I think the big deal is perhaps 34 dB feedback could sound worse than 28 dB or whatever. Whereas 22 dB might not control crossover distortion enough. Some local feedback can arrive at an excellent book specification whilst making that intermediate stage easier to drive. One sometimes underestimates how bad high damping factors can sound. The cure is simple. Fit a small resistor ( 0R22, I seem to rememberyou do Mr D ). Being serrious I don't really mean that. What I mean is the pursuit of high damping factors is somewhat questionable if that's the prime goal. In fact some looseness is heard as better bass as long as not gross. Damping factor > 8 should be OK with the hose pipe cables most use ( not me ).
What Michael said was this " Suppose an amplifier might be 3 or 4 stages and what you really want is 3.6 ". Exactly.
I was somewhat surprised to see Tracy Chapman as an unworthy test material. To be frank all and any music is ideal. My hi fi I have been making will play anything. Bad things still need the best.
Some years ago when at a very bad time in my life I went with Colleen to Madrid. This was where Colleen couldn't help herself and showed her true colours. She knows plenty about plenty. She is an anti-snob usually.
In the art galleriers I had a big surprise. Constable, Mondrian and Picasso. Constable face to face is just wonderful. Mondrian to me is better in art books. Picasso was a shock. I really like it and don't know why. I was moderately OK with it when books. Blind listening tests don't mean much. It's all an art book when hi fi.
I have develloped more of an understanding about art. Caravaggio and Valasquez come to mind. Colleen fooled me to spend 6 Euro to go into a cathedral in Malta. I only did it to shut her up. She fooled me into seeing a very large collection of Caravaggio. Wow is all I can say.
Some years ago when at a very bad time in my life I went with Colleen to Madrid. This was where Colleen couldn't help herself and showed her true colours. She knows plenty about plenty. She is an anti-snob usually.
In the art galleriers I had a big surprise. Constable, Mondrian and Picasso. Constable face to face is just wonderful. Mondrian to me is better in art books. Picasso was a shock. I really like it and don't know why. I was moderately OK with it when books. Blind listening tests don't mean much. It's all an art book when hi fi.
I have develloped more of an understanding about art. Caravaggio and Valasquez come to mind. Colleen fooled me to spend 6 Euro to go into a cathedral in Malta. I only did it to shut her up. She fooled me into seeing a very large collection of Caravaggio. Wow is all I can say.
I remember a text in the sales literature for the reVox A78 integrated amp. They flatly said that with a damping factor of 20 dB the amp did what it could with the speaker's Q factor and pushing for more as a design goal was actually meaningless. That amp was a bit like Quad designs. measurement results were mediocre even for those times, but it sounded better than many of its competitors with better nominal specs.
What never fails to save my hide is my insistence on very large output currents. Lie it or not, for that you have to use paralleled pairs of output devices, and this also helps improve their natural damping factor.
What never fails to save my hide is my insistence on very large output currents. Lie it or not, for that you have to use paralleled pairs of output devices, and this also helps improve their natural damping factor.
That's even more true when Audio MOSFET's ( BUZ 900/905 ). The typical Ron of a MOSFET is about 1R5. If three devices used it will be more like 0R3 as the devices are not having to work so hard when paralelled. Buying matched gate sets means they will share and not hog. If buying 25 from Exicon they come in these sets.
A misconception of many is the Cgd of nearing 1nF will prevent a 8 mA VAS driving them. A bit of stydy will show that capaitance is bootstrapped to the load and not 0V. The seldom quoted Cgd is important. None the less the standard VAS will drive them. At this point a set of FET's should start to have some kick. The choice is then how much bias. It can be done by ear and heatsink temperture alone. 20 mA per pair is OK and 100 mA is about the upper limit. Mostly this sets the class A to B changover point. Being that FET's are untidy from A to B it can be best higher up. That moves the Ron up so some kick is lost. If in critical bias AB the Bipolar will be best. If we place the distortion at about 2 watts most will never hear the distortion compromise. In fact very like a tube amp with kick. The distortion either side of that point will be like very expensive op amps considering the voltage swing possible. As far as I know MOS FET amps do not show slewing distortion when using modest driving stages. The reason being MOSFET's are very fast and help the driver stage do it's job. The switch off times ( the ones that really matter ) are lightning fast. These amps can work with very high feedback. They sound nothing like a bipolar amp with high feedback. Most tube amps with feedback sound awful. If a very good transformer used and careful filtering results can be good. Dynaco comes to mind. I would say a tube feedback amp is the hardest amp to make. I mean loop feedback if asking.
A misconception of many is the Cgd of nearing 1nF will prevent a 8 mA VAS driving them. A bit of stydy will show that capaitance is bootstrapped to the load and not 0V. The seldom quoted Cgd is important. None the less the standard VAS will drive them. At this point a set of FET's should start to have some kick. The choice is then how much bias. It can be done by ear and heatsink temperture alone. 20 mA per pair is OK and 100 mA is about the upper limit. Mostly this sets the class A to B changover point. Being that FET's are untidy from A to B it can be best higher up. That moves the Ron up so some kick is lost. If in critical bias AB the Bipolar will be best. If we place the distortion at about 2 watts most will never hear the distortion compromise. In fact very like a tube amp with kick. The distortion either side of that point will be like very expensive op amps considering the voltage swing possible. As far as I know MOS FET amps do not show slewing distortion when using modest driving stages. The reason being MOSFET's are very fast and help the driver stage do it's job. The switch off times ( the ones that really matter ) are lightning fast. These amps can work with very high feedback. They sound nothing like a bipolar amp with high feedback. Most tube amps with feedback sound awful. If a very good transformer used and careful filtering results can be good. Dynaco comes to mind. I would say a tube feedback amp is the hardest amp to make. I mean loop feedback if asking.
Nigel,
pardon me for saying, but Ron is the source-drain resistance when fully switched on, hence R-on.
It's relevance is for switching MOSFET purposes, but of no consequence for audio output stages.
It's easy enough to calculate source-drain resistance for various conditions.
What is of importance for current hogging is the resistance spread between devices.
10 percent difference is of greater significance for a vertical MOSFET with a source-drain resistance level that differs by an order, than it is for a lateral MOSFET.
That is far a from the common view. I accept what you say as strictly that's true. If you think about it is matters nothing where a resistance is in a series circuit. If banging big currents ( kick ) it will matter. This is even more so with a FET as isn't two back to back diodes, the source doesn't start at the gate. The gate is a neck region. I think you might have been thinking this is damping factor? Mostly feedback would influence that. All the same the point you make is fair. Strange that all Audio FET makers talk about Ron. I suppose their pulsed output can be of use to some. Dejan and I were talking of banging out some power.
The hogging is mostly Vgs. If as you seem not to accept Ron was of some importance, to stop one device turning on sooner and hooging will matter. I only mention it as commonly multiple FET amps are shown with source resistors. These are not required if the Vgs is matched. Why somethings hogs matters nothing, it is if they do. It seems a shame with FET's to fit a source resistor. Doubtless in near constant power running it has proved to be useful. I have no interest in IRF540/9540 or the like. They seem the worst of all worlds. T0220 a prime reason.
A commonly accepted view is unmatched MOS FET's will reduce hogging due to internal resistance ( Ron ) raising with heat. This then forces another device to pass more current that is beside it despite it's Vcs being higher ( 1.2V 1.8 V perhaps ). This is fine up to a point. It will possibly makes three devices work as well as two matched devices. The absolute result will be a reduction in series loss between PSU and speaker.
If you would like to devellope this a bit please do. I might be learning something worth thinking about. In my bones what I have just said it just ohms law. Perhaps I have overlooked something. Looking back I did say enough to avoid confussion. Maybe I used terms with usually with very exact meanings. Hogging to me would be fair enough.
The hogging is mostly Vgs. If as you seem not to accept Ron was of some importance, to stop one device turning on sooner and hooging will matter. I only mention it as commonly multiple FET amps are shown with source resistors. These are not required if the Vgs is matched. Why somethings hogs matters nothing, it is if they do. It seems a shame with FET's to fit a source resistor. Doubtless in near constant power running it has proved to be useful. I have no interest in IRF540/9540 or the like. They seem the worst of all worlds. T0220 a prime reason.
A commonly accepted view is unmatched MOS FET's will reduce hogging due to internal resistance ( Ron ) raising with heat. This then forces another device to pass more current that is beside it despite it's Vcs being higher ( 1.2V 1.8 V perhaps ). This is fine up to a point. It will possibly makes three devices work as well as two matched devices. The absolute result will be a reduction in series loss between PSU and speaker.
If you would like to devellope this a bit please do. I might be learning something worth thinking about. In my bones what I have just said it just ohms law. Perhaps I have overlooked something. Looking back I did say enough to avoid confussion. Maybe I used terms with usually with very exact meanings. Hogging to me would be fair enough.
FET manufacturers also talk about Rds(on)max
Knock yourself out, download the datasheet of a Siemens BUZ90.
MOSFET in a TO-220 package, 6 pages of data with a dozen Rds(on) curves.
http://products.semelab-tt.com/pdf/ApplicationNoteAlfet.pdf
https://books.google.co.uk/books?id...IzAA#v=onepage&q=ron of audio mosfets&f=false
Ben Duncan offers an excellent overal answer. He does support the idea that feedback is the key element ( hope link 2 works ). I will have to add that to my thinking more than I did. He put's it forward another way. He like myself calls it I squared R losses. To put it simply 100 watts 8R 150 watts 4R is typical for 2 devices. Ron is about all we can refer to and must be related to real world losses. The 50 watts not seen at 4R must be in the device, it isn't 1R5. What I might have to imagine is that this can not be heard. I suspect it might, looks I might be wrong. Happy man if so because it was my only doubt about the devices ( big thanks ). Maybe if feedback is available it's my ears that have let me down. My friend John is very fond of his C-Audio amps. His living was PA work. The C-Audio too heavy to use these days. I wonder what he will have to say? C-Audio I seem to remeber is many FET's. He works building scientific devices so isn't given to hype. No money in music, these days just a hobby which pays for itself.
One thing I might have to revise is more bias sounds like Ron went up. From what you say and common sense maybe it didn't. That is weird as for all the world it sounds like not enough current when the sound is sweetest. The heat sinks are OK. Two possibilities. I instantaneous into 6R it far greater than I imagine. Or, lower distortion means it sounds less powerful. I suspect instantaneous power is higher. We are talking between 20mA and 120 mA bias. When 120 mA we are talking about 60C at the TO3 can top. 120 mA sounds really nice but lacks punch.
https://books.google.co.uk/books?id...IzAA#v=onepage&q=ron of audio mosfets&f=false
Ben Duncan offers an excellent overal answer. He does support the idea that feedback is the key element ( hope link 2 works ). I will have to add that to my thinking more than I did. He put's it forward another way. He like myself calls it I squared R losses. To put it simply 100 watts 8R 150 watts 4R is typical for 2 devices. Ron is about all we can refer to and must be related to real world losses. The 50 watts not seen at 4R must be in the device, it isn't 1R5. What I might have to imagine is that this can not be heard. I suspect it might, looks I might be wrong. Happy man if so because it was my only doubt about the devices ( big thanks ). Maybe if feedback is available it's my ears that have let me down. My friend John is very fond of his C-Audio amps. His living was PA work. The C-Audio too heavy to use these days. I wonder what he will have to say? C-Audio I seem to remeber is many FET's. He works building scientific devices so isn't given to hype. No money in music, these days just a hobby which pays for itself.
One thing I might have to revise is more bias sounds like Ron went up. From what you say and common sense maybe it didn't. That is weird as for all the world it sounds like not enough current when the sound is sweetest. The heat sinks are OK. Two possibilities. I instantaneous into 6R it far greater than I imagine. Or, lower distortion means it sounds less powerful. I suspect instantaneous power is higher. We are talking between 20mA and 120 mA bias. When 120 mA we are talking about 60C at the TO3 can top. 120 mA sounds really nice but lacks punch.
I looked at BUZ90. It has a much more useful spec sheet than many. I suspect there is no BUZ90P? The T0220 limitation I suspect is still there as 4A is it's rated current. As peak current is stated 4 A seems to be T0220 limitations. It reminds me of other FET's of the older type, FQA12N60. It has the more useful T03P pakage. They do have similar P types if 200 V is enough. I suspect they bias much like IRF540/9540.
FQA12N60 pdf, FQA12N60 description, FQA12N60 datasheets, FQA12N60 view ::: ALLDATASHEET :::
FQA12N60 pdf, FQA12N60 description, FQA12N60 datasheets, FQA12N60 view ::: ALLDATASHEET :::
You do not like the scent of freedom of thought?
Due to the pseudo CFA design ?
I used in my studio the Studer A68 at this time and I can confirm it was the best sounding amplifier i could found at this time and it remained my prefered amp during 10 years.
Ah Christophe, but the A68 is the biggie pro amp, rated at 100W/8 Ohms, using 3 series/prallel pairs of 250W TO-3 Motorola power devices. The old A78 was an integrated amp using just one pair of output devices, which were Nigel's favorite 2N3055/2N2955 types, from RCA if memory serves. Also, it preceeded the A68 by at least 12 years.
However, in terms of sound, I agree the A68 was a very good sounding amp indeed, and bult like a German Tiger tank as well. To harm it, you needed at least a sledge hammer.
However, in terms of sound, I agree the A68 was a very good sounding amp indeed, and bult like a German Tiger tank as well. To harm it, you needed at least a sledge hammer.
Last edited:
SIPMOS (Siemens Power MOS) is a development of the late '70s, no P-channels. Higher powered successor devices in SIPMOS have been on the market for a number of years, IRFBC40, no P-channels either.
For more Rds(on) curves, see BUZ23 (also SIPMOS, but looks a bit more like a vertical MOSFET)
Both BUZ23 and BUZ90 have been used as output stage devices of a couple of quasi-comp. amp designs, mostly German.
I'm curious, how was the comparing process done when you made that decision?
Note that just about any question of substance is deflected, not answered.
- Status
- Not open for further replies.