Three main blocks:
- one NPN differential input pair
- two identical all NPN Output power modules
Each input transistor drives one half of the Output
.. via Light
using two Analog OptoCouplers.
I can see two main issues:
1. Finding analog optocouplers. With good linearity + good speed, bandwidth.
2. Arrange thermal tracking and temperature compensation.
--------------------------------------------------------------------------------------
Maybe in future there will be some great and fast analog optocouplers
suitable for audio applications, like this Lineup idea.
Right now there are not many good analog optos around. Not what I know.
However, there are a few, that can be used.
To make one power amplifier with moderate but maybe acceptable performance.
Lineup - Sweden autumn 2008
- one NPN differential input pair
- two identical all NPN Output power modules
Each input transistor drives one half of the Output
.. via Light
using two Analog OptoCouplers.
I can see two main issues:
1. Finding analog optocouplers. With good linearity + good speed, bandwidth.
2. Arrange thermal tracking and temperature compensation.
--------------------------------------------------------------------------------------
Maybe in future there will be some great and fast analog optocouplers
suitable for audio applications, like this Lineup idea.
Right now there are not many good analog optos around. Not what I know.
However, there are a few, that can be used.
To make one power amplifier with moderate but maybe acceptable performance.
Lineup - Sweden autumn 2008
Attachments
But Why?
Hello,
Why use opto couplers? Is there any advantage (imagining the linearity is great)?
Hello,
Why use opto couplers? Is there any advantage (imagining the linearity is great)?
Optocouplers are naughty, trust me...
HCNR200, IL300 and similar ones allow for some controlled (linear) transfer due to feedback and the matched pair of photodoiodes. However, there are too many unknown variables in straight open loop optocouplers, including the working point.
HCNR200, IL300 and similar ones allow for some controlled (linear) transfer due to feedback and the matched pair of photodoiodes. However, there are too many unknown variables in straight open loop optocouplers, including the working point.
As with many ideas there is not much new under this sun
I have designed and built such a device about a year ago! and can find no real advantage due to poor linearity and transfer ratio of most opto devices !
I thought I had struck gold when I conceived it up but as ever its never that simple . most opto have a poor vce rating on the transistor side of things the output transistor can not source or dissipate much This was the idea I conceived when I issued a challenge for a dc coupled amplifier without any pnp devices but as results were dissapointing I droped the design
regards Trev
I have designed and built such a device about a year ago! and can find no real advantage due to poor linearity and transfer ratio of most opto devices !
I thought I had struck gold when I conceived it up but as ever its never that simple . most opto have a poor vce rating on the transistor side of things the output transistor can not source or dissipate much This was the idea I conceived when I issued a challenge for a dc coupled amplifier without any pnp devices but as results were dissapointing I droped the design
regards Trev
Hello lineup,
I certainly don't mean to shoot down your idea before you start it, but I have seen analog optocouplers with their share of problems in being used with audio (that is, using an analog optocoupler directly in the audio signal path).
--> One of the lesser known problems is optocoupler aging - yes, the "driver" side LED ages over time, depending upon how often the devices are used. So if you "tune" or calibrate your optocouplers to get the right forward gain when you build your amplifier, they will probably drift in the future because the LED driver "weakens" (I'm talking over about a year or two of fairly heavy use). My old company experienced this problem with their optocouplers that they were using in a linear fashion - they made high voltage amplifiers (+/- 30kV at ~30mA) for capacitive loads.
--> Similar to what was already suggested, have you thought about driving your "NPN" output module and your "PNP" output module with optocouplers using PWM? This requries "relatively" fast optocouplers, but high currents are really not needed, and it can help fix the problem of linearity and aging (I.E. if the driver side of the otocoupler is driven "high", then the optocoupler output voltage may be 5V, or 4.9V, or 4.2V, or 4V. This variation in output voltage could be due to drift, or aging, or non-linearity, but it doesn't matter, because a logic "high" is still a logic "high").
Sounds like a great project lineup! I think we CERTAINLY don't mean to rain on your idea at all - we are just offering constructive criticism based on our own experiences. Please keep us posted on your progress!
I certainly don't mean to shoot down your idea before you start it, but I have seen analog optocouplers with their share of problems in being used with audio (that is, using an analog optocoupler directly in the audio signal path).
--> One of the lesser known problems is optocoupler aging - yes, the "driver" side LED ages over time, depending upon how often the devices are used. So if you "tune" or calibrate your optocouplers to get the right forward gain when you build your amplifier, they will probably drift in the future because the LED driver "weakens" (I'm talking over about a year or two of fairly heavy use). My old company experienced this problem with their optocouplers that they were using in a linear fashion - they made high voltage amplifiers (+/- 30kV at ~30mA) for capacitive loads.
--> Similar to what was already suggested, have you thought about driving your "NPN" output module and your "PNP" output module with optocouplers using PWM? This requries "relatively" fast optocouplers, but high currents are really not needed, and it can help fix the problem of linearity and aging (I.E. if the driver side of the otocoupler is driven "high", then the optocoupler output voltage may be 5V, or 4.9V, or 4.2V, or 4V. This variation in output voltage could be due to drift, or aging, or non-linearity, but it doesn't matter, because a logic "high" is still a logic "high").
Sounds like a great project lineup! I think we CERTAINLY don't mean to rain on your idea at all - we are just offering constructive criticism based on our own experiences. Please keep us posted on your progress!
Hello again,
Sorry to ask again and risk to be rude but can someone point the advantage of using these opto-couplers, would they be linear. If it's a DC question, one can use a good cap, it would be still more linear with today technology aand the relatively high impedance would only require a small value and even allow some styro cap. Can someone explain please?
Sorry to ask again and risk to be rude but can someone point the advantage of using these opto-couplers, would they be linear. If it's a DC question, one can use a good cap, it would be still more linear with today technology aand the relatively high impedance would only require a small value and even allow some styro cap. Can someone explain please?
When I looked at this idea the main advantage to me appeared to be that the amplifier would have a minimum damage scenario id an output device failed i.e the damage would be only the output transistors and maybe drivers . however due to low vce of the opto transistors there would have to be an increased no of power supplies or increased driver complexity
I got so far as to set the output current bias via altering the standing current through the difff pair and temperature compensation would be by alering the diff standing current
the dc offset would be by altering the ratio of current between the diff pair however I dont think that this would be a low distortion solution
I could see no point in making an amplifier that would have a higher inherent distortion
As others have said this could be done with coupling capacitor as per the HK and Armstrong 521 circuit all though in those cases output capacitors are also involved
regards Trev
I got so far as to set the output current bias via altering the standing current through the difff pair and temperature compensation would be by alering the diff standing current
the dc offset would be by altering the ratio of current between the diff pair however I dont think that this would be a low distortion solution
I could see no point in making an amplifier that would have a higher inherent distortion
As others have said this could be done with coupling capacitor as per the HK and Armstrong 521 circuit all though in those cases output capacitors are also involved
regards Trev
latala said:
But it isn't that hard to make the VAS protected from output stage failures. So damage due to blown output stage is limited to output transistors, drivers, predrivers if used and bias network.
It's not so interesting to fight like that to save the 2 half dollar transistors of the VAS knowing that the 20 dollars of the output stage blew up... in my opinion... Or just use mosfets and they will never fail in a catastrophic way.
Eva said:
Exactly. Quasi-complementary designs were used back in the 70's because good PNP power transistors didn't exist at the time. This is not the case anymore. There are several excellent complementary pairs available now.
I think there is some interest in quasi designs, it's the distortion pattern which tends to be more natural espescially if the feedback is kept reasonable (see pass designs too). Because of the asymmetrical transfert function... I think...
But PNP is always the bad one. They are not the mirror of each other.
Someone perhaps can tell us, why it (even today) is so hard to make the PNP exactly as good as the NPN partner ?
Someone perhaps can tell us, why it (even today) is so hard to make the PNP exactly as good as the NPN partner ?
Simply speaking, PNP requires more complex technology in case of Silicone transistors. In case of Germanium ones, PNP are easier to make.
Ragnwald,
I would recommend you to read Kjell O Jeppson book Halvledarteknink sid 512, Studenliteratur, ISBN 91-4421101-5.
It has to do with mobility, P-type silicon have lower mobility than N-type, which in turn requires more die area for a given current.
I would recommend you to read Kjell O Jeppson book Halvledarteknink sid 512, Studenliteratur, ISBN 91-4421101-5.
It has to do with mobility, P-type silicon have lower mobility than N-type, which in turn requires more die area for a given current.
- Status
- Not open for further replies.