| alaskanaudio |
I have read with some interesting comments on the high output impedance of output stages using common emitter or common source circuits. These of course use the collector or drain as the output instead of the emitter or source.
One person commented that nobody uses the common source configuration. This of course is an error in thinking. All of my amplifier designs for example use common source configuration. Naturally the output impedance of these and designs that may be similar circuits are going to have higher output impedances than those using common drain configuration. This rise in output impedance is primarily due to the fact that the common source configuration does not use 100 percent negative self-feedback in the output stage that a source or emitter follower configuration has. The great benefit of using a common source configuration in a amplifier that drives a loudspeaker is that you gain the ability to control the output impedance via a selected amount of negative feedback around the stage. This is because the restriction of 100 percent self-negative feedback is gone. This has some major benefits when one wishes to achieve maximum performance when driving loudspeakers..
Many persons have been lead to believe that an audio power amplifier must have very low output impedance and thus able to provide high damping factors. This line of thinking is incorrect in my opinion since it totally disregards what it takes to make loudspeakers perform the best. There are designs that actually include positive current feedback to achieve near zero output impedance over a good portion of their operating range.
If one has the ability to adjust the output impedance while listening to music played though your favorite loudspeakers you may find that output impedances between .2 and .5 ohms actually provide the best sound quality. And also that low feedback factors around the output stage in the order of 6 to 15 Db are best. This should provide damping factors in the order of 12 to 20 and is similar to what a tube amplifier would provide. Feedback factors of 6 to 15 Db cannot be achieved with circuits that use a source follower in the output stage since they run at close to 100 percent negative feedback already.
The source follower output stage is more popular for two main reasons. It is easy to use and second it has lower distortion. Unfortunately the low distortion and thus its linearity is provided by 100 percent negative feedback. Considering that negative feedback is undesirable in such large amounts makes me wonder why the configuration is used at all to drive loudspeakers.
For every one who is seeking the sound of a tube amplifer in a solid state design the first thing you have to get rid of is the source follower in the output stage. Then adjust the negative feedback around the output stage to reasonable values.
John Fassotte
Alaskan Audio |
|
|
| Circlotron |
| quote: | Originally posted by alaskanaudio
The great benefit of using a common source configuration in a amplifier that drives a loudspeaker is that you gain the ability to control the output impedance via a selected amount of negative feedback around the stage. This is because the restriction of 100 percent self-negative feedback is gone.
The source follower output stage is more popular for two main reasons. It is easy to use and second it has lower distortion. Unfortunately the low distortion and thus its linearity is provided by 100 percent negative feedback. Considering that negative feedback is undesirable in such large amounts makes me wonder why the configuration is used at all to drive loudspeakers. |
Another benefit of a common source output stage is that it can swing virtually from rail to rail, something a bipolar follower and especially a mosfet follower can't touch without bootstrapping the drive.
Negative feedback is only undesirable in combination with it's enemy phase shift. If you had an amp with no phase shift or signal path delay you could wind up the feedback all you wanted and it would sound great, but it is phase shift and similar things that make nfb look bad because it lures it over to the dark side and it becomes positive feedback or part thereof.
An emitter follower can tolerate 100% nfb because phase shift is virtually nonexistent in that one stage. Try it over several stages and that's when the wheels fall off.
GP. |
|
|
| sonnya |
With BJT stage i think that there is a small problem when working in emitter follower.
1) You do have phase shift but it is at a much higher frequency.
2) The biggest problem i have discovered lately when designing a NFB amp is that when driving a capacitive load where the current is 90 degress ahead of the voltage you would be able to see this at the bases of the BJT!. Which mean that you turn the phase of the previous stage too much when the idle "current or outputimpedance" of the previous stage is too "low or high"!!! That will make a lot of trouble in a NFB amp.
;)
Here i think the common emitter/source has some great benefit because of the load placement in the signal path. The load is no longer part of "gm" stage wich make up the voltage-current conversion.
Can you follow me?
Sonny |
|
|
| Dave |
A couple of questions,
The open loop output impedance of a common source/emitter output stage will be very high meaning that when connected to a reactive load the gain of the output stage will change with varying load impedances. Therefore with any given loudspeaker the level of global feedback will be quite variable and also frequency dependent, depending on what the input impedance graph of the loudspeaker looks like.
Isn't this quite undesirable?
The amplifier might be a quite unstable when driving a capacative load as a result?
Also some consider only global feedback to be musically destructive while local is not. An emitter follower operates with only local feedback and thus should give better objective and subjective performance? |
|
|
| AMPMAN |
| You are quite right, electronics world december 1999 [new class AB power amp] published a no crossover distortion, common emitter amp with a new current mode class ab driver stage. it swings rail to rail,and the ocsiloscope waveforms show no switching distortion and utterly stable into 1uf. if your interested HPotter has the schematic,and a digital camera[i have no cam or scanner] so maybe he can post them. this may sound like heresy but can we get rid of class A with all its wasteful heat and aluminum. yes, I am building a stereo amp and i'll let you know how it turns out |
|
|
| jam |
Yes, Peter (Harry Potter) it would be great if you could post the schematic.
Jam |
|
|
| P.Lacombe |
Sonny,
You are right. Furthermore, additionnal difficulty with BJTs is caused by the very high transconductance. This is not the case with FETs. My ears are saying that the bests results are in common collector for BJTs, and common source with FETs. Design of a high quality BJT amplifier working in class AB with common emitter output stage is a venturesome challenge. (Or very high emitter resistors are to be used...)
With BJT common emitter darlington output configuration, some low value resistor is to be connected between the bases of the drivers and ground, in order to define the overall open loop gain of the amplifier, independently of the loudspeaker impedance variations. Generally speaking, emitter follower must always be driven by a moderate impedance.
Regards, P.Lacombe. |
|
|
| alaskanaudio |
I have seen a lot of excellent comments to my original post on this subject. I would likel to respond to the one by AMPMAN for now.
AMPMAN,
I’m looking forward to looking at the circuit in the article you mentioned.
In reference to class A heat:
I suspect that this amplifier you referenced likely uses BJT’s because my experience with MOSFETs is that you have to run them in class A for maximum performance, or close to 500MA of idle current per device. Thus whether we can get rid of class A or not may depend on the type of output devices used. I would certainly like to get rid of 800 or so watts of heat in my amplifiers. Its ok in the winter but in summer it can make the listening room extremely warm. But since I like MOSFET’s I live with it.
Driver stages:
No matter what circuit arrangement an amplifier uses the driver stage must be configured to drive the output devices properly. This will likely add some complexity to both the power supply, driver and other circuits, but the results are well worth it. Many amplifiers are lacking in this regard and attempt to keep things to simple. Then they likely have to depend on excessive negative feedback as a cure.
More Common emitter and common source circuits:
I would really like to see more circuits used that use the output stage running common emitter or source mode. As far as stability goes I have tested my amplifiers with all sorts of capacitive loads for .1 to 10 uf and have never had any oscillation or instability problems with them. I contribute this to multiple feedback loops instead of one over all loop. The effects of driving such a large capacitive loads at high frequencies can be seen at the output when the waveform is viewed. But no instabilitie occurs due to good phase margins.
John Fassotte
Alaskan Audio |
|
|
| AMPMAN |
| Dear sir first of all I believe in economy in all things. We should be intelectually capable of designing amps that do not waste so much power. I understand that of course, this is an ideal.Re your comments about negative feedback,I'll give you the authors view for now as I don't have a scanner. so lets begin,An important problem encountered in class-AB audio amp design concerns thebias-control loop.Often a complementry commen-collector output stage is used and the power transistors are included in the bias control loop.This can easily cause thermal instability due two the large temperature variations in the output transistors.Thermal coupling of all diodes and transistors in the class-AB control loop can improve the thermal stability of the quiescent current in the output stage. but this is in most cases too slow to react to burst signals. As a result , emitter resistors are usually added to the power transistors to improve thermal stability.However the voltage drop across the emitter resistors can switch off the transistor that is conducting the residual current.Because of the limited bandwidth of the distortion reduction by means of using negative feedback, transistor switching can be a source of high-frequency distortion. Additional circuitry is necessary to prevent this. Moreover, a common-collector stage is not able to reach a rail to rail voltage output swing due to the bass-emitter voltages. Common-emitter output stages are usually based on a complementary feedback pair. However, the local feedback loop around the pair can be a source of HF oscillation. In order to achieve thermal stability without switching problems, and to allow maximum output voltage swing, we designed a common-emitter power amplifier based on a new current-mode class-AB driver circuit. Due to the absence of local feedback at the output, the stability of the amplifier is only dependent on the global feedback-loop. The open-loop output impedance of a common-emitter amplifier is inherently high, but can be lowered by applying negative feedback. As a result, the output resistance becomes inversely proportional to the transconductance gm, and the feedback factor. Rout=1/gmB In order to obtain a closed-loop gain of 34 the feedback factor must be 1/34. For an output resistance of 30m-ohms the transconductance should be approx. 1,000A/V. This requires a cascade of at least 3 gain stages: an input transcondance stage and 2 current gain stages. end quote. Ihope this has been of use |
|
|
| Tube_Dude |
Yes Ampman!
And more:as the openloop output impedance is high and if you have a load with a phase shift (thats you have with a speaker)the feedback will be not in phase wih the input signal,but with the phase of the load inprinted in it!
Second and more important all the EMF of the peaker is not dissipated in the low output impedance of the output but will be conected to the input stage where it can intermodulate with the input signal...puting in another way the amplifier will be subjected
at two input signals...the normal input signal and a delayed and distorted that comes from the speaker...is obvious that one will intermodulate with the other.
This is nothing new as Matti Otalla have call it interface intermodulation distortion
This is a great forum! :)
Regards
Jorge |
|
|
| AMPMAN |
| first of all,signals sent back in phase with the input will be positive,signals sent back negativly reduce gain and distortion. regarding interface modulation this amp's output impedance is 30 milli ohms, it has not been proven that low output impedance amplifiers sound any better.So tell me what is wrong with this topology. |
|
|
| Peter Daniel |
| So here it is: |
|
|
| AMPMAN |
| further to my reply to you I browsed through my old wireless world mag's and found an article by robert cordell which refutes what you say.its too long repeat here but I'll fax it to you or anybody else who is SERIOUSLY INTERESTED. OK thanks to HP for the schematic,I would point out that the power tranny's are obsolete, replace with 2sc5200/2sa1943, the ssm input transistors are matched pairs in integrated circuit form. I used 2n3904/2n3906 transistors replacing the two 100 ohm resistors with a 200 ohm pot,is interes if anyone is interested in this circuit I'll be happy to supply more info. |
|
|
| sonnya |
| quote: | Originally posted by HPotter
So here it is: |
I wish it was me who had come up with this outputstage!!! ;)
Sonny |
|
|
| alaskanaudio |
| quote: | Originally posted by AMPMAN
further to my reply to you I browsed through my old wireless world mag's and found an article by robert cordell which refutes what you say.its too long repeat here but I'll fax it to you or anybody else who is SERIOUSLY INTERESTED. OK thanks to HP for the schematic,I would point out that the power tranny's are obsolete, replace with 2sc5200/2sa1943, the ssm input transistors are matched pairs in integrated circuit form. I used 2n3904/2n3906 transistors replacing the two 100 ohm resistors with a 200 ohm pot,is interes if anyone is interested in this circuit I'll be happy to supply more info. |
AMPMAN
I'm interested in more information on the article you mention and the circuit. Perhaps you could mail me the information you have available or send them to me via fax at a pre-arranged time.
Please use my email johnf@audioamps.com to respond.
Best regards,
John Fassotte
Alaskan Audio |
|
|
| janneman |
Ampman,
I know some articles by Bob Cordell and have a high regard for him, but I don't recognise your reference. Can you give my magazine & issue or publishing date?
Thanks, Jan Didden |
|
|
| hugobross |
| quote: | | For every one who is seeking the sound of a tube amplifer in a solid state design the first thing you have to get rid of is the source follower in the output stage. Then adjust the negative feedback around the output stage to reasonable values. |
I've just read an article on rod's page about variable amplifier impedance (project 56: http://sound.westhost.com/project56.htm)
He's comparing voltage amplifiers with current amplifiers and amps with the two compared. I'm just wondering if anyone here has experienced with that?? Although, the conclusion of Rod himself about the sound of tubes and the output impedance of amps is interesting.
I'm going to read it again a few times because my English isn't that good.
best regards,
HB. |
|
|
| AMPMAN |
| the article appeard inW.W. feb 1983 |
|
|
| Ian Macmillan |
Would anyone care to summarise how this circuit works for those of us who do not have access to the article in question? I recognise various parts of the topology and the front end is conventional enough, but I am not clear how the driver and output stage operate.
Ian. |
|
|
| ergo |
Have you seen this kind of output stage and how do you feel about this? (think IGBT into bjt or MOSFET)
I think there are several plusses here. One can set the output stage for 2X or 4X voltage amplification for example. This in turn means that voltage amp stage does not have to have as much voltage swing and can be cascoded or have a stabilised supplies taken from primary supply....
And then think if there would only be the final and driver stage which would form a power amp with 14-20X voltage gain..... A friend of mine came up with such circuit and it's in my next four months or so todo list :)
Ergo |
|
|
| Ian Macmillan |
I've no seen anything quite the same as this before but it looks like a variation of the Class A+B topology. T15 and 16 look to be a fairly standard output stage (probably class AB) whilst T16 and 17 presumably provide additional (Class B?) power. I note the somewhat unusual inclusion of R30 though, which I assume allows T16 and T17 to provide voltage gain too. It is this aspect that I regard as the most novel.
Note that the driver is already cascoded by T12 and 13.
Ian. |
|
|
| AMPMAN |
Here is the information you requested. Also look at my earlier thread to Alaskanaudio. The imput stage, consisting of a differential transistor pair TR1 & 2 and a current mirror, TR3,4, converts the differential imput voltage to a single output current. This current feeds the base of driver transistor TR9 and via the common base transistor TR8 feeds the base of the driver transistor TR10. The driver transistors supply their emitter currents to power transistors TR11 and TR12 respectively. Biasing and Class-AB control are achieved by means of a bias-control loop formed by TR6-9. Due to the buffer function of TR6 the base-emitter voltage of the power transistor TR11 is isolated from the bias control loop. This is done to avoid thermal or HF switching distortion problems mentioned earlier. This design is based on complementary/pnp transistors. Their parameters can be assumed equal to make the equations easier. The Class AB control is based on the well-known geometric Class AB control law. Ic8 x Ic9 = I²r (2) The DC collector current of the driver transistors is given by Ic9 = Ic10 which = Ir x the square root of HFE (3).
Collector terminals of the driver transistors TR9,10 connect to the output terminal to minimize driver dissipation and to prevent the power transistors TR10,12 from saturating. Power dissipation in the bias-control loop transistors is low compared to the dissipation in the output transistors. Hence, if TR6-9 share a small heat sink, thermal stability is achieved without emittor degeneration and switching distortion. The remaining dominent source of temperature dependence is the temperature coefficient of the power transistor forward current gain. This coefficient is approximately 0.6%/K to the power of 4. Maximum output current is determined by emitter current and the current gain of TR9,11 or TR10,12 respectively, Io (max) = plus or minus Ieh²fe (4).
It can be seen that, in contrast with many other designs, the maximum output current capability is symetrical. A problem with power transistors driven by a current source is that there is no turn-off resistor for them. Under high frequency, high-amplitude drive there will be a tendency for the effective bias current to rise dynamically. By using HF power transistors with a ft of 80MHz, this bias current rise is reduced to 60% at 20kHz at full drive.
In summary this Class AB common-emitter power amplifier incorporates a new current mode Class AB driver circuit to obtain good thermal stability of the quiescent current in the output stage. It also guarantees none-zero currents in the output transistor that is conducting the resisdual current, avoiding HF switching distortiion. Maximum output voltage is near to the rail-to-rail limit. Saturation in power transistors is avoided, resulsting in fast recovery from clipping. The circuit has an excellent stability due to a phase margin of 85 degrees with a B 1/34. There is more but I hope that this gives the salient points. So Ian all we need now is for you to put a su/sy on the front pretty please. |
|
|
| Circlotron |
Output stage voltage gain is approx (R30+R29)/R29. Snip out R29 and s/c R30 and the gain drops back to unity. The ETI-480 amp used such a configuration but with bipolar outputs. I would expect the higher the gain the more critical the bias setting, i.e. you wind the pot to get some current and when it does start you would only have to *look* at the pot and it would alter the current a fair bit, depending on the gain.
GP. |
|
|
| janneman |
Another advantage of the gain in the power stage is that the signal swing in the voltage amplifier stage is reduced. In this amp, it allows them to feed the Vas with an LM317/337. This makes for very cheap and simple stabilised supply for the pre-stages, rather than a more complex 40 or 50V stabilised supply if the output stage didn't have gain. Could that be actually the reason they put in the output stage gain?
Cheers, Jan Didden |
|
|
| Ian Macmillan |
AMPMAN, thanks very much for the detailed answer. I haven't had time to digest it yet but will read your reply avidly just as soon as I am able. No doubt I will then have more questions.
Ian. |
|
|
| Ian Macmillan |
AMPMAN, I've now read your explanation of how the above circuit works pretty thoroughly. I get the general jist but sadly can't claim to have understood all the details. In particular I not't know to what the "r" in the various corresponds. I studied Electronics at college but that was a long time ago. Also you say that the bias control loop comprises Tr6-9. Did you mean this or should it be TR5-8? I have to admit that BJT and current-mode operation give me a headache. I'm more comfortable with FETs and voltage mode. Fortunately I also happen to prefer the sound of the latter, at least in general.
I also missed your plea for SuSy front end on the first reading. In principle this would not be difficult but I fear that practice might well be very different. SuSy requires simple circuits and this one does not really qualify. However, if you wish to try, simply follow the principle in the Aleph-X thread since the general topology of this and an Aleph are similar. You will of course have to produce a bridge design, i.e. replicate the output stage. Also the current mirror on the front end will have to go, or be modified since you will require both output phases. If you decide to go ahead then I wish you the best of luck. BTW, I'm not the expert on SuSy. Perhaps you had better ask Nelson (as the inventor) for his opinion?
Ian. |
|
|
| Tube_Dude |
Ampman
sorry but only now i can anserwer you, because i´m was far from home all this week.
Regarding the schematic i already know it because i´m a long time reader of Electronic Wireless World. It´s in fact a realy enovatif disign but what i have said regarding ampilfiers with high open loop output impedance i still mantain it.
When you talk about the article by Robert Cordell, i don´t know it, but i don´t belive in everythig that is written.
If you can prove my point make that experience with your amplifier.
- connet another amplifier through a 8 Ohm resistor to the output of your amplifier...Put some signal in the other amplifier and put the prob point of a osciloscope in the collector of Tr1.
you will see that the same signal is bieng amplified by your amplifier in reponse of a disturbance in the output.
Thats what i said about the intermodulation betwen the original signal and the reaction from the speaker.
But you d'ont need to belive me because i d'ont write in EWW.:D
regards
Jorge |
|
|
| Tube_Dude |
AMPMAN
One more point!
When you said that your amp have 0,030 Ohm output impedance and because that it will not be subjected to interface intermodulation distortion you are forgetting that, that impedance is obtained from the feedback . So as the open loop is very high the reactions from the speakers "fell free" for reaching the input LTP ...the ideal conditions for the interface intermodulation distortion to happen...this is not new...as i said Matti Otalla also have remarked that many years ago...
Regards
Jorge |
|
|
| sonnya |
TubeDude
I hate to say it but a common collector will be subjected to the same phase shift in another way. The phase shift will show up in the previous stage.
A bipolar Junction Transistor is in general a current - current converter which in the perfect world is a linear transfer slope. So on the base of a BJT the phase shift will show up. Not as large but it will be there.
So if you Feedback loop has to compensate for the phase shift on the output or a phase shift in the the VAS stage because of a capacitive load.... This will be be the same result.....
Only a Common Emitter stage will be subjected to saturation. and it should be easier to control/minimize those swith glitches presented by class B and AB mode.
Sonny |
|
|
| Tube_Dude |
Yes Sonny but what about the EMF from the Speker??
Jorge |
|
|
| sonnya |
In common collector/emitter one of you Transistor will turn of and create switch glitches.
Both circuits will have the same control over the speaker when they have the same output impedance, frequency response and phase margin..
Only then you wil be able to compare those two types of output stages.
Sonny |
|
|
| AMPMAN |
I quote cordell directly, Interface intermodulation was next measured in a manner equivalent to the procedure outlined in reference 1. Equal-level 1000 and 60Hz signals were applied to opposite ends of an 8 ohm load resistor by the amplifier under test and a second power amplifier. A spectrum analyzer was placed across the output of the amplifier under test and the R.M.S. sum of the distortion products was referred to the 1kHz level. The operating level of each amplifier was 25W. The similar levels of interface intermodulation in all 3 cases confirm that open-loop output impedence and feedback factor have virtually nothing to do with it in amplifiers properly constructed at the same cost.
This is the summary of a highly detailed and exhaustive analysis investigating interface intermodulation by an expert in this field. |
|
|
| Samuel Jayaraj |
Ampman,
Have you actually built the circuit (of the schematic posted in this thread) from EW&WW using 2SC5200 and 2SA1943 transistors? How does this amp sound?
As a starting point, I built this amp soon after publication of the article and since the output devices were not available(even then), I used MJE3055/2955 transistors. Soon after powering up, the bias shot up and blew the output devices. Although I had to hand, quite a large number of 2SC3281 and 2SA1302 transistors, I did not try the amp with these. I thought that the 60MHz fT criteria was quite critical to the stability of the design while the transistors I had were only about 30MHz, and gave up completely on the design.
Input from your success may cause me to give this amp a try again. Thanks. |
|
|
|
