"Patrick Turner" <info@[EMAIL PROTECTED]
> wrote in message
news:47DD0B84.E12FE8B9@[EMAIL PROTECTED]
>
>
> Iain Churches wrote:
>>
>> It is not too difficult to build a tube powered system
>> which is totally silent with the CD player on pause,
>> even ear-against-the-speaker. "Inky blackness" my
>> pal Richard calls it.
>>
>> My 50W PPP EL34 amp has a SNR of 106dB that's
>> 10dB better than a CD:-)))
>
> 50 watts into 8 ohms = 20Vrms of signal,
> and if the noise was 106dB below this, unweighted, 20Hz to 20kHz -3dB
> BW,
> then noise = 20V / 200,000 = 0.2mV.
>
> To achieve the above good noise figures, the power supply filtering does
> not have to be so
> good for PP circuits and a basic CLC for OPT CT supply is usually very
> adequate,
> and Quad-II with their appalling toy like PS simply relied on common
> mode rejection
> to allow a good noise performance.
Yes. One often sees PP output stages fed straight from the
reservoir cap!
>
>> Also special attention needs to be paid to the grounding
>> scheme (there is a good thread on this topic with plenty
>> of info on RAT at the moment) and also to the heater
>> and anode supplies. It can be done. It is well worth
>> the effort - especially if it's a homebrew amp:-)
>
> All that is basic good practice.
Easy when you know how:-) It took me a long time
by trial and error, and discussion with others, to find
out what works and what doesn't.
As someone said just recently, the schematic tells you
only part of the story. The implementation is a totally
different thing.
>>
> Amplifier noise can be signal related if the noise increases with
> increased signal.
> Class AB amps are subject to this phenomena since the PSU has to provide
> additional
> Idc to the output stage as the amp crosses over from the class A to AB
> threshold.
Hmm. Yes I see. But one cannot measure the noise floor
if the amp is playing music. What would be a typical increase
as we move into class AB?
>
> Most certainly, the noise of any amp should be measured with its input
> terminal
> grounded. For quietest noise performance, series resistance between the
> input terminal and grid
> should be minimal, about 3k3 maximum.
Yes. That's how I do it.
>
> The noise of a typical 3 stage amplifier is usually determined by the
> FIRST tube in the line up.
I found that out, too a long time ago, when pentodes were
in vogue!
>
> Adding NFB does nothing to reduce the noise in the input tube grid
> circuit which is responsible for
> most noise in most power and preamps.
Agreed/understood. Wehave discussed this before.
>
> SO, if we have done our design homework we will have applied DC to the
> heaters of at least V1 of a 3 stage amp.
> We will have chosen the input tube carefully for low grid noise and
> microphony,
> especially im****tant with a phono or microphone amp.
>
> Suppose we have chosen a quiet 6CG7 for the input tube to a power amp.
>
> The grid noise with dc to the heaters will be perhaps 2uV.
>
> If the gain with NFB of the power amp is say 20x, or 26dB, then the
> noise at the output from grid input
> noise will be 20uV x 20 = 400uV, or 0.4mV.
> Usually hum from other sources within the amp might equal this hiss
> noise from the grid and
> the final outcome could thus be 1.41 x 0.4mV = 0.56mV and quite OK.
Thanks. That's good info. I am working on a power amp with a 6CG7
front end. I will take somne measurements and see how it adds up.
> Phono preamps and mic amps have lots more gain
> yet the grid noise is the same as above with 2uV being typical noise.
> So you need a large input signal to get a 60dB SNR.
I use 5mV input and 44dB gain (0.775V out) as a design
target. It seems to be difficult to better SNR 72dB
without a FET or transformer at the input.
>
> I have said a lot more on how to measure preamp tube noise at
> rec.audio.tubes, and how to test
> each triode for its noise by amplifying the anode output noise with grid
> grounded using a
> typical well measuring opamp preamplifier, gain = 1,000,
> and with 20Hz to 20kHz BW. Its noise will uisually be well below the
> tube noise which is being
> tested.
Yes. I built a 60dB "measuring amp" with a Hardy type 990
discrete Op-Amp. It rather overkill but I had a box of then. Then
I got the Radford ANM3 psophometer (audio noise meter) which
has a resolution better than 10µV full scale, on the most sensitive
setting. That was a real eye opener!
http://www.kolumbus.fi/iain.churches/Pics/TubeAmpNoise/RadfordANM3.jpg
It can also measure wide band, audio band, and with
various weighting factors.
> Most ppl wouldn't use a tube to amplify a low level mic signal.
There are some very good tube mic preamps available.
Many old Neumann studio mics, type U47, U49, U50
which were converted to FET in the 1970s have since
been converted back again:-)
>
> The bandwidth is reduced by a factor of (50 - 20) / (20,000 = 20) =
> approx to 30/20,000 = 0.0015,
> and noise is reduced by the square root of this bandwidth reduction
> factor, ie, by a factor of 0.038
>
> So, if the 20kHz BW noise at the input was 2uV, and gain was say 10,000
> without RIAA
> then noise at the output = 2uV x 10,000 = 20mV, which is loud as bugary!
>
> But with RIAA, noise becomes 20mV x 0.038 = 0.76mV, and mainly all
> rumble at LF
> and which does not offend the ear as much as it would if the noise was
> unattenuated at say 1 kHz.
Yes indeed. When I was at Decca, we trainees cut some
vinyl with the RIAA record curve switched out, so that it
could be replayed through a mic preamp. The surface noise
(normally attenuated by the RIAA repro curve) was horrendous.
In addition, the LF took up so much lateral space (as it was
not attenuated on record) that onlyt about 10-12 mins of
playing time was possible.
>
> I have cross posted to rec.audio.tubes where there may be some other ppl
> able to comment > and stay on topic, apart from the usual useless crew
> of no hopers who
> warp the topic
> into an abortion.
>
Thanks! Is nice to find someone is still interested.
It was alarming to see how fast the thread on Ground busses was
morphed into an argument about the relative merits of PC and Mac.
Regards
Iain
|
