I think the RFI immunity of a degenerated bipolar input pair is similar to a JFET pair.
For a power amp, no one today would run the input stage undegenerated. We've known about this for at least 15 or 20 years, or maybe even longer. Sure, they may be a bit more noisy (1nV/rHz vs 5nV/rHz) because of the degen resistors, but in a power amp that is of no consequence, and given the trend towards the use of servo's, whether FET input or bipolar input, I don't think the bias current differences between the two technologies make any material differences.
For small signal stages (line amps etc), FET's do have some advantages, and for MM input stages if you select the JFET carefully, you can get some solid noise advantages. MC is a different story, and here I think bipolar's are the way to go.
No doubt there will be a slew of people who read this thread and will suddenly declare all bipolar input stages no good, citing lots of material out of context, not appreciating any of the tradeoffs, and 'subjectivizing' the issues. There are dozens of examples on the forum.
If you are not predjudiced against op-amps, then there are many fine JFET input devices to choose from. That said, I am using a bipolar input stage op-amps on my current line amp. I don't think it sounds mid-fi. But then again, on this point I am probably predjudiced.
For a power amp, no one today would run the input stage undegenerated. We've known about this for at least 15 or 20 years, or maybe even longer. Sure, they may be a bit more noisy (1nV/rHz vs 5nV/rHz) because of the degen resistors, but in a power amp that is of no consequence, and given the trend towards the use of servo's, whether FET input or bipolar input, I don't think the bias current differences between the two technologies make any material differences.
For small signal stages (line amps etc), FET's do have some advantages, and for MM input stages if you select the JFET carefully, you can get some solid noise advantages. MC is a different story, and here I think bipolar's are the way to go.
No doubt there will be a slew of people who read this thread and will suddenly declare all bipolar input stages no good, citing lots of material out of context, not appreciating any of the tradeoffs, and 'subjectivizing' the issues. There are dozens of examples on the forum.
If you are not predjudiced against op-amps, then there are many fine JFET input devices to choose from. That said, I am using a bipolar input stage op-amps on my current line amp. I don't think it sounds mid-fi. But then again, on this point I am probably predjudiced.
Last edited:
Do you really expect that the RFI/EMI on the input is that great that it rectifies this unwanted signal? I thought the issue here was that on an undegenerated bipolar input, the gm was much higher than either a JFET or degenrated bip input, hence the greater susceptability - i.e it has greater gain right at the input stage. Maybe someone else can comment on this.
Most preamps/power amps incorprate some form of bandwidth input filtering - as to exactly what the correct level is, I guess that's up to the designer. I think 150-200KHz -3db is about right, and that's what I am using now.
Most preamps/power amps incorprate some form of bandwidth input filtering - as to exactly what the correct level is, I guess that's up to the designer. I think 150-200KHz -3db is about right, and that's what I am using now.
Last edited:
Thanks for that, Bob. But would the same mechanism also not occur in the OL situation? The AM distortion will modulate the Gm (or rather is a result of the modulation of Gm) which will then also cause PIM if only from the miller cap downstream. It will also wobble the OL cut-over point about, no? Or is that effect much less pronounced?
jan
I must yield to Bob here - I haven't done any practical PIM measurements. I know Bob has a purpose-designed measurement set for that.
jan
Last edited:
Hi Bonsai,
If the voltage at the emitter of the input transistor of a BJT input pair follows the base voltage just as well at very high frequencies as at low frequencies, then these degeneration arguments for decent BJT LTP immunity to RFI may hold. However, many effects can come into play at very high frequencies that may cause a departure from the way the emitter should swing. In this case, the nonlinearity of the turned on BE junction can come into play. With a JFET, the junction is reverse-biased and there is far more "wiggle room" for voltages between the gate and source before any nonlinearity gets serious.
Cheers,
Bob
I must yield to Bob here - I'haven't done any practical PIM measurements. I know Bob has a purpose-designed measurement set for that.
jan
Hi Bonsai,
Some measured numbers for real amplifiers are available here
http://www.cordellaudio.com/papers/phase_intermodulation_distortion.pdf
and here
http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
The PIM was measured in accordance with Matti Otala's technique, where a SMPTE-IM signal is applied to the amplifier and phase modulation of the 7kHz carrier is measured instead of amplitude modulation, yielding PIM instead of AIM. PIM was measured with a highly sensitive coherent IM analyzer that used a PLL for carrier recovery and synchronous detection with in-phase and quadrature detectors (described in the first link above).
I express PIM in rms nanoseconds. Since the period of the 7kHz carrier is 143us, and since this corresponds to 360 degrees, we have 2.5 degrees per us of PIM. Thus, 1ns of PIM corresponds to approximately 0.0025 degrees.
A 1970's vintage quasi-complementary amplifier with an architecture somewhat like a Citation 12 measured about 7ns of PIM while exhibiting about 0.1% of SMPTE IM (AIM).
My MOSFET power amplifier with error correction measured 0.04ns of PIM (yes, 40 picoseconds) while exhibiting SMPTE IM of about 0.00015%.
Cheers,
Bob
Of course there are the digital is terrible folks who claim to be able to hear 100 picoseconds of clock jitter! So I guess it is a good thing I sold off my horrible old Citation 12. Interestingly enough it had a number of modern features, the dual transformers were isolated from the chassis by rubber washers, the ground buss was isolated from the chassis, low noise resistors were used throughout and of course the innovations it is best know for, differential input, single Vas stage and no output capacitor.
Last edited:
Bob,
Back around 1979 when I was in charge of our hybrid JFET/biploar op-amps I had thousands of JFET samples around and tried a lot of all NFET circuits including Barrie's cascomp. The best was the amp where you put FET's in the feedback current nulled against the input pair. Barrie called it the "ADS" amp.
Good luck on the PIM stuff, when asked directly JC still will not back down on the impossible frequencies.
Back around 1979 when I was in charge of our hybrid JFET/biploar op-amps I had thousands of JFET samples around and tried a lot of all NFET circuits including Barrie's cascomp. The best was the amp where you put FET's in the feedback current nulled against the input pair. Barrie called it the "ADS" amp.
Good luck on the PIM stuff, when asked directly JC still will not back down on the impossible frequencies.
simon,
The Citation 12 was a good amplifier for its time, and I did not intend to put it down. Indeed the amp I referred to was not a Citation 12, but just one with a similar topology - I just used the term Citation 12 to conveniently invoke the image of the topology. However, the Citation 12 did have many of the shortcomings of the day, including no IPS degeneration and a quasi-complementary output stage with slow output transistors.
I'm pretty sure that the issue of 100ps of clock jitter in a D/A process is a much different animal than 100ps of jitter in a PIM situation, and that the clock jitter is worse and generates much different artifacts.
Cheers,
Bob
Bob,
No offense was taken. That was meant as humor. As you are aware the Citation 12 was a groundbreaking amplifier. It was the first use of transistors with extend flat Hfe made for audio. The output transistors were state of the art for the day. The basic circuit topography was the basis of the mass-fi boom of the 70's. Before I sold it I did listen to it. It still holds it's own.
As to jitter vs PIM they are different issues but isolating those differences would shed more light on each form of distortion.
ES
Hi Scott,
Did you mean to say cascomp? Pat Quinn invented the cascomp at Tek. Pat's an old friend of mine. We were together at Stanford getting our Masters' together in 1970, then both worked at Bell Labs Holmdel until he left for Tek. Great guy.
Did you do a JFET cascomp?
Cheers,
Bob
Yes. Sorry I mix up some of the circuit names. The Ft doublers and cascomp circuits sometimes look similar.
Last edited:
- Status
- Not open for further replies.