Well actually..........
Reference the current output's load impedance to the supply rails which ideally are AC ground and it looks exactly like a folded cascode. I would call any cascode where the DC current through the gain and cascode transistor is different a folded cascode although I don't know if that follows a strict definition of the "folded cascode" ...........
Reference the current output's load impedance to the supply rails which ideally are AC ground and it looks exactly like a folded cascode. I would call any cascode where the DC current through the gain and cascode transistor is different a folded cascode although I don't know if that follows a strict definition of the "folded cascode" ...........
Re: Well actually..........
Certainly a folded cascode allows you to use different currents in each leg (the input transistor and the cascode transistor). But to me a "folded cascode" is one that uses opposite polarity devices, whereby the signal current "folds" back in the other direction.
Fred Dieckmann said:
Certainly a folded cascode allows you to use different currents in each leg (the input transistor and the cascode transistor). But to me a "folded cascode" is one that uses opposite polarity devices, whereby the signal current "folds" back in the other direction.
jam said:
Hi Jam,
no degeneration emitter resistors, why? As far as I understand the stage can be even more linear with them if the gain is not that necessary irrespective of CCT.
What is the base voltage at the cascode?
Took a look at those MPS transistors ON Semi datasheet and seemed quite ok, a bit low hfe but maybe not an issue!?
Have you ever tried BC550C/560C, wonder if they could not be a better choice, or what was the choise to go with MPS??
Regards,
Michael
Michael,
I was a lot younger then and had not read Borbely's article on degeneration and linearity. In later experiments I found that you could adjust the amount of degeneration, and there was usually a sweet spot where it sounded best. Cascodes still had the same effect on the sound.
I have never tried the BC550/560 but I am sure they are good devices. I usually use Mosfets and Jfets for differentials now, these devices seldom need degeneration in differentials because their transconductance is much lower than a bipolar.
In the bipolar differential circuit I set the base of the cascode pair at about half the rail voltage. The biasing consisted of a zener and 4ml current source with a 0.47 uF cap across the zener to get rid of noise.
Regards,
Jam
I was a lot younger then and had not read Borbely's article on degeneration and linearity. In later experiments I found that you could adjust the amount of degeneration, and there was usually a sweet spot where it sounded best. Cascodes still had the same effect on the sound.
I have never tried the BC550/560 but I am sure they are good devices. I usually use Mosfets and Jfets for differentials now, these devices seldom need degeneration in differentials because their transconductance is much lower than a bipolar.
In the bipolar differential circuit I set the base of the cascode pair at about half the rail voltage. The biasing consisted of a zener and 4ml current source with a 0.47 uF cap across the zener to get rid of noise.
Regards,
Jam
Re: Re: Well actually..........
That's pretty much my understanding, and it's typically used to
accomplish level shifting as well, mostly in video circuits, where
it's quite popular. Folded cascode does not preclude equal
currents for the two devices, in fact it seems to me that typically
they are set the same.
Charles Hansen said:
That's pretty much my understanding, and it's typically used to
accomplish level shifting as well, mostly in video circuits, where
it's quite popular. Folded cascode does not preclude equal
currents for the two devices, in fact it seems to me that typically
they are set the same.
Hello Nelson,
The variable bias patent was a good one and spawned many imitations. My favorite story about that is regarding the 1981 AES paper from a Sansui engineer that presented a slight variation on your circuit. Then 15 years later, Madrigal copied the Sansui circuit exactly (for the Mark Levinson No 33 and later models) and claimed it as their own. Here's the Madrigal circuit:
This is what Madrigal says in the owner's manual for the No 434:
"Mark Levinson amplifiers use a proprietary adaptive biasing scheme
developed by Madrigal (and first introduced in the Nº33 Reference
Monaural Amplifiers)..."
And here is the Sansui circuit. Does anyone else spot any similarities?
The variable bias patent was a good one and spawned many imitations. My favorite story about that is regarding the 1981 AES paper from a Sansui engineer that presented a slight variation on your circuit. Then 15 years later, Madrigal copied the Sansui circuit exactly (for the Mark Levinson No 33 and later models) and claimed it as their own. Here's the Madrigal circuit:
An externally hosted image should be here but it was not working when we last tested it.
This is what Madrigal says in the owner's manual for the No 434:
"Mark Levinson amplifiers use a proprietary adaptive biasing scheme
developed by Madrigal (and first introduced in the Nº33 Reference
Monaural Amplifiers)..."
And here is the Sansui circuit. Does anyone else spot any similarities?
Attachments
Jam,
Ok I see!
But do I now understand you right that in your present/latest(?) design/(philosophy) you use some FET at input and the given MPS transistors as cascodes?
If you find some interest and time try BC550/560 as cascodes, I would be curious to hear from you how they would sound in your design!
Regards,
Michael
Ok I see!
But do I now understand you right that in your present/latest(?) design/(philosophy) you use some FET at input and the given MPS transistors as cascodes?
If you find some interest and time try BC550/560 as cascodes, I would be curious to hear from you how they would sound in your design!
Regards,
Michael
I fold on the definition....remind me never to play cards with you guys
"Folded cascode does not preclude equal currents for the two devices, in fact it seems to me that typically they are set the same."
I think the ability make the DC current different is one of the best features of the topology. Level shifting and the ability to use a larger load resistor for a given rail voltage are real handy. Maybe the best thing about circuit is that it is very easy to use the same node (ground typically) for the reference point for the load and degeneration resistors often improving power supply rejection. I will even confess to using both a regular and folded cascode in the same gain stage for the added headroom for voltage swing while keeping the voltage across the gain transistors low, very useful for jfets..........
"Folded cascode does not preclude equal currents for the two devices, in fact it seems to me that typically they are set the same."
I think the ability make the DC current different is one of the best features of the topology. Level shifting and the ability to use a larger load resistor for a given rail voltage are real handy. Maybe the best thing about circuit is that it is very easy to use the same node (ground typically) for the reference point for the load and degeneration resistors often improving power supply rejection. I will even confess to using both a regular and folded cascode in the same gain stage for the added headroom for voltage swing while keeping the voltage across the gain transistors low, very useful for jfets..........
Charles Hansen said:The variable bias patent was a good one and spawned many imitations. My favorite story about that is regarding the 1981 AES paper from a Sansui engineer that presented a slight variation on your circuit. Then 15 years later, Madrigal copied the Sansui circuit exactly (for the Mark Levinson No 33 and later models) and claimed it as their own. Here's the Madrigal circuit:
An externally hosted image should be here but it was not working when we last tested it.
This is what Madrigal says in the owner's manual for the No 434:
"Mark Levinson amplifiers use a proprietary adaptive biasing scheme
developed by Madrigal (and first introduced in the Nº33 Reference
Monaural Amplifiers)..."
And here is the Sansui circuit. Does anyone else spot any similarities?
It's fair to say that Hawksford and Cordell did a better job, as
the circuit turned out to be most useful with Mosfets.
I admire your erudition, inasmuch as I was unaware of most of
this. Such a good competitor, keeps me sharp.
Schwing!!!
Re: I fold on the definition....remind me never to play cards with you guys
I agree, although I think the really best feature is the level
shifting part, so that a single stage can have a DC coupled
input and output and voltage gain with only a single stage.
(Of course that's because I don't consider a cascode a gain
stage). The X1000 and X600 used both types of cascodes also.
Fred Dieckmann said:
I agree, although I think the really best feature is the level
shifting part, so that a single stage can have a DC coupled
input and output and voltage gain with only a single stage.
(Of course that's because I don't consider a cascode a gain
stage). The X1000 and X600 used both types of cascodes also.
Why the output stage cascode sounds compressed?
Grollins says "What you measure is not what you hear". Cascodes in output stages can make measurement figures better, but not a better sound.
What about this Mr.Pass patent? The device 13 and 11 are cascoded so the current source 33 can give big bias current in low voltage (less heat). Will this makes the sound better or the same "Compressed" inspite of there is a big bias current flowing in the output transistor due to this cascode? Has anyone have tried this / simulate this patent?
Grollins says "What you measure is not what you hear". Cascodes in output stages can make measurement figures better, but not a better sound.
What about this Mr.Pass patent? The device 13 and 11 are cascoded so the current source 33 can give big bias current in low voltage (less heat). Will this makes the sound better or the same "Compressed" inspite of there is a big bias current flowing in the output transistor due to this cascode? Has anyone have tried this / simulate this patent?
Attachments
Charles Hansen said:
One way of thinking about this is by viewing it as the dual of the power amp output stage error correction, except it's operating in current mode.
With the voltage mode output stage error correction, we ideally have a unity gain voltage amplifier. The difference between the output voltage and input voltage is computed and subtracted from the input drive voltage in such a way that the error ideally cancels out. This makes the voltage gain very near 1, and reduces distortion and output impedance.
With the enhanced cascode, the current-mode equivalent of this is occurring. We ideally have a unity gain current amplifier. The error current is sitting there waiting for us already. It's the base current of the common base amp. It consists not only of linear errors due to finite beta, but also distorted currents due among other things to the modulation of the collector-base capacitance by the varying collector-base voltage. So we take this current and subtract it from the drive current (by feeding it around to the emitter of the common emitter amp) such that the error (almost) cancels out. So now you have a current amplifier whose current gain is closer to one, its distortion is reduced, and its output impedance is increased.
There seems to be at least two disadvantages of this circuit. Clipping behavior is not too good. When the common base stage saturates, the base current goes way up and causes the base voltage to "bounce", since it's not fixed as with a common base amp. Second, the base current feedback mechanism seems to cause the potential for oscillation in the 50-100 MHz range.
Charles Hansen said:
Yes Andy's explanation will.
Another way is to visualise the circuit working in a real
application, for example, folded cascode voltage stage in an amp
or pre amp.
20k load R to ground and assume an ideal current sink to neg
supply. If we modulate the cascodes current with an ideal
sinusoidal wave the resistor should swing an ideal voltage sine
wave. But it doesn't because some of the current modulated into
the cascode is lost through the base of the P ch bjt and there will
also be a reactive current between the collector and base due to
miller C.
With Hawksfords cascode this reactive current is fed straight
back into the circuit at the upper P ch BJT's emitter and is
not "lost". It is in effect + feedback and is modulating the
top current source to correct the lower one.
It is in effect doing a similar thing that the super
pair current source does, but with 1 less device.
Regards,
Terry
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.