I should have mentioned that the driven cascode mainly helps the signal-related drain-source capacitance issue in the case where feedback is used in the first stage in a single-ended input design. In the non-feedback design it might help some with common-mode effects.
It is also worth noting that the gate-source capacitance issue, which can be more significant with multiple paralleled JFETs, will be less in a differential input stage that uses feedback, since the feedback effectively bootstraps the sources to follow the signal, reducing the effective gate-source capacitance seen by the cartridge.
Cheers,
Bob
It is also worth noting that the gate-source capacitance issue, which can be more significant with multiple paralleled JFETs, will be less in a differential input stage that uses feedback, since the feedback effectively bootstraps the sources to follow the signal, reducing the effective gate-source capacitance seen by the cartridge.
Cheers,
Bob
Hi Georg,
Thanks for bringing these Goldring specs to our attention. Its nice to see at least one manufacturer specifying inductance. It is significant to note that even the high-output Goldring, with a nominal output of 2.5mV, has inductance of only 200uH and resistance of 77 ohms.
I agree that unless we can gain higher confidence in those other cartridges that were cited to have L in the range of 50 uH, we can probably not worry too much for most MC cartridges frequency response being unduly influenced even when driving a virtual ground, as the pole of DCR and L will be at a fairly high frequency (as an example, 5 ohms and 10 uH will be at about 80 kHz).
Cheers,
Bob
Bappe, there is only one potential noise contributor that I see, and that is the use of 47 ohms as the current source Gm reducing component. I would go for 100 ohms or more. You might look at the 'improvement' on Spice, regarding noise contribution of an active load. I avoid this, myself, by just using a resistor, rather than a current source, but it does complicate the power supply.
You are of course correct reg noise performance and other comments. I did want to keep the component count down so i skipped the first cascode and you are right, the voltage should be taken down on the input devices. I do not consider the capacitance to be much of an issue though and i do think it would perform quite well open loop.
BR,
Anders
Hi bappe,
I'm not sure of the numbers, but the capacitance issue is likely to be more important for an MM application, given the much higher cartridge impedance of an MM cartridge. This will not be as bad if paralled devices are not used to get the noise unnecessarily low for the MM situation, although the use of even large single JFETs should be perhaps viewed with caution, especially in a no NFB MM input stage. We really don't want any MM cartridge interactions with the input stage.
BTW, this is why MM/MC preamps that use the same first stage for both types of cartridges are not a good idea. The optimum input stage for these two applications is quite different. A stage that has low enough noise for the MC application will tend to be quite sub-optimal for the MM application.
Cheers,
Bob
Oh so true! This one is sort of universal but probably best siuted for highish output mc's. If going for a low mc to mm it would take parallell devices and maybe a driven cascode.
//A
Bappe, YOUR design could be used effectively with BOTH MC or MM cartridges. However, it is the current source loading that is the potential problem.
I would hope that you can look at the active loads as potential noise amplifiers, injecting their noise into the second stage transistor bases. Kind of like competing signal sources.
I solved this problem, decades ago, BUT my solution would 'compromise' your circuitry with more parts, caps especially. Let us find a QUIET LED replacement with 3.3-6V drop. Hopefully 2 terminal, and low Z.
I would hope that you can look at the active loads as potential noise amplifiers, injecting their noise into the second stage transistor bases. Kind of like competing signal sources.
I solved this problem, decades ago, BUT my solution would 'compromise' your circuitry with more parts, caps especially. Let us find a QUIET LED replacement with 3.3-6V drop. Hopefully 2 terminal, and low Z.
Attachments
No, too much input capacitance from numerous paralleled JFETs. If you get the noise low enough for MC use, you have probably ended up with too much JFET input capacitance for MM use. Bear in mind, this capacitance is nonlinear.
Cheers,
Bob
Yes, you exceed the recommended C with the FET's alone let alone the cable. The non-linearity at these levels is a problem overstated IMNSHO.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.