hitsware said:
Not in this case. Mains hum (actually 100Hz) was due to the circuits crappy PSRR. That was later fixed with +/- rails provided by a pair of TO-3 regulators.
I guess that most of what I have said so far will be ignored, but issue of an amplifiers immunity if RF interference extends will beyond the input stage and with adequate discrete design it is not necessarily a BJT Vs JFET issue. The frequency compensation technique employed combined with the input stage overload margin, linearity and the ability of the input stage to amplify RF signals and pass them on the following stage (as well as the behaviour of the following stage) are every bit as important.
Cheers,
Glen
scott wurcer said:
.. and it could be much older.
A complete analysis of this capacitive feedback is in
James G. Holbrook:
LaPlace Transformation
I have only the german translation ISBN 3 528 13535 2,
bought it in january 1976 and it has abt. 10 pages on this
topic.
The english version is:
Laplace transformation for Electronic Engineers
(c) 1959, 1969 Pergamon Press, Oxford
The german translation is 1970, 1973
I really loved that book. Math for the solder fans.
(i could scan it within fair use constraints, but the back translation
of the 10 p. text would take some time. I assume AD has a nice library..)
regards, Gerhard
AM demodulation.
Hi Glen,
In the meantime I've done some simulations by myself. But first, I have to admit that I have created some confusion. I didn't mean single ended vs differential, but a non-complementary vs complementary VAS. Sorry.
In order to figure out whether a complementary VAS might reduce the susceptibility to HF, I've subjected a blameless like amp to a 0.1V 1MHz sine and fiddled with complementary input stages and VAS's. Also with ideal versions of the input stage or VAS, as well as ideal versions of both of them (at the same time). In all cases the DC level at the output was shifted by about 3.5mV by that HF signal. This clearly demonstrates that there was AM demodulation, but it doesn't stem from the input stage, neither from the VAS. Needless to say that I was puzzled.
Finally, I've replaced the output stage by an ideal 1:1 buffer and... the DC shift was (almost) gone.
Tomorrow, I will proceed further with AM susceptibility issues.
Cheers,
Edmond.
G.Kleinschmidt said:
Hi Glen,
In the meantime I've done some simulations by myself. But first, I have to admit that I have created some confusion. I didn't mean single ended vs differential, but a non-complementary vs complementary VAS. Sorry.
In order to figure out whether a complementary VAS might reduce the susceptibility to HF, I've subjected a blameless like amp to a 0.1V 1MHz sine and fiddled with complementary input stages and VAS's. Also with ideal versions of the input stage or VAS, as well as ideal versions of both of them (at the same time). In all cases the DC level at the output was shifted by about 3.5mV by that HF signal. This clearly demonstrates that there was AM demodulation, but it doesn't stem from the input stage, neither from the VAS. Needless to say that I was puzzled.
Finally, I've replaced the output stage by an ideal 1:1 buffer and... the DC shift was (almost) gone.
Tomorrow, I will proceed further with AM susceptibility issues.
Cheers,
Edmond.
john curl said:
jcx said:
This is most likely the J74. They are rated at 25V but I have it from
good authority that this is conservative. So individual devices may
go considerably higher.
A question to Scott or others, what is the best non destructive way
to test for this?
scott wurcer said:
Technically, yes it's a little market oriented to put things nicely.
However, Grant Carpenter the designer went into the design of
the product on a pro audio forum and I'd have to say he's technically
very good.
The relevant thing here, JC are you listening, is that the gordon mic
pre is pretty much the closest equivalent of the Blowtorch design
concept applied to a mic pre for pro audio.
You might be surprised that it has gained a reputation of being one
of, if not the best mic pre for transparent, non coloured recording.
T
Terry Demol said:
I just stick them into a Tek 576 curve tracer. Here are some notes from some testing I did several years ago:
Toshiba 2SK170 is rated at 40 volts. One random sample measured at 73 volts (182%).
Toshiba 2SJ74 is rated at 25 volts. One random sample measured at 42 volts (168%).
Edit: Don't forget that the gate leakage current rises significantly with Vds, especially with the N-channel parts. So if you are trying to make a low-noise design, you are better off to keep Vds below 10 or 15 volts.
Re: AM demodulation.
Hi Edmond
If I could propose a more meaningful test as a particular amplifier design may be good in terms if immunity to the demodulation of AM signals, but its linearity can be significantly degraded by an RF signal that the listener may not even be aware of.
The typical non-inverting opamp topology used in most power amps can have a second Rf resistor added, connected to the inverting input and fed from an RF signal generator.
Standard steady state audio signal THD tests can then be performed with zero and progressively increased levels of RF carrier injected, with the degradation in linearity noted against the RF carrier level.
Amplifiers with input stages that maintain a high voltage gain out to high frequencies (>500kHz) generally perform poorly when tested with RF carriers in the AM broadcast frequency range. The input stage amplified RF signal readily overloads the second stage (typically the VAS)
Amplifiers that are predominantly compensated by rolling off the input stage gain generally perform better. Folded cascode amplifiers fit into this category and are generally also very good performers in this regard, especially so as there isn’t a second stage that contributes voltage gain to overload.
Miller compensated amplifiers fall into this category to a degree, as the VAS and LTP gain is rolled off in unison.
Cheers,
Glen
Edmond Stuart said:
Hi Edmond
If I could propose a more meaningful test as a particular amplifier design may be good in terms if immunity to the demodulation of AM signals, but its linearity can be significantly degraded by an RF signal that the listener may not even be aware of.
The typical non-inverting opamp topology used in most power amps can have a second Rf resistor added, connected to the inverting input and fed from an RF signal generator.
Standard steady state audio signal THD tests can then be performed with zero and progressively increased levels of RF carrier injected, with the degradation in linearity noted against the RF carrier level.
Amplifiers with input stages that maintain a high voltage gain out to high frequencies (>500kHz) generally perform poorly when tested with RF carriers in the AM broadcast frequency range. The input stage amplified RF signal readily overloads the second stage (typically the VAS)
Amplifiers that are predominantly compensated by rolling off the input stage gain generally perform better. Folded cascode amplifiers fit into this category and are generally also very good performers in this regard, especially so as there isn’t a second stage that contributes voltage gain to overload.
Miller compensated amplifiers fall into this category to a degree, as the VAS and LTP gain is rolled off in unison.
Cheers,
Glen
- Status
- Not open for further replies.