Dear Nick,
I didn't understand your video and the subtitle is more confusing than helpful.
What is it? Again, nothing more than a voltage divider!
u2_ = (u1_ * R2) / (R2 + R1 + j (Omega * L1) or also u2_ = u1_ * [(1+R1/R2) + j (XL1/R2)]^-1. So, at f=1kHz your model achieves a gain-advantage of 0,3631dB.
It might be more helpful to represent the source differently, with
Z(jOmega) = 1 / [(R1 + j(Omega * L1) )^-1 + R2^-1]. The smaller (the resulting amount' impedance), the less thermal noise, right?
Perhaps you should explain once again in written words why, in your opinion, the termination with 150kOhm is better in terms of noise than the termination with 47kOhm.
You consider the input of the subsequent amplifier to be almost infinitely large and purely "reell". At the same time, your generator is an ideal voltage source with a source resistance of zero ohms. Now you put your equivalent voltage divider in between ... but we all know that R1 and L1 are not constant values in reality, just as an aside.
Best regards,
HBt.
I didn't understand your video and the subtitle is more confusing than helpful.
What is it? Again, nothing more than a voltage divider!
u2_ = (u1_ * R2) / (R2 + R1 + j (Omega * L1) or also u2_ = u1_ * [(1+R1/R2) + j (XL1/R2)]^-1. So, at f=1kHz your model achieves a gain-advantage of 0,3631dB.
It might be more helpful to represent the source differently, with
Z(jOmega) = 1 / [(R1 + j(Omega * L1) )^-1 + R2^-1]. The smaller (the resulting amount' impedance), the less thermal noise, right?
Perhaps you should explain once again in written words why, in your opinion, the termination with 150kOhm is better in terms of noise than the termination with 47kOhm.
You consider the input of the subsequent amplifier to be almost infinitely large and purely "reell". At the same time, your generator is an ideal voltage source with a source resistance of zero ohms. Now you put your equivalent voltage divider in between ... but we all know that R1 and L1 are not constant values in reality, just as an aside.
Best regards,
HBt.
When you are talking about an input, you are also talking about the load for the previous device.
In this case, you are talking about a cartridge going into a stage and comparing 150K to 47K input impedance.
So the passive current of the cartridge is working into a lighter load, and a higher voltage potential can be developed, that in turns causes the input sensitivity to be greater.
Without wanting to open an old barrel here, but the simplest thing is: the measurement!
Take an official test record or simply your favorite record and digitally record a track. a) at 150k and b) at 47k, and now the extensive analysis follows with c) -> the tools and algorithms required for this are certainly known to developers.
Who actually defines what is meant by a "high quality phono preamp input"? What is that - noiseless with 47k||125pF?
Take an official test record or simply your favorite record and digitally record a track. a) at 150k and b) at 47k, and now the extensive analysis follows with c) -> the tools and algorithms required for this are certainly known to developers.
Who actually defines what is meant by a "high quality phono preamp input"? What is that - noiseless with 47k||125pF?
There have since been several helpful threads about this proposal, all more generally accessible. Nick Sukhov is trying to inform DIYers about alternative MM phono cartridge loadings, although this video and its English translation are not the best possible proponent.
Since the mid-1970s some have argued for a much lower than the "standard" 47K Ohm + a couple hundred pF loading, with correction for the cartridge's inductive source / loading R rolloff - maybe a coupla K Ohms. Here, heavy damping removes the LCR resonance. N. Sukhov instead argues for reducing the C of the loading, by eliminating most of the coax cabling between cartridge and (pre)amplifier input. He approaches the LC resonance issue from the other direction, by moving the LCR above audio range.
Both approaches make a MM cartridge approach the frequency response of a (comparatively very low Z) MC cartridge, removing the LCR effects expected for classical MM cartridges, and leaving their raw mechanical response - including scanning (geometric) losses and vinyl compliance x stylus moving mass resonance. N. Sukhov's retains some (but significantly less) of the classic frequency response shaping of LCR loading but without a noise penalty. Both approaches do however give significantly less flat frequency response within the most audible range. The classic Shure/Stanton model is flatter ("in-band"), for whatever that's worth. Nothing is ideal.
All good fortune,
Chris
Since the mid-1970s some have argued for a much lower than the "standard" 47K Ohm + a couple hundred pF loading, with correction for the cartridge's inductive source / loading R rolloff - maybe a coupla K Ohms. Here, heavy damping removes the LCR resonance. N. Sukhov instead argues for reducing the C of the loading, by eliminating most of the coax cabling between cartridge and (pre)amplifier input. He approaches the LC resonance issue from the other direction, by moving the LCR above audio range.
Both approaches make a MM cartridge approach the frequency response of a (comparatively very low Z) MC cartridge, removing the LCR effects expected for classical MM cartridges, and leaving their raw mechanical response - including scanning (geometric) losses and vinyl compliance x stylus moving mass resonance. N. Sukhov's retains some (but significantly less) of the classic frequency response shaping of LCR loading but without a noise penalty. Both approaches do however give significantly less flat frequency response within the most audible range. The classic Shure/Stanton model is flatter ("in-band"), for whatever that's worth. Nothing is ideal.
All good fortune,
Chris
Since the Boltzmann is so popular here:
I_noise2 = sqrt( 4*k*T*b / R2)
I_noise1 = sqrt( 4*k*T*b / (|R1+jOmega*L1| ))
I_sum_noise = sqrt( I_noise2^2 + I_noise1^2 )
The corresponding voltage, which is what we're talking about here, is calculated quite brashly as U_noise using Ohm's law. However, we still have to form the necessary resistance (R1+jOmega*L1) || R2.
This is the correct way.
I_noise2 = sqrt( 4*k*T*b / R2)
I_noise1 = sqrt( 4*k*T*b / (|R1+jOmega*L1| ))
I_sum_noise = sqrt( I_noise2^2 + I_noise1^2 )
The corresponding voltage, which is what we're talking about here, is calculated quite brashly as U_noise using Ohm's law. However, we still have to form the necessary resistance (R1+jOmega*L1) || R2.
This is the correct way.
Chris,
isn't Nick's point rather to prove that a quasi real (in german "reell") input impedance of 150kOhm +/- j0Ohm has less noise in total than one of 47kOhm -j X Ohms (In combination with the tactile pickup, the signal-source)?
😉
isn't Nick's point rather to prove that a quasi real (in german "reell") input impedance of 150kOhm +/- j0Ohm has less noise in total than one of 47kOhm -j X Ohms (In combination with the tactile pickup, the signal-source)?
😉
Do you mean to say a 150K loading would be better than 47k? I have a phonostage with selective loading. I can change the resisters to 150k
Please suggest.
Regards
Sachin
Please suggest.
Regards
Sachin
Better might be subjective, but it would increase the voltage and sensitivity on the front end. Try it, if you don't like it, change it back.
Me?
No, 47kOhm is the right answer.
But in the end, everything really depends on your MM system. However, it now revolves around the resulting TP2, with a Q of >0.5 but <1 (and the achievable upper corner frequency).
Nick's topic, however, is the simulated noise - which he thinks would be lower if he terminates the transducer with high impedance. 47k compared to 150k.
It is also possible that he means something completely different.
It's best to agree on the models first. In fact, our forefathers already did that for us. They were able to distinguish between equivalent sources. Strangely enough, I can do the same. Perhaps I had good teachers.
Best wishes,
HBt.
I quickly created a table in the hope that I hadn't made any mistakes - let's just assume that everything is ok. You can still throw tomatoes later 😉.
The last two columns represent the noise voltage source applied to the input. R2 = 150kOhm unfortunately leads to a little more noise. It seems to be exactly the opposite of what Nick Sukhov assumes.
Sorry about that.
The last two columns represent the noise voltage source applied to the input. R2 = 150kOhm unfortunately leads to a little more noise. It seems to be exactly the opposite of what Nick Sukhov assumes.
Sorry about that.
1000 | 0,5 | 47000 | 150000 | b=20kHz | T=293k | | | | | | |
f | xL1 | riq47 | riq150 | In47 | In150 | |Z1| | InZ1 | I47sum | I150sum | UN47 | UN150 |
1000 | 3141,7 | 3080,8888303319 | 3226,1006781501 | 8,29770059272585E-011 | 4,64474384066016E-011 | 3239,1165438841 | 3,16077611353762E-010 | 3,26787759733672E-010 | 3,19472097269284E-010 | 1,00679675885265E-006 | 1,03064914965047E-006 |
2000 | 6283,4 | 5603,8707771338 | 6103,5843813143 | 8,29770059272585E-011 | 4,64474384066016E-011 | 5692,3956017492 | 2,38429223597922E-010 | 2,52455299366363E-010 | 2,42911216702817E-010 | 1,41472687465172E-006 | 1,48262910831335E-006 |
3000 | 9425,1 | 7887,4261423445 | 8914,7104527686 | 8,29770059272585E-011 | 4,64474384066016E-011 | 7950,5654610814 | 2,01747430863894E-010 | 2,18144927451531E-010 | 2,07025096050495E-010 | 1,72060200360106E-006 | 1,84556878774677E-006 |
4000 | 12566,8 | 9940,2986469696 | 11629,168591343 | 8,29770059272585E-011 | 4,64474384066016E-011 | 9990,4723207136 | 1,79975757996811E-010 | 1,98182887705218E-010 | 1,85872639194319E-010 | 0,00000197 | 2,16154425770861E-006 |
5000 | 15708,5 | 11791,3688594488 | 14245,4471857636 | 8,29770059272585E-011 | 4,64474384066016E-011 | 11833,6967841659 | 1,65366221514172E-010 | 1,85016676898399E-010 | 1,71765397424537E-010 | 2,18159988245848E-006 | 2,44687489735293E-006 |
6000 | 18850,2 | 13467,6617170432 | 16766,7051577389 | 8,29770059272585E-011 | 4,64474384066016E-011 | 13504,7366551411 | 1,5479750691824E-010 | 1,75634426183351E-010 | 1,61615694419316E-010 | 2,36538503770436E-006 | 2,70976269719189E-006 |
7000 | 21991,9 | 14992,2909803783 | 19197,1677208674 | 8,29770059272585E-011 | 4,64474384066016E-011 | 15025,6044417632 | 1,46754371222649E-010 | 1,68588341784381E-010 | 1,53929237013279E-010 | 2,52752547593091E-006 | 0,000002955 |
8000 | 25133,6 | 16384,7085160601 | 21541,2377064448 | 8,29770059272585E-011 | 4,64474384066016E-011 | 16415,1964093109 | 1,40405454620975E-010 | 1,63091615970825E-010 | 1,47888661573015E-010 | 2,67220858909517E-006 | 3,18570481303227E-006 |
9000 | 28275,3 | 17661,2745843535 | 23803,2182553834 | 8,29770059272585E-011 | 4,64474384066016E-011 | 17689,5624576735 | 1,3525348878518E-010 | 1,58677943461639E-010 | 1,43006540980117E-010 | 2,80245472995652E-006 | 0,000003404 |
10000 | 31417 | 18835,8024195937 | 25987,2181638923 | 8,29770059272585E-011 | 4,64474384066016E-011 | 18862,3289333521 | 1,30981320576507E-010 | 1,55052539007325E-010 | 1,38972914175751E-010 | 2,92053898939832E-006 | 3,61151943955713E-006 |
11000 | 34558,7 | 19920,0160734936 | 28097,1221539739 | 8,29770059272585E-011 | 4,64474384066016E-011 | 19945,1006607699 | 1,27376380981462E-010 | 1,52019491988978E-010 | 1,35580629023728E-010 | 3,02823072390477E-006 | 3,80942549539229E-006 |
12000 | 37700,4 | 20923,9221077984 | 30136,5868376282 | 8,29770059272585E-011 | 4,64474384066016E-011 | 20947,8045716781 | 1,24290451314349E-010 | 1,49443299617537E-010 | 1,32685646633159E-010 | 3,12693996072972E-006 | 3,99869251186706E-006 |
13000 | 40842,1 | 21856,1085422436 | 32109,0474092833 | 8,29770059272585E-011 | 4,64474384066016E-011 | 21878,9734816406 | 1,21616794074357E-010 | 1,47227131037648E-010 | 1,30184519569186E-010 | 3,21781215632196E-006 | 4,18010091080175E-006 |
14000 | 43983,8 | 22723,9852562355 | 34017,728776589 | 8,29770059272585E-011 | 4,64474384066016E-011 | 22745,9777966481 | 1,19276453167204E-010 | 1,45299882287635E-010 | 1,28000925054014E-010 | 3,30179238283697E-006 | 4,35430075163995E-006 |
15000 | 47125,5 | 23533,9779942541 | 35865,6583004755 | 8,29770059272585E-011 | 4,64474384066016E-011 | 23555,2142897074 | 1,17209683178476E-010 | 1,43608124225098E-010 | 1,26077255543313E-010 | 3,37967043530955E-006 | 4,52184376677819E-006 |
16000 | 50267,2 | 24291,6857246574 | 37655,6788176197 | 8,29770059272585E-011 | 4,64474384066016E-011 | 24312,2601858717 | 1,15370388310404E-010 | 1,42110907433405E-010 | 1,24369172359667E-010 | 3,45211350141814E-006 | 4,68320560918879E-006 |
17000 | 53408,9 | 25002,0089985001 | 39390,4613147102 | 8,29770059272585E-011 | 4,64474384066016E-011 | 25021,9993997498 | 1,13722400169588E-010 | 1,40776304160126E-010 | 1,22841966912237E-010 | 3,51969042338705E-006 | 4,83880174547937E-006 |
18000 | 56550,6 | 25669,2552627107 | 41072,5169630233 | 8,29770059272585E-011 | 4,64474384066016E-011 | 25688,7264328577 | 1,12236918599349E-010 | 1,39579043589391E-010 | 1,21468063420851E-010 | 3,58289009922111E-006 | 0,000004989 |
19000 | 59692,3 | 26297,2257629662 | 42704,2083929143 | 8,29770059272585E-011 | 4,64474384066016E-011 | 26316,232306856 | 1,10890710487718E-010 | 1,3849885625926E-010 | 1,20225264428929E-010 | 3,64213569096237E-006 | 5,13412474626622E-006 |
20000 | 62834 | 26889,2876555867 | 44287,7601758783 | 8,29770059272585E-011 | 4,64474384066016E-011 | 26907,8759961631 | 1,09664815675291E-010 | 1,37519290682245E-010 | 1,19095492490819E-010 | 3,69779576534714E-006 | 5,27447260946148E-006 |
Disclaimer
Mistakes made by me are careless mistakes due to lack of attention and prudence.
If I should have made any! But actually it is quite clear that 150k can never lead to less noise. That is simply impossible.
Repetition
I don't understand the video. It would be very nice of Nick Sukhov if he could explain to us again in words what the video is about. It would also be advisable to use technical terms.
Excuse me,
HBt.
(Any fatal errors and I will correct them - don't use a stupid spreadsheet program.)
Mistakes made by me are careless mistakes due to lack of attention and prudence.
If I should have made any! But actually it is quite clear that 150k can never lead to less noise. That is simply impossible.
Repetition
I don't understand the video. It would be very nice of Nick Sukhov if he could explain to us again in words what the video is about. It would also be advisable to use technical terms.
Excuse me,
HBt.
(Any fatal errors and I will correct them - don't use a stupid spreadsheet program.)
Deterioration of the signal-to-noise ratio compared to traditional termination with 47kOhm.
Please don't hit me, I already apologize.
f in kHz | +SNR in dB |
1 | 0,203 |
2 | 0,407 |
3 | 0,609 |
4 | 0,806 |
5 | 0,997 |
6 | 1,181 |
7 | 1,357 |
8 | 1,527 |
9 | 1,689 |
10 | 1,845 |
11 | 1,993 |
12 | 2,136 |
13 | 2,273 |
14 | 2,403 |
15 | 2,529 |
16 | 2,649 |
17 | 2,765 |
18 | 2,876 |
19 | 2,982 |
20 | 3,085 |
Please don't hit me, I already apologize.
I'm not Nick but I have a preamp/cartridge model in spice whose simulation shows a couple dB lower total rms noise over the audible band when I compare 150k to 47k cartridge termination.
It also seems to alter high frequency response somewhat and not totally sure how close the simulation is is what I would measure if I built it and tried it.
Well Ugly,
I didn't expect any different - unfortunately it is impossible to generate less noise at the input with 150kOhm than with 47kOhm.
But I will submit a PSPICE simulation of the TP2 (because that's exactly what you're looking at), of course a noise analysis, I'm curious as hell.
It is obvious that the terminating resistor, i.e. the load, has a decisive influence on the cut-off frequency of the TP1.
xL = Omega * L -> real part and imaginary part (in terms of magnitude) are equal -> Omega = R2 / L1.
In our case, R1 and the input capacitance of the amplifier as well as the capacitance of the connecting cables also come into play.
Well, and L1 and the resulting real part are unfortunately not constant values for the pickup. Only the pure ohmic wire resistance is constant (but also only if we keep the temperature constant).
L1=0.5H
-> fg150k = 47.746kHz
->fg47k = 14.961kHz
did you think it so?
That model is very incomplete. Shunt capacitance in the small 100s of pF range are part of the classic model, and the loading resistance is proposed to be synthesized "cold" resistance in N. Sukhov's model, along with the mandatory reduction in shunt capacitance. Without understanding his whole proposal, errors in judgement will be made. There is some difficulty because of language translation issues, but it's worth the effort for serious students.
All good fortune,
Chris
All good fortune,
Chris
That is absolutely correct.
Let's ignore R1 and assume L1 is constant:
fres = 1/ (2*PI* sqrt(L1*C2))
C2=200pF -> fres = 15.92kHz
C2=100pF -> fres = 22.51kHz
C2=47pf -> fres = 32.83kHz
So far, so good.
|Z| = sqrt(L1/C2)
50kOhm, 70.71kOhm, 103.14kOhm ... If we now conclude our three example cases with the calculated impedance, we achieve a Q of 1.
This is a second-order passive low-pass filter.
My God, you can simply set up the transfer function in Nick's example, it's not that difficult.
What we are missing, however, is C2. And C2 is guaranteed to be present, please don't forget: L1 & R1 are not constant numbers.
Dear Chris,
Nick's circuit contains no secrets and no tricks. It's absolutely nothing special.
Of course, we all have a hurdle to overcome, and that is the English language. But electrical engineering, electronics or communication have their own language, spoken, written and described.
I am very sorry, but I cannot take up the cudgels for Nick's presentation.
Regards,
HBt.