I recently encountered a request on another forum as to the suitability of Quad 405 for driving Stax headphones as a Class A amplifier. There seems to be a number of about 10 Watts class A output floating around, based on the quiescent dissipation of the "class A" amp of around 5 watts (which is hopelessly wrong in this circuit). I have long been interested in the circuit, impressionably growing up in the 70's 80's. I dusted off the simulation that I did 10+ years ago, of the 306 circuit. I approximately calculated the class A output into 8 ohms of 100uW of the 306 which has a bias of one diode drop between the basses of the "dumpers".
Unfortunately the thread is in a bit of a backwater of that forum and hasn't been critiqued sufficiently for my satisfaction, of if I am in the right ball park, it is a bit of a controversial claim. So peer review please. I am quite happy for my slightly flippant comment about the bridge inductor filtering out the crossover glitch to be criticised, and I know there is more than just that going on with the bridge feedback, but that's what it looks like.
Cut and paste from "another Forum"
Morning, I too have been fascinated by the current dumping spiel of the 405, and a general lover of pre 90's Quad in general. So much so that I did try and build one when I was young (didn't work) and more recently as a grown up engineer decided to simulate the circuit (of the 306, because the audio is not going through an op-amp). Stable biasing of the class AB amplifier is a problem and the Quad seems to present a neat circumnavigation, hence my curiosity.
Most transistor amplifiers have a class A Vas stage, the 405 / 306 is no exception, except it is a bit beefier and is connected to the output by a resistor of around 47 ohms. The voltage generated across this resistor determines when the current dumpers are turned on. Due to the biasing arrangement that is approximately +- 300mV.
300mV / 47ohms is about 6.4mA.
6.4 mA x 8 Ohms = 51mV
Convert that into RMS about 36mV of class A at the 8 ohm speaker 100uW (if my maths is correct) before current dumping occurs. I don't know what the input impedance of the headphones is but presumably significantly more than 8 ohms and everything works out!!
I have not built or tested a current dumping amp in person, yet, but in my simulation of the 306, the area where the magic happens, in my opinion, is the bridge. The basic current dumping amplifier is a class B (sorry) output stage rescued by feedback, to plaster over the switching discontinuity. Yes the super fast class A amplifier huffs and puffs through its 47 ohm resistor, but the magic happens in the bridge.
The switching discontinuity appears to be filtered out by the inductor. I don't really understand what precisely is going on, but it does the trick. It is a remarkable piece of work IMO, especially considering they didn't have the tools available we have today.
Voltage across the inductor simulating 1KHz 8W.
All comments are for entertainment purposes only.
https://www.hifiwigwam.com/threads/rekindling-interest.124470/post-2529901
Thanks
Unfortunately the thread is in a bit of a backwater of that forum and hasn't been critiqued sufficiently for my satisfaction, of if I am in the right ball park, it is a bit of a controversial claim. So peer review please. I am quite happy for my slightly flippant comment about the bridge inductor filtering out the crossover glitch to be criticised, and I know there is more than just that going on with the bridge feedback, but that's what it looks like.
Cut and paste from "another Forum"
Morning, I too have been fascinated by the current dumping spiel of the 405, and a general lover of pre 90's Quad in general. So much so that I did try and build one when I was young (didn't work) and more recently as a grown up engineer decided to simulate the circuit (of the 306, because the audio is not going through an op-amp). Stable biasing of the class AB amplifier is a problem and the Quad seems to present a neat circumnavigation, hence my curiosity.
Most transistor amplifiers have a class A Vas stage, the 405 / 306 is no exception, except it is a bit beefier and is connected to the output by a resistor of around 47 ohms. The voltage generated across this resistor determines when the current dumpers are turned on. Due to the biasing arrangement that is approximately +- 300mV.
300mV / 47ohms is about 6.4mA.
6.4 mA x 8 Ohms = 51mV
Convert that into RMS about 36mV of class A at the 8 ohm speaker 100uW (if my maths is correct) before current dumping occurs. I don't know what the input impedance of the headphones is but presumably significantly more than 8 ohms and everything works out!!
I have not built or tested a current dumping amp in person, yet, but in my simulation of the 306, the area where the magic happens, in my opinion, is the bridge. The basic current dumping amplifier is a class B (sorry) output stage rescued by feedback, to plaster over the switching discontinuity. Yes the super fast class A amplifier huffs and puffs through its 47 ohm resistor, but the magic happens in the bridge.
The switching discontinuity appears to be filtered out by the inductor. I don't really understand what precisely is going on, but it does the trick. It is a remarkable piece of work IMO, especially considering they didn't have the tools available we have today.
Voltage across the inductor simulating 1KHz 8W.
All comments are for entertainment purposes only.
https://www.hifiwigwam.com/threads/rekindling-interest.124470/post-2529901
Thanks
@hbc - the Current Dumping concept is different. See attachment.
In a nutshell - the class A amp drives the bridge to the dumpers. When the dumpers switch on, there is a jump in the loop gain which would normally cause distortion.
But with the bridge balanced, the conduction of the dumpers actually increases the feedback factor which decreases the gain and thereby cancels the jump in loop gain. The result is that (ideally) the switching of the dumpers does not impact the performance.
See fig 2 & 3 in the attached.
Jan
In a nutshell - the class A amp drives the bridge to the dumpers. When the dumpers switch on, there is a jump in the loop gain which would normally cause distortion.
But with the bridge balanced, the conduction of the dumpers actually increases the feedback factor which decreases the gain and thereby cancels the jump in loop gain. The result is that (ideally) the switching of the dumpers does not impact the performance.
See fig 2 & 3 in the attached.
Jan
Last edited:
I thought I had attached an image to yesterdays post.
1 watt into 8 ohm with dumpers removed. The output current ability (so -/+ 500ma here) seems to decrease a little at higher voltage swings into higher impedance loads. Its progressive but falls to around -/+300ma at -/+30 volts output into 100 ohm (second image)
LT sim file attached but you will have to change it to use your own transistor models.
1 watt into 8 ohm with dumpers removed. The output current ability (so -/+ 500ma here) seems to decrease a little at higher voltage swings into higher impedance loads. Its progressive but falls to around -/+300ma at -/+30 volts output into 100 ohm (second image)
LT sim file attached but you will have to change it to use your own transistor models.
@Mooly Thanks, that's very kind. I put some generic sensible to me guesses in and off we go 
Indeed taking out the dumpers and it give one watt into 8 ohms, which is very pleasing, actually. I haven't full thought this through yet, but am conscious that the bottom "dumper" is a CFP, and with the NPNs removed the PNP device is still acting like a dumper. Removing that and it all falls to pieces, leaving the bootstrapped resistor to pull the 8 ohms....
This is my problem with the 1 watt output (with dumpers connected). That (current through R38) should be a sine wave if it is solely coming from the class A not driving the dumpers.
If I wind down the input to 1.5mV I get this:
Where I R38 (blue trace) is starting to look "fairly" sinusoidal as the dumpers are pretty much not doing anything. Which gives me 400uW output in to 8 ohms.
Indeed taking out the dumpers and it give one watt into 8 ohms, which is very pleasing, actually. I haven't full thought this through yet, but am conscious that the bottom "dumper" is a CFP, and with the NPNs removed the PNP device is still acting like a dumper. Removing that and it all falls to pieces, leaving the bootstrapped resistor to pull the 8 ohms.... This is my problem with the 1 watt output (with dumpers connected). That (current through R38) should be a sine wave if it is solely coming from the class A not driving the dumpers.
If I wind down the input to 1.5mV I get this:
Where I R38 (blue trace) is starting to look "fairly" sinusoidal as the dumpers are pretty much not doing anything. Which gives me 400uW output in to 8 ohms.
Last edited:
That makes sense in what you are saying.
So cutting all the lower stage out leaving only the two 560 ohms for current and it delivers 280mv rms into 8 ohm (10 milliwatt)
Another idea for you to explore might be something much simpler like a higher voltage ACA amp delivering higher voltage swing but running at lower bias current as I can not see your headphone loading being all that low. Yes its AC coupled but is that really an issue... I just think for revealing headphones it might be a better fit sonically. It would be dead easy to add simple circuit to take any swithc on thumps out of the equation.
So cutting all the lower stage out leaving only the two 560 ohms for current and it delivers 280mv rms into 8 ohm (10 milliwatt)
Another idea for you to explore might be something much simpler like a higher voltage ACA amp delivering higher voltage swing but running at lower bias current as I can not see your headphone loading being all that low. Yes its AC coupled but is that really an issue... I just think for revealing headphones it might be a better fit sonically. It would be dead easy to add simple circuit to take any swithc on thumps out of the equation.
Hi, thanks. I started this mini exploration thinking I was responding to a request about how much class A output came from a Quad 405 on the "other forum" by a poster who was looking forward to using it for running his Stax headphones. I was a bit surprised at what I was calculating, but now am at one. I have been thinking quite a bit about the circuit and I would not claim total mastery, but my own personal take home is that:
It is a class B amp who's crossover is tamed by the addition of a resistor across its base to emitters. There is plenty of feedback at low frequencies, but as the frequency rises, the cap in the "bridge" starts to roll back the overall feedback, leaving the Class B to increasingly fend for it self. But meanwhile, the inductor in the "bridge" is increasingly filtering the high frequencies, including the crossover artefacts. This is actually quite a nice balancing act as all those nasty family of ultrasonic harmonics (or glitch) that we all love looking at on FFTs are lost in the inductor, and don't make it to the outside world. Quite clever really. And yes there is a "null" to be found by tweaking the inductor and capacitor in the bridge to where the glitch is pretty much eliminated from the output.
So I don't really find anything at odds in general with the utterings of PW, but only the output power of class A output being 10 watts, Maybe this was a misunderstanding perhaps, that in a 405 amplifier the total power consumed by the both the class A amplifiers would be around 10 watts?
It is a class B amp who's crossover is tamed by the addition of a resistor across its base to emitters. There is plenty of feedback at low frequencies, but as the frequency rises, the cap in the "bridge" starts to roll back the overall feedback, leaving the Class B to increasingly fend for it self. But meanwhile, the inductor in the "bridge" is increasingly filtering the high frequencies, including the crossover artefacts. This is actually quite a nice balancing act as all those nasty family of ultrasonic harmonics (or glitch) that we all love looking at on FFTs are lost in the inductor, and don't make it to the outside world. Quite clever really. And yes there is a "null" to be found by tweaking the inductor and capacitor in the bridge to where the glitch is pretty much eliminated from the output.
So I don't really find anything at odds in general with the utterings of PW, but only the output power of class A output being 10 watts, Maybe this was a misunderstanding perhaps, that in a 405 amplifier the total power consumed by the both the class A amplifiers would be around 10 watts?
Last edited:
Hi, thanks Jan.
Unfortunately at this point in my life maths makes my head hurt and that document, especially today, I cannot read it and gain any useful information, but thanks for posting. I prefer the above representation, which is blissfully free of bridge representations (!!). I do feel that the circuit works like it works, and I am probably describing the same thing from a different perspective? I can see you are knowledgeable about this circuit. If you could simply without reference to balancing bridges or maths, give me a pointer to where a am confused that would be interesting (entirely optional). I have run multiple simulations and chopped up the circuit and investigated the various parts, combined with my previous experiences of amplifier experimentation, have come to my above understanding, unusual as it may be?
Simulated current dumping amplifier at 500kHz.
Blue trace Class A amplifier output
Red trace Class B output
Green trace speaker output
Unfortunately at this point in my life maths makes my head hurt and that document, especially today, I cannot read it and gain any useful information, but thanks for posting. I prefer the above representation, which is blissfully free of bridge representations (!!). I do feel that the circuit works like it works, and I am probably describing the same thing from a different perspective? I can see you are knowledgeable about this circuit. If you could simply without reference to balancing bridges or maths, give me a pointer to where a am confused that would be interesting (entirely optional). I have run multiple simulations and chopped up the circuit and investigated the various parts, combined with my previous experiences of amplifier experimentation, have come to my above understanding, unusual as it may be?
Simulated current dumping amplifier at 500kHz.
Blue trace Class A amplifier output
Red trace Class B output
Green trace speaker output
@hbc, your post above shows the bridge in all it's glory: the two resistors plus L & C! (Except the input R).
No math necessary.
When the dumpers switch on, current is flowing through the L which generates a feedback signal back to the input, in addition to the feedback through C already present.
Thus switching on the dumpers, which increases the gain, also increases the feedback factor.
The trick is in balancing the bridge: When the bridge is balanced, the extra feedback exactly compensates for the extra gain. QED.
In the attached figure, the balance condition is Z1*Z3 = Z2 * Z4 and Z1 and Z3 are the resistors, Z2 is the cap and Z4 is the L.
Since Z2 = 1/wC and Z4 = wL, the balance becomes R1/R2 = L/C.
Your sim looks OK.
Jan
No math necessary.
When the dumpers switch on, current is flowing through the L which generates a feedback signal back to the input, in addition to the feedback through C already present.
Thus switching on the dumpers, which increases the gain, also increases the feedback factor.
The trick is in balancing the bridge: When the bridge is balanced, the extra feedback exactly compensates for the extra gain. QED.
In the attached figure, the balance condition is Z1*Z3 = Z2 * Z4 and Z1 and Z3 are the resistors, Z2 is the cap and Z4 is the L.
Since Z2 = 1/wC and Z4 = wL, the balance becomes R1/R2 = L/C.
Your sim looks OK.
Jan
Attachments
Hi Jan,
I think I a sparked a response from HBC, as indeed I did ask on another forum about wanting to drive my 405/909 at a low level in the hope that it would remain in Class A - sufficient to drive my headphones. However, this had led on to me reading as much about Quad's Current Dumping technique. This has only led to further confusion, particularly as so much emphasis is given to gain in the feed - back loop.
My problem comes from the fact that Quad themselves state, it is the error that is fed to neutralize the distortion created when the dumpers switch on. The question I keep asking myself is ''Why would Peter Walker use an ''Impedance Bridge'' if it was simply a question of using negative feedback to minimise distortion? The consequences of all this leads me to postulate an alternative reason - which may be entirely erroneous. One cannot argue with the issue of ''balancing the bridge''. But is that the sole purpose? I think, for me a clue came when comparing my 405 to my 909, which is virtually new (Old stock). What stands out is the perceived speed of the 909 in terms of 'Attack Time''. It is very fast, when compared to the 405, which in itself is very good. The 909 is lightening - quick.
Logic suggests, the ''Impedance Bridge'' cannot handle the full spectrum of audio, yet distortion will occur at all frequencies when the Dumpers are invoked. So, that leads me back to concluding, Walker is indeed only talking about the error signal - the composition of which will only be high order components. It also leads me to the conclusion, that the amplifiers response time is so quick because of the absence of high levels of feedback. This, I believe is because, to ''Arrest'' distortion, any feedback must be injected in a very short space of time.
However, in doing so, it would only act to minimize distortion, not as Walker states, eliminate it - specifically cross - over, although the dumpers conduct at some distance away from where conventional crossover distortion occurs. So it is my contention that, a), The Class A section is both very high quality and is lightening quick in responding to signal changes, and b) The error signal fed to it, is a function of the distortion introduced when the dumpers start to conduct., not the whole audio envelop.
At the time the original 405 was designed, Walker must have had his original electrostatic loudspeakers in mind as the load. This device is a full - range element, and it's ''Reactive Characteristics'' well researched. So much so, that it would have been viewed as ''Part of the 405's Output Stage''. That being the case, he would have taken into consideration the loading characteristics it would have on any ''Fed-back Error'' signal.
It is for this reason, I conclude, c) '' The function of the Impedance Bridge is to Modify the Phase of the Error Signal - at all frequencies to ensure it is inserted into the Class A section to ''Add Neutralization''; as best it can, to the feed - forward error and thus achieve extremely low distortion figures''
So, it is both the Phase - Shift across the Impedance Bridge, in conjunction with the extremely short Attack Time, which allows makes this amplifier to function as it does.
Although, this is a very broad hypothesis, it has evolved over many years as I sought out an original 405, where there are parallels with a 200kW broadcast transmitter I was involved with in the early eighties, the output stages of which were formed by a Carrier Valve and a Peaking Valve. The Carrier Valve, dealt with power up to 60% of PEP, whilst the remaining was supplied by the Peaking Valve. To achieve ''Broadcast Standard Distortion Figures'', power in excess of 60% had to be introduced accurately with minimum ''Phase - Error''.
Because Walker was of that generation - this Broadcast Transmitter was and engineering exercise first prototyped in the 1920's, but not until the '80's were component tolerances good enough for the method to be developed for commercial transmitters - I cannot help but make a connection. The Communications World was very small in those days and I am sure research papers were widely circulated within it.
Kind Regards,
I think I a sparked a response from HBC, as indeed I did ask on another forum about wanting to drive my 405/909 at a low level in the hope that it would remain in Class A - sufficient to drive my headphones. However, this had led on to me reading as much about Quad's Current Dumping technique. This has only led to further confusion, particularly as so much emphasis is given to gain in the feed - back loop.
My problem comes from the fact that Quad themselves state, it is the error that is fed to neutralize the distortion created when the dumpers switch on. The question I keep asking myself is ''Why would Peter Walker use an ''Impedance Bridge'' if it was simply a question of using negative feedback to minimise distortion? The consequences of all this leads me to postulate an alternative reason - which may be entirely erroneous. One cannot argue with the issue of ''balancing the bridge''. But is that the sole purpose? I think, for me a clue came when comparing my 405 to my 909, which is virtually new (Old stock). What stands out is the perceived speed of the 909 in terms of 'Attack Time''. It is very fast, when compared to the 405, which in itself is very good. The 909 is lightening - quick.
Logic suggests, the ''Impedance Bridge'' cannot handle the full spectrum of audio, yet distortion will occur at all frequencies when the Dumpers are invoked. So, that leads me back to concluding, Walker is indeed only talking about the error signal - the composition of which will only be high order components. It also leads me to the conclusion, that the amplifiers response time is so quick because of the absence of high levels of feedback. This, I believe is because, to ''Arrest'' distortion, any feedback must be injected in a very short space of time.
However, in doing so, it would only act to minimize distortion, not as Walker states, eliminate it - specifically cross - over, although the dumpers conduct at some distance away from where conventional crossover distortion occurs. So it is my contention that, a), The Class A section is both very high quality and is lightening quick in responding to signal changes, and b) The error signal fed to it, is a function of the distortion introduced when the dumpers start to conduct., not the whole audio envelop.
At the time the original 405 was designed, Walker must have had his original electrostatic loudspeakers in mind as the load. This device is a full - range element, and it's ''Reactive Characteristics'' well researched. So much so, that it would have been viewed as ''Part of the 405's Output Stage''. That being the case, he would have taken into consideration the loading characteristics it would have on any ''Fed-back Error'' signal.
It is for this reason, I conclude, c) '' The function of the Impedance Bridge is to Modify the Phase of the Error Signal - at all frequencies to ensure it is inserted into the Class A section to ''Add Neutralization''; as best it can, to the feed - forward error and thus achieve extremely low distortion figures''
So, it is both the Phase - Shift across the Impedance Bridge, in conjunction with the extremely short Attack Time, which allows makes this amplifier to function as it does.
Although, this is a very broad hypothesis, it has evolved over many years as I sought out an original 405, where there are parallels with a 200kW broadcast transmitter I was involved with in the early eighties, the output stages of which were formed by a Carrier Valve and a Peaking Valve. The Carrier Valve, dealt with power up to 60% of PEP, whilst the remaining was supplied by the Peaking Valve. To achieve ''Broadcast Standard Distortion Figures'', power in excess of 60% had to be introduced accurately with minimum ''Phase - Error''.
Because Walker was of that generation - this Broadcast Transmitter was and engineering exercise first prototyped in the 1920's, but not until the '80's were component tolerances good enough for the method to be developed for commercial transmitters - I cannot help but make a connection. The Communications World was very small in those days and I am sure research papers were widely circulated within it.
Kind Regards,
Last edited:
Because it isn’t. One of the many original motivations was to build an amplifier that didn't have any setup adjustments, unlike, say the 303 which has 5. You are conducting a straw man argument.
Could you expand on that? We all know the 303 needs to be biased correctly, but nothing has come close to adequately explaining how it is achieved with the 405. Other than that, the class A amplifier can stand alone without the current dumpers and indeed is used to control the dumpers 'at all times'. The class A amp does not need any feedback. However, the dumpers do, when they come into play. But Walker calls it feed-forward, which I understand to be an error correction signal, based on the discrepancy in phase between the Class A signal and the insertion of ''Top - up power'', supplied by the dumpers. I add this further proviso, which is, the point at which the dumpers switch on is well away from the non - linear region of the transistors junction and the action is very swift.
In essence, all that is required is for the dumpers to be switched - on cleanly (and off again) on the linear part of their transfer characteristic, and in - phase.
In essence, all that is required is for the dumpers to be switched - on cleanly (and off again) on the linear part of their transfer characteristic, and in - phase.
Last edited:
Foot-note, why would anyone produce a very high quality Class A amplifier, only to add distortion to it? - madness! It is my contention that the feed back is simply derived from the distortion generated when the dumpers switch on. In essence, it plays no part in the performance of the Class A amplifier and is simply dealt with, in the same way the input audio is.
Morning,
In my subjective experience, "speed" is a low frequency phenomenon, probably related to phase in the the low Hz.
Foot note to the above. I got to a point where I realised that to constructively proceed with the analysis, I need to fully understand the complex imaginary maths. It is difficult trying to dust off that part of my brain, and resurrect the rusty half understood from 30 years ago, so got stalled.... easily distracted by designing / building (high quality) class A valve amps.
In more positive mathematical news, I have remembered / worked out how to do long division, fourth time lucky doing primary school home work... Small person 5 is pretty bright and will probably not benefit form my newly rediscovered talent, sigh.
aka dave
Last edited:
If the Class A (current output) amplifier was to "stand alone" which it could, as discussed previously, it would absolutely need feedback to bring its output "impedance" down, if it was to stand any chance of behaving like a "normal" audio amplifier.