To be honest mjona, while I acknowledge that this is a 40+ year old circuit, for today's standards there nothing special here. Besides R5 (which you mentioned before), the next weakest link for DC gain is Q5. A Wilson mirror will give the next step up in DC gain. Even then, I still don't see anything to be mesmerized.
So ... yes, cascodes boost the output impedance of a node by ~gm*r_pi (i.e. Beta). Note that this not good enough by itself since the impdedance of a node becomes the lowest of what is attached to that node. In the case of the node at the collector of Q5, it will be Q5 itself since the impedance into the collector of Q5 is:
(1+gm*RV1/2)*ro
For reference, the impedance into the collector of Q3 is ~(1+gm*r_pi)*ro or ~(1+beta)*ro and into the MOSFET is infinite.
Finally, this is all good at low frequency. At frequency, the C_GS of the MOSFET will dominate the impedance at the node.
Bottomline: Cascode boost node impedances, impedance at a node is based on the lowest impedance, over frequency all bets are off.
So ... yes, cascodes boost the output impedance of a node by ~gm*r_pi (i.e. Beta). Note that this not good enough by itself since the impdedance of a node becomes the lowest of what is attached to that node. In the case of the node at the collector of Q5, it will be Q5 itself since the impedance into the collector of Q5 is:
(1+gm*RV1/2)*ro
For reference, the impedance into the collector of Q3 is ~(1+gm*r_pi)*ro or ~(1+beta)*ro and into the MOSFET is infinite.
Finally, this is all good at low frequency. At frequency, the C_GS of the MOSFET will dominate the impedance at the node.
Bottomline: Cascode boost node impedances, impedance at a node is based on the lowest impedance, over frequency all bets are off.
Last edited:
The image I posted was in response to kokoriantz initial request - he was interested in MOSFETs and cascode circuit. I left it up to him to decide what to make of that and whether he could make something better out of it. I knew others would not like this part of the image - hence my take it or leave it comment.
As for the first stage I am surprised that you think that on today's standards there is nothing special there . You ran a SPICE tutorial recently - this based on a popular power amplifier project. The first stage structure in that looks the same albeit in complementary form.
Anyway I think we should just let this one slide.
As for the first stage I am surprised that you think that on today's standards there is nothing special there . You ran a SPICE tutorial recently - this based on a popular power amplifier project. The first stage structure in that looks the same albeit in complementary form.
Hi mjona, you are referring to the circuit in post #136 or something else?
Hi mjona, you are referring to the circuit in post #136 or something else?
Got it. BTW, I am not saying cascodes are not good, they are great, they are just not novel, they have been around forever (since the era of tubes) hence my original comment.
Also there are tons of variations:
- Bootstrap versions where the base is driven so that Q1 and Q2 VCE stays constant - improves distortion due to VA.
- With current recycling where you take the current of the cacode transistor base and re-inject it back into the collector (think of a darlington) - Improves beta induced distortion.
- Mosfet and JFET based cascoded - JFET are interesting since they do not need a referece due to the Source being at a lower voltage than the Gate in a p-channel.
- etc.
But yes... you are right, the simple cascode has been discussed to death, so no need to go further on it.
Also there are tons of variations:
- Bootstrap versions where the base is driven so that Q1 and Q2 VCE stays constant - improves distortion due to VA.
- With current recycling where you take the current of the cacode transistor base and re-inject it back into the collector (think of a darlington) - Improves beta induced distortion.
- Mosfet and JFET based cascoded - JFET are interesting since they do not need a referece due to the Source being at a lower voltage than the Gate in a p-channel.
- etc.
But yes... you are right, the simple cascode has been discussed to death, so no need to go further on it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.