I dont know what was the 3 poles 2pos switch or relay used in JC-2 (revA) schematic John provided, but for sure, the surface it implies for the CR loop is not something I would have done. I would try here the less distances and surface as possible. It explain too, may-be, why John is so upset about quality of contacts: i would not had put any moving contact at such low signal level neither. With the saving of the demanding relay/contactor, i suppose you can provide the extra OPA and the stuff around to get rid of this CR commutation with better results.
John, 100pf is the usual capacitance... of the cable between the cartridge and the Preamp. ( measured average cable run)
If you double-it, the risk is high to have a nice peak resonance between 10 and 20K. As the usual inductances of MM heads are between 400 to 950mH.
A way to get nice reviews in audiophile magazines about " analytic & detailed high end of the spectrum " ?
Last edited:
Gee SY,
Lets go over the basics.
We have power supplies to allow us to use common 120V (or 240...) to provide DC power to our sound systems.
When you turn AC into DC you produce amazing amounts of harmonics. These go both into the power supply and back into the AC power cord etc.
When you step down AC voltage you create magnetic fields. These can be localized but are always present to some extent.
When you filter the rectified AC to get DC you will always have some "ripple" or noise. There is also noise making it through the power line directly into your circuit. (I did make an AC line noise generator to allow quantification of these levels vs circuit techniques.)
There are techniques using additional windings on the transformer, more diodes and multiple capacitors to reduce these effects.
Different transformer types have different noise and rectified voltage characteristics.
Increasing the places where you put diodes can decrease pass through noise. How the transformer is split and more diodes can provide better results. Changing something as unimportant as the phasing of a winding will also have significant results in pass through noise. It will also affect the ground currents between two pieces of gear.
Transformers have capacitance to their cores. This passes noise. It can be reduced by better primary connections.
Diodes make noise during rectification (I have not yet expanded on this and the types of rectifiers used, all 6 types.)
The noise from the primary power supply may be additionally filtered. (I haven't gone into this, but George has shown one way to add resistors.)
Capacitors have serious issues. Using multiple small value radial capacitors is one method of improvement.
Regulators are often used in power supplies to improve apparent source resistance and reduce "Ripple."
Due to the "Magnetic" field radiated by the power supply components there is more noise that would be apparent from just considering current flow through the physical components.
Due to "Hidden" capacitors there is more coupling than would be apparent from just considering current flow through the physical components.
In manufacturer's data sheets they show data under controlled conditions. Sometimes the data presented is suspect! (The LM4562 data sheet shows the test circuit using an AP system 2722 It has a maximum single ended output voltage of 12.8 volts, but the data presented does not make it clear if the "device" distortion is real or a limit of the test signal in all cases.)
A power supply does not ever have a true "0" volt referenced to "Earth" ground condition.
The output of any circuit connected to a power supply will always have a noise current flow when it is connected to any other gear using a power supply.
Now just for you SY, you will never be able to achieve paper results in the real world.
Now not yet up are the techniques to:
Reduce the influence of DC on transformers,
Add inductive filters to reduce noise,
Techniques to minimize resonances in particular components.
Techniques to reduce the rectification noise back into the power line,
Reducing power factor in linear power supplies,
Rectification techniques to reduce generated noise.
And a bit more.
Now which part of that is unclear to you?
I have used the AD797 as an example on this thread only. as it is a common choice here.
In conclusion you will never see -200 db AC power line noise in a real preamp. There are issues besides the paper warrior methods you seem to be using that show up in reality.
Last edited:
Please allow me to explain the input switching, Esperado. First a GOLD-ON-GOLD contacts multiple pole switch is used. There are two gain ranges, one for MC and one for MM
The feedback resistor to ground for MM is 499 ohms and for MC is about 40 ohms.
Since most MM cartridges have a DC resistance of about 499 ohms, you get the best bias offset with this value. However the noise will go up a few dB, with such a feedback resistor to ground. However, to LOWER value of the feedback resistor will LOAD the input IC significantly, or clip it with too much gain. It is already loaded with 800 ohms effective load above a few thousand Hz. Should we load it more? This is a decision that is called: TRADEOFFS something an analog designer must take seriously.
The feedback resistor to ground for MM is 499 ohms and for MC is about 40 ohms.
Since most MM cartridges have a DC resistance of about 499 ohms, you get the best bias offset with this value. However the noise will go up a few dB, with such a feedback resistor to ground. However, to LOWER value of the feedback resistor will LOAD the input IC significantly, or clip it with too much gain. It is already loaded with 800 ohms effective load above a few thousand Hz. Should we load it more? This is a decision that is called: TRADEOFFS something an analog designer must take seriously.
Simon,
It's been almost 40 years since I was in Architectural school, SciArch. but I don't remember anyone not connecting the corner lines. Perhaps when doing inking with a Rapidograph you would want to do that so the ink wouldn't smudge but I never did that, just waited for the vertical lines to dry before doing the horizontal lines. I still have my Mayline table and K&E arm attached but I finally went the way of CAD drafting and design a few years ago and took the plunge, Not doing architecture but I still like to draw in full scale sometimes for a design just to get a better feel of the proportions. But I sure can't do surface design with a pencil drawing, yes I can still draw a perspective drawing, but it isn't the same as being able to do a 3D drawing and take it into a photo realistic program and see what it will look like completed. Sometimes you would sit and other times you would stand while drafting and you are correct it wasn't easy on the back when doing a "D" scale or larger drawing. Oh the memories.........
It's been almost 40 years since I was in Architectural school, SciArch. but I don't remember anyone not connecting the corner lines. Perhaps when doing inking with a Rapidograph you would want to do that so the ink wouldn't smudge but I never did that, just waited for the vertical lines to dry before doing the horizontal lines. I still have my Mayline table and K&E arm attached but I finally went the way of CAD drafting and design a few years ago and took the plunge, Not doing architecture but I still like to draw in full scale sometimes for a design just to get a better feel of the proportions. But I sure can't do surface design with a pencil drawing, yes I can still draw a perspective drawing, but it isn't the same as being able to do a 3D drawing and take it into a photo realistic program and see what it will look like completed. Sometimes you would sit and other times you would stand while drafting and you are correct it wasn't easy on the back when doing a "D" scale or larger drawing. Oh the memories.........
I used to have a friend who was a judge for architectural drafting contests. He taught me that trick. So it may not be universal.
And how often when you were almost finished did you get a smudge?
Thanks
ES
Did your schematic was not clear enough for us, unwashed people, can't understand-it at first sight ?
Please, read again my critics about CR loop surface and contact-quality/signal-level ratio both in the input and CR. One separate OPA for MC with 10/15 db of gain would had been a better decision on my opinion. Its noise weighted by the following RIAA curve, saving the GOLD on GOLD for your Lady.
Last edited:
Wow, that jello sure does splatter. I used your calculation, just without cooking the books like you did. That's apparently "paper warrior" when someone other than you does it.
Neither I nor Scott (who has considerably more experience than both you and I put together) see any signs of the 3k/6k/whatever-new-one ripple harmonics in the output of real-world 797 circuits using basic raw supplies and basic regulators. Apparently you haven't either, or you'd have added that at the end of your lecture.
Simon,
Way to many times, but we didn't have plotters then to do it for us. I still have those pens and they were a major pain in the behind. Smudges always pissed me off to no end.,........ But pen plotters could get you also, I still have a Houston Instruments plotter with printer port connection, have to keep legacy computers just for that type of stuff.
Way to many times, but we didn't have plotters then to do it for us. I still have those pens and they were a major pain in the behind. Smudges always pissed me off to no end.,........ But pen plotters could get you also, I still have a Houston Instruments plotter with printer port connection, have to keep legacy computers just for that type of stuff.
SY,
Are you really that confused?
The 3K figures are based on weighting because human hearing appears to be most sensitive at that frequency.
The 6K peak in power supply noise is from the single .01F capacitor power supply shown.
Those are separate issues. Why do you confuse them.
The AD797 can reject all the power supply noise that ever will be, when the power supply moves the ground around that is your limit.
But just in case you missed it you will not achieve -90 db PSSR in a circuit with more than one active component, or even one with gain.
Now why does Scott never see 6K in the output, most likely because he doesn't use bad power supplies. He won't see 3K unless he has a Fletcher Munson weighted analyzer.
I gave an example of a real power supply, Waly got confused about ripple before and after regulators. So I gave a bit more of the example. You keep chiming in with paper calculations, that in your opinion disprove my measurements. I think that since you can't even calculate a ripple voltage your opinion is worthless.
Simon,
Way to many times,
I know! Mylar and ink remover were great steps forward.
That's a statement that can not be proven, first off you don't even specify frequency. At DC it's easy.
Even if I couldn't, PSUD2 certainly can.
So, got that data showing the 3k/6k/whatever-your-latest-dodge from PS ripple on the output of a competently laid out 797 circuit? Or is that hidden away in the lockbox with John's 7th and 9th harmonic data?
Just to refresh your memory, Ed. If we change the PSRR to your new number of 95dB, the noise is 5dB worse. But that's based on my earlier estimate of 120mV ripple for the raw supply, which is too high (based on measurements of the components I have on hand). The actual number, as I showed, is about 12dB lower. So change my -164dB to -171dB. The 6kHz numbers are even lower. Drop in a more modern reg than the bell-bottoms and platform shoes 7815 and you can take out another 20dB or more, as if you needed that.
You'd do better worrying about the effect of room light on the Seebeck coefficient of the speaker terminals.
Ed
You stuffed that soup with all the herbs of the world.
Each paragraph in your post asks for a dedicated thread.
As is the case with this thread, every subject has a half life of a few hours. Two days later everything is buried deep down in the noise.
So although I agree on most with your points, I will not expand on any but on this .
In that work I was cascading RC filter stages between the rectifying diodes and the voltage stabilizer trying to see the noise effect of capacitance distribution increasing from rectifiers to stabilizer (something that can be reliably tested with the PSUD2 too)
Thank you for this intriguing post and from your side try to find a way to settle this issue. I would paraphrase you:
Life is so short SY and the level of noise is destined to increase till the end, all along the universe (you know that).
George
Very true
Now lets try a simple question. If my PSSR at 120 Hz is 90 dB, There is .1 V P-P ripple (120 Hz.) on my negative supply, none on the positive supply and my gain is 60 dB. Can you predict from the data sheet what the output ripple will be?
Can you offer a value based on experience?
Now if I have have the same voltage of ripple on the positive rail, will there be more ripple out if the ripples are in phase or out of phase? Can you answer that question from the data sheet? From experience?
If you want to, assume a chip you are familiar with.
In my article I am trying to avoid brand names as some will jump to conclusion based on a limited sample size of a particular production run that A is better than B. Since sample size is limited and processes change I wish to avoid that.
ES
Last edited:
... an thus useless. Well said, George. We are moving above thousands of surfaces, nothing in deep. Real world results are not appreciated.
- Status
- Not open for further replies.