For THD analysis I use this set of commands (cant remember where I got them, was from SOMEONE here 🙂
.options plotwinsize=0
.options numdgt=8
.options noopiter
.param freq 10K (or whatever the frequency you want to check is)
.param num_fft_pts 4096
.param timestep {1/{ {freq} * {num_fft_pts}} }
.four {freq} 9 8 V(OUT)
.tran 0 {20/freq} 0 {timestep}
The signal source is then SINE(0 <voltage> {freq})
.options plotwinsize=0
.options numdgt=8
.options noopiter
.param freq 10K (or whatever the frequency you want to check is)
.param num_fft_pts 4096
.param timestep {1/{ {freq} * {num_fft_pts}} }
.four {freq} 9 8 V(OUT)
.tran 0 {20/freq} 0 {timestep}
The signal source is then SINE(0 <voltage> {freq})
The SA2014 has a lot more things added, like a cascode on the LTP and a lot more things, that would make consider it a blameless design a bit wrong.
Carlos,
This circuit arrangement or topology do not belong to Doug Self. He used this topology to create a amplifier design method - Blameless.
This method become very popular, but I believe there are other better topologies, like the one you open a thread here.
Yamaha and Sony used this circuit arrangement in very old amplifier and now Yamaha use it with small variation in VA stage(In the simulation file there is the amplifier model with service manual in the web). Have a look in simulation attached.
Ronaldo
Last edited:
Thanks, Ronaldo.
I was just curious as how that amp Self wrote a book around as a very low distortion project, how it did sound.
I was just curious as how that amp Self wrote a book around as a very low distortion project, how it did sound.
It started as a set of articles in Electronic World a couple of years after he write the book. Find all D Self writings in this web site.
Have a look in this simulation of a Marantz Module MA700 that is more modern design with other topology (I prefer this one).
Yamaha amp is a simple schematic and can be assembled with proto-board or perfored board. You can do tests and listen to get an personal impression.
Regards
Ronaldo
Yes, I read the original EWW articles at the time they were published. Also bought his book in '96.
At the time I was intending to modify a Pioneer receiver I had that didn't sound 'interesting". By that I mean it didn't make you want to continue listening to music, which I think is an important subjective reaction that people should pay attention to.
That was a surround receiver, which I could compare to my DIY main amplifier, same power, made from a design by Graham Nalty. That amplifier did sound exciting, completely natural.
Well, at the time I was curious about this Doug Self design that attacked distortion in several areas.
I am very familiar with the Marantz MA-700, and it seemed to be a good option to the Luxmans I was simulating.
As I never listened to an MA-700, I looked at some reviews of it. Comments found it to be a good sounding amp, even if a little aggressive.
I noticed you increased the output stage DC power considerably, from +/-44v to +/-65v, which I think is a good move. You also used two transistors for the bias, instead of the usual Darlington Marantz puts there.
Marantz recommends 37mA bias adjustment on each emitter resistor, but I'm not sure that's what you used on your simulation. I say that because I could not use your bias pot on my sim, so I just changed bias resistor values.
What I watched is dissipation on each output transistor. For 3.75W I got 0.00062% THD @ 10KHz 100W @ 8 ohms. Which I think is the output you aimed at.
But for that I had to increase bias to 58mA on emitter resistors, which is above what Marantz suggested.
Bias adjustment should be tuned using a scope, of course, even if I guess Marantz used that 37mA setting because of the heatsink they used on the MA-700 monoblocks.
The Marantz MA-700 architecture is quite similar to the one used by the Luxmans I am also simulating. The Luxman curves and specs were better on the Luxmans.
At the time I was intending to modify a Pioneer receiver I had that didn't sound 'interesting". By that I mean it didn't make you want to continue listening to music, which I think is an important subjective reaction that people should pay attention to.
That was a surround receiver, which I could compare to my DIY main amplifier, same power, made from a design by Graham Nalty. That amplifier did sound exciting, completely natural.
Well, at the time I was curious about this Doug Self design that attacked distortion in several areas.
I am very familiar with the Marantz MA-700, and it seemed to be a good option to the Luxmans I was simulating.
As I never listened to an MA-700, I looked at some reviews of it. Comments found it to be a good sounding amp, even if a little aggressive.
I noticed you increased the output stage DC power considerably, from +/-44v to +/-65v, which I think is a good move. You also used two transistors for the bias, instead of the usual Darlington Marantz puts there.
Marantz recommends 37mA bias adjustment on each emitter resistor, but I'm not sure that's what you used on your simulation. I say that because I could not use your bias pot on my sim, so I just changed bias resistor values.
What I watched is dissipation on each output transistor. For 3.75W I got 0.00062% THD @ 10KHz 100W @ 8 ohms. Which I think is the output you aimed at.
But for that I had to increase bias to 58mA on emitter resistors, which is above what Marantz suggested.
Bias adjustment should be tuned using a scope, of course, even if I guess Marantz used that 37mA setting because of the heatsink they used on the MA-700 monoblocks.
The Marantz MA-700 architecture is quite similar to the one used by the Luxmans I am also simulating. The Luxman curves and specs were better on the Luxmans.
Carlos,
Simulation is only a math analises that gives you references of your work. MA-700 have a microprocessor that changes the transformer tap to bigger voltages. It is a compact and low cost design so you got well the trick.
Here I sent you and all a DIY Chinese Design using the aproach of your thread. Have a look in results. The diference is a well designed versus a ... you know
Simulation is only a math analises that gives you references of your work. MA-700 have a microprocessor that changes the transformer tap to bigger voltages. It is a compact and low cost design so you got well the trick.
Here I sent you and all a DIY Chinese Design using the aproach of your thread. Have a look in results. The diference is a well designed versus a ... you know
And if you apply more changes you get it working well in simulation.
Real world is a little more complicated.
You do not have equal transistors - simulation uses same model so you have perfect matched devices.
My basic simulation file do not show assembly errors or other issues that happens in real world so it is a little easy to get very low values if you do the right think.
By the way: Thank you so much to Mooly that make a very good job introducing LTSpice to all forum user.
Ronaldo
Real world is a little more complicated.
You do not have equal transistors - simulation uses same model so you have perfect matched devices.
My basic simulation file do not show assembly errors or other issues that happens in real world so it is a little easy to get very low values if you do the right think.
By the way: Thank you so much to Mooly that make a very good job introducing LTSpice to all forum user.
Ronaldo
Yes, I have seen that MX50 amp kit selling on eBay, and I believe there's a thread here on the forum of people who built it.
AFAIK, most if not all eBay amp kits threads I have read here show people not being too satisfied with the results they get.
I am not sure to which thread of mine the MX50 approaches to. Certainly not the Self either, and I was just curious about it.
You are certainly right that a simulation is just that, and presupposes that all transistors are matched and all that, particularly the LTP input pair.
AFAIK, most if not all eBay amp kits threads I have read here show people not being too satisfied with the results they get.
I am not sure to which thread of mine the MX50 approaches to. Certainly not the Self either, and I was just curious about it.
You are certainly right that a simulation is just that, and presupposes that all transistors are matched and all that, particularly the LTP input pair.
Carlos,
The Self book schematics are only for reference to illustrate the book text. Normally they have issues.
Lots of issues appears in amplifier saturation and square wave amplifier answer both in simulator and real world. My template is arranged to test it and also to use a speaker model as output load.
I normally use to test commercial products that use to be more reliable than circuit of books. Lots of theses amplifier do not gives low distortion value.
The Bob Cordell Book is very interesting and useful to read. The Leach book is fantastic to all that loves to calculate amplifier parameters.
Regards
Ronaldo
The Self book schematics are only for reference to illustrate the book text. Normally they have issues.
Lots of issues appears in amplifier saturation and square wave amplifier answer both in simulator and real world. My template is arranged to test it and also to use a speaker model as output load.
I normally use to test commercial products that use to be more reliable than circuit of books. Lots of theses amplifier do not gives low distortion value.
The Bob Cordell Book is very interesting and useful to read. The Leach book is fantastic to all that loves to calculate amplifier parameters.
Regards
Ronaldo
The so Called Self arrangement or topology is:
First stage with a differential pair as main gain device, excited by a Constant Current Source (CCS) with a current mirror load. Can be used resistive degeneration in the differential pair or transistor cascode on each arm as variation.
Second Stage using an active device loaded by current load. In this stage is applied Miller compensation and also can be applied another current gain stage. Yamaha simulation file shows a good alternative.
Both Current sources for first and second stages can be tied together in lots of ways. Self show his choice, but there are other options.
I hope to clarify my point of view to Carlos and also all.
First stage with a differential pair as main gain device, excited by a Constant Current Source (CCS) with a current mirror load. Can be used resistive degeneration in the differential pair or transistor cascode on each arm as variation.
Second Stage using an active device loaded by current load. In this stage is applied Miller compensation and also can be applied another current gain stage. Yamaha simulation file shows a good alternative.
Both Current sources for first and second stages can be tied together in lots of ways. Self show his choice, but there are other options.
I hope to clarify my point of view to Carlos and also all.
Last edited:
- Status
- Not open for further replies.