Progg70,
There is nothing wrong with NP's current sinks. Very few people have more experience in this area, and they sound terrific.
I designed a SE amplifier of 28W rms some years back using two IRFP150s with 0R47 source resistors and a simple LED voltage reference. It reliably sinks 2.8A combined to ground with no hint of thermal instability and sounds wonderful with a suitable front end - a tube 6SL7.
It can be theoretically improved upon with a bipolar transistor sensing device, but frankly I have not found thermal instbility to be an issue in ten years and ten of these amps, six commercially.
Epops,
The man asked for a current sink. He wants to try it, listen to it, yet you condemned his choice, deriding NP amps and SE in general. While I am prepared to accept you might be a passable engineer, you are opinionated and elitist and your comments are absurd, misleading and discouraging to Progg70.
All topologies have their distinctive 'sound', but so do operating points, layouts, and component choices. All amps sound different, and some PPs sound like SE and vice versa. I don't believe denigrating NP is either helpful or useful to your cause. Have you anything better? If yes, please show it........
Why must you be so rigid, so negative, and so ineffective answering the question put to you? Is this really helpful?
Hugh
There is nothing wrong with NP's current sinks. Very few people have more experience in this area, and they sound terrific.
I designed a SE amplifier of 28W rms some years back using two IRFP150s with 0R47 source resistors and a simple LED voltage reference. It reliably sinks 2.8A combined to ground with no hint of thermal instability and sounds wonderful with a suitable front end - a tube 6SL7.
It can be theoretically improved upon with a bipolar transistor sensing device, but frankly I have not found thermal instbility to be an issue in ten years and ten of these amps, six commercially.
Epops,
The man asked for a current sink. He wants to try it, listen to it, yet you condemned his choice, deriding NP amps and SE in general. While I am prepared to accept you might be a passable engineer, you are opinionated and elitist and your comments are absurd, misleading and discouraging to Progg70.
All topologies have their distinctive 'sound', but so do operating points, layouts, and component choices. All amps sound different, and some PPs sound like SE and vice versa. I don't believe denigrating NP is either helpful or useful to your cause. Have you anything better? If yes, please show it........
Why must you be so rigid, so negative, and so ineffective answering the question put to you? Is this really helpful?
Hugh
Progg70,
My circuit is commercial and while the support circuits are openly available, the drive is proprietary, and I do not release it.
However, try this design, with full construction details, from Andreas Ciuffoli of Rome. It's very cleverly designed:
http://www.audiodesignguide.com/PowerFollower/index.html
Hope this helps, it's a very good sounding circuit from all reports, but you will need to drive it with a beefy source. If you want to use a tube, I'd suggest using a strong triode with a low plate impedance like a 5687.
CS: You won't need a Vbe multiplier to make a mosfet current sink or source. Look at the above link to check out how Ciuffoli does it; the mosfet is thermally very stable and ideal in this role.
Cheers,
Hugh
My circuit is commercial and while the support circuits are openly available, the drive is proprietary, and I do not release it.
However, try this design, with full construction details, from Andreas Ciuffoli of Rome. It's very cleverly designed:
http://www.audiodesignguide.com/PowerFollower/index.html
Hope this helps, it's a very good sounding circuit from all reports, but you will need to drive it with a beefy source. If you want to use a tube, I'd suggest using a strong triode with a low plate impedance like a 5687.
CS: You won't need a Vbe multiplier to make a mosfet current sink or source. Look at the above link to check out how Ciuffoli does it; the mosfet is thermally very stable and ideal in this role.
Cheers,
Hugh
Hugh, as you know, my " brother in fight " is your compatriot Bruce Candy, not Nelson Pass, although personally I very appreciate his knowledge. I don't like class A and SE still more. You, like good technician, certainly understand why. By my opinion is all only " Class A mania " and I say, that good amp don't make working class, but quite different things. And you know the same. But fight with this superstition is lost fight - many men and firms make with this thing big money and they are very strong...sometimes I feel like Don Quijote 
... BTW, stability of current source with mosfet isn't caused by mosfet ( this type have high positive Tc of Ids ), but by local feedback with silicon transistor...
... BTW, stability of current source with mosfet isn't caused by mosfet ( this type have high positive Tc of Ids ), but by local feedback with silicon transistor...
Epops,
Thank you for your email; you explain yourself much better.
Bruce Candy was born in South Africa, and emigrated to Australia as a teenager. He was educated at Adelaide University, but IIRC dropped out of an engineering degree there as a young man. Bruce is a celebrated and famously generous eccentric, maintains an apartment in Paris, and is crazy about the Tour de France, which he attends every year. He is renowned for the way he plays the marketing game, is a very clever tactician and a damn good designer. Some of his patent, however, has been criticized for being the work of others, but aren't they all???
There is indeed much more to an amplifier than the Class. I agree with this emphatically. Much of the sound quality comes from the voltage amplifier.......
Yes, the bipolar transistor on a CCS confers most of the stability, well aware of that, with its 1.2mV/degree Vbe shift, but in truth the LED driven mosfet is still very stable in practice.
Once again, and I've said this to your friend PMA, there is no need for negative, patronising tones when a novice asks for opinions/circuits. Strangely enough, Pavel, people do not see you as an individual, but rather as an ambassador for your country. A presence on this forum is a dynamic, world view of your nation, as others never see us as we see ourselves.......
Cheers,
Hugh
Thank you for your email; you explain yourself much better.
Bruce Candy was born in South Africa, and emigrated to Australia as a teenager. He was educated at Adelaide University, but IIRC dropped out of an engineering degree there as a young man. Bruce is a celebrated and famously generous eccentric, maintains an apartment in Paris, and is crazy about the Tour de France, which he attends every year. He is renowned for the way he plays the marketing game, is a very clever tactician and a damn good designer. Some of his patent, however, has been criticized for being the work of others, but aren't they all???
There is indeed much more to an amplifier than the Class. I agree with this emphatically. Much of the sound quality comes from the voltage amplifier.......
Yes, the bipolar transistor on a CCS confers most of the stability, well aware of that, with its 1.2mV/degree Vbe shift, but in truth the LED driven mosfet is still very stable in practice.
Once again, and I've said this to your friend PMA, there is no need for negative, patronising tones when a novice asks for opinions/circuits. Strangely enough, Pavel, people do not see you as an individual, but rather as an ambassador for your country. A presence on this forum is a dynamic, world view of your nation, as others never see us as we see ourselves.......
Cheers,
Hugh
Progg70,
The resistor in the emitter of a CCS is an essential element in the topology. You cannot strictly remove it......
A CCS relies on a bias voltage applied to the control electrode, usually a gate or base, and an ohmic load in the source/emitter.
Let's choose a mosfet, which might have a bias voltage of 3V8 at 3A with say 15V across the drain/source.
If we put a 4V46 voltage at the gate with respect to rail (say ground in our case), then the mosfet carries 3V8 volts of bias, and the 0R22 source resistor drops the remainder, (4.46 - 3.8) = 0.66V. The 4V46 volts can be applied by a string of LEDs, a zener, or an active device like a bipolar transistor. Since we now have a 0R22 resistor between source and rail with 0.66V across it, then the current flow through the mosfet AND the resistor is precisely 3A.
So, the resistor is crucial to correct CCS operation and can't be left out unless the fine control over the gate voltage is incredibly tightly controlled (which is possible with an opamp, incidentally).
Cheers,
Hugh
The resistor in the emitter of a CCS is an essential element in the topology. You cannot strictly remove it......
A CCS relies on a bias voltage applied to the control electrode, usually a gate or base, and an ohmic load in the source/emitter.
Let's choose a mosfet, which might have a bias voltage of 3V8 at 3A with say 15V across the drain/source.
If we put a 4V46 voltage at the gate with respect to rail (say ground in our case), then the mosfet carries 3V8 volts of bias, and the 0R22 source resistor drops the remainder, (4.46 - 3.8) = 0.66V. The 4V46 volts can be applied by a string of LEDs, a zener, or an active device like a bipolar transistor. Since we now have a 0R22 resistor between source and rail with 0.66V across it, then the current flow through the mosfet AND the resistor is precisely 3A.
So, the resistor is crucial to correct CCS operation and can't be left out unless the fine control over the gate voltage is incredibly tightly controlled (which is possible with an opamp, incidentally).
Cheers,
Hugh
HI Progg,
You are partly right.
The source/emitter resistor in current source/sink does indeed set the current (together with the reference voltage) but it is important to the impedance of the current source seen at the drain/collector of the active device.
The load side of the current source will have an active device passing the audio; normally a current amplifier. This device wants to see a constant current over a wide voltage range (the voltage range is the audio signal). If we think of the CCS just as a circuit element, then if we can vary the voltage across it over a range yet the current stays constant it must in fact be simulating a very high impedance, since variations in voltage show no variation in current (remember, Ohms Law states that deltaV/deltaI = RESISTANCE).
It turns out that this resistor has considerable influence over the impedance; making it larger (and dropping more voltage across it) increases the impedance of the current source seen by the load at the collector/source.
Send me your email, and I will include schematics; I have software problems putting graphics into this forum.
Cheers,
Hugh
You are partly right.
The source/emitter resistor in current source/sink does indeed set the current (together with the reference voltage) but it is important to the impedance of the current source seen at the drain/collector of the active device.
The load side of the current source will have an active device passing the audio; normally a current amplifier. This device wants to see a constant current over a wide voltage range (the voltage range is the audio signal). If we think of the CCS just as a circuit element, then if we can vary the voltage across it over a range yet the current stays constant it must in fact be simulating a very high impedance, since variations in voltage show no variation in current (remember, Ohms Law states that deltaV/deltaI = RESISTANCE).
It turns out that this resistor has considerable influence over the impedance; making it larger (and dropping more voltage across it) increases the impedance of the current source seen by the load at the collector/source.
Send me your email, and I will include schematics; I have software problems putting graphics into this forum.
Cheers,
Hugh
AKSA said:
Hi Hugh,
not sure of which resistor you are meaning, if you mean the "degeneration" (source/emitter)resistor of the CCS it doesn't have any affect on the CCS's output impedance, only the voltage gain as per R.Elliot.
EDIT: I conclude that it depends actually on the early effect when Vds/Vce is varying.
But which is what, how can we determine on what is affectimg the Zout?
Because if there would be no early effects the current would be absolutely stable independently of Vds/Vce and SR/ER.
Cheers Michael
Michael,
Yes, I am referring to the degeneration resistor.
The Art of Electronics, 2nd edition, specifies around 0.5% Early effect variation of current with a simple, bipolar current source across the whole load voltage range. hfe variation adds another 0.05% due to the non-zero impedance of the voltage reference, and temperature effects add another 0.2%.
These are very small variations, though of course the mosfet calculations are necessarily very different. Even if they were an order of magnitude greater, they are still small in the overall picture. What is clear is that the Wilson current mirror and the bipolar sense concept you espoused earlier are much superior; but in practice, and in an output stage where the operating impedances are down to 8R, the improvements by using these topologies would be technically negligible. After all, an improvement in impedance from 40K to 300K would pale into insignificance if the load is 8R, and I very much doubt you'd hear a sonic difference at a 3A source current!!
When someone asks for a current source in a high current, non-instrument application like this one, a simple solution is often better to establish the working principle. Refinements are possible, but should be approached only after the original, simple solution is found wanting. The truth is that improvements to the current source I proposed are tiny in the sonic/stability sense, so I go for simplicity, known reliability, and low parts count. I try to include LEDs where possible because they are cheap, quiet, and give visual indication of operation, which for the beginner is actually very important because you can't see electricity.
Cheers,
Hugh
Yes, I am referring to the degeneration resistor.
The Art of Electronics, 2nd edition, specifies around 0.5% Early effect variation of current with a simple, bipolar current source across the whole load voltage range. hfe variation adds another 0.05% due to the non-zero impedance of the voltage reference, and temperature effects add another 0.2%.
These are very small variations, though of course the mosfet calculations are necessarily very different. Even if they were an order of magnitude greater, they are still small in the overall picture. What is clear is that the Wilson current mirror and the bipolar sense concept you espoused earlier are much superior; but in practice, and in an output stage where the operating impedances are down to 8R, the improvements by using these topologies would be technically negligible. After all, an improvement in impedance from 40K to 300K would pale into insignificance if the load is 8R, and I very much doubt you'd hear a sonic difference at a 3A source current!!
When someone asks for a current source in a high current, non-instrument application like this one, a simple solution is often better to establish the working principle. Refinements are possible, but should be approached only after the original, simple solution is found wanting. The truth is that improvements to the current source I proposed are tiny in the sonic/stability sense, so I go for simplicity, known reliability, and low parts count. I try to include LEDs where possible because they are cheap, quiet, and give visual indication of operation, which for the beginner is actually very important because you can't see electricity.
Cheers,
Hugh
Thanks now we are cooking.
AKSA I tried to mail you but your settings don’t allow that. You can mail me at progg70@mysigt.com.
Thanks Amplifierguru for your schematic.
AKSA I tried to mail you but your settings don’t allow that. You can mail me at progg70@mysigt.com.
Thanks Amplifierguru for your schematic.
amplifierguru said:Hi Progg70,
If your after a 3A CCS that simple with high regulation here's one -
and it lights up.
Cheers,
greg
Oh Greg...
May I throw back an old bone...
Thanks Hugh for your input!
AKSA said:
I happend to have the same book, I looked through it but couldn't find particularly your statements, well it's not a big issue.
But in the same book one finds the saying about Early effect gives up to 25% variation in current, but it's on the other hand without R degen.
I still wonder about Rod Elliots statement about output impedance, as I mentioned in post #35.
Is it so easy that we can estimate ouput impedance on the formula you gave earlier, (remember, Ohms Law states that deltaV/deltaI = RESISTANCE)?
Note that I'm not trying to drive you down Hugh, I'm just curious if there's other explanations.
I have actually Googled a bit but couldn't so far find anything with deeper insight CCS's, may anyone like to add something to the discussion?
Cheers Michael
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.