... I don't think the CFB version will have such a high PSRR since the LTP has a high CMRR by itself ???
Of course, to know if the CFB sounds better than the original SKA you'd have to build both....
....
I just noticed that you had modified your post...
Probably lower PSSR than differential input but also probably high enough for the job because of the very high gain of the input stage (increased by the bootstrap).
Just to be clear for everyone, looking at it in a positive way it could be more an improved CFB amp than an inferior voltage feedback type...
And yes why not build both and LISTEN
Fab
Last edited:
Fab, why are you bootstrapping the input stage - the output stage has gain, so I am trying to see what you are gaining here. I would also comment that your current sources are now quite complex (4 transistors) - could you not simplify and use the transistors in the gain stages. One option to consider is to use a conventional CFB structure with source follower output (so no gain on output stage) but bootstrap the mosfet gate drive to give you rail to rail output swing.
Anyway, looks like an interesting project - good luck!
(I doubt I will go back to VFA again - accept for small signal - despite the very good sound I am getting from my e-Amp
Anyway, looks like an interesting project - good luck!
(I doubt I will go back to VFA again - accept for small signal - despite the very good sound I am getting from my e-Amp
Fab, why are you bootstrapping the input stage - the output stage has gain, so I am trying to see what you are gaining here. I would also comment that your current sources are now quite complex (4 transistors) - could you not simplify and use the transistors in the gain stages. One option to consider is to use a conventional CFB structure with source follower output (so no gain on output stage) but bootstrap the mosfet gate drive to give you rail to rail output swing.
Anyway, looks like an interesting project - good luck!
(I doubt I will go back to VFA again - accept for small signal - despite the very good sound I am getting from my e-Amp
Hi Bonsaï
I am "gaining" gain
thus even more feedback. I already have CFB amps such as 3 variant of VSSA, F5, F5 Juma with source follower and so on. The current sources are 3 transistors each for now instead of usually 2 with BJT. I am open to suggestions but not sure to understand how to compensate the thermal drift of MOSFET using the gain stages...The bootstrap goal in this case is for more gain and not mainly for close to rail-to-rail operation.
Like you I like CFB amps a lot
Thanks for your comment and please more info on your idea of the simplification of the current sources
Fab
As we all are CFB fans here it is normal to support each other. So for both CCS you can use many well known schematics, but since CCS currrent for your amp is very small, the best solution would be to use one j-fet adjustable current source. Small constant current means small thermal dissipation, so there will be only small thermal drift expected from this kind of CCS. Since j-fet-s usually have Vds 30 Vmax you can insert drain serial resistor to take an excess voltage drop for j-fet to operate in normal conditions.
Regarding thermal stability one simple solution would be to thermally couple j-fet CCS to the main heatsink, since j-fets have negative tempco, and there you have a negative thermal feedback to stabilize output's bias current.
Regarding thermal stability one simple solution would be to thermally couple j-fet CCS to the main heatsink, since j-fets have negative tempco, and there you have a negative thermal feedback to stabilize output's bias current.
CCS
Yes we are fans
I thought about single Jfet for CCS but the wide spreading of the IDSS among the same Jfet part number is a constraint because you end up with a very high value of potentiometer to cope for the extreme values of IDSS... A very high value for pot makes it difficult to tweak if one of the Jfets has a high IDSS... The solution would be to select a specific value of high IdSS , select lower value than IDSS current and then screen the Jfets for that value (to use a lower pot value..).
Also, the Jfet ( usually to-92 package ?) does not attach so easily as a to-220 device on the MOSFET or heat sink for the compensation. However I may still consider the screened per-defined IDSS but it is more a pain to find them...
Thanks
Fab
Fab
More CFB amps --- YES!
Thx-RNMarsh
As we all are CFB fans here it is normal to support each other. So for both CCS you can use many well known schematics, but since CCS currrent for your amp is very small, the best solution would be to use one j-fet adjustable current source. Small constant current means small thermal dissipation, so there will be only small thermal drift expected from this kind of CCS. Since j-fet-s usually have Vds 30 Vmax you can insert drain serial resistor to take an excess voltage drop for j-fet to operate in normal conditions.
Regarding thermal stability one simple solution would be to thermally couple j-fet CCS to the main heatsink, since j-fets have negative tempco, and there you have a negative thermal feedback to stabilize output's bias current.
Yes we are fans
I thought about single Jfet for CCS but the wide spreading of the IDSS among the same Jfet part number is a constraint because you end up with a very high value of potentiometer to cope for the extreme values of IDSS... A very high value for pot makes it difficult to tweak if one of the Jfets has a high IDSS... The solution would be to select a specific value of high IdSS , select lower value than IDSS current and then screen the Jfets for that value (to use a lower pot value..).
Also, the Jfet ( usually to-92 package ?) does not attach so easily as a to-220 device on the MOSFET or heat sink for the compensation. However I may still consider the screened per-defined IDSS but it is more a pain to find them...
Thanks
Fab
Fab
Last edited:
Jfet CCS
I have attached a version with jfet current source. So far the only info I have obtained for the jfet tempco is typical -0.4% ID variation per deg C increase from a curve in a book... Is that correct?
If so and mosfet tempco is about 6mv/deg C then
0.004 * ID = 0.006 / (R5 + R7) thus with R5=R7=1.5K then Iccs must be set at 500 ua. But that would mean screening mosfets with selected VGSoff to get the desited bias current...
Note that the schematics is not set at 500ua but more 800 ua or so.
Any comments?
Thanks
Fab
....
....
Regarding thermal stability one simple solution would be to thermally couple j-fet CCS to the main heatsink, since j-fets have negative tempco, and there you have a negative thermal feedback to stabilize output's bias current.
I have attached a version with jfet current source. So far the only info I have obtained for the jfet tempco is typical -0.4% ID variation per deg C increase from a curve in a book... Is that correct?
If so and mosfet tempco is about 6mv/deg C then
0.004 * ID = 0.006 / (R5 + R7) thus with R5=R7=1.5K then Iccs must be set at 500 ua. But that would mean screening mosfets with selected VGSoff to get the desited bias current...
Note that the schematics is not set at 500ua but more 800 ua or so.
Any comments?
Thanks
Fab
Attachments
Hi Fab
As long as you have VSSA input stage with CCS arrangement as it is now you have no worries at all, since this one is perfectly thermally stable and it acts as DC servo too.
Regarding thermal stability you have two mechanisms of possible thermal regulation in the circuit:
- CCS, the trick is to use lower Idss j-fet since these have greater delta Idds around point (current) of interest and its setup is simple, just calculate appropriate serial plus parallel resistors around Rgs trimmer to cover the whole range of Idss for the selected group
- Q3, Q4, they act in negative tempco here since by heating them they have higher Ic and thus actually lower Vgs of the outputs. So they must be mounted on the same heatsink as the outputs and will bring certain amount of tempco regulation
For the start you have something to consider, but I would definitely stuck on j-fet current sources as they bring much more to musical reproduction and PSRR than classic two BJT CCS or plain resistor, since they have very high output impedance and are very fast - almost ideal CCS.
Regards, LC
As long as you have VSSA input stage with CCS arrangement as it is now you have no worries at all, since this one is perfectly thermally stable and it acts as DC servo too.
Regarding thermal stability you have two mechanisms of possible thermal regulation in the circuit:
- CCS, the trick is to use lower Idss j-fet since these have greater delta Idds around point (current) of interest and its setup is simple, just calculate appropriate serial plus parallel resistors around Rgs trimmer to cover the whole range of Idss for the selected group
- Q3, Q4, they act in negative tempco here since by heating them they have higher Ic and thus actually lower Vgs of the outputs. So they must be mounted on the same heatsink as the outputs and will bring certain amount of tempco regulation
For the start you have something to consider, but I would definitely stuck on j-fet current sources as they bring much more to musical reproduction and PSRR than classic two BJT CCS or plain resistor, since they have very high output impedance and are very fast - almost ideal CCS.
Regards, LC
Hi Fab
As long as you have VSSA input stage with CCS arrangement as it is now you have no worries at all, since this one is perfectly thermally stable and it acts as DC servo too.
Regarding thermal stability you have two mechanisms of possible thermal regulation in the circuit:
- CCS, the trick is to use lower Idss j-fet since these have greater delta Idds around point (current) of interest and its setup is simple, just calculate appropriate serial plus parallel resistors around Rgs trimmer to cover the whole range of Idss for the selected group
- Q3, Q4, they act in negative tempco here since by heating them they have higher Ic and thus actually lower Vgs of the outputs. So they must be mounted on the same heatsink as the outputs and will bring certain amount of tempco regulation
For the start you have something to consider, but I would definitely stuck on j-fet current sources as they bring much more to musical reproduction and PSRR than classic two BJT CCS or plain resistor, since they have very high output impedance and are very fast - almost ideal CCS.
Regards, LC
Hi Lazy Cat
I think you have convinced me
I guess the PSRR of this CFB will not be that far away of the original SKA ( VFA diff input)...
Thanks for your input
Fab
Okay, so no one liked the idea of a cascoded j-fet front end because the j-fets are no longer easily available. How about using a cascoded opamp with its power supply pins connected as signal outputs and its output connected as the feedback node. Configure it as a buffer or voltage follower. If you insert a cap (1uF) in the feedback loop with a resistor (2M2) between the inverting input of the opamp and the output of the amplifier it will self servo. If you use a j-fet input opamp you can then be rid of both the input cap and the feedback dc blocking cap.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.