Thank yougo for that bless...one for flower to our garden
One girl here is so rare as a "white fly"... some miracle from heaven.
Lets go guys!... shave yourself, control the words.
One more lady in our place.
My wish is that alá, Mahatma, Jesus, Mahomad, and all others bless us.
Iuuuuuhú....own Yes!!!!!
Carlos
One girl here is so rare as a "white fly"... some miracle from heaven.
Lets go guys!... shave yourself, control the words.
One more lady in our place.
My wish is that alá, Mahatma, Jesus, Mahomad, and all others bless us.
Iuuuuuhú....own Yes!!!!!
Carlos
My english is very poor Jam
If you use some different words and put dots...hehe...i turn confused....understood nothing...but must be good thing for a man from Charllotsville.
I feel very good seeing girls over here, in my place i do not find too much girls with that kind of interest.
And i like very much the feminine way... i really like them...not the secual way...... alike some complementary object of pleasure....not this way....i think they are wonderfull.. the capacity to forgive, the capacity to reproduce and create a family for us, their courage.
I love woman by a hundred of reasons.... and 99 have not related to sex... i really apreciatte them
My welcome has the intention of welcome her really, the second intention is to say that she is well come... and the third intention is to make clear that this tuns me happy.
Just simple this way.... i am happy, have 4 girls in my house, this is great....no beard monster talking strong here....just flowers in my own garden.
I used to make games, and things i talk have first, second, third intention...must think a little to see what i mean.
But this way...really... i mean....happy!!!!
regards,
Carlos
If you use some different words and put dots...hehe...i turn confused....understood nothing...but must be good thing for a man from Charllotsville.
I feel very good seeing girls over here, in my place i do not find too much girls with that kind of interest.
And i like very much the feminine way... i really like them...not the secual way...... alike some complementary object of pleasure....not this way....i think they are wonderfull.. the capacity to forgive, the capacity to reproduce and create a family for us, their courage.
I love woman by a hundred of reasons.... and 99 have not related to sex... i really apreciatte them
My welcome has the intention of welcome her really, the second intention is to say that she is well come... and the third intention is to make clear that this tuns me happy.
Just simple this way.... i am happy, have 4 girls in my house, this is great....no beard monster talking strong here....just flowers in my own garden.
I used to make games, and things i talk have first, second, third intention...must think a little to see what i mean.
But this way...really... i mean....happy!!!!
regards,
Carlos
Hi Susan,
Novel circuit - outstanding! Have you considered running a phase splitter tube for input voltage amplification? This would enable lower impedance drive to the gates, keeping input impedance high....... And would it be possible to drive the speaker from across the mosfet sources, eschewing the secondary of the transformer, using it only as a centre tapped choke?
I'm sure you considered this, but I'd be interested in your response.
Cheers,
Hugh
Novel circuit - outstanding! Have you considered running a phase splitter tube for input voltage amplification? This would enable lower impedance drive to the gates, keeping input impedance high....... And would it be possible to drive the speaker from across the mosfet sources, eschewing the secondary of the transformer, using it only as a centre tapped choke?
I'm sure you considered this, but I'd be interested in your response.
Cheers,
Hugh
New Thread
Hugh,
Susan's kind of busy since she started her new thread on the subject :
http://www.diyaudio.com/forums/showthread.php?s=&threadid=42259
Too many guys here and no girl. : )
Patrick
Hugh,
Susan's kind of busy since she started her new thread on the subject :
http://www.diyaudio.com/forums/showthread.php?s=&threadid=42259
Too many guys here and no girl. : )
Patrick
Hi Patrick (EUVL)
I feel somewhat parasitic because it was John (jcx) who started this thread with some genuinely useful simulations of the JLH output stage, but I do agree with you and Upupa Epops regarding the use of transformers along an audio path.
Hi Susan,
I like the simplicity of your input isolated design and you appear to have achieved a decent power capability, but it is my experience that there is a greater loss with bass power and image definition when using transformers (whatever the loudspeaker type) than when using large value series decoupling capacitors. I too have 600 ohm driven power amplifiers to overcome cable losses, and clearly your loudspeaker sited output transformer is driven at low impedance (which valves cannot easily do), but what about the input transformer - that would need to be rather special ?
Cheers ............ Graham.
I feel somewhat parasitic because it was John (jcx) who started this thread with some genuinely useful simulations of the JLH output stage, but I do agree with you and Upupa Epops regarding the use of transformers along an audio path.
Hi Susan,
I like the simplicity of your input isolated design and you appear to have achieved a decent power capability, but it is my experience that there is a greater loss with bass power and image definition when using transformers (whatever the loudspeaker type) than when using large value series decoupling capacitors. I too have 600 ohm driven power amplifiers to overcome cable losses, and clearly your loudspeaker sited output transformer is driven at low impedance (which valves cannot easily do), but what about the input transformer - that would need to be rather special ?
Cheers ............ Graham.
Hi everyone, thank you for your welcome (blushing slightly).
Thank you - the simplicity was of necessity.
I am not worried with higher levels of low bass as I believe that this should be handled by a seperate power amp and speaker to avoid the Doppler effects of a cone when operating as a piston not just vibratory mode.
I use 200 VA mains transformer cores size for 50 Watts peak power as I have determined that one should have 4 x VA per Watt.
In a domestic living environment (which here in the UK is a room of 14 x 16 feet) the bass is more than adequate - particually as being in appartments one wants to be able to listen to music at night without upsetting the neighbours.
Image definition is phenominal with my amplifier, it beats anything I have ever heared, at any price. Not only can one precisly locate the position of an instrument, but one can even hear the solo violinist swaying as she/he plays their instrument.
I can paly JMJ's Concerts in China and hear things that just don't resolve with my Quad 34-303 system.
The key to the output transformer is that it is quad filar wound in heavy gauge 0.8 mm enamelled wire. It is quite a different beast to a valve output transformer and took me two years to develop and understand as there is very little literature on the subject (for example things like inter-winding capacitance don't figure). At 1 watt the output transformers run from under 25 Hz to over 250 KHz.
I used 200VA mains transformer kits and removed the mains windings and rewound the bobbins. It is important to ensure that the laminations are M6 material, otherwise there is nothing special in the construction and I even stack the EI laminations alternaltivly rather than any special sequence.
http://www.susan-parker.co.uk/zeus-out-tx.htm
The input transformer was made for me by E A Sowter Ltd. as it has lots of turns of fine wire. It is a simple split bobbin with each half having one secondary (one ends up with two primaries, one on each half which are then wires in parrallel). The bobbin is wound as a (S/2)
P)S/2) (i.e. split secondary wound either side of the primary) with a 1:10 turns ratio.
Sowter designated it as part # 8160 and they recently quoted me UKP 38.25 (plus VAT) each ex-works for some more (they list specials they design for people, but they don’t charge a big up front design fee).
The imput transformer could be made more exotic by increasing the number of interleaves e.g. (S/4)P/2)S/2)P/2)S/4) but I wonder if the extra complexity is necessary.
However as a cheep alternative for proving the circuit it would be possible to use one of those low profile encapsulated PCB transformers of around 10 VA with two 120 primaries and two 12 volt secondaries connected backwards to try out the power stage.
e.g. http://www.airlinktransformers.com/lowprofile.html
10VA 0-12V,0-12V LP1012
Within the frequency bandwidth this should enable one to experiment with bias levels and measure distorson etc.
Graham, should you be in West London you are welcome to come round and bring a few DVD (movies) and CDs to listen to. My setup isn't special, just a good middle of the range Sony DVD player and a pre-amp box I made myself with a volume control and 600 ohm line drivers. Supper could be provided.
=========
Hi Hugh, reply in "Zero Feedback Impedance Amplifiers"
=========
Thank you for your comments. Any further posts I will make in reply will be on the seperate "Zero Feedback Impedance Amplifiers" thread.
===========
Best wishes,
Susan.
Graham Maynard said:
Thank you - the simplicity was of necessity.
I am not worried with higher levels of low bass as I believe that this should be handled by a seperate power amp and speaker to avoid the Doppler effects of a cone when operating as a piston not just vibratory mode.
I use 200 VA mains transformer cores size for 50 Watts peak power as I have determined that one should have 4 x VA per Watt.
In a domestic living environment (which here in the UK is a room of 14 x 16 feet) the bass is more than adequate - particually as being in appartments one wants to be able to listen to music at night without upsetting the neighbours.
Image definition is phenominal with my amplifier, it beats anything I have ever heared, at any price. Not only can one precisly locate the position of an instrument, but one can even hear the solo violinist swaying as she/he plays their instrument.
I can paly JMJ's Concerts in China and hear things that just don't resolve with my Quad 34-303 system.
The key to the output transformer is that it is quad filar wound in heavy gauge 0.8 mm enamelled wire. It is quite a different beast to a valve output transformer and took me two years to develop and understand as there is very little literature on the subject (for example things like inter-winding capacitance don't figure). At 1 watt the output transformers run from under 25 Hz to over 250 KHz.
I used 200VA mains transformer kits and removed the mains windings and rewound the bobbins. It is important to ensure that the laminations are M6 material, otherwise there is nothing special in the construction and I even stack the EI laminations alternaltivly rather than any special sequence.
http://www.susan-parker.co.uk/zeus-out-tx.htm
The input transformer was made for me by E A Sowter Ltd. as it has lots of turns of fine wire. It is a simple split bobbin with each half having one secondary (one ends up with two primaries, one on each half which are then wires in parrallel). The bobbin is wound as a (S/2)
P)S/2) (i.e. split secondary wound either side of the primary) with a 1:10 turns ratio. Sowter designated it as part # 8160 and they recently quoted me UKP 38.25 (plus VAT) each ex-works for some more (they list specials they design for people, but they don’t charge a big up front design fee).
The imput transformer could be made more exotic by increasing the number of interleaves e.g. (S/4)
P/2)S/2)P/2)S/4) but I wonder if the extra complexity is necessary.However as a cheep alternative for proving the circuit it would be possible to use one of those low profile encapsulated PCB transformers of around 10 VA with two 120 primaries and two 12 volt secondaries connected backwards to try out the power stage.
e.g. http://www.airlinktransformers.com/lowprofile.html
10VA 0-12V,0-12V LP1012
Within the frequency bandwidth this should enable one to experiment with bias levels and measure distorson etc.
Graham, should you be in West London you are welcome to come round and bring a few DVD (movies) and CDs to listen to. My setup isn't special, just a good middle of the range Sony DVD player and a pre-amp box I made myself with a volume control and 600 ohm line drivers. Supper could be provided.
=========
Hi Hugh, reply in "Zero Feedback Impedance Amplifiers"
=========
Thank you for your comments. Any further posts I will make in reply will be on the seperate "Zero Feedback Impedance Amplifiers" thread.
===========
Best wishes,
Susan.
Graham's class A: thermal distortion?
Hi Graham,
I only recently saw your set of articles in Electr. World,
and also didn't look in diyAudio for a while, so am joining in somewhat late...
I have several years of experience now in creating audio DACs and my own tube amplifiers, and am lately thinking agian on solid-state amplifiers. I admire your attempts in speeding-up the amplifier feedback loop: I have learned that the human ear seems unbelievably sensitive to timing problems.
In wondering why good tube amplifiers sound so much superior to their silicon counterparts, I believe that timing plays a strong role.
One other aspect that people brought up on potential flaws in solid-state amplifiers, in comparison with tubes, is thermal distortion. In particular, thermal distortion in the differential input stage of the amplifier, which is not attenuated by the feedback loop.
It seems to me that your (basic) schematics as published in EW sept.2004 is really sensitive to this phenomenon:
The input transistors run at relatively high current, and have significant voltage drop. As result, AC current variations would give rise to temperature fluctuations in these transistors, leading to (unplanned) Vbe voltage fluctuations.
Do you believe that such thermal distortion could noticably affect audio performance?
I guess this effect does not pop up in typical Spice simulations, as dynamic thermal effects are normally not modelled. It would probably require quite some effort to add this to the transistor model....
Of course, your schematic could be slightly updated to reduce thermal effects. Options are adding cascode transistors or jfets to reduce voltage drop (thus power) in the differential pair, or changing to a real single-die dual-transistor so that the power variations cancel out, and probably more options. Do you have an opinion regarding such modifications?
Looking forward to your feedback,
Jos van Eijndhoven
Hi Graham,
I only recently saw your set of articles in Electr. World,
and also didn't look in diyAudio for a while, so am joining in somewhat late...
I have several years of experience now in creating audio DACs and my own tube amplifiers, and am lately thinking agian on solid-state amplifiers. I admire your attempts in speeding-up the amplifier feedback loop: I have learned that the human ear seems unbelievably sensitive to timing problems.
In wondering why good tube amplifiers sound so much superior to their silicon counterparts, I believe that timing plays a strong role.
One other aspect that people brought up on potential flaws in solid-state amplifiers, in comparison with tubes, is thermal distortion. In particular, thermal distortion in the differential input stage of the amplifier, which is not attenuated by the feedback loop.
It seems to me that your (basic) schematics as published in EW sept.2004 is really sensitive to this phenomenon:
The input transistors run at relatively high current, and have significant voltage drop. As result, AC current variations would give rise to temperature fluctuations in these transistors, leading to (unplanned) Vbe voltage fluctuations.
Do you believe that such thermal distortion could noticably affect audio performance?
I guess this effect does not pop up in typical Spice simulations, as dynamic thermal effects are normally not modelled. It would probably require quite some effort to add this to the transistor model....
Of course, your schematic could be slightly updated to reduce thermal effects. Options are adding cascode transistors or jfets to reduce voltage drop (thus power) in the differential pair, or changing to a real single-die dual-transistor so that the power variations cancel out, and probably more options. Do you have an opinion regarding such modifications?
Looking forward to your feedback,
Jos van Eijndhoven
i believe thermal modulation in the input could be virtually eliminated with a bootstrapped cascode; ie drive the cascode with the input common mode voltage by using a divider with the same ratio as the feedback network between the output and some bias voltage
with near constant V on the input transistors and high loop gain reducing current modulation the power in the input transistors is nearly constant
in simulation bootstrapped cascode on the input diff pair cleans up some second harmonic from Ccb modulation once you get near -120 dB distortion - this nonlinearity is more pronounced with fet inputs
while spice doesn’t model coupled thermal behavior it is possible to add a thermal model and use controlled sources to partially simulate Vbe modulation - see fairchild's thermal modeling in power mosfets, also Cyril Batemann is using this modeling technique in recent WW/EW articles
thermal modulation should result in easily measured IM distortion with a very low frequency, high amplitude signal to cause the thermal modulation and a high frequency probe signal - in a test with a resistive load the low frequency signal could be subsonic
with near constant V on the input transistors and high loop gain reducing current modulation the power in the input transistors is nearly constant
in simulation bootstrapped cascode on the input diff pair cleans up some second harmonic from Ccb modulation once you get near -120 dB distortion - this nonlinearity is more pronounced with fet inputs
while spice doesn’t model coupled thermal behavior it is possible to add a thermal model and use controlled sources to partially simulate Vbe modulation - see fairchild's thermal modeling in power mosfets, also Cyril Batemann is using this modeling technique in recent WW/EW articles
thermal modulation should result in easily measured IM distortion with a very low frequency, high amplitude signal to cause the thermal modulation and a high frequency probe signal - in a test with a resistive load the low frequency signal could be subsonic
Hi Jos,
It is not just decreasing the propagation delay that improves reproduction, but minimising loudspeaker induced fractional output stage reverse commutation which leads to NFB loop induced fractional interstage current overshoot - - which is minimal in a JLH class-A.
There is nothing wrong with a silicon bipolar junction running hot, and yes the differential input stage can be part of a high frequency distortion mechanism. However, the input stage current variation in my amplifier circuit is a tiny fraction of its quiescent, and therefore I am at a loss to understand how thermal effects could arise. I am not aware that any differential stage effects are observable either, actually quite the opposite due to the way that high frequencies remain clear during simultaneous bass drive.
Vbe does not increase with Vce !
I would look forwards to seeing any improved circuit you can think of, but remember, if you add more devices you will introduce high frequency phase shift, which is presently not a real-world problem with the circuit. Also any stability compensation that becomes necessary will then affect performance because it will slow the amplifier down !
As it stands, this class-A circuit is excellent where high power is not required. It is also very simple to build. As JCX suggests, you would need to be thinking about THD levels below -100dB for improvement changes to be observable, and yet the effects of delayed loudspeaker back EMF NFB control can be much greater in non class-A circuits, so are sonic colourations being noted for their true cause ?
Cheers ......... Graham.
It is not just decreasing the propagation delay that improves reproduction, but minimising loudspeaker induced fractional output stage reverse commutation which leads to NFB loop induced fractional interstage current overshoot - - which is minimal in a JLH class-A.
There is nothing wrong with a silicon bipolar junction running hot, and yes the differential input stage can be part of a high frequency distortion mechanism. However, the input stage current variation in my amplifier circuit is a tiny fraction of its quiescent, and therefore I am at a loss to understand how thermal effects could arise. I am not aware that any differential stage effects are observable either, actually quite the opposite due to the way that high frequencies remain clear during simultaneous bass drive.
Vbe does not increase with Vce !
I would look forwards to seeing any improved circuit you can think of, but remember, if you add more devices you will introduce high frequency phase shift, which is presently not a real-world problem with the circuit. Also any stability compensation that becomes necessary will then affect performance because it will slow the amplifier down !
As it stands, this class-A circuit is excellent where high power is not required. It is also very simple to build. As JCX suggests, you would need to be thinking about THD levels below -100dB for improvement changes to be observable, and yet the effects of delayed loudspeaker back EMF NFB control can be much greater in non class-A circuits, so are sonic colourations being noted for their true cause ?
Cheers ......... Graham.
Hi jcx, Graham,
you apparently seem to feel that thermal distortion is probably not a significant effect.
OK, I don't know, so lets try some back-of-the-envelope calculations on a virtual experiment....
Suppose we start with input=0V, output=0V, thermally steady state, simple 8-ohm resistive load connected.
We provide a (bandwidth limited) input voltage step, leading to an 8V,1A output.
As the amplifier gain is about 14x, this input step is 0.56V.
This leads to the following current changes in the amplifier:
TR3Ic +0.5A, TR4Ic -0.5A.
HFE=150 for TR3,TR4, so changes TR2Ic -3.3mA.
HFE=150 for TR2, so TR2Ib -22uA.
This current change is created by the differential input:
TR1aIc -11uA, TR1bIc +11uA.
With about 25V on Tr1a,b:
TR1a -0.29mW, TR1b +0.29mW
The (Philips) datasheet on BC556 specifies 250K/W and Vbe -2mV/K
So temperature change is: TR1a -0.07K, TR1b +0.07K
and thermal voltage drift becomes: TR1a +0.14mV, TR1b -0.14mV
The induced offset of 0.28mV is 0.05% or -66dB of the 0.56V input step.
This is indeed extremely small, and is doubtful to be noticable.
However we are not there yet: music is not a DC step.....
The thermal warm-up speed is not given in the datasheet.
We would need something as xx K/s per Watt, or K/J, which I don't know.
Lets estimate a warming-up time constant of the BC556 die of 1 second. That corresponds to a thermal corner-frequency of 1/6 Hz.
For a 30Hz sinewave signal, the thermal variations would hence be attenuated with about -46dB.
For this 30Hz sinewave (and above 8 ohm load) the thermal effect would hence be -66dB -46dB = -112dB.
This calculation seems to indicate that this schematic indeed has neglectible thermal distortion from its input stage.
Do you think this reasoning is OK?
(just trying to understand the issue.....)
Greetings,
Jos
you apparently seem to feel that thermal distortion is probably not a significant effect.
OK, I don't know, so lets try some back-of-the-envelope calculations on a virtual experiment....
Suppose we start with input=0V, output=0V, thermally steady state, simple 8-ohm resistive load connected.
We provide a (bandwidth limited) input voltage step, leading to an 8V,1A output.
As the amplifier gain is about 14x, this input step is 0.56V.
This leads to the following current changes in the amplifier:
TR3Ic +0.5A, TR4Ic -0.5A.
HFE=150 for TR3,TR4, so changes TR2Ic -3.3mA.
HFE=150 for TR2, so TR2Ib -22uA.
This current change is created by the differential input:
TR1aIc -11uA, TR1bIc +11uA.
With about 25V on Tr1a,b:
TR1a -0.29mW, TR1b +0.29mW
The (Philips) datasheet on BC556 specifies 250K/W and Vbe -2mV/K
So temperature change is: TR1a -0.07K, TR1b +0.07K
and thermal voltage drift becomes: TR1a +0.14mV, TR1b -0.14mV
The induced offset of 0.28mV is 0.05% or -66dB of the 0.56V input step.
This is indeed extremely small, and is doubtful to be noticable.
However we are not there yet: music is not a DC step.....
The thermal warm-up speed is not given in the datasheet.
We would need something as xx K/s per Watt, or K/J, which I don't know.
Lets estimate a warming-up time constant of the BC556 die of 1 second. That corresponds to a thermal corner-frequency of 1/6 Hz.
For a 30Hz sinewave signal, the thermal variations would hence be attenuated with about -46dB.
For this 30Hz sinewave (and above 8 ohm load) the thermal effect would hence be -66dB -46dB = -112dB.
This calculation seems to indicate that this schematic indeed has neglectible thermal distortion from its input stage.
Do you think this reasoning is OK?
(just trying to understand the issue.....)
Greetings,
Jos
The approach seems ok, some older transistor data sheets had single pulse “equivalent thermal impedance” curves that could be used to fit a thermal model
there is actually some recent research into thermal modulation distortion at RF where thermal modulation throws off the predistortion approach that is increasingly used to reduce distortion
The resistor tail “current source” has an advantage from the thermal point of view; since nearly equal V is dropped across the resistor and the input Qs you maintain a near const power dissipation in the Qs for small input Vcommon changes – the resistor tail source does lead to increased 2nd harmonic distortion from other mechanisms compared to the usual active tail current source
The largish diff pair bias current in Graham’s design puts this power vs Vcommon compensation and thermal symmetry of the 2 input Qs to a rather more severe test than typical lower bias current – a small Vcommon to differential power dissipation conversion comes from the slight V difference imposed by the 2 Q current mirror’s 1 Vbe input drop vs the output stage’s 2 Vbe bias point (easily fixed with a extra dropping diode or a more symmetric 3 Q feedback current mirror)
You could take a lesson from instrumentation designers striving for low offset drift; the (well matched for Vbe and Hfe) diff pair Qs can be put in close thermal contact – gluing to-92s together with a wrapping of heavy gauge bare copper wire over the plastic bodies and paying attention to equalizing thermal terminations of the leads
But the basic conclusion that thermal modulation distortion in the input diff pair isn’t likely to be the 1st or 2nd most important distortion mechanism seems right to me
there is actually some recent research into thermal modulation distortion at RF where thermal modulation throws off the predistortion approach that is increasingly used to reduce distortion
The resistor tail “current source” has an advantage from the thermal point of view; since nearly equal V is dropped across the resistor and the input Qs you maintain a near const power dissipation in the Qs for small input Vcommon changes – the resistor tail source does lead to increased 2nd harmonic distortion from other mechanisms compared to the usual active tail current source
The largish diff pair bias current in Graham’s design puts this power vs Vcommon compensation and thermal symmetry of the 2 input Qs to a rather more severe test than typical lower bias current – a small Vcommon to differential power dissipation conversion comes from the slight V difference imposed by the 2 Q current mirror’s 1 Vbe input drop vs the output stage’s 2 Vbe bias point (easily fixed with a extra dropping diode or a more symmetric 3 Q feedback current mirror)
You could take a lesson from instrumentation designers striving for low offset drift; the (well matched for Vbe and Hfe) diff pair Qs can be put in close thermal contact – gluing to-92s together with a wrapping of heavy gauge bare copper wire over the plastic bodies and paying attention to equalizing thermal terminations of the leads
But the basic conclusion that thermal modulation distortion in the input diff pair isn’t likely to be the 1st or 2nd most important distortion mechanism seems right to me
Hi Jos,
From your Post #92, I now see where you are coming from, and you raise a genuinely noteworthy point.
I think your calculations are most reasonable, though they also suggest that differential transient thermal distortion levels at frequencies where our hearing is most sensitive really are quite negligible. My old Motorola databooks show TO92 thermal responses, and the heating effect is not linear with time, but follows an inverse square law with the sort of thermal corner frequency you suggest.
For your exampled DC step you arrive at a 0.05% response aberration over 166mS.
For asymmetrical waveform distortion ith a modern TO92 device I would further divide your figure by an approximated rule of thumb, of 10 for 10Hz, and 100 for 1kHz. Also, with that miniscule introduced offset distortion waveform having such a long time period I doubt that loudspeakers would introduce any audibly recognisable sonic characteristics.
I admire your thinking on this one Jos, and I agree with JCX.
I also use those clip-on 'frilly hat' TO5 heatsinks on my TO92s by clipping the gapped open ends over their epoxy bodies.
Cheers ........... Graham.
From your Post #92, I now see where you are coming from, and you raise a genuinely noteworthy point.
I think your calculations are most reasonable, though they also suggest that differential transient thermal distortion levels at frequencies where our hearing is most sensitive really are quite negligible. My old Motorola databooks show TO92 thermal responses, and the heating effect is not linear with time, but follows an inverse square law with the sort of thermal corner frequency you suggest.
For your exampled DC step you arrive at a 0.05% response aberration over 166mS.
For asymmetrical waveform distortion ith a modern TO92 device I would further divide your figure by an approximated rule of thumb, of 10 for 10Hz, and 100 for 1kHz. Also, with that miniscule introduced offset distortion waveform having such a long time period I doubt that loudspeakers would introduce any audibly recognisable sonic characteristics.
I admire your thinking on this one Jos, and I agree with JCX.
I also use those clip-on 'frilly hat' TO5 heatsinks on my TO92s by clipping the gapped open ends over their epoxy bodies.
Cheers ........... Graham.
"When the load is reactive the 'dynamic' output device base currents are not a simplistic split of this 'bias' current passing through the driver transistor collector resistor, which is why a standard JLH loses clarity when it is run above approximately half of the expected maximum listening power with a real world loudspeaker."
Hi Graham,
This was your answer (partial) to one question on the AABB JLH thread.
I would like to ask you if your findings on this 25 W class A JLH output present a similar situation to the "standard" JLH?
I guess you are refering to de 10W class A double rails.
The reason to my question is that I'm planning to build your 25W class A and would like to know what to expect from it.
I appreciate your extraordinary contribution to the DIY community
Hi Graham,
This was your answer (partial) to one question on the AABB JLH thread.
I would like to ask you if your findings on this 25 W class A JLH output present a similar situation to the "standard" JLH?
I guess you are refering to de 10W class A double rails.
The reason to my question is that I'm planning to build your 25W class A and would like to know what to expect from it.
I appreciate your extraordinary contribution to the DIY community
Hi Working,
The 2SC4793 is TO-220 ? Larger transistor, higher dissipation - higher capacitance ?
The splitter/driver transistor should have good gain and speed at low current, also low capacitance. The 2SC3421 is listed with a typical Cob of 15pF.
A BD139 works fine, though some have poor gain, and yet with the lower gain ones I have not heard reduced quality in JLH output circuitry.
Hi Tony,
This 25W design is so easy to throw together that you have nothing to lose by trying it in a rough form before building a neat and finished version. Come on; enjoy the adventure !
The JLH output stage is good to its rated current output, but I am being no more than brutally honest in warning prospective constructors that this pure class-A output stage has no reserve to cope with dynamic loudspeaker impedance dips below nominal !
That said it, this simple version with the mirrored differential input stage ( that can lead to reproduction degradation in class-AB amplifiers due to subsequent circuit phase changes and delays - including other class-A types with compensated VAS+CCS push-pull ) remains my favourite. As JCX has simulated, the products are fundamentally low, and the sound is extremely clean on transients.
Now that my JLH based higher power versions have proved workable, but not genuinely Hi-Fi, I am looking at other driving modes, my concern being that it is near impossible to have higher power non-class-A without substantial odd harmonic distortion, which must then by minimised with NFB, which then degrades music reproduction.
Tony, if you build this circuit using modern devices, then do not follow instructions for the earlier JLH design. The leads to the output devices should be spaced at least 1cm/half inch; the lead to the lower output base especially, also keep input and output well apart. Note that my Zobel resistor is lower to cope with the higher hf gain; keep this near the small signal citcuitry.
From my own point of view I will go a long way to have decent power without 300W of heat, but I am not willing to sacrifice audio quality to do so, which indicates that a triple or quad version of this 25W amplifier might yet still arise for what I regard to be ultimately clean amplification to a genuine 50W with awkward loudspeakers.
My work is not yet finished.
Cheers ........... Graham.
The 2SC4793 is TO-220 ? Larger transistor, higher dissipation - higher capacitance ?
The splitter/driver transistor should have good gain and speed at low current, also low capacitance. The 2SC3421 is listed with a typical Cob of 15pF.
A BD139 works fine, though some have poor gain, and yet with the lower gain ones I have not heard reduced quality in JLH output circuitry.
Hi Tony,
This 25W design is so easy to throw together that you have nothing to lose by trying it in a rough form before building a neat and finished version. Come on; enjoy the adventure !
The JLH output stage is good to its rated current output, but I am being no more than brutally honest in warning prospective constructors that this pure class-A output stage has no reserve to cope with dynamic loudspeaker impedance dips below nominal !
That said it, this simple version with the mirrored differential input stage ( that can lead to reproduction degradation in class-AB amplifiers due to subsequent circuit phase changes and delays - including other class-A types with compensated VAS+CCS push-pull ) remains my favourite. As JCX has simulated, the products are fundamentally low, and the sound is extremely clean on transients.
Now that my JLH based higher power versions have proved workable, but not genuinely Hi-Fi, I am looking at other driving modes, my concern being that it is near impossible to have higher power non-class-A without substantial odd harmonic distortion, which must then by minimised with NFB, which then degrades music reproduction.
Tony, if you build this circuit using modern devices, then do not follow instructions for the earlier JLH design. The leads to the output devices should be spaced at least 1cm/half inch; the lead to the lower output base especially, also keep input and output well apart. Note that my Zobel resistor is lower to cope with the higher hf gain; keep this near the small signal citcuitry.
From my own point of view I will go a long way to have decent power without 300W of heat, but I am not willing to sacrifice audio quality to do so, which indicates that a triple or quad version of this 25W amplifier might yet still arise for what I regard to be ultimately clean amplification to a genuine 50W with awkward loudspeakers.
My work is not yet finished.
Cheers ........... Graham.
Hi Graham-
Here are links to the two Toshiba datasheets:
3421: http://www.semicon.toshiba.co.jp/td...Transistors/en_20040804_2SC3421_datasheet.pdf
4793: http://www.semicon.toshiba.co.jp/td...Transistors/en_20040820_2SC4793_datasheet.pdf
I am trying to A) figure out which parts I can use to build your design and B) learn a bit about substitutions, but don;t want to drag down your thread either.
The 4793 has a capacitance of only 20pf (slightly higher) but also higher gain (100-320 vs 80-240). Many other ratings are close or identical, or better. Is an ft difference of 100 vs 120 significant in this application?
I think maybe I should just build your circuit and try it out.
Here are links to the two Toshiba datasheets:
3421: http://www.semicon.toshiba.co.jp/td...Transistors/en_20040804_2SC3421_datasheet.pdf
4793: http://www.semicon.toshiba.co.jp/td...Transistors/en_20040820_2SC4793_datasheet.pdf
I am trying to A) figure out which parts I can use to build your design and B) learn a bit about substitutions, but don;t want to drag down your thread either.
The 4793 has a capacitance of only 20pf (slightly higher) but also higher gain (100-320 vs 80-240). Many other ratings are close or identical, or better. Is an ft difference of 100 vs 120 significant in this application?
I think maybe I should just build your circuit and try it out.
Hi Graham,
Thanks for your reply and glad to know you continue to work on these JLH designs.
I'm not an EE so usualy preffer to work on designs that have previouly been built with god results and avoid test beds on Power amps (most of the time), this does not say i'm not a tweeker or that i would not like to test a different set of transistors but usualy keep that to a minimum.
This would be my first atempt for an audio project using BJT's, have done before a couple of 150W PP tube, Borbely's and Pass clones.
For this design (yours) 25W class A JLH I have in mind using some components I have at hand on a double mono case using
Hitachi's 2SC1890AE for the VAS
Hitachi's 2SD669AC for the driver
Motorola's MJ15003 for the output
For the power supply most likly will use 300/350VA EI trafos and a LC or CLC filtering with no other regulation. Coils will be 18mH. As with my previous amps the trafos and coils will be homebrewed.
If you are kind enough please let me know if you see anythig that may not work properly. This could avoid some frustration from my side.
Cheers...
Thanks for your reply and glad to know you continue to work on these JLH designs.
I'm not an EE so usualy preffer to work on designs that have previouly been built with god results and avoid test beds on Power amps (most of the time), this does not say i'm not a tweeker or that i would not like to test a different set of transistors but usualy keep that to a minimum.
This would be my first atempt for an audio project using BJT's, have done before a couple of 150W PP tube, Borbely's and Pass clones.
For this design (yours) 25W class A JLH I have in mind using some components I have at hand on a double mono case using
Hitachi's 2SC1890AE for the VAS
Hitachi's 2SD669AC for the driver
Motorola's MJ15003 for the output
For the power supply most likly will use 300/350VA EI trafos and a LC or CLC filtering with no other regulation. Coils will be 18mH. As with my previous amps the trafos and coils will be homebrewed.
If you are kind enough please let me know if you see anythig that may not work properly. This could avoid some frustration from my side.
Cheers...
Hi Working,
I cannot see why the 2SC4793 will not work just fine.
When you wrote TO220 I imagined a heatsink tab, but maybe this is a plastic 'ISO' version.
Thanks for the datasheet link. I note that the transition frequencies are quoted at different voltages, so the 2SC3421 will be a faster device within the amplifier.
Hi Tony,
Not sure what you mean by 'VAS' and 'driver'. There are;- an input pair, a mirror pair, a current splitter, and an output pair. 2SD okay for splitter, and MJs okay for outputs.
The circuit really is not fussy, but I tested and listed new component types in case anyone was buying new.
Cheers ........... Graham.
I cannot see why the 2SC4793 will not work just fine.
When you wrote TO220 I imagined a heatsink tab, but maybe this is a plastic 'ISO' version.
Thanks for the datasheet link. I note that the transition frequencies are quoted at different voltages, so the 2SC3421 will be a faster device within the amplifier.
Hi Tony,
Not sure what you mean by 'VAS' and 'driver'. There are;- an input pair, a mirror pair, a current splitter, and an output pair. 2SD okay for splitter, and MJs okay for outputs.
The circuit really is not fussy, but I tested and listed new component types in case anyone was buying new.
Cheers ........... Graham.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.