The "x100 Coolest" PB amp for taperecorders

Jump to Latest
N

Nick Sukhov

Member
Joined 2022
patreon.com/nick_sukhov
#1
The "x100 Coolest" PB amp for taperecorders
coolPBAmp_circ.jpg
cooling_comparison.png

Simple "standard" IEC I cassette version
© Nick Sukhov, 1986-2020-2023
Detailed description see at https://www.patreon.com/posts/usilitel-s-uv-80024956
 
#3
Intended to reduce noise. But at first glance it doesn't look quite right - gain of shunt circuit seems low. The idea is that you have x100 gain to the loading resistor, so to make it look like the required resistance, it is 100x required value, hence the large value. But although resistance is now x100, its noise has only gone up by x10, so we have one tenth the noise of a conventional shunt resistance. However, that assumes that in fitting the x100 resistor you didn't add any extra noise in extra amplifiers. 5532 has fairly low voltage noise, but significant current noise.
 
#4
stellavox said:
link is in Russian (I guess) and costs $ to "join".
Click to expand...

It is needed simply one mouse click to translate:

Playback amplifier with proper cooling - "coolPBAmp-XXI"
For those of you who first learned about the term "chilled" in relation to low-noise amps, I recommend that you first read the article by Marcel Van De Gevel here:

https://www.patreon.com/posts/68141299

https://www.patreon.com/posts/68555442

In a nutshell, cooling is a way to reduce the input current noise of low-noise amplifiers by increasing the resistance of the resistor that forms the input resistance, while increasing the voltage across this resistor with a special additional cascade (so that the input impedance of the amplifier apparent to the signal source remains the same). The principle is based on the fact that the equivalent noise current of a resistor (Johnson-Nyquist noise) is inversely proportional to the root of its resistance. This noise current, flowing through the internal resistance of the signal source (in our case, the inductance of the playback head), in accordance with Ohm's law, creates an additional noise voltage that worsens the signal-to-noise ratio, especially at higher audio frequencies. Back in 1985, I showed [p.69 of my book "Technique of High-Quality Sound Reproduction", - Kiev, Technique, 1985 - it is here https://www.patreon.com/posts/67628599] that a typical 47 kΩ input resistor makes a little noise or not 2 times more than the base noise current of the low noise transistor.

Marcel Van De Gevel was the first to use such a topology for audio preamplifiers in his original vinyl stage in 2003 (see the article quoted above), but due to the strong drop in the frequency response at higher audio frequencies, the cooling in the vinyl stage provides a noise gain of only 1-1.5 dBA, but in In the playback amplifier of tape recorders, the frequency response of which at higher frequencies does not decrease, but even picks up, the gain from cooling can reach 5-6 dBA. But this is only if it is implemented correctly😉 An example of incorrect implementation of cooling is discussed in detail here:

Below is my schematic file for the Microcap 12 playback amplifier coolPBAmp-XXI, which is a chilled upgraded version of my PBamp-1986-87 [Radio Yearbook-1986, p.51, fig.3], [Radio No. 6/1987, p.31, Fig.1]. The actual cooling is formed by a cascade at the X3 op-amp with a resistor R15. For the correct operation of the cooler, it is very important that the phase of the current launched by the cooler into the resistor R15 coincides with the phase of the input voltage. Otherwise, R15 will no longer be a pure resistor, but with a fair amount of additional capacitance or inductance, which will lead to a frequency detuning of the input LC circuit L1C4 , and a gyrator).

In coolPBAmp-XXI, the exact phase correction is created by the R12R13C5 chain, which forms the AFC, the reverse of the main AFC R7R8C3. Thus, the entire circuit of the SW from the base Q1 to the output of the X3 op-amp forms a linear frequency response and phase response with a gain equal to the ratio R14/R1=100. The resistance of the resistor R15 = 5M1 is chosen 100 times greater than R10 = 51 kOhm (in fact, R10 is not installed in the circuit - R15 already performs its functions), i. the input impedance of the HC, apparent for the playback head, remains equal to a typical 51 kOhm, and the current noise of the resistor is reduced to the root of 100, i.e. 10 times. By running the schematic file in the Microcap 12 program in the Analysis \ AC mode (and also placing the file of the previous analysis with the ANO extension next to the CIR file - I attached it next to the CIR), you can compare the noise spectral densities in the normal mode (red color) and chilled (green).

Initially, the R10 resistor was used as an input in the file (which corresponds to the usual SW circuit), and R15 and all cooling are disabled. To enable the cooling mode, click on R10 and in the Enabled window set the FALSE option, then click on OK. And then click on R15 and in the Enabled window set the TRUE option, then click on OK.

stellavox said:
What is the "shunt impedance circuit" supposed to do?
Click to expand...
See Marcel van de Gevel`s article in EW October 2003^ https://www.patreon.com/posts/68141299 In my preamp shunt impedance circuit create noise with much lower level (green line in screenshot) than typical resisitive input (red line in screenshot).
 
Last edited:
#5
EC8010 said:
Intended to reduce noise. But at first glance it doesn't look quite right - gain of shunt circuit seems low.
Click to expand...

The gain is quite right. I need x100 (51 kOhm x100 = 5M1) and have it, Ku=R14/R1 . You can download and check my circuit file here : https://www.patreon.com/posts/usilitel-s-uv-80024956

EC8010 said:
5532 has fairly low voltage noise, but significant current noise.
Click to expand...
It is used in second stage after low noise BC557 and therefore does not add noise at all.
 
#8
Delighted to be corrected. Well, well, well. Mr Percival, he of "Percival peaking". Percival peaking was a trick used in plumbicon television cameras. The plumbicon tube was connected to a charge amplifier (near-zero input resistance) but there was inevitably shunt detector capacitance and series inductance between detector and charge amplifier, which gave a resonant LC circuit. Percival peaking exploited this resonance by arranging for it to occur towards the upper limit of video (5.5MHz for 625 line television), then fitted a complementary trap circuit after the charge amplifier to restore flat video frequency response. The cunning bit was that in doing so, some high frequency noise was attenuated. Neat. EMI2001 colour cameras had Percival peaking, and I expect the much earlier Emitron monochrome 405 line cameras did too.

Thank you kindly for the correction, and especially for the original paper.
 
  • Thank You
Reactions: Nick Sukhov
#9
You must log in or register to reply here.
Top Bottom