For anyone interested in Electrocompaniet amplifiers, former Electrocompanion Terje Sandstrøm has a nice website complete with articles on Sven-Erik Børja (whose brother Bjørn is presently one of the chief designers at SEAS), Matti Otala and his influence on Electrocompaniet designs, TIM distortion and NFB, a history of the early days of Electrocompaniet, schematics of various prototype circuits as well as production amplifiers, mechanical drawings, troubleshooting tips, and quite a bit more.
http://home.online.no/~tsandstr/TableOfContent.htm
But you don't have to be an Electrocompaniet fan to enjoy this website. Personally, I think that it is a fun read for anyone interested in solid-state amplifier design.
Enjoy!
jonathan carr
http://home.online.no/~tsandstr/TableOfContent.htm
But you don't have to be an Electrocompaniet fan to enjoy this website. Personally, I think that it is a fun read for anyone interested in solid-state amplifier design.
Enjoy!
jonathan carr
Just hanging on like many in the industry.
Tell me about it............
"I am a victim of circumstances, as are we all."
Right, Phred???????
Jocko
Tell me about it............
"I am a victim of circumstances, as are we all."
Right, Phred???????
Jocko
Hi,
So, I wasn't the only one.
Bias was a big problem with those amps...
When working correctly they where not bad at all compared to what was around in those days.
Cheers,😉
So, I wasn't the only one.
Bias was a big problem with those amps...
When working correctly they where not bad at all compared to what was around in those days.
Cheers,😉
Thanks Jonathan. I have always Been a fan of Matti Otala. His idea of having the Open loop gain of the amplifier cover the entire Audio band is somthing i have proven time and time again as producing Better sound that when the open loop gain starts to roll off below 20 KHz. Some Audio EE types I know hear in the US refer to Matti Otala as DR. T.I.M. I read the theroy on the link you provided and also read the AES artical from Matti Otala on Low TIM and one thing i noticed from the link that i think might not have been in the AES Artical was the fact that increcing the openloop gain so as to provide about 30 dB of Feedback was ideal. this is what i have also discovered. My method is to increce the Open loop gain untill THD at 20 KHZ with a resistive load no longer Improves. this has always resulted in 30-40 dB of Negitive feedback. interestingly Both Op amp and discreet component designs show similar results. i use the Multi-loop feedback method as done by walt jung to reduce the open loop gain of an op amp to the 30- 40 db range.
regardless of the Active devices used the method of using local Feedback Augmented By an overall loop around the complete Amp Is a valid method of making sure the Amp has enought available feedback to deal with any Audio signal presented to its inputs. the method used to apply local feedback is IMHO unimportent be it from a local feedback loop around just one stage of the Amp or using Emitter resistors on that stage to create local feedback the results are similer.
the school Amp i found most interesting as one of my Discreet component headphone amps used almost the same complimentry front end although mine was cascoded to Bipolars using almost the same transistors although my Jfet compliment was the 2N5458 and the 2N5461 rather than the 2N5459 and the 2N5462. I wounder if the 5459/5462 would be better than the two i use?I have been thinking of revising that design and might post a schematic in these fourms to solisit any thoughts of improvements.
regardless of the Active devices used the method of using local Feedback Augmented By an overall loop around the complete Amp Is a valid method of making sure the Amp has enought available feedback to deal with any Audio signal presented to its inputs. the method used to apply local feedback is IMHO unimportent be it from a local feedback loop around just one stage of the Amp or using Emitter resistors on that stage to create local feedback the results are similer.
the school Amp i found most interesting as one of my Discreet component headphone amps used almost the same complimentry front end although mine was cascoded to Bipolars using almost the same transistors although my Jfet compliment was the 2N5458 and the 2N5461 rather than the 2N5459 and the 2N5462. I wounder if the 5459/5462 would be better than the two i use?I have been thinking of revising that design and might post a schematic in these fourms to solisit any thoughts of improvements.
Please note that Williamsson said the same thing long before Otala, He said "An amplifier must have ample bandwidth before any feedback is applied covering at least the audio band" He also said "It is important that an amplifier must have as low distorsion as possible before any feedback is applied"
It is interesting to note that it took solid state designers 30 years, (with Otala one of the first) to reach the same level on enlightment as Williamsson did in the fifties and is interesting to note that still some designers refuse to see the importance of the above.
For me as a designer I feel uneasy if I not try as much as possible to increase open loop bandwidth and lowering distorsion before appying feedback, (I have never liked distorsion curves where distorsion increase with increasing frequency).
Regards Hans
Here we go again......
Lucky for us, one of the biggest PITAs on the forum who doesn't believe that is in the SinBin.
Thank goodness.
Anyway......those guys knew what they were talking about, despite all the negative press to the contrary from the eggheads in the AES. Glad to see someone else here that agrees with those postions.
Jocko
Lucky for us, one of the biggest PITAs on the forum who doesn't believe that is in the SinBin.
Thank goodness.
Anyway......those guys knew what they were talking about, despite all the negative press to the contrary from the eggheads in the AES. Glad to see someone else here that agrees with those postions.
Jocko
tubetvr said:
I feel the same way at times it is quite tempting to go for those Ultra low THD numbers because they look better on Paper. I have also found that the Primary audible benefit of having a constant amount of THD over the complete 20-20,000 Hz. bandwidth is a lack of Glaring Brightness that not only makes the Lower Midrange and bass sound thin but also robs the Music of it's micro dynamics and heightens sibilance. The real Art of Audio Amplifier design is to find that Ideal spot where the Amplifier really sings and still remain transparent. loss of transparency IMHO will result if not enough feedback is applied to reduce the Distortions of the output stage to a sufficiently low number, However the point of diminishing returns will happen at some point and further Increase of negative feedback will start to induce Transient errors and ringing that also reduces transparency in addition to inducing a lot more sonic problems.
Ah Williamson. This brings me back to My old tube days with one of my first Amplifier designs using his ultra-linear output design based around Gasp! 6L6GC Power tubes. Williamson is another visionary for sure. As for those People that have a different approach to Amplifier design! Hopefully after years of practicing there craft I am sure they will find that Electronic design is not just science but also an Art form that demands creativity and vision. In these days of computer simulation it is Easy for Electronic engineers to get seduced into purely theoretical analyzation of test results and simulated behavior. This seams to result when creativity and inspiration is lacking. It is all too easy to let ones tools define ones work when inspiration is lacking. Tools are what they are, and like any other Tool is only as good as the operator. No one denies the Great help Tools are to our Craft, however A designer of electronic equipment should have the Insight to understand the Concept of what the Circuit is supposed to do in order to properly employ any Tool to the task.
The Above are just my way of doing Amplifier design. There are lot's of alternate Views from highly qualified people to the Contrary. They are entitled to an opinion just lake anyone else. I used to get into debates with those of differing views than mine, However I now just leave them to there own way and ignore those that force there view as an absolute truth. I believe that there is no Absolutes in anything and the universe is all just relative.
ppl:
Although I do keep Otala's concepts in mind, I don't take all of his suggestions as the gospel truth. In my applications, the global NFB range that leads to results that I consider acceptable is usually between 50~60dB. This is with multi-layer circuit boards, full ground and power planing/bussing, comprehensive bypassing, attention paid to parasitic inductance and capacitance, 3-dimensionally stacked construction, the use of air-dielectric, teflon standoffs and floating isolated pads for circuit nodes that warrant it, and usually the topology will be designed for low open-loop distortion with as few dominant poles as I can get away with. I invariably place an LPF at the input, but there typically will be no or at least minimal add-on phase compensation components.
When a topology with a greater number of dominant poles is used, parasitic inductance and capacitances are not properly addressed, bypassing is inadequate, or ground and power planes are not used, etc., my observations are that the amount of global NFB should be kept considerably lower, but precisely how much lower seems to depend on the individual situation.
I don't have any preference for multi-loop NFB or local degenerative feedback. Each circuit and application is somewhat different, and I use whatever does the job best.
When I have to use topologies that entail multiple dominant poles, I usually arrange things so that the gain progresses from highest at the input to lowest at the output, while the bandwidth is the opposite - lowest at the input progressing towards highest at the output. When such an arrangement is not feasible, I will put in an extra effort on arranging the stagger ratios of the dominant poles.
Rather than looking at the THD, I find it much more productive to study the individual harmonic order distortion curves.
One more important thing to realize is that in today's world, every amplifier is constantly being asked to amplify far more signal content than the audio signal itself, and the increased signal content consists of unwanted conducted and radiated noise from the environment.
I believe that it is a very good idea to keep in mind how the amplifier circuitry, regulators and power supplies will be affected by the constant and unavoidable presence of wide-band noise and spurious signals, and design accordingly.
hth, jonathan carr
Although I do keep Otala's concepts in mind, I don't take all of his suggestions as the gospel truth. In my applications, the global NFB range that leads to results that I consider acceptable is usually between 50~60dB. This is with multi-layer circuit boards, full ground and power planing/bussing, comprehensive bypassing, attention paid to parasitic inductance and capacitance, 3-dimensionally stacked construction, the use of air-dielectric, teflon standoffs and floating isolated pads for circuit nodes that warrant it, and usually the topology will be designed for low open-loop distortion with as few dominant poles as I can get away with. I invariably place an LPF at the input, but there typically will be no or at least minimal add-on phase compensation components.
When a topology with a greater number of dominant poles is used, parasitic inductance and capacitances are not properly addressed, bypassing is inadequate, or ground and power planes are not used, etc., my observations are that the amount of global NFB should be kept considerably lower, but precisely how much lower seems to depend on the individual situation.
I don't have any preference for multi-loop NFB or local degenerative feedback. Each circuit and application is somewhat different, and I use whatever does the job best.
When I have to use topologies that entail multiple dominant poles, I usually arrange things so that the gain progresses from highest at the input to lowest at the output, while the bandwidth is the opposite - lowest at the input progressing towards highest at the output. When such an arrangement is not feasible, I will put in an extra effort on arranging the stagger ratios of the dominant poles.
Rather than looking at the THD, I find it much more productive to study the individual harmonic order distortion curves.
One more important thing to realize is that in today's world, every amplifier is constantly being asked to amplify far more signal content than the audio signal itself, and the increased signal content consists of unwanted conducted and radiated noise from the environment.
I believe that it is a very good idea to keep in mind how the amplifier circuitry, regulators and power supplies will be affected by the constant and unavoidable presence of wide-band noise and spurious signals, and design accordingly.
hth, jonathan carr
jonathan > Thanks for your detailed reply. I also hold the View that circuit layout And highly stress Quality High frequency Local Bypassing with at a Min an R/C Bypass topology for the gain stage and Perfer separate regulation with separate transformer windings for The Output and Voltage gain stages. I use Multi-loop feedback networks with Op Amp Based Amps because of the Ability to control the Open loop band width within the local loop. Also since the Op amp receives its feedback from its own output like it will if a cap is placed around the opamp then the additional phase shift of the output stage has minimil effect upon the stability of the system. I still like the output stage to have as wide of a bandwidth as possible and at least several times wider than the Voltage gain stages driving it. this can allow the removal of the Phase leed and noise gain compensation typicaly employed to obtain stability and as such tends to minimize the number of poles if Parrisitics are properly accounted for and acomidated.
For discreet component designs i perfer to control the Open loop gain with local degeration properly chosen to obtain good noise performance for the Application. For line level stages I as you put most of the gain at the first stage for good niose performance while in High power Amps i like most of the Voltage gain at the second stage so as to use quality low noise low voltage transistors as diff amps.
For discreet component designs i perfer to control the Open loop gain with local degeration properly chosen to obtain good noise performance for the Application. For line level stages I as you put most of the gain at the first stage for good niose performance while in High power Amps i like most of the Voltage gain at the second stage so as to use quality low noise low voltage transistors as diff amps.
- Status
- Not open for further replies.