If only education was free and spoonfed.
The book is informative, BTW.
Amazing...
Almost all countries have a free education system...
In Europe, i had to pay about 100$/year as education
fees in an university that is on the world Top100.....
Yeah, not exactly free as in 'costless' then.But internet is free at any public library.
Problem is there is lots of false info floating around out there. That makes it a bit confusing.
Hi CBS240,
I have actually found Wikipedia to be remarkably good, as long as one filters it carefully. The references that are often provided also lead to other material that can be helpful as a sanity check or for a deeper coverage of the topic. Of course, we always have to avoid the trap of thinking that anything that is "published" on the Internet is true.
Cheers,
Bob
Hi masood,
The retail price is $60, but it can be had in the US on Amazon for about $38. Not sure what kind of pricing is in place overseas, however.
Cheers,
Bob
Hi Pete,
These are good questions and I don't have a handy answer. However, I seem to recall having gotten some Fairchild parts that were 9240's a few years back (I think) so that I could verify for myself that the effect existed and that it did not exist in the Fairchild parts. That verification was successful in both regards.
I don't know for sure, but I'd guess that the Vishay parts have the issue, since I doubt that they would change the process, but one never knows. I have not measured any of the Vishay 9240's.
A few years back I corresponded with a guy from IRF about the issue but he was of no help. The mechanism for the effect is unclear, but in the attached notes I show how I measured it, what the results were, and speculation on the nature of the mechanism.
The measurements were done in a common-source configuration, where they show up well.
I must admit that I have not made measurements on its effect in a typical amplifier source-follower output stage configuration.
Cheers,
Bob
Thanks very much Bob that looks like an easy test so I might buy or sample some of the Vishay parts, or call them and ask. Hansen mentioned some odd behavior on a curve tracer do you know anything about that issue? Thanks again, just trying to learn as much as possible about the issue.
Hansen also mentioned that all the P-type IR devices that they measured had it and that all from other companies did not.
I've only been able to find IR and Vishay parts, and some Fairchild parts but only from one questionable source.
Last edited:
Hi Pete,
Yes, I do seem to remember Charles Hansen mentioning something about curve tracer results on those parts. I haven't looked at curve traces on them, however. I think Nelson Pass may have been the first one to bring it to people's attention. I definitely encourage you to get some Vishay versions and see if they still behave that way. Its a shame to see some parts disappearing.
I suspect that the degree of the effect of the p-channel anomalous behavior may depend somewhat on what circuit they are used in, so some measurement of output stages in open-loop with anomolous and non-anomalous parts might be helpful. As I mentioned, it looked like the behavior was characterized by some reduction in ac transconductance at higher audio frequencies, particularly at higher drain currents. While we always like to avoid frequency-dependent nonlinearities (like nonlinear junction capacitance), we also need to put this in perspective with the fact that the transconductance of the MOSFETs is varying all over the place in a class AB output stage anyway, especially in the crossover region where they suffer transconductance droop.
Cheers,
Bob
Hi Samuel,
I was referring to the resistance that I placed in series with the gate of the MOSFET. If the anomaly was somehow due to a feedback mechanism to the gate (such as Miller effect via Cgd), then I would have expected to see the effect decrease substantially as the resistance in series with the gate was reduced below 100 ohms.
The effect almost seems like there is some gate charge that is associated with a long time constant. If this gate charge moved with the gate at low frequencies, then it might tend to contribute to transconductance, while if it did not move with higher frequencies it would tend not to contribute to transconductance at higher frequencies.
Cheers,
Bob
I would appreciate some comments from Samuel Groner on a output triple inspired by his comments on D.Self book.
http://www.diyaudio.com/forums/solid-state/188143-groner-triple.html
dado
http://www.diyaudio.com/forums/solid-state/188143-groner-triple.html
dado
Hi dado,
This is a Triple implemented with a diamond driver instead of a pair of emitter followers (driver/pre-driver) up front. It is a good circuit and has some nice current-limiting and biasing characteristics. I think I showed a Quad output stage based on a diamond driver in my book. The pre-driver in this arrangement is sometimes referred to as a folded emitter follower. I have used the folded emitter follower as the driver in some of my MOSFET amps.
I don't see much advantage to replacing the current sources with bootstrapped resistors just to gain a volt or so of swing. Sometimes the extra natural drop of the regular Triple helps by keeping the output transistors a bit away from the rails when there is clipping.
However, note that the diamond driver does buy the VAS a bit more headroom.
Notice also that the pre-driver in the diamond driver does not have to have a high-voltage rating and its collector is bootstrapped with the signal. This means the transistor can be fast, needs to dissipate little power, has its collector-base capacitance largely bootstrapped out of the picture, and is less subject to Early effect. The "heavy lifting" is done by the current source instead of the pre-driver, and the current source is not directly involved in passing the signal.
Cheers,
Bob
Hi Bob,
Thank you a lot for explanation. Ordinary CCS prefered over this bootstraped one?
Deplation-Mode Vertical DMOS FETs from Supertex inc. look very promising to be used for simple CCS(DN2535, DN2540 in TO220). I tried to use Supertex Spice model in LTspice with no cuccess, so I used JFET in simulation. Looking in Supertex datasheets those FETs are better for CCS.
dado
Thank you a lot for explanation. Ordinary CCS prefered over this bootstraped one?
Deplation-Mode Vertical DMOS FETs from Supertex inc. look very promising to be used for simple CCS(DN2535, DN2540 in TO220). I tried to use Supertex Spice model in LTspice with no cuccess, so I used JFET in simulation. Looking in Supertex datasheets those FETs are better for CCS.
dado
Thanks for the clarification--I thougth so, but I'd have called this "source impedance", so I wasn't sure.
Maybe I'm off here, but is this effect not more like an output conductance effect? It looks to me as if the drain resistor were replaced with an active transimpedance stage (opamp IV-converter), the effect would not be observed. However, if it were indeed attributed to transconductance, there would be no sensitivity to the drain load impedance.
Samuel
Hi dado,
I've used both BJT and HEXFET MOSFETs with success for these current sources. It is, of course, important that they have adequate voltage, current, dissipation and SOA ratings for the application.
Cheers,
Bob