The state variable filter osciillator I show was for 1N4148. The 47 K current limiting resistor and 22K bleed resitor were for them. To my great surprise the LED's do much the same as proved by the less than square wave in the limiter circuit ( good ). Removing the 22 K pushes output to 6.02 V rms ( 5.29 with 22 K ). For a loss of 0.7 V less distrotion. There is the faintest red glow. Doubtless more current would help temperature stability. Cold to warm circuit 5.10 V to 5.3 V rms. That is fine. Notice how well the 10.2 nF ( hand picked to +/- 0.5% ) 10 K 22 K filter works. It's 49 Hz roll off suits the +0.2 dB output at Fo. It gives exactly the same output. This was a fluke. As most of the caps are 10.2 nF this suited very well. They are the cheapest possible mylar types at 5 cents 200 ppm. Some COG 1205 SMD 10 nF 5% 30 ppm would be a great idea. As most people don't have frequency meters ( mine to 0.0002 % ) to tweek the 300 + 18 K is the thing to do. A 3.2768 MHz (+ 1M ohm ) crystal divided down to 100 Hz ( 2 x 74HC4060 ) to power a strobe if a turntable will do it cheaply and very well. Or divide again to 50 Hz and use a dual beam scope to set. If one side has a pot this will do it well. It will slightly change the output. It you want to offset the inductance of a motor with no extra circuitry make a large change to one side only. As is logical it follows a very similar law. For a motor I designed this means for example 13 V rms 33 1/3 and 16.5 V rms 45 plus 25.3 Vrms at 96 rpm. The motor wouldd be a 24 V 50 Hz type . Power used is 10 watt, 9 watts and 7 watts. At 78 rpm the motor is still in it's ideal range. Although it is more ideal for 33 1/3. I notice no real change in distortion at 45/78. My motor is a hysteresis type so will allow this working. A synchronous being a stepper usually hits resonance at about 90 Hz. The motor I use could safely quadrupple it's input power. It was designed for medical use. I retain the big bits. We bought 160 . This means if we can not source another motor only 150 Garrard 501/601 will be built. It has to rotate backwards to add to it's problems. A syncronous like Crouzet sounds great if the - 40 dB rumble can be ignored. We get - 79 dB weighed. Mostly due to very low distortion and motor design. The Crouzet has in itself > 7 % THD ( reading the current input waveform, mild guess, it might be more ). This is due to the magnetic window shape as my best guess as to cause. Ideal for triangle waves but not sine. Some think the phase capacitor is the main cause. Not really. The use of a triangle wave might give slightly more torque. If a square wave is used it draws a triangular current. It had occured to me to feed the sine wave and current waveform into a differential circuit and extract a composite anti distortion. Record this onto CD and use the CD to see if an opposite waveform could work. My guess is it will help a bit as it might be the ideal wave shape. I strongly doubt it will be a cure. This is pre distortion. As the wave is repetitive it is the easiest use of pre distortion. What is might do is offer the most torque. For an LP12 that would be welcolm.
An interesting version uses a LM285 20 ppm regulator. This is is a precission voltage reference in one direction and a standard diode in the other ( not shown in tech spec so might not be all types ). If using the 1.2 V version the SVF will cope very well with the offset ( 1.2/0.7 V ). This makes it almost a zener in action. This halves the tempertaure drift. Alas in this simple circuit one can't use two face to face because they turn into 1N4148 more or less. If two LM285-2.5 were configured as two back to back zeners that would drop the error to perhaps 30 %. That would need an op amp that can swing >12 V rms. That can be done if collector output dumpers are added connected bases between op amp V in and a resistor to the rails. TL071 and BC327/337 seem possible.
With a bit of work the oscillator is for AC synchronous via sine and cos outputs. Lenco/ Garrard/ TD124/ BSR via Cos filtered. With COG caps as stable as crystals ( 6 seconds a day using 100 ppm in the past if room temperature driving a clock, resistors are about 50 ppm ).
An interesting version uses a LM285 20 ppm regulator. This is is a precission voltage reference in one direction and a standard diode in the other ( not shown in tech spec so might not be all types ). If using the 1.2 V version the SVF will cope very well with the offset ( 1.2/0.7 V ). This makes it almost a zener in action. This halves the tempertaure drift. Alas in this simple circuit one can't use two face to face because they turn into 1N4148 more or less. If two LM285-2.5 were configured as two back to back zeners that would drop the error to perhaps 30 %. That would need an op amp that can swing >12 V rms. That can be done if collector output dumpers are added connected bases between op amp V in and a resistor to the rails. TL071 and BC327/337 seem possible.
With a bit of work the oscillator is for AC synchronous via sine and cos outputs. Lenco/ Garrard/ TD124/ BSR via Cos filtered. With COG caps as stable as crystals ( 6 seconds a day using 100 ppm in the past if room temperature driving a clock, resistors are about 50 ppm ).
Dejan you migh like this. Figure four is me to a tee. What pure genius.
http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf
I was looking for a circuit that I saw some time ago that looks like a cascode and or Darlington. The search question was " transition miller compensation ".
http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf
I was looking for a circuit that I saw some time ago that looks like a cascode and or Darlington. The search question was " transition miller compensation ".
Yes, you have an interesting shed.
For the same kind of fun, goto -31.554760, 115.637988.
Not nearly as green as your area, but perfectly fresh Indian Ocean air, and less than 1000m to the beach.
Dan.
Would you believe the ex helped build it. Taking her to hospital today so still friendly. Will zoom in on your beach.
Just re-ran my little oscillator with TL074. LM324 was slighlty better ( gnats whisker ). It proves the power of negative feeback at low frequency. The LM324 is Darlington input. It almost loads the circuit like the JFET's of TL074. LM324 looks so like LM339 comparator. I use LM324 as one as it both sinks and sources. Mostly any LM339 circuit works with LM324, hence I have plenty. LM324 sometime is Cinderella. It is a tough device when abused. Real audio is not it's thing.
I will try a null between op amp No 1 and No 3 . A 100 K pot with about 10 K to op amp 3 should be a near perfect null. If passed through the filter I suspect - 80 dB might be possible. Certainly - 70 dB. MC33079 also as op amp if wanting > 10 kHz.
Will look at -31.554760, 115.637988
I repaired the shed roof for $15 . First skin DPC ( plastic for floors under cement ) for $0. Then next to worst grade felt. Retained 50 % of the original , why take it off. The capping was gone with the wind so to speak. Looks standard and is much better. 20 minutes work and the rain no longer gets in. The replacement was right to about 20 cm. Even that was used as the edges get damaged a bit. That's when life goes right. Cars for all of their problems used to be like that.
Just re-ran my little oscillator with TL074. LM324 was slighlty better ( gnats whisker ). It proves the power of negative feeback at low frequency. The LM324 is Darlington input. It almost loads the circuit like the JFET's of TL074. LM324 looks so like LM339 comparator. I use LM324 as one as it both sinks and sources. Mostly any LM339 circuit works with LM324, hence I have plenty. LM324 sometime is Cinderella. It is a tough device when abused. Real audio is not it's thing.
I will try a null between op amp No 1 and No 3 . A 100 K pot with about 10 K to op amp 3 should be a near perfect null. If passed through the filter I suspect - 80 dB might be possible. Certainly - 70 dB. MC33079 also as op amp if wanting > 10 kHz.
Will look at -31.554760, 115.637988
I repaired the shed roof for $15 . First skin DPC ( plastic for floors under cement ) for $0. Then next to worst grade felt. Retained 50 % of the original , why take it off. The capping was gone with the wind so to speak. Looks standard and is much better. 20 minutes work and the rain no longer gets in. The replacement was right to about 20 cm. Even that was used as the edges get damaged a bit. That's when life goes right. Cars for all of their problems used to be like that.
Dejan you migh like this. Figure four is me to a tee. What pure genius.
http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf
I was looking for a circuit that I saw some time ago that looks like a cascode and or Darlington. The search question was " transition miller compensation ".
Dan. I see you can rent a classic motorcycle up the road. Cool.
http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf
I was looking for a circuit that I saw some time ago that looks like a cascode and or Darlington. The search question was " transition miller compensation ".
Dan. I see you can rent a classic motorcycle up the road. Cool.
That's news to me...please give links.....just searched, just a few doors up the road.
I have never been keen on motorcycles.
That said I have ridden what seems like a million miles on high performance bicycles....got a silver medal in state road titles after changing a blown tyre back in the day.
I used to wear my front tyres past the tread and up into the side walls, every morning and afternoon both tyres sliding mid corners and roundabouts, regularly replaced pedals because the alloy/dustcaps was ground away exposing the outer bearings.
My mother worked in a rehab ward and forbid that I ever own a motor bike....she is correct.
Dan.
My friend Martin Rewick died on a bicycle. He was revived. He now makes $30 000 bicycles in Syndey. He held the record in one of the Sydney races before the crash.
I think a nice 250 cc trials bike should be OK. The idea is to go slow.
He was importer for Epos speakers also.
Horses are the worse. Mile per death that is.
I think a nice 250 cc trials bike should be OK. The idea is to go slow.
He was importer for Epos speakers also.
Horses are the worse. Mile per death that is.
Sometimes a horse rider will "die" for other reasons, 🙂 ... https://www.youtube.com/watch?v=DYM69N8V5J0
A funny thing happened. I was using my slow cooker to check the temp coe of LEDs. I went to sleep and melted the thermometer. Lets speculated 100 C plus. For fun I blew on the oscillator and sure enough it drifted back to life, The heat stalled it. However an interesting change. The output voltage had dropped from 4.19 to 4.09 V rms at 21 C. The frequency is unchanged so the caps were as before. Now the crunch . The distortion is 3 dB better ( - 64 dB LM324, fourth harmonic only ). This is for a motor drive so rather excellent. I have seen small changes go this way before. Never when something nearly died. I think the band gap must have slightly changed and maybe I made them equal by cooking them ? That LM324 is tough. The TL074 measures no better in this circuit.
The oscillator is for a friend. She asked how low distortion should be. I thought I should throw it open .
" The theoretical zero distortion has been said to be 20 dB ( 10 ) times lower than the device used after it. So if a speaker has 3 % the ideal is better than 0.3% for the system behind it. It then is zero with respect to that device. This seems to be born out by the 1947 idea of 0.1% THD as ideal ( Hafler , Williamson, Leak ). I put Hafler first as his ideas powered the industry. Williamson was only 19 ! Leak had a product to sell.
Your friend Mr Hiraga likes an exponential decay of harmonics. If we say the second harmonic is 0.08% and all harmonics in perfect exponential decay, then that really is hi fi. The Magneplanar speakers we both like are almost that good.
Now what amplifier would suit them to be zero distortion ? The 1967 Quad 303 can be exactly that. Typical distortion is 0.01% or 20 db better than the best possible speakers. 303 is also in exponential decay ( The single input transistor helps ensure that ). Hiraga suggests the non exponential is a source of highly disagreeable distortion. This seems likely to be true. The inner ear produces about 30 % THD in exponential decay ( Radcliffe where you were did the research ). The ear being mostly analogous to a digital system that knows it's own distortion. If a similar distortion is added all is OK . Now if total addition is about - 20 dB or 3% all is OK. Exactly as most true hi fi is. Vinyl is about 3% if lucky. Thus a good amp and speakers are required.
CD although very low in distortion when loud starts to brake some of these rules when quiet. I think we all hear that. It's like unnatural lighting.
Now to motors. The oscillator I have just designed is - 65 dB and only 4 th harmonic. If we measure the motor in current mode we can find that if it were a speaker it's distortion is at - 52 dB or about 0.6%. In the ideal world we need - 72 dB. However we already have that. The motor in itself is a filter. It will reduce the the 4 th harmonic. In the very ideal world we would rewind the motor for 24 V .
The LP12 motor measures about - 39 dB or worse. The simple cosine function is at the ideal - 59 dB. The sine is at - 49 dB. As the sine cosine phase error is usually dominant and we would have none, I would call that job done. The belt is the No 1 problem which in my opinion kills ALL belt drive systems. It is for cheap decks only like AR . Only the originator understood that. He was infuriated to to be asked how he made sure his turntable imaged so well ? I would answer for him. It was stable in the vertical direction where out of phase echo information is kept.
BBC Radio 3 had a Vinyl day !
Rosen had to design an oscillator to measure down at 6.02 x 16 ( + 20 ) = - 116.32 dB. I have stolen a section of it to do this. He used the despised NE 5534 op amp to do it !"
" The theoretical zero distortion has been said to be 20 dB ( 10 ) times lower than the device used after it. So if a speaker has 3 % the ideal is better than 0.3% for the system behind it. It then is zero with respect to that device. This seems to be born out by the 1947 idea of 0.1% THD as ideal ( Hafler , Williamson, Leak ). I put Hafler first as his ideas powered the industry. Williamson was only 19 ! Leak had a product to sell.
Your friend Mr Hiraga likes an exponential decay of harmonics. If we say the second harmonic is 0.08% and all harmonics in perfect exponential decay, then that really is hi fi. The Magneplanar speakers we both like are almost that good.
Now what amplifier would suit them to be zero distortion ? The 1967 Quad 303 can be exactly that. Typical distortion is 0.01% or 20 db better than the best possible speakers. 303 is also in exponential decay ( The single input transistor helps ensure that ). Hiraga suggests the non exponential is a source of highly disagreeable distortion. This seems likely to be true. The inner ear produces about 30 % THD in exponential decay ( Radcliffe where you were did the research ). The ear being mostly analogous to a digital system that knows it's own distortion. If a similar distortion is added all is OK . Now if total addition is about - 20 dB or 3% all is OK. Exactly as most true hi fi is. Vinyl is about 3% if lucky. Thus a good amp and speakers are required.
CD although very low in distortion when loud starts to brake some of these rules when quiet. I think we all hear that. It's like unnatural lighting.
Now to motors. The oscillator I have just designed is - 65 dB and only 4 th harmonic. If we measure the motor in current mode we can find that if it were a speaker it's distortion is at - 52 dB or about 0.6%. In the ideal world we need - 72 dB. However we already have that. The motor in itself is a filter. It will reduce the the 4 th harmonic. In the very ideal world we would rewind the motor for 24 V .
The LP12 motor measures about - 39 dB or worse. The simple cosine function is at the ideal - 59 dB. The sine is at - 49 dB. As the sine cosine phase error is usually dominant and we would have none, I would call that job done. The belt is the No 1 problem which in my opinion kills ALL belt drive systems. It is for cheap decks only like AR . Only the originator understood that. He was infuriated to to be asked how he made sure his turntable imaged so well ? I would answer for him. It was stable in the vertical direction where out of phase echo information is kept.
BBC Radio 3 had a Vinyl day !
Rosen had to design an oscillator to measure down at 6.02 x 16 ( + 20 ) = - 116.32 dB. I have stolen a section of it to do this. He used the despised NE 5534 op amp to do it !"
Not all 🙂. You're talking about the fairly common type of distortion in digital playback - say, equivalent to inner groove distortion of LPs. A fixable problem, but one needs to be aware of what's going on.
I'm playing with tweaking of an ordinary laptop's inbuilt sound, using internal speakers, at the moment - this is purely an exercise to see what's possible. Of course normally it suffers from that same symptoms that you speak of, and the challenge is to "fix" this. Showing excellent signs so far of being able to do this, but may require a number of rounds of tuning to get a consistency of good behaviour, 😉 ...
Quite so. CD makes it easier to have the problem. Hiraga has a point. Keeping that natural structure is the hardest thing.
Analogy. My old Sony TV was getting old. My brother showed me how to get a few more months. We fitted a small transformer and boosted the red gun. I then had to adjust the white balance. This was done both at low and high levels. The problem was sometimes a green white and other times a blue white at the intermediate levels. My brother said the basic linearity was showing up and a nearly dead red gun making it obvious. His comment to me I would now always see it and I did. New or old CRT will do that. People who watched football often lost the green gun first. The 32 inch jobs more than most.
Analogy. My old Sony TV was getting old. My brother showed me how to get a few more months. We fitted a small transformer and boosted the red gun. I then had to adjust the white balance. This was done both at low and high levels. The problem was sometimes a green white and other times a blue white at the intermediate levels. My brother said the basic linearity was showing up and a nearly dead red gun making it obvious. His comment to me I would now always see it and I did. New or old CRT will do that. People who watched football often lost the green gun first. The 32 inch jobs more than most.
The LED thing has a new bizare twist. Output back to normal overnight and distortion as before. Same devices that I think saw 120 C. It was something to watch. After the body of the LM324 was at a normal temperature nothing worked. Then I saw 10 mV. I blew on it then very slowly saw it got back to 4.09 Vrms ( now 4.19 ). It stopped at 2 V for a while. If someone told me this I would doubt it. Lazarus picked up his bed.
Speculation. Given the right choice of parts could running at >80 C help the sound ? A preamp perhaps? Someone said the raised levels inside valves are part of the magic. I have never asked myself if transistors would like it? Instinct is to say the opposite, slightly more hiss. The parts had seemed to cool down making it more interesting. Could the LED band gap take all night to restore ? Over to you Mr Firme.
Speculation. Given the right choice of parts could running at >80 C help the sound ? A preamp perhaps? Someone said the raised levels inside valves are part of the magic. I have never asked myself if transistors would like it? Instinct is to say the opposite, slightly more hiss. The parts had seemed to cool down making it more interesting. Could the LED band gap take all night to restore ? Over to you Mr Firme.
I said the other day a hi fi system in expontial decay of harmonics would have zero distortion if pretending the ear was the reference device ( 3 % THD). This was using standard theory of measuring. This says any device to be used as a measuring tool must be 20 dB better than the thing being measuerd. It is implied this is a minimum. Most people mistake damping factor and THD. They hear a so called zero feedback amp and take the booming bass to be distortion. Yes and no. Mostly no.
It is very important to say more. This is only true if exponential harmonics. The ear because of the way it is made will not " see " a difference if the relative difference of the two slopes is zero.
Conversely and at a level I can only make a stab at - 40 dB at a fifth harmonic will be very unpleasent. A system of seemingly impecible preformance could have an error of that type if not the exact type.
This might explain things. If a weighting curve was used we might get a very different picture.
People who will object to this will carefully ignore the speakers they own. Much in the way children of ours are OK whilst of others are not.
Most of the electrostatic or isodynamic speakers are able to be compared with 1950's amplifers when talking distortion. If so even when vinyl the zero perceived distortion is possible. Any shift from exponential might produce distortion many times greater, even if the lumped THD is less.
The doistortion I refered to for my friend was motor vibration. Just holding the motor would tell us what we want to know. My conjecture was to say the distortion of the driving amplifer should be 20 dB lower than that of using the motor as an altenator with load. It's not an exact sceince, a good throw of the dice I think ?
It is very important to say more. This is only true if exponential harmonics. The ear because of the way it is made will not " see " a difference if the relative difference of the two slopes is zero.
Conversely and at a level I can only make a stab at - 40 dB at a fifth harmonic will be very unpleasent. A system of seemingly impecible preformance could have an error of that type if not the exact type.
This might explain things. If a weighting curve was used we might get a very different picture.
People who will object to this will carefully ignore the speakers they own. Much in the way children of ours are OK whilst of others are not.
Most of the electrostatic or isodynamic speakers are able to be compared with 1950's amplifers when talking distortion. If so even when vinyl the zero perceived distortion is possible. Any shift from exponential might produce distortion many times greater, even if the lumped THD is less.
The doistortion I refered to for my friend was motor vibration. Just holding the motor would tell us what we want to know. My conjecture was to say the distortion of the driving amplifer should be 20 dB lower than that of using the motor as an altenator with load. It's not an exact sceince, a good throw of the dice I think ?
So how might we apply the 20dB test to other factors? An amplifier is more than THD and damping factor. IE how would you settle on a figure of merit for phase and gain margin?
Good point on zero feedback amps, Nige. They always sound a little rich for me, mostly in the bass, and they always make me think that somehow they are unfinished.
Sounding rich and beefy is also what I usually hear from amps using Darlington transistors for te output - not always, but often.
Sounding rich and beefy is also what I usually hear from amps using Darlington transistors for te output - not always, but often.
Spladski, the answer to your question hinges on one as yet undetermined (here, at least) point, and that is at what level and under which conditions is the threshold of hearing harmonic and IM distortion, and all other forms of distortion for that matter.
I raised the question here several times, and while discussing at length what's better, 16 or 24 bit resolution, everybody prentended not to have noticed it. It's like discussing what the 8th floor should look like without having determined what the first floors should be like.
I raised the question here several times, and while discussing at length what's better, 16 or 24 bit resolution, everybody prentended not to have noticed it. It's like discussing what the 8th floor should look like without having determined what the first floors should be like.
That's an excellent question. Dave Mate of SSL thinks humans process phase shift very well. This means the sound is not obviously wrong. However the sound without phase shift is less fatiguing. Dave was thinking of phase shift in speakers. He thought speaker squarewave tests should be taken seriously. Given a less expensive speaker try to brake at 7 kHz if possible said Dave. A speaker I helped design does this. It was very hard to do. I think it is great to listen to. My girlfriend has a pair of them , modified Rotel RA931 amp, JVC L3-E turntable and Sony DVD as CD player. All these items are giant killers, The Sony may not be worse than most audiophile CD players. The speakers would if allowed roll off at 45 kHz ( about 24 kHz ). We spent all yesterday listening to her 45's played on a Dansette record player in the 1960's. Some real gems like first pressing 1958 Buddy Holly. Although the sound was as you would imagine it was fantastic. The openess and speed of the speakers make it excellent to listen too. The Rotel mods are 33 uF non polar caps with 0.1 uF 250 V op-amp in and out and power amp in. 2 x 1 uf to PSU caps. 2 x 33 uF nonpolar to output fuses. 22K x 4 pull down resistors to - ve NE5532 and OPA2604 ( SE class A ). The cap at OPA 2604 output replaced with a wire link. The only reason it was there was the polar caps would become reversed polarised if not. The feedback arm 100 uF 6.3 V polar replaced with 100 uF 100 V non polar and 0.22 uF ( run out of 0.1 uF ). There was just space to fit them. I can not say if phase shift is the key ? It is despite the wrinkles of age like the best Junk box I ever heard. The near DC frequecy responce of the amp is good news. The speakers will bottom on warps. Truth is it is too load and needs to be turned down at that point. I told her I was jealous, I wasn't kidding . The JVC is better by a very long way than the TD145 I lent her with DL110. It is based on the JVC TT 71. It looks a pile of junk. It is a turntable you can use after 2 bottles of wine. Wow and flutter 0.03% and - 76 dB rumble. Motor more so than a Denon seems cogging free,more so than 1210 and perhaps better than SP10 in the treble. If I am honest better than a LP12. It has tighter bass.
Dejan, for me the big, bad answer is that the important distortion - that is, that which is audible - is a very complex mess, and coming out with fixed numbers for acceptable levels of whatever specific one you happen to be looking at will be almost as useless as all the 'magic' numbers and ways of measuring that have been thought of to date. I would suggest a far better way is to work in reverse - use recordings and special test tracks that deliberately stress systems and highlight flaws, and then measure the behaviour of a misbehaving setup in every conceivable way to determine what is actually occurring at that moment of poor performance.
Just my notes to myself. This was a £0 upgrade out of parts I had in the shed. WW ( Electronics Today ? ) showed that non ploar caps used with care are better. Better by a big margin over 63 V polyester ( 100 V types OK ) and Audiophile polar types like Black Gates or Oscon. 0.4 V peak if possible. The DC offset on the NE5532 is high, even so the 0.4V peak is rarely exceeded. Class A also is marginally better, don't use too much. Coloured bits are updates. The RA 931 is a giant killer. Took me about 2 hours to do. The fuse mod is the big deal. At 3 watts the 100 uF passes the 0.4 V peak. That is very OK as our ears start to do the same. Oscon / Black Gate would be no better. What might work is change the feedback resistor from 8K2 to 16 K. That would exceed 8 watts at 0.4v peak . One little Rotel thing I liked. The chassis is broken at one point and isolated so as to brake the loop. Just two plastic washers.
- Status
- Not open for further replies.