When and WHERE did you do THAT on this Thread?
😕
Sorry, I made a mistake, you did do so back at thread 256. BUT read what was discussed when you posted that... THANKS
Last edited:
seems like some people like to pile on and stir things up?
so i dont see a comittee anywhere. and how can people both accuse tis to be done via comittee as well as people being ignored? i call ** on this one... its still work in progress.
nico has shown a screen of a simulation that seems to indicate a problem. it hasnt been responded to by any of the people doing any of the circuits or other simulations. nor have they responded to any other post, so again i say there is no evidence of intent to ignore him.
the same goes for lazy cats observation.
i for one is happy that things are moving along and its great that there is an open and collabrative discussion. to say that this project is not inclusive or making progress is just disengenious in my opinion.
so i dont see a comittee anywhere. and how can people both accuse tis to be done via comittee as well as people being ignored? i call ** on this one... its still work in progress.
nico has shown a screen of a simulation that seems to indicate a problem. it hasnt been responded to by any of the people doing any of the circuits or other simulations. nor have they responded to any other post, so again i say there is no evidence of intent to ignore him.
the same goes for lazy cats observation.
i for one is happy that things are moving along and its great that there is an open and collabrative discussion. to say that this project is not inclusive or making progress is just disengenious in my opinion.
To my limited understanding, DiyAudio owns the thread. Kean started it.
If I were impatient with the progress, I wouldn't depose him (a few posts back attempted coup d'etat), I'd just go ahead and build it on my own.
As for my involvement, that is immaterial. I'm as free to post here as you are.
Mostly I just watch and try to learn a thing or two, and if I have some helpful input I may share it.
BTW, you might see a couple of non-sin binable comments that I made over there at the first thread. I was just trying to be helpful there too.
🙂
Nico,
Can you be a little more specific about the instability you stated in post 1071? Why is it unstable? What's wrong? Constructive criticism is fine, but useless comments don't help.
Rick
Can you be a little more specific about the instability you stated in post 1071? Why is it unstable? What's wrong? Constructive criticism is fine, but useless comments don't help.
Rick
Okay, The thread is evidently back up to a little over its former speed...
Firstly, you guys all come in the night when everyone's sleeping, make a HUGE tally of posts into the next page, and then after a little while wonder "hey where are the simulator guys?". Then you go all emo and gloomy and say things like "design by committee isn't working". Patience please, I don't ignore anyone on purpose, but perhaps if there is a flood of posts.
Firstly, Nico saw oscillation on his simulator. Well, I chose the stability compensation values to be the least that would give adequate stability, and if your models are different enough it might mean you need to adjust Cdom. However for phase margin to go from near 90 to below zero, there MUST be something significantly different between Nico's simulation and mine. Nico only gives sparing comments about his simulator setup, and rarely gives the full details. So just like previous times this has happened, we might have no chance to figure out what Nico is telling us, if we can't reproduce his results.
I understand the line of thinking that one doesn't necessarily have to build it to predict how it will behave. But let's build the prototype, and determine what needs to be done after we test it. Until we do that everything else is just a guess, even if an educated one, and I want to be SURE that the modifications I make is an improvement, and not just a bloat.
Another interesting point was made about VAS loading. It will reduce "oscillation and frequency related distortions". I translate this to mean that loading the VAS will reduce OLG which will slow down the amp to a manageable speed and increase OLG bandwidth. OLG bandwidth improves frequency related distortions. However I think the best way to do this is to increase LTP degeneration, not loading the VAS as the former will not increase the distortions of the LTP or VAS. So from an engineering standpoint, the solution suggested is to degrade the amp to make it manageable. I prefer to avoid this until we know it's necessary.
I agree with Lee Knatta on the grounding and stuff. That is how I would do my prototype. C8 and C9 are different because the loading at these points are different because of the parasitic capacitance of the FETs.
I can test versions with and without the drivers, but whether they are "necessary" could be a tricky topic. They help with the slew rate though.
Now as suggested, it looks like we need:
1: A series resistor to R12. (51R is suggested for max 12mA, I agree)
2: TMC optional (easy to implement really).
Since my proposal I have been offered assistance by no less than 4 (5?) people. One of them contacted me privately and is offering cash assistance. This is probably best since I can order parts from more places than Mouser. However he is sending it by mail, which means it can take several weeks to arrive.
Here is the parts list per board:
8x KSC1845
2x KSA1381
2x KSC3503
2SC4793
2SA1837
2x 10uF 120V lytic
100uF 100V lytic
3x 100n 150V film
330n 150V film
1.5u 150V film
6x 10n 100V film
2x 1u 150V film
suitable dual or matched jfet (surprise me I guess)
1x 8.6p COG cap
several values of COG caps between 10p and 68p to test as Cdom.
8x 100R resistors
10K resistor, 2W
56k resistor, 2W
12k resistor, 2W
33k resistor, 2W
4.7R resistor, 2W
100R 5 turn trimmer
1k 5 turn trimmer
20k 5 turn trimmer
I think I can handle the rest of the resistors. I have 2N5769 to use, but I added 2 extra KSC1845 to test out with.
Not sure what to do about the power supply. A small 9V transformer would work to boost the upper rails above the lower ones, and a lower power 80V transformer could be used for small power testing. However I can test at 60V or below with a transformer I have.
- keantoken
Firstly, you guys all come in the night when everyone's sleeping, make a HUGE tally of posts into the next page, and then after a little while wonder "hey where are the simulator guys?". Then you go all emo and gloomy and say things like "design by committee isn't working". Patience please, I don't ignore anyone on purpose, but perhaps if there is a flood of posts.
Firstly, Nico saw oscillation on his simulator. Well, I chose the stability compensation values to be the least that would give adequate stability, and if your models are different enough it might mean you need to adjust Cdom. However for phase margin to go from near 90 to below zero, there MUST be something significantly different between Nico's simulation and mine. Nico only gives sparing comments about his simulator setup, and rarely gives the full details. So just like previous times this has happened, we might have no chance to figure out what Nico is telling us, if we can't reproduce his results.
I understand the line of thinking that one doesn't necessarily have to build it to predict how it will behave. But let's build the prototype, and determine what needs to be done after we test it. Until we do that everything else is just a guess, even if an educated one, and I want to be SURE that the modifications I make is an improvement, and not just a bloat.
Another interesting point was made about VAS loading. It will reduce "oscillation and frequency related distortions". I translate this to mean that loading the VAS will reduce OLG which will slow down the amp to a manageable speed and increase OLG bandwidth. OLG bandwidth improves frequency related distortions. However I think the best way to do this is to increase LTP degeneration, not loading the VAS as the former will not increase the distortions of the LTP or VAS. So from an engineering standpoint, the solution suggested is to degrade the amp to make it manageable. I prefer to avoid this until we know it's necessary.
I agree with Lee Knatta on the grounding and stuff. That is how I would do my prototype. C8 and C9 are different because the loading at these points are different because of the parasitic capacitance of the FETs.
I can test versions with and without the drivers, but whether they are "necessary" could be a tricky topic. They help with the slew rate though.
Now as suggested, it looks like we need:
1: A series resistor to R12. (51R is suggested for max 12mA, I agree)
2: TMC optional (easy to implement really).
Since my proposal I have been offered assistance by no less than 4 (5?) people. One of them contacted me privately and is offering cash assistance. This is probably best since I can order parts from more places than Mouser. However he is sending it by mail, which means it can take several weeks to arrive.
Here is the parts list per board:
8x KSC1845
2x KSA1381
2x KSC3503
2SC4793
2SA1837
2x 10uF 120V lytic
100uF 100V lytic
3x 100n 150V film
330n 150V film
1.5u 150V film
6x 10n 100V film
2x 1u 150V film
suitable dual or matched jfet (surprise me I guess)
1x 8.6p COG cap
several values of COG caps between 10p and 68p to test as Cdom.
8x 100R resistors
10K resistor, 2W
56k resistor, 2W
12k resistor, 2W
33k resistor, 2W
4.7R resistor, 2W
100R 5 turn trimmer
1k 5 turn trimmer
20k 5 turn trimmer
I think I can handle the rest of the resistors. I have 2N5769 to use, but I added 2 extra KSC1845 to test out with.
Not sure what to do about the power supply. A small 9V transformer would work to boost the upper rails above the lower ones, and a lower power 80V transformer could be used for small power testing. However I can test at 60V or below with a transformer I have.
- keantoken
Last edited:
Hi guys, I don't want to criticise, but one thing is for sure, one does not go from simulation to PCB design and I have said this on many occasions and even more threads. What happened to the old proto-boards, where you plug in a components see if the idea is working and verify the simulator.
It was noteworthy to mention John's thread when he designed his amp, the thread "Patchworks" was actually one of the best I followed. Comments came from all over, he tried it some worked some did not and he gave constant feed-back as to the results and I think he ended up with an excellent amp, and I am sure that he uses it until this day.
I have also said that simulators offer a perfect environment, not a practical one. You cannot tweak a simulation to death and by considering the 9th digit in distortion or trying to just achieve one more nano volt rise time, you are entering the realm of instability, and that is exactly what you guys have done here.
A few hundred threads ago, I offered a working amplifier that was pretty close to the Goldmund, with some mods that made it an excellent contender, and worked. The PCBs where available and quite a number was built since its launch, (it will be replaced by a upgraded version) and hence me not minding sharing it. It simulated well but not perfect, it measures and sounds excellent.
I think the problem lies in if you did not design it. it cannot possibly be right and hence the thread went on and on and here we are, a few thousand posts and still no tangible results.
I am not in the mood to redesign the amp in post 1037, I already have a working one for several years. Like John and Andrew, I am here out of interest, nothing more.
The hint is to look at your front end very carefully, the other is to create lesser poles by throwing away most of the capacitors.
Kind regards
Nico
It was noteworthy to mention John's thread when he designed his amp, the thread "Patchworks" was actually one of the best I followed. Comments came from all over, he tried it some worked some did not and he gave constant feed-back as to the results and I think he ended up with an excellent amp, and I am sure that he uses it until this day.
I have also said that simulators offer a perfect environment, not a practical one. You cannot tweak a simulation to death and by considering the 9th digit in distortion or trying to just achieve one more nano volt rise time, you are entering the realm of instability, and that is exactly what you guys have done here.
A few hundred threads ago, I offered a working amplifier that was pretty close to the Goldmund, with some mods that made it an excellent contender, and worked. The PCBs where available and quite a number was built since its launch, (it will be replaced by a upgraded version) and hence me not minding sharing it. It simulated well but not perfect, it measures and sounds excellent.
I think the problem lies in if you did not design it. it cannot possibly be right and hence the thread went on and on and here we are, a few thousand posts and still no tangible results.
I am not in the mood to redesign the amp in post 1037, I already have a working one for several years. Like John and Andrew, I am here out of interest, nothing more.
The hint is to look at your front end very carefully, the other is to create lesser poles by throwing away most of the capacitors.
Kind regards
Nico
Last edited:
Thank you Nico 🙂
I did learn a lot and got some very well informed advice, much of it I couldn't understand at the time (some, even now). I did put it all to good use and the amp(s) is still going today.
Wether it s the models or something else, the schematic
in post 1038 doesn work in my simulator..
I ll look further later..
Besides, there s some strange things, as the feedback
resistor that will stand 3W dissipation at full power...
Connecting two current mirrors in serial fashion will bring
nothing, just useless complexity, if ever it works..
Also, the way the VAS is enanhced with baxandall super pairs
make this amp prone to instability, that beg ther question
where exactly is the compensation in this schematic...?.........
in post 1038 doesn work in my simulator..
I ll look further later..
Besides, there s some strange things, as the feedback
resistor that will stand 3W dissipation at full power...
Connecting two current mirrors in serial fashion will bring
nothing, just useless complexity, if ever it works..
Also, the way the VAS is enanhced with baxandall super pairs
make this amp prone to instability, that beg ther question
where exactly is the compensation in this schematic...?.........
Last edited:
Well, to improve stability why not get rid of the drivers and enhanced vas. My bet is that we have introduced too many poles in the design and massive compensation is not the answer neither is chasing the lowest distortion number. We are overthinking this design.
Jam
Jam
Last edited:
Wahab, the instability from the Baxandall pair comes from the connection between the driver's collector and the slave's emitter. In this case there is a 180 ohm resistor between the two, eliminating interaction. One of Yamaha's designs linked to earlier uses it the same way.
Once the prototypes are built we can test a lot of things.
It was never my intent to have to dumb this thing down with massive compensation. If that's necessary, then I've failed, or we need to look at the layout.
I increased the feedback resistors by a factor of 10 to lower dissipation.
Harharhar, it's coming along. It will make sense after I build it...
- keantoken
Once the prototypes are built we can test a lot of things.
It was never my intent to have to dumb this thing down with massive compensation. If that's necessary, then I've failed, or we need to look at the layout.
I increased the feedback resistors by a factor of 10 to lower dissipation.
Harharhar, it's coming along. It will make sense after I build it...
- keantoken
Just thinking, is there any way, knowing this stuff without a formal education in electronics, is there any way that my first job could be more than burger-flipping?
- keantoken
- keantoken
I tried simulation with/without TMC on schematic similar to keantoken, but with morre LTP tail current, and here are result.
At 0dB loop gain there is 102 phase shift.
It is simulation only, but for me it looks very stable.
dado
22pF only 10kHz
Fourier components of V(vin)
DC component:1.78813e-005
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 5.059e-01 1.000e+00 -0.01° 0.00°
2 2.000e+04 6.067e-08 1.199e-07 -175.53° -175.51°
3 3.000e+04 1.460e-07 2.886e-07 -89.97° -89.96°
4 4.000e+04 1.803e-08 3.564e-08 -151.75° -151.73°
5 5.000e+04 3.980e-08 7.866e-08 89.70° 89.71°
6 6.000e+04 1.333e-08 2.634e-08 -156.11° -156.09°
7 7.000e+04 9.605e-08 1.898e-07 90.41° 90.43°
8 8.000e+04 7.722e-09 1.526e-08 -164.74° -164.72°
9 9.000e+04 9.682e-08 1.914e-07 90.77° 90.79°
Total Harmonic Distortion: 0.000042%
Fourier components of V(vout)
DC component:-0.0148973
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 1.358e+01 1.000e+00 -1.14° 0.00°
2 2.000e+04 2.989e-05 2.201e-06 -34.49° -33.35°
3 3.000e+04 3.368e-04 2.480e-05 -89.98° -88.84°
4 4.000e+04 4.137e-05 3.047e-06 -151.83° -150.69°
5 5.000e+04 9.086e-05 6.691e-06 89.90° 91.04°
6 6.000e+04 3.063e-05 2.256e-06 -156.17° -155.03°
7 7.000e+04 2.193e-04 1.615e-05 90.60° 91.74°
8 8.000e+04 1.792e-05 1.320e-06 -164.86° -163.72°
9 9.000e+04 2.210e-04 1.628e-05 91.01° 92.15°
Total Harmonic Distortion: 0.003474%
22p/68p/680R 10kHz
Fourier components of V(vin)
DC component:1.78814e-005
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 5.059e-01 1.000e+00 -0.01° 0.00°
2 2.000e+04 7.710e-08 1.524e-07 173.95° 173.96°
3 3.000e+04 2.067e-09 4.086e-09 -94.37° -94.35°
4 4.000e+04 4.686e-10 9.262e-10 -148.86° -148.84°
5 5.000e+04 1.301e-09 2.570e-09 98.78° 98.79°
6 6.000e+04 3.696e-10 7.305e-10 -131.71° -131.69°
7 7.000e+04 3.381e-09 6.682e-09 113.77° 113.79°
8 8.000e+04 2.088e-10 4.127e-10 -113.14° -113.12°
9 9.000e+04 3.728e-09 7.368e-09 123.18° 123.19°
Total Harmonic Distortion: 0.000015%
Fourier components of V(vout)
DC component:-0.014897
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 1.358e+01 1.000e+00 -1.11° 0.00°
2 2.000e+04 1.755e-05 1.292e-06 134.68° 135.80°
3 3.000e+04 8.222e-06 6.055e-07 -94.70° -93.59°
4 4.000e+04 1.266e-06 9.321e-08 -154.45° -153.33°
5 5.000e+04 3.008e-06 2.215e-07 102.36° 103.48°
6 6.000e+04 1.005e-06 7.400e-08 -141.27° -140.16°
7 7.000e+04 7.830e-06 5.766e-07 115.63° 116.74°
8 8.000e+04 6.525e-07 4.805e-08 -137.96° -136.85°
9 9.000e+04 8.701e-06 6.407e-07 125.17° 126.29°
Total Harmonic Distortion: 0.000169%
At 0dB loop gain there is 102 phase shift.
It is simulation only, but for me it looks very stable.
dado
22pF only 10kHz
Fourier components of V(vin)
DC component:1.78813e-005
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 5.059e-01 1.000e+00 -0.01° 0.00°
2 2.000e+04 6.067e-08 1.199e-07 -175.53° -175.51°
3 3.000e+04 1.460e-07 2.886e-07 -89.97° -89.96°
4 4.000e+04 1.803e-08 3.564e-08 -151.75° -151.73°
5 5.000e+04 3.980e-08 7.866e-08 89.70° 89.71°
6 6.000e+04 1.333e-08 2.634e-08 -156.11° -156.09°
7 7.000e+04 9.605e-08 1.898e-07 90.41° 90.43°
8 8.000e+04 7.722e-09 1.526e-08 -164.74° -164.72°
9 9.000e+04 9.682e-08 1.914e-07 90.77° 90.79°
Total Harmonic Distortion: 0.000042%
Fourier components of V(vout)
DC component:-0.0148973
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 1.358e+01 1.000e+00 -1.14° 0.00°
2 2.000e+04 2.989e-05 2.201e-06 -34.49° -33.35°
3 3.000e+04 3.368e-04 2.480e-05 -89.98° -88.84°
4 4.000e+04 4.137e-05 3.047e-06 -151.83° -150.69°
5 5.000e+04 9.086e-05 6.691e-06 89.90° 91.04°
6 6.000e+04 3.063e-05 2.256e-06 -156.17° -155.03°
7 7.000e+04 2.193e-04 1.615e-05 90.60° 91.74°
8 8.000e+04 1.792e-05 1.320e-06 -164.86° -163.72°
9 9.000e+04 2.210e-04 1.628e-05 91.01° 92.15°
Total Harmonic Distortion: 0.003474%
22p/68p/680R 10kHz
Fourier components of V(vin)
DC component:1.78814e-005
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 5.059e-01 1.000e+00 -0.01° 0.00°
2 2.000e+04 7.710e-08 1.524e-07 173.95° 173.96°
3 3.000e+04 2.067e-09 4.086e-09 -94.37° -94.35°
4 4.000e+04 4.686e-10 9.262e-10 -148.86° -148.84°
5 5.000e+04 1.301e-09 2.570e-09 98.78° 98.79°
6 6.000e+04 3.696e-10 7.305e-10 -131.71° -131.69°
7 7.000e+04 3.381e-09 6.682e-09 113.77° 113.79°
8 8.000e+04 2.088e-10 4.127e-10 -113.14° -113.12°
9 9.000e+04 3.728e-09 7.368e-09 123.18° 123.19°
Total Harmonic Distortion: 0.000015%
Fourier components of V(vout)
DC component:-0.014897
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+04 1.358e+01 1.000e+00 -1.11° 0.00°
2 2.000e+04 1.755e-05 1.292e-06 134.68° 135.80°
3 3.000e+04 8.222e-06 6.055e-07 -94.70° -93.59°
4 4.000e+04 1.266e-06 9.321e-08 -154.45° -153.33°
5 5.000e+04 3.008e-06 2.215e-07 102.36° 103.48°
6 6.000e+04 1.005e-06 7.400e-08 -141.27° -140.16°
7 7.000e+04 7.830e-06 5.766e-07 115.63° 116.74°
8 8.000e+04 6.525e-07 4.805e-08 -137.96° -136.85°
9 9.000e+04 8.701e-06 6.407e-07 125.17° 126.29°
Total Harmonic Distortion: 0.000169%
Also, I have to amend the delivery time mentioned in post 1085. I've been told it should be 3-6 business days.
- keantoken
- keantoken
I agree with mace1337,
Make First the Original Goldmund clone “make it stable” and listening to that Amplifier.
Than build the one with the Goldmund Mods, than you can make the last one better in sound quality.
Sorry I cannot help you because I am not a electro engineer, just my 2 Euro cents 😉
The one that building and testing the Amplifier must have a compensating for the cost he make, when there is a group buy for the boards.
Rudy
Hi all,
firstly I think that the guy that is building the prototype should be one of the members that would like to have such an amp. He/she should be compensated by receiving two final boards for free.
In my opinion it is not that the circuit won't work at all, but that it would need some serious practical intervension to make it work properly.
Once working to expectation the schematic and PCB can be updated "as built" then the expensive components (MOSFETs) can be removed and reused in the upgraded boards, the rest of the components are worth peanuts.
Besides, the builder should be someone who can make a practical contrinbution and at least be familiar with fault finding a design. This is very different from someone who can fix something that is broken.
I have a few queistions that may shed some light on the problems, please do not take it as derogatory or defaming anyone's character.
Why chose such a high tail current, is it really necessary, you are forcing FT up, why? Play with the front-end I feel that 2mA tail current is quite sufficient. Also look at all the unnecessary poles that is created, every emitter resistor aplies NFB to that stage, now shunting it with a capacitor is another matter alltogether.
I honestly feel that the driver transistors does not do much good and just another stage that adds complications. Get rid of it and see if it was needed in the first place.
Lateral MOSFETs in their own right is quite fast, what does the driver do, make it faster, or overcomes the gate capacitance?
Then why place a series resistor in the gate. Does the theory apply equally well to lateral and vertical FETs, if you run a simulation of gate current you may be surprised at how little drive the lateral MOSFET requires.
Current mirrors can be a nightmare, see if you cannot get good specification and stability without them, use them as neccessary not because you can.
Why use FETs in the input, why not just transistors that can operate at the rail voltage and you can get rid of the cascode. If it is for lowering the noise then I disagree as the noise of the preceeding stage (before you power amplifier) will deterime the predominant noise factor of the entire system.
One can consider distortion in the same light, if the preceeding stage has 1% distortion, it is very unlikely that an amp with 0.001% distortion would be of any benefit.
The simplest amplifier normally is the easiest to linearise and also in my opinion sounds the best. Where you should pay the most attention is in the power supply. You can accpet this it is fact. A mediocre amplifier will sound good on a god power supply, but a good amplifier will sound mediocre on a mediocre power supply.
Please accept that my questions are not intended to point a finger at the problems for all I know there is no problem at all, it is merely intended to ask did you think out the system properly before embarking on improving the Goldmund. Throwing components at a problem does not necessarily solve it.
Go back to the Hitachi app note. Using that mediocre amp or the Goldmund circuit with the exact same power supply, I will be stumped if Nagys or anyone can actually tell which one is best, or even if there is a difference.
Kind regards
Nico
firstly I think that the guy that is building the prototype should be one of the members that would like to have such an amp. He/she should be compensated by receiving two final boards for free.
In my opinion it is not that the circuit won't work at all, but that it would need some serious practical intervension to make it work properly.
Once working to expectation the schematic and PCB can be updated "as built" then the expensive components (MOSFETs) can be removed and reused in the upgraded boards, the rest of the components are worth peanuts.
Besides, the builder should be someone who can make a practical contrinbution and at least be familiar with fault finding a design. This is very different from someone who can fix something that is broken.
I have a few queistions that may shed some light on the problems, please do not take it as derogatory or defaming anyone's character.
Why chose such a high tail current, is it really necessary, you are forcing FT up, why? Play with the front-end I feel that 2mA tail current is quite sufficient. Also look at all the unnecessary poles that is created, every emitter resistor aplies NFB to that stage, now shunting it with a capacitor is another matter alltogether.
I honestly feel that the driver transistors does not do much good and just another stage that adds complications. Get rid of it and see if it was needed in the first place.
Lateral MOSFETs in their own right is quite fast, what does the driver do, make it faster, or overcomes the gate capacitance?
Then why place a series resistor in the gate. Does the theory apply equally well to lateral and vertical FETs, if you run a simulation of gate current you may be surprised at how little drive the lateral MOSFET requires.
Current mirrors can be a nightmare, see if you cannot get good specification and stability without them, use them as neccessary not because you can.
Why use FETs in the input, why not just transistors that can operate at the rail voltage and you can get rid of the cascode. If it is for lowering the noise then I disagree as the noise of the preceeding stage (before you power amplifier) will deterime the predominant noise factor of the entire system.
One can consider distortion in the same light, if the preceeding stage has 1% distortion, it is very unlikely that an amp with 0.001% distortion would be of any benefit.
The simplest amplifier normally is the easiest to linearise and also in my opinion sounds the best. Where you should pay the most attention is in the power supply. You can accpet this it is fact. A mediocre amplifier will sound good on a god power supply, but a good amplifier will sound mediocre on a mediocre power supply.
Please accept that my questions are not intended to point a finger at the problems for all I know there is no problem at all, it is merely intended to ask did you think out the system properly before embarking on improving the Goldmund. Throwing components at a problem does not necessarily solve it.
Go back to the Hitachi app note. Using that mediocre amp or the Goldmund circuit with the exact same power supply, I will be stumped if Nagys or anyone can actually tell which one is best, or even if there is a difference.
Kind regards
Nico
Last edited:
Actually, the first Hitachi app note, back from 1978 or so, is no worse and even better than the goldmund ,provided we correct the tiny compensation, since it s on the wrong bjt.
Adding a complementary emitter follower to drive the laterals, it let the goldmund
in the dust..
Attachments
Last edited:
Good points Nico but I still like the idea of fet inputs because you could direct couple the amp anot have to deal with the base bias current. I don't think we need the drivers either.
Jam
Jam
Hi Jam, I would agree with you provided that the preceding stage or source has zero off-set.
Think about it if the source has a tiny off-set of just 1mV and the amp gain is 31 times then the off-set would only be 31 mV.
What if the input off-set was 100 mV? Maybe it is unimportant as many sources have capacitively coupled output.
Yes, a capacitor is said to introduce distortion, but has anyone quantified it.
Nico
Nico
I am uncomfortable with your OFFER...
You are offering a 2007 designed amplifier, that has been presently upgraded to the 2011 Feb edition.
You and or your company reached a conclusion that the 2007 design could be improved on. SO I presume that the 2001 is better and so forth.
The assembled group here is striving for THE BEST designed amp we can muster PERIOD, whether it be a improved version of the ST-120 from Dynaco or MisMash from Goldmucky. Maybe even the improved version of what you offered or are going to offer in 2007 or 2011.
I am failing in eyesight in real life sightly. I think I can see thru your offer and think that we can do better, THANKS, but NO THANKS. At least from my point of view.
Some have offered criticisms left and right up and down. But NO ONE has offered to help build. Some have said they look at what is going on, and offer IDEAS, BUT have no intentions of building it when it goes to Group Buy...
Positive is the mood and Positive are the thoughts...That is where I try to stay. I posted some thoughts on some other threads and I was trying to be helpful, at least in my own mind. Some POSITIVE suggestions, so said I.
Your lack of response to what were the particulars of your simulations puzzles me and others. I will wait till you decide to share those specificities with us.
I hope time will let this thread bear some Delirious Delicious fruit of love and labor...without the Criticisms and cheap shot critics that have NO intention of tasting the mentioned FRUIT.
Thanks
I am uncomfortable with your OFFER...
You are offering a 2007 designed amplifier, that has been presently upgraded to the 2011 Feb edition.
You and or your company reached a conclusion that the 2007 design could be improved on. SO I presume that the 2001 is better and so forth.
The assembled group here is striving for THE BEST designed amp we can muster PERIOD, whether it be a improved version of the ST-120 from Dynaco or MisMash from Goldmucky. Maybe even the improved version of what you offered or are going to offer in 2007 or 2011.
I am failing in eyesight in real life sightly. I think I can see thru your offer and think that we can do better, THANKS, but NO THANKS. At least from my point of view.
Some have offered criticisms left and right up and down. But NO ONE has offered to help build. Some have said they look at what is going on, and offer IDEAS, BUT have no intentions of building it when it goes to Group Buy...
Positive is the mood and Positive are the thoughts...That is where I try to stay. I posted some thoughts on some other threads and I was trying to be helpful, at least in my own mind. Some POSITIVE suggestions, so said I.
Your lack of response to what were the particulars of your simulations puzzles me and others. I will wait till you decide to share those specificities with us.
I hope time will let this thread bear some Delirious Delicious fruit of love and labor...without the Criticisms and cheap shot critics that have NO intention of tasting the mentioned FRUIT.
Thanks