Most compensation caps usually limit the top frequency (bandwith/slew rate), hence the PRICE to pay
. I found that Pavel Dudek is fond of using RC filters in several part of the amplifier. Other than that, look at the LTP of attached 300V/us circuit.I think that you designers must use other compensation techniques beyond my knowledge, not just LP filter. May be by trying to watch out for phase shift in every stage, or to tailor a "flowing" slew rate from the first stage to the second stage and so on till the last stage. I don't even know what I'm talking about
Attachments
May be you think that you can hear the high THD in the tube amp and your friend cannot. But... what if he can hear the nasty distortion in the tube amp as well as you do but he also can hear something in the solid state that you cannot. May be. Might be. This is not about you and your friend, just an analogy of something that may happen between two people with different preference.
Let me give a subtle "fact"...
Knowledgeable diyers may spend their time (30%) playing with simulations. But un-knowledgeable diyers may spend their time (70%) trying to listen to the different sound between kiwame, riken and caddock. I have to say that it is not strange or illogical if the later group often have more "trained" ears.
And I DOUBT that there will be anyone out of 15 blind test fanatics around here who cannot differentiate a Pioneer from a Halcro
I have been waiting for lateral amp with the topology used by Goldmund for so long. Yes, I drew my "future" amps, something that I would build if I can design one. Wahab's Luxman58? Keep the output stage but change the input stage! Your CX input/voltage/driver stage plus FET output? I would love to build it if you show a working schematic! Your original CX BJT? From looking at the schematics, I'm more interested to build the Mirrand!
If you want to be objective, you have to accept that there are different "preferences", and then do accordingly. How?
If you want to beat a Marshall tube amp that is claimed to be the best sounding amp in the world, build an amp that sound better in terms used or agreed by the tube lovers, don't build a different amp and say that an orange is better than an apple.
In short, please finalise the CXFET
Last edited:
Revisit first thread
I have spent much time and effort on the Goldmund schematic as it appears in the other thread and make the following conclusions. If some have other views please feel free to comment. These are my views only.
The opening thread states as follows:
This thread is to discuss improvements in performance, stability, and reliability.
Among the issues which were brought up:
1: The circuit can be unstable:- This is speculation. Any circuit can be made unstable and there is nothing to suggest that the Goldmund is less stable than any other amp provided it is loaded as specified.
2: Output stage bias has no clear adjustment:- Replace the bias resistor with a trim pot of 1K and bias is adjustable.
3: The topology can be improved:- I think a number of alternative schematics were offered, whether these would be perceived to sound better or worse is debatable.
4: The front end CCS is strange, it is hard to know what it was designed for. A more conventional CCS will be less confusing and more flexible."]Regardless if it is strange or confusing, it obviously works. If anyone is uncomfortable with Goldmund's CCS then replace it with any of those discussed earlier in the thread. Whether this impacts on the perceived sound is hard to tell by just looking at a simulation.
5: Front end rails may have sub-optimal power supply: - They incorporate a separate source with capacitance multiplier, nothing sub-optimal about that.
6: Transistors used don't have adequate voltage specs for the rails: - As far as I can see all the transistors are adequately rated in the application that they are used. Unless for some catastrophic condition all the semiconductors will operate properly.
Several options have been offered to enhance the amplifier using similar topologies. Some have suggested other topologies as well. Kean, it is your thread so concoct one from all the replies that you feel is the best and then make a project of it.
I have spent much time and effort on the Goldmund schematic as it appears in the other thread and make the following conclusions. If some have other views please feel free to comment. These are my views only.
The opening thread states as follows:
This thread is to discuss improvements in performance, stability, and reliability.
Among the issues which were brought up:
1: The circuit can be unstable:- This is speculation. Any circuit can be made unstable and there is nothing to suggest that the Goldmund is less stable than any other amp provided it is loaded as specified.
2: Output stage bias has no clear adjustment:- Replace the bias resistor with a trim pot of 1K and bias is adjustable.
3: The topology can be improved:- I think a number of alternative schematics were offered, whether these would be perceived to sound better or worse is debatable.
4: The front end CCS is strange, it is hard to know what it was designed for. A more conventional CCS will be less confusing and more flexible."]Regardless if it is strange or confusing, it obviously works. If anyone is uncomfortable with Goldmund's CCS then replace it with any of those discussed earlier in the thread. Whether this impacts on the perceived sound is hard to tell by just looking at a simulation.
5: Front end rails may have sub-optimal power supply: - They incorporate a separate source with capacitance multiplier, nothing sub-optimal about that.
6: Transistors used don't have adequate voltage specs for the rails: - As far as I can see all the transistors are adequately rated in the application that they are used. Unless for some catastrophic condition all the semiconductors will operate properly.
Several options have been offered to enhance the amplifier using similar topologies. Some have suggested other topologies as well. Kean, it is your thread so concoct one from all the replies that you feel is the best and then make a project of it.
Nico Ras:
In response to just two of your comments...
1. The circuit itself may be okay, but the PRINTED CIRCUIT board proposed in the other thread is an unknown... IMHO it aint what I would build, with or with out a ground plane.
2. I would never ever build that amp with the output transistors of that rating, due to the complex loads it will be subjected to AND the unknowness of the PCB used to build it on.
In addition to stability I think RELIABILITY should be a big part of the equation.
I also feel a good candidate should be built and played with listen to and hopefully that will be soon. BUT that aspect should not be RUSHED either.
In response to just two of your comments...
1. The circuit itself may be okay, but the PRINTED CIRCUIT board proposed in the other thread is an unknown... IMHO it aint what I would build, with or with out a ground plane.
2. I would never ever build that amp with the output transistors of that rating, due to the complex loads it will be subjected to AND the unknowness of the PCB used to build it on.
In addition to stability I think RELIABILITY should be a big part of the equation.
I also feel a good candidate should be built and played with listen to and hopefully that will be soon. BUT that aspect should not be RUSHED either.
I concur. PCB layout can only be proven in practice. You may be right to suspect that Goldmund discontinued the product based on its reliability issues, we don't know.
In a commercial environment stability and reliability is of major concern. You would lose market place if you produced a junk expensive product unless the product is purely a status symbol.
BMW has recalled 35 000 units because of probably brake problems. There are rumors Mercedes has steering problems.
Nothing in this world is guaranteed unless you have huge samples and data gathered over time, which is a luxury DIY do not really have.
Give you an example, one of our products will be 50 years old next year. The actual design has never changed, but its reliability has deteriorated over the past ten years simply because component manufacturers are cutting corners and their products are becoming critically close to the minimum requirement to remain competitive.
In a commercial environment stability and reliability is of major concern. You would lose market place if you produced a junk expensive product unless the product is purely a status symbol.
BMW has recalled 35 000 units because of probably brake problems. There are rumors Mercedes has steering problems.
Nothing in this world is guaranteed unless you have huge samples and data gathered over time, which is a luxury DIY do not really have.
Give you an example, one of our products will be 50 years old next year. The actual design has never changed, but its reliability has deteriorated over the past ten years simply because component manufacturers are cutting corners and their products are becoming critically close to the minimum requirement to remain competitive.
Krisf, If you look at my amp virtually identical topology, it is a slight over design, four devices per rail and 65V max, rated 200 watt, why? I felt this would be okay, more reliable but not infaluable.
Goldmund uses three devices, 80V, why? They must have made a commercial decision that the number of failures that they will encounter under warranty would not eat into their margins and that they can afford it judging to the selling price.
Maybe Kean should consider adding more devices, those who want to can fit more, those comfortable with less can do so. I would like to add anyone who plays music continuously at 300 watts has rocks in his head.
Goldmund uses three devices, 80V, why? They must have made a commercial decision that the number of failures that they will encounter under warranty would not eat into their margins and that they can afford it judging to the selling price.
Maybe Kean should consider adding more devices, those who want to can fit more, those comfortable with less can do so. I would like to add anyone who plays music continuously at 300 watts has rocks in his head.
Hi Nico,
Regarding stability of the Goldmund I have tested several samples, granted from a few years ago and all showed signs of instability into a capacitive load. I have compared them to the Thresholds, Levinson and Audio Research units of the same time period, none of which exhibited this behavior. Most people don't realize that the amp can be helped by attaching another Zobel at the speaker end which usually improves it's sound considerably.
Speaker cable choice used to be a problem such also was the case with Spectral amplifiers.
My main gripe is , what did you get for this elevated price and my conclusion is....a fancy chassis. A timex keeps better time than a Rolex, imagine that but a Rolex looks nicer and hence..............
Jam
Regarding stability of the Goldmund I have tested several samples, granted from a few years ago and all showed signs of instability into a capacitive load. I have compared them to the Thresholds, Levinson and Audio Research units of the same time period, none of which exhibited this behavior. Most people don't realize that the amp can be helped by attaching another Zobel at the speaker end which usually improves it's sound considerably.
Speaker cable choice used to be a problem such also was the case with Spectral amplifiers.
My main gripe is , what did you get for this elevated price and my conclusion is....a fancy chassis. A timex keeps better time than a Rolex, imagine that but a Rolex looks nicer and hence..............
Jam
Attachments
Last edited:
Hi Jam, it seems too simple but place a 47 pF capacitor across the feedback resistor, it is then stable, but you forfeit some volts in slew rate. Toss a coin, what is more important to perceived sound; infinite slew or infinite stability - I would like the coin to land on its side. But if it was my call adequate slew rate high stability.
The problem with the designer is probably that he was hooked on slew rate being the highest regard to sound quality, like some of us are hooked on THD or phase or what not.
Unfortunately his design landed up here where nothing is sacred and everyone is fair game for hunting down.
The problem with the designer is probably that he was hooked on slew rate being the highest regard to sound quality, like some of us are hooked on THD or phase or what not.
Unfortunately his design landed up here where nothing is sacred and everyone is fair game for hunting down.
Last edited:
There was a good article about that in Jan's bookzine Linear Audio Vol 0.
It might be nice to try it out on this design.
It is also very easy to make it an optional mod on the PCB. So you can either put it in or leave it out.
Well say, Nico.
While I don't have the background to fully understand all the technical aspects, I follow both thread and hope I will be able to build one some one. Being purely a builder without any electronic background and lack of testing equipment, the very last thing that I want to tackle is instability.
There are far too many personal preferences, it is impossible to have one size fits all. I can only hope that this thread will reach a point that I can build one and draw my own conclusion whether this indeed sounds good to me. Whilel I am at it, I can express the
2 major reasons that I don't want to participate the GB:
1. There are too many expensive and discontinued parts.
** I hope this thread will find some good alternatives.
2. No offense to Alex who did a good job to build the PCB, but it is untested.
Don't ask me how I know, a bad PCB can cause endless grief.
** If this thread will end up with a final product/PCB, I have no doubt it will be
tested before mass production.
My 2 cents and kudos to Kean to start this thread as well as those have been contributing numerous ideas.
While I don't have the background to fully understand all the technical aspects, I follow both thread and hope I will be able to build one some one. Being purely a builder without any electronic background and lack of testing equipment, the very last thing that I want to tackle is instability.
There are far too many personal preferences, it is impossible to have one size fits all. I can only hope that this thread will reach a point that I can build one and draw my own conclusion whether this indeed sounds good to me. Whilel I am at it, I can express the
2 major reasons that I don't want to participate the GB:
1. There are too many expensive and discontinued parts.
** I hope this thread will find some good alternatives.
2. No offense to Alex who did a good job to build the PCB, but it is untested.
Don't ask me how I know, a bad PCB can cause endless grief.
** If this thread will end up with a final product/PCB, I have no doubt it will be
tested before mass production.
My 2 cents and kudos to Kean to start this thread as well as those have been contributing numerous ideas.
There are different loudspeaker THD metrics which quantify some parameters of importance based on psychoacoustics and there is straight THD reported like in amps. If we talk electrodynamic speakers sinking 200W+ amps like the one discussed here, supposed to play loud, say around 100dB SPL median, the THD can be even more than 1-3%. In the lowest lows 10-20% even. I checked the Yamaha sound reinforcement handbook says, many over 1%. Those are large motor horn loaded pro stuff with much more control and less travel than hi-fi at same levels, they cruise there. A wiki article on audio measurements says ''Essentially all loudspeakers produce more distortion than electronics, and 1–5% distortion is not unheard of at moderately loud listening levels''. Audio system measurements - Wikipedia, the free encyclopedia
But they all argue that the nature of loudspeaker THD is not like in amps, Geddes even says its largely irrelevant. We can hear amps through them.
Attachments
Where to now oh Great One?
Attached are the mods to satisfy your requirements.
The circuit mods incorporates a feed back cap in parallel with the 10K NFB resistor. Unfortunately you cannot bandwidth limit and have high slew rate, they are related.
Slew rate is still >40V/uS and I think this is adequate for 300K bandwidth. Phase change over the bandwidth is very acceptable. 1.5 deg at 10 kHz and 17 deg at 100kHz, better than most amps around.
I have added the bias adjustment pot.
The feedback NFB cap brings 0 dB gain to within the phase inversion thus it will now be more stable for 20 uF in series with 2 ohm loads much worse than several parallel connected electrostatic speakers.
Distortion was improved a little by shorting the ac path between the input Diff amp legs. Thus degeneration does not worsen the THD but remains dc stable.
I really think that Nagy's got the power supply wrong, so I adjusted the driver PSU for 80V and output at 65V, makes more sense and there is not much difference between the Goldmund original output voltage swing with the modified one one.
I am not sure why they used a reverse biased transistor in the CCS, but it seems to work for a wide temperature range and I left it alone.
The zeners and diodes are removed they have no purpose because these lateral Mosfets already has them built into the die.
So Kean buddy - where do we go from here?
Attached are the mods to satisfy your requirements.
The circuit mods incorporates a feed back cap in parallel with the 10K NFB resistor. Unfortunately you cannot bandwidth limit and have high slew rate, they are related.
Slew rate is still >40V/uS and I think this is adequate for 300K bandwidth. Phase change over the bandwidth is very acceptable. 1.5 deg at 10 kHz and 17 deg at 100kHz, better than most amps around.
I have added the bias adjustment pot.
The feedback NFB cap brings 0 dB gain to within the phase inversion thus it will now be more stable for 20 uF in series with 2 ohm loads much worse than several parallel connected electrostatic speakers.
Distortion was improved a little by shorting the ac path between the input Diff amp legs. Thus degeneration does not worsen the THD but remains dc stable.
I really think that Nagy's got the power supply wrong, so I adjusted the driver PSU for 80V and output at 65V, makes more sense and there is not much difference between the Goldmund original output voltage swing with the modified one one.
I am not sure why they used a reverse biased transistor in the CCS, but it seems to work for a wide temperature range and I left it alone.
The zeners and diodes are removed they have no purpose because these lateral Mosfets already has them built into the die.
So Kean buddy - where do we go from here?
Where is the TOTAL schematic? Power supply for the front end for example.
Can the schematic you posted, VAS drive 4 output pair with out major changes to the circuit?
YES it is time to get out the arc welders, got some soldering to do. The lump of copper at the end of the stick is getting oxidized....
Can the schematic you posted, VAS drive 4 output pair with out major changes to the circuit?
YES it is time to get out the arc welders, got some soldering to do. The lump of copper at the end of the stick is getting oxidized....
