Two dodes in series could be used and then there is more voltage over the emitter resistor. From the standpoint of output impedance it is highest when the voltage over the emitter resistor is beta x Ut = ca.200 x 26mV = 5.2V . Then the product beta x Ua ( early voltage ) is constant. I do not know if a high output impedance is desirable here and a loss
of 5.2V plus Uce is a lot. One way whould be to use a transistor with a low beta.
of 5.2V plus Uce is a lot. One way whould be to use a transistor with a low beta.
Personally, I normally use resistors, BUT that makes for a problem of greatly increased power supply sensitivity. So, why not use 100 ohms or so, BUT change the input devices to 2SK369 and run about 5 ma per device? Then the Gm will jump to 40K from 28K. We don't need an extreme voltage swing, because the second stage will make the output about 30 times higher, in any case.
Joachim,
Emitter resistors, you are referring to the current sources for the folded cascode, yes?
In this case, why not use a current mirror with low value resistors and bias it with a J-Fet that also can supply the bias for the CCS in the tail of the J-Fets which also could be a low value resistor based Current Mirror?
Ciao T
Emitter resistors, you are referring to the current sources for the folded cascode, yes?
In this case, why not use a current mirror with low value resistors and bias it with a J-Fet that also can supply the bias for the CCS in the tail of the J-Fets which also could be a low value resistor based Current Mirror?
Ciao T
No, I am thinking it through, just like everyone else. I do know, as a building block, what it should do. In fact, different versions might be more optimum for different applications. However, for a phono input application: Low input noise, decent output current swing, and no extra caps would be nice.
If I were to make an all out approach with this design, I would add bias string bypass caps, as I did with the JC-80.
A comp fet output stage might be ideal for an op amp like configuration, such as was used in the Levinson JC-2 phono stage. Etc.
If I were to make an all out approach with this design, I would add bias string bypass caps, as I did with the JC-80.
A comp fet output stage might be ideal for an op amp like configuration, such as was used in the Levinson JC-2 phono stage. Etc.
Joachim,
Sure. I take Jonathan Carr's point about "use CCS in folded cascode" reasonably serious.
By using a J-Fet CCS near zero Tempco as bias source and using current mirrors on top and bottom of the input the whole thermal issues should be minimised. One would of course have to temperature compensate the base voltage for the two cascode transistors as well. Noise from the resistors in the circuit can be also minimised.
Complexity is only marginally increased.
I have been thinking about using Mosfet's in the folded cascode, but what easily available p-channel types are available which do not need extreme matching? Then adding an N-Channel Mosfet current mirror is trivial.
Using the current mirror on the folded cascode double slew rate and transconductance at the output node, over using a pure CCS load, though that could be employed, using the same bias source as the tail CCS.
Ciao T
Sure. I take Jonathan Carr's point about "use CCS in folded cascode" reasonably serious.
By using a J-Fet CCS near zero Tempco as bias source and using current mirrors on top and bottom of the input the whole thermal issues should be minimised. One would of course have to temperature compensate the base voltage for the two cascode transistors as well. Noise from the resistors in the circuit can be also minimised.
Complexity is only marginally increased.
I have been thinking about using Mosfet's in the folded cascode, but what easily available p-channel types are available which do not need extreme matching? Then adding an N-Channel Mosfet current mirror is trivial.
Using the current mirror on the folded cascode double slew rate and transconductance at the output node, over using a pure CCS load, though that could be employed, using the same bias source as the tail CCS.
Ciao T
Hi John,
I'm glad to see we are agreeing on using an N channel JFET design, but I have been pushing that from Day One, not coming around to it
. I'm glad to see you are coming around to it. However, in your post 461 you claimed that one had to buffer the input with a 5564 source follower, which is not what I advocate, so we may still see things differently. I speculated that you were buffering with a JFET source follower so that you could follow with a complementary BJT stage, but you did not confirm that. Is that the case?Cheers,
Bob
Bob, I prefer complementary, and continue to use it, but it is almost impossible to obtain the low noise P channel parts in the quantity and Idss matching that make them practical for most people. Even Parasound has asked me to limit their use, when possible.
This design is an exercise of making a 'special', i.e. a phono input gain block that is linear, yet not complementary. The approach used by Scott Wurcer in the AD797 is pretty good, and with a few changes, it could be used optimally.
I do use complementary devices for the output stage, but they can be lower Gm parts, if necessary, just like the Levinson JC-2 used, back in 1974. I have used this simple, but elegant output stage for a number of pro products, including master recorders and studio boards.
This design is an exercise of making a 'special', i.e. a phono input gain block that is linear, yet not complementary. The approach used by Scott Wurcer in the AD797 is pretty good, and with a few changes, it could be used optimally.
I do use complementary devices for the output stage, but they can be lower Gm parts, if necessary, just like the Levinson JC-2 used, back in 1974. I have used this simple, but elegant output stage for a number of pro products, including master recorders and studio boards.
Last edited:
Well-stated, John. It is indeed a shame that complementary JFETs are so difficult to obtain.
Cheers,
Bob
I know they did not treat you right but here is some information on the new version of the Ortofon MC25E. This is a rework of a 25 year old design. It has quite low impedance and inductance and a decent output voltage. In Germany it costs somewhat less then 400,-€ so it is semi affordable too.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.