John Curl's Blowtorch preamplifier

[Edmond Stuart]

Hi Glen,

Thanks for your suggestions. Doing it this way is a lot of work. So, regarding the results, have some patience with me.

Regards,
Edmond.

edit: >PS, maybe continue this discussion in the Negative Feedback thread?

That's a good idea.

[dimitri]

Apply 1V rms 70% 1kHz modulation 150kHz-1GHz carrier to input terminals, then to output terminals and measure the audio signal (demodulated) on the output.

Do 0.1MHz to >100MHz in 1,2,5 steps from 100uV to >100mV also in 1,2,5 steps. Mix this with a 1kHz sinewave input signal of an amplitude that produces, say 1/4 of full power output. Just measure the 1kHz THD

These efforts will be wasted. Unmodulated carrier will be detected and shifts device(s) idle voltage/current, and it will give some shift in THD. So what?

[GK]

Originally posted by dimitri
Apply 1V rms 70% 1kHz modulation 150kHz-1GHz carrier to input terminals, then to output terminals and measure the audio signal (demodulated) on the output.


These efforts will be wasted. Unmodulated carrier will be detected and shifts device(s) idle voltage/current, and it will give some shift in THD. So what?

No they won't be wasted and demodulation performance is not an entirely reliable indicator/measure of RF immunity (eg fully complementary circuits)

Go back an read the previous discussion on input stage gain and signal handling capability Vs VAS signal handling capability and compensation methods.

Some amplifiers will overload and produce gross THD with very little injected RF carrier and some will do similar with significantly higher amounts of RF carrier.

[Edmond Stuart]

Originally posted by dimitri
Apply 1V rms 70% 1kHz modulation 150kHz-1GHz carrier to input terminals, then to output terminals and measure the audio signal (demodulated) on the output.

about THD:

These efforts will be wasted. Unmodulated carrier will be detected and shifts device(s) idle voltage/current, and it will give some shift in THD. So what?

Hi Dimitri,

Looking for AM demodulation was also my first choice (although I've simmed it in a slightly different way), but don't you think 1V rms HF is a bit unrealistic? With such high amplitude almost every amp will get berserk, unless the amp is provided with a low pass filter at the input.
However, to make fair comparisons between different topologies, we should disable such filter.


Cheers,
Edmond.

PS: If no AM demodulation takes place, doesn't necessarily mean that no other kind of distortions may rise under influence of HF signals.

[GK]

Originally posted by Bob Cordell

In referring to amplifiers in which the input stage is rolled off as being more immune, I'm wondering if the topology I use (miller input compensation? - where the Miller cap is returned to the input differential pair) falls into that category as being more immune.

Hello Bob.

Thinking about this a bit more, I think that your amp is saved from gross VAS overload at RF by the input stage (100R+150pF) RC.

Suppose a 100mV RF carrier of 2MHz is injected into your amplifiers input. At 2MHz the loop gain is unity so the differential error signal ( E=Vin/(1+AB) ) = 50mV. At 2MHz your input stage has ~unity gain, so the VAS only sees ~50mV differential input. At higher frequencies the input stage gain heads well below unity.
I think that's about right, but it is nearly midnight now

Cheers,
Glen

[Bob Cordell]

Originally posted by Terry Demol


So, getting back on topic, what about the RF susceptibility of an
open loop (locally degenerated) circuit such as Blowtorch style, as
compared to circuits with GFB?

From everything that has been said so far, it all appears to point
to the fact that they are potentially better.

- IP stage usually has more degeneration than closed loop designs.
- Each following stage has lower gain, if any.
- There can be multiple RF attenuation points through the circuit
without affecting stability, phase margin etc.

T

Hi Terry,

These are very good (and interesting) points. It is a particularly good point that a no-NFB approach forces you to do right by the input stage. The inverse of this is that the use of negative feedback can allow some less skilled designers (especially in the past) to use an un-degenerated input stage with all of its consequent badness.

Cheers,
Bob

[Bob Cordell]

Originally posted by homemodder
Edmond

I hope you place your findings here, i am very interested in the results. Ive used that arrangement since i can remeber, i have no idea about its rf immunity effects but there are many other benefits including reactive loads and most importantly sonic|subjective benefits. Found this out when i was kid starting building kit amps and thanks to John linsey Hood. One does not always get the best distortion results but the sonics is in my opinion more important.

John i thought of mentioning this type of compensation, am not sure what the right term for it is around here, phase lead???, regarding lowering higher order harmonics on the ad797. Ive always used this type of compensation scheme on amplifiers and using this type of compensation with experimetal amp based on the ad797 topology with very interesting results. Did not mention it for fear of being lamblasted and asked many questions which i dont have all the answers for, i dont have 30,40 years of experience designing audio.

Alex

Hi Alex,

Some on the NFB thread have described it as phase lead compensation in the past, but I have always thought of it as just another way of doing dominat-pole integrating feedback compensation. In essence, we are still building a Miller integrator with it. We are just feeding back the capacitor current one stage earlier. Doing that encloses the input stage in the local Miller compensation feedback loop, with significant input stage linearization benefits.

Cheers,
Bob

[Bob Cordell]

Originally posted by G.Kleinschmidt



Hello Bob.

Thinking about this a bit more, I think that your amp is saved from gross VAS overload at RF by the input stage (100R+150pF) RC.

Suppose a 100mV RF carrier of 2MHz is injected into your amplifiers input. At 2MHz the loop gain is unity so the differential error signal ( E=Vin/(1+AB) ) = 50mV. At 2MHz your input stage has ~unity gain, so the VAS only sees ~50mV differential input. At higher frequencies the input stage gain heads well below unity.
I think that's about right, but it is nearly midnight now

Cheers,
Glen

Hi Glen,

That is a good point. I should remind others here that the purpose of the 100R + 150 pF RC across the cascode collectors of the input stage is to frequency compensate the loop that is formed locally by the Miller compensation capacitor feeding back to the input stage. I think that loop was designed to have a unity gain frequency in the range of 10-20 MHz.

Cheers,
Bob

[1audio]

While I think designing the circuitry for EMI resistance is important my experience points to layout and stability as being the dominant issues in emi/rfi resistance. Since your dealing with RF it can get into the circuit after the input stage if the "window" (the area inside the wire loop at the input of the VAS) is large enough. Even an output follower is a candidate. And as higher frequency energy becomes more common at 850/900/1800/1900 MHz (cell phones) then even a small window is a good antenna. And that window could be a slot in a ground plane.
I have used both fast rise pulses and RFI generators (walkie-talkies are great for this) to find local instabilities and fixing those will usually eliminate excess RFI sensitivity.
I have measured over a volt of RFI at 80 MHz in a customer's house near a TV transmitter. TV signals have a characteristic buzz from the vertical refresh that sounds like bad PS hum. This will be gone next year when TV becomes digital in the US. It will be a hash instead. I told the customer to move, fixing the problem on a moving coil input was almost impossible, and I didn't think it would be healthy to live in such a high RF field.

[john curl]

Up near Sutro tower, Damien?