the signal parts. This ST120 had suffered a one channel fire. I purchased it in 1985, and replaced the big caps (output, power supply) the output transistors (NTE60 one side, NTE181 other side) and on the damaged side, the predriver transistors (NTE 49,SK3440) and the phase splitter (SK3275). The other side has the original 1970 RCA TO5 predriver transistors and splitter. Due to no schematic and fire damaged components, I put a resistor and regulator TO3 on the power board that limited the ST120 output to 2 amps or 16 watts. I was drawn to DIYaudio by DJoffe's ST120 improvement article, which allowed me to tune up the power supply to 6.5 amps per spec, and install his bias boards, which make the highs sound better at very low volume. I originally had found this ST120 to have an annoying low frying crackle sound after the 1985 repair, both with my PAS2 preamp on LP's, and with an Ampeg mixer as a keyboard amp. After I replaced all the other electrolytic caps in 2009, the amp is now dead quiet. It still has unmatched transistors on the sides, but scope traces show the power output to be equal on a quasi-sine wave from a phase shift oscillator.
I'm still wresting with the weird connectors of the CS800S to make a full test of cap amp vs direct coupled through relay amp. I can't even get a 16 ga zip cord to fit the phone plugs that I bought for the CS800s to SP2 cables, whereas the old phone plugs easily take 10 ga 4SO cord. And the input requires stereo RCA to XLR cables, and it is taking UPS 7 days to get RCA plugs from Phoenix to Indiana.
And the transformer? I was driven to upgrade my speakers because my Hammond H182 with a push pull 7591 tube output through transformers (35 watt?) puts out way better bass through it's 15" speaker than my amp could put out on the same piece from a pipe-organ LP (E Power Biggs Bach Favorites #1, Colombia) on the 10" woofers of Peavey T300 Hi-Freq Projector speakers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.