AudioNervosa

Systemic Development => Amplification and Preamplification => Topic started by: steve on January 21, 2022, 01:31:13 PM

Title: RIAA Inaccuracies
Post by: steve on January 21, 2022, 01:31:13 PM
Hi Gents,

I finally found the AES article covering RIAA equalization by Stanley Lipshitz I lost many moons ago. According to the article, there is a problem with correct RIAA equalization design in some/many phono stages. (I am wondering about recording as well.)

No need for math etc, the introduction and references 1-18 at the end of the article give a sense of the scope of the problem. Granted the article is from 1976, but who knows who is really following the true RIAA curve?

https://forums.melaudia.net/attachment.php?aid=1301

Interestingly, the computer program I used for designing the RIAA in my test phono stage was up to 0,7 db off in the highs and 0,3 db off in the bass. After correcting the problem, what a sonic improvement. Of course I had to reset the TT tracking, rake angle, and anti skate.

For those who diy, here is a website whose calculator is based on Stanley's design equations.

http://www.mh-audio.nl/Calculators/CalculateRIAA.html

cheers

steve
Title: Re: RIAA Accuracy
Post by: P.I. on January 21, 2022, 06:50:41 PM
Hi Gents,

I finally found the AES article covering RIAA equalization by Stanley Lipshitz I lost many moons ago. According to the article, there is a problem with correct RIAA equalization design in some/many phono stages. (I am wondering about recording as well.)

No need for math etc, the introduction and references 1-18 at the end of the article give a sense of the scope of the problem. Granted the article is from 1976, but who knows who is really following the true RIAA curve?

https://forums.melaudia.net/attachment.php?aid=1301

Interestingly, the computer program I used for designing the RIAA in my test phono stage was up to 0,7 db off in the highs and 0,3 db off in the bass. After correcting the problem, what an sonic improvement. Of course I had to reset the TT tracking, rake angle, and anti skate.

For those who diy, here is a website whose calculator is based on Stanley's design equations.

http://www.mh-audio.nl/Calculators/CalculateRIAA.html

cheers

steve
The my-audio.nl site is a great resource for all kinds of things audio!
Title: Re: RIAA Accuracy
Post by: _Scotty_ on January 21, 2022, 08:14:30 PM
Even if the designers math is right when calculating the parts values used in their phono stage,
parts tolerances can ruin the RIAA accuracy. Every single resistor and capacitor used
 in the network has to be measured and matched if errors are to be minimized.
How often does anyone think that happens in a phono stage that is not custom made.
Tolerance stacking is a problem in many fields of endeavor including our hobby.
In some cases it can actually lead to products that are unsafe to operate.
Scotty
Title: Re: RIAA Accuracy
Post by: P.I. on January 21, 2022, 10:37:04 PM
Even if the designers math is right when calculating the parts values used in their phono stage,
parts tolerances can ruin the RIAA accuracy. Every single resistor and capacitor used
 in the network has to be measured and matched if errors are to be minimized.
How often does anyone think that happens in a phono stage that is not custom made.
Tolerance stacking is a problem in many fields of endeavor including our hobby.
In some cases it can actually lead to products that are unsafe to operate.
Scotty
Exactly and precisely (redundant?) said.  For critical applications, even when it comes to power amps where tolerances "could be" wider, it is essential to match within at least 1%.

I have found this to be true even with loudspeakers.  Details, details, details:  the quest for excellence!
Title: Re: RIAA Accuracy
Post by: HAL on January 22, 2022, 09:03:29 AM
A straight gain cartridge preamp and DSP RIAA EQ work well here.  As accurate as the RIAA math for both channels. 

Had my DSA Phono One preamp moded for flat gain to drive the stereo ADC. 
Title: Re: RIAA Accuracy
Post by: tmazz on January 22, 2022, 09:58:37 AM
Even if the designers math is right when calculating the parts values used in their phono stage,
parts tolerances can ruin the RIAA accuracy. Every single resistor and capacitor used
 in the network has to be measured and matched if errors are to be minimized.
How often does anyone think that happens in a phono stage that is not custom made.
Tolerance stacking is a problem in many fields of endeavor including our hobby.
In some cases it can actually lead to products that are unsafe to operate.
Scotty

And the problem even goes further than that. the tolerance rating n a resistor is a tolerance across the lifetime of the piece. it is not uncommon for a resistor to shift value over time. A 1% resistor is rated to be no more than 10% away from its rated value at all times. but this means that it could be 10% when it is brand new and end up at -10% as it aged, meaning it can have a 20% swing over time and still be within it's rated specs. Of course you would not use 10% resistors if you were building a precision phono stage, but just the same, no matter how high precision resistors you use they can all shift over time and not necessarily in the same direction. This was a very big problem with some of the early CD players when resistor value shifts in the servo feedback curcuits caused big problems with CD players starting to mistrack.

So as close as you get it to the curve when it is built, resistor value shifts can easily get it out of spec over time. How audible will this be, who knows, but my point is if you are trying to get a perfect match to the RIAA curve it is very difficult to do (and maintain over time) in the analog realm.
Title: Re: RIAA Accuracy
Post by: P.I. on January 22, 2022, 02:33:05 PM
[quote author=tmazz link=topic=8387.msg103763#msg103763 date=1642874317
And the problem even goes further than that. the tolerance rating n a resistor is a tolerance across the lifetime of the piece. it is not uncommon for a resistor to shift value over time. A 1% resistor is rated to be no more than 10% away from its rated value at all times. but this means that it could be 10% when it is brand new and end up at -10% as it aged, meaning it can have a 20% swing over time and still be within it's rated specs. Of course you would not use 10% resistors if you were building a precision phono stage, but just the same, no matter how high precision resistors you use they can all shift over time and not necessarily in the same direction. This was a very big problem with some of the early CD players when resistor value shifts in the servo feedback curcuits caused big problems with CD players starting to mistrack.

So as close as you get it to the curve when it is built, resistor value shifts can easily get it out of spec over time. How audible will this be, who knows, but my point is if you are trying to get a perfect match to the RIAA curve it is very difficult to do (and maintain over time) in the analog realm.
[/quote]

That is why you buy stupidly expensive resistors from these guys:

http://texascomponents.com/pdf/tx2575.pdf

Average price is ~ $16.00 each.

Since preamps with inverse RIAA networks average between 14-22 resistors, if your goal is perfect(as near to as possible) the cost of the resistors is acceptable given the cost of most phono stages, I would think.
Title: Re: RIAA Accuracy
Post by: tmazz on January 22, 2022, 07:07:53 PM
Dave actually I think the best solution is to just get it as close as you can and once it sounds good stop stressing over how close it measures on paper and just sit back and enjoy the music. (https://groups.tapatalk-cdn.com/smilies/1733/1534579949.3596-smiley.gif)
Title: Re: RIAA Accuracy
Post by: P.I. on January 22, 2022, 09:13:39 PM
Dave actually I think the best solution is to just get it as close as you can and once it sounds good stop stressing over how close it measures on paper and just sit back and enjoy the music. (https://groups.tapatalk-cdn.com/smilies/1733/1534579949.3596-smiley.gif)
Well, there's that!   :thumb:
Title: Re: RIAA Accuracy
Post by: steve on January 23, 2022, 07:06:49 PM
[quote author=tmazz link=topic=8387.msg103763#msg103763 date=1642874317
And the problem even goes further than that. the tolerance rating n a resistor is a tolerance across the lifetime of the piece. it is not uncommon for a resistor to shift value over time. A 1% resistor is rated to be no more than 10% away from its rated value at all times. but this means that it could be 10% when it is brand new and end up at -10% as it aged, meaning it can have a 20% swing over time and still be within it's rated specs. Of course you would not use 10% resistors if you were building a precision phono stage, but just the same, no matter how high precision resistors you use they can all shift over time and not necessarily in the same direction. This was a very big problem with some of the early CD players when resistor value shifts in the servo feedback curcuits caused big problems with CD players starting to mistrack.

So as close as you get it to the curve when it is built, resistor value shifts can easily get it out of spec over time. How audible will this be, who knows, but my point is if you are trying to get a perfect match to the RIAA curve it is very difficult to do (and maintain over time) in the analog realm.

That is why you buy stupidly expensive resistors from these guys:

http://texascomponents.com/pdf/tx2575.pdf

Average price is ~ $16.00 each.

Since preamps with inverse RIAA networks average between 14-22 resistors, if your goal is perfect(as near to as possible) the cost of the resistors is acceptable given the cost of most phono stages, I would think.
[/quote][/quote]



To whom it concerns.

I started this string to help the public maximize their LP experience, to help diyers, but what is with the negative comments. Why hate accuracy and resistors while exhorting expensive NOS tubes and capacitors costing hundreds of dollars, sometimes for each tube and capacitor? No one has to spend big bucks in the phono design to obtain the correct RIAA part values. So what is it to you guys what we do?

To the general public.

I checked the link (PI's post) and compared the passive capacitor values to Stanley's, and the discrepancy was 8.5%, the same or more error than my computer program. That is a whopping 0,7% db off the RIAA standard and clearly perceivable because a discrepancy exists over many octaves (similar to preamplifiers and amplifiers), affecting the music in virtually every category from bass through treble.
So why dice accurate RIAA design?

(The link I listed in my initial post provides resistor values to obtain very very close to optimum values using Stanley's Lipshitz's equations. It does not cost any more if one does not wish to spend more.)

In response, a measurement is not just a spec on paper, as the RIAA audio response curvature across the entire audio band changes, degrading the natural music in virtually all ways. Why purchase expensive capacitors just to have the sound degraded due to improper RIAA response? It does not make sense.

For Tom, are you stating that a 1% tolerance resistor's ohmage changes 10% of the 1% tolerance over time, or that a 1% tolerance resistor changes 10% of its total ohmage over time?

If changing 10% of 1% of a 10K resistor, the change would only be 10 ohms, which would result in a very minor RIAA db change, in the low milli dbs (thousandths of 1 db). 

In my own, 25 year old test phono stage, the parts stayed extremely accurate. I did recently change the value of the parts when I used the calculator based on Stanley Lipshitz's equations. The music is clearly better. (pun intended)

I started the string so as to help the public, to ask questions when purchasing, and diyers to obtain the most from their designs. Why settle for less when one can have more?

cheers

steve
Title: Re: RIAA Accuracy
Post by: P.I. on January 23, 2022, 07:18:35 PM
No.

The link is to naked resistors with tolerances as close as 0.001% with TCR of 1ppm.

Are you sure you went to the right site?
Title: Re: RIAA Accuracy
Post by: steve on January 23, 2022, 07:55:03 PM
No.

The link is to naked resistors with tolerances as close as 0.001% with TCR of 1ppm.

Are you sure you went to the right site?

I went to your link. I also edited my previous post. Again, are you claiming a 1% resistor changes 10% of its total resistance over time, or 10% of the 1% tolerance over time? I thought about purchasing super tight tolerant resistors, but I don't play the TT that often. The RIAA is extremely accurate.

One does not have to purchase naked, expensive resistors to be RIAA accurate.

Cheers

steve
Title: Re: RIAA Accuracy
Post by: steve on January 23, 2022, 08:06:43 PM
Even if the designers math is right when calculating the parts values used in their phono stage,
parts tolerances can ruin the RIAA accuracy. Every single resistor and capacitor used
 in the network has to be measured and matched if errors are to be minimized.
How often does anyone think that happens in a phono stage that is not custom made.
Tolerance stacking is a problem in many fields of endeavor including our hobby.
In some cases it can actually lead to products that are unsafe to operate.
Scotty

It is called trimming, to obtain accuracy. Resistors change virtually zero, at least in my designs. Yes, one measures each part and matches, which is not your run of the mill phono section. So we should also not? But then many purchase expensive tubes and capacitors, for what purpose. In a passive design, there is minimal tolerance stacking. My designs stayed tight tolerance for 2 decades or longer, until I adjusted using Stanley's designs.

Maybe, just maybe, some of us like higher quality.

cheers
steve
Title: Re: RIAA Accuracy
Post by: P.I. on January 23, 2022, 10:30:22 PM
No.

The link is to naked resistors with tolerances as close as 0.001% with TCR of 1ppm.

Are you sure you went to the right site?

I went to your link. I also edited my previous post. Again, are you claiming a 1% resistor changes 10% of its total resistance over time, or 10% of the 1% tolerance over time? I thought about purchasing super tight tolerant resistors, but I don't play the TT that often. The RIAA is extremely accurate.

One does not have to purchase naked, expensive resistors to be RIAA accurate.

Cheers

steve
Steve, at no time did I ever claim ANY percentage changes over time.  That was Tom's point.  For whatever reason, my response was not broken out from his.

I can tell you this:  in the long in the tooth dB Audio Labs DAC, changing the I/V resistor and a couple of others in that DAC made profound improvements in SQ.  A very much modified version of that DAC is one of the 3 that I rely upon. NOS DACs for Redbook (the only format I use), when done right, are my overwhelming choice for musical ENJOYMENT.  I'm not a 'more data' freak.  But that is a totally different discussion.  Perhaps I'll create a thread on "What, to you, comprises a musical DAC?"
Title: Re: RIAA Accuracy
Post by: steve on January 24, 2022, 08:01:27 AM
No.

The link is to naked resistors with tolerances as close as 0.001% with TCR of 1ppm.

Are you sure you went to the right site?

I went to your link. I also edited my previous post. Again, are you claiming a 1% resistor changes 10% of its total resistance over time, or 10% of the 1% tolerance over time? I thought about purchasing super tight tolerant resistors, but I don't play the TT that often. The RIAA is extremely accurate.

One does not have to purchase naked, expensive resistors to be RIAA accurate.

Cheers

steve
Steve, at no time did I ever claim ANY percentage changes over time.  That was Tom's point.  For whatever reason, my response was not broken out from his.

I can tell you this:  in the long in the tooth dB Audio Labs DAC, changing the I/V resistor and a couple of others in that DAC made profound improvements in SQ.  A very much modified version of that DAC is one of the 3 that I rely upon. NOS DACs for Redbook (the only format I use), when done right, are my overwhelming choice for musical ENJOYMENT.  I'm not a 'more data' freak.  But that is a totally different discussion.  Perhaps I'll create a thread on "What, to you, comprises a musical DAC?"

I noticed that quoting seems to be messed up a bit. That was Tom's point; I understand and will alter my previous post to Tom mentioning it, not you.

In general, I do support custom manufacturers, I was one myself, as I think many advances come from our segment of the industry. Large manufacturers do have a role to contribute as expensive improvements, such as cone materials require a substantial investment.

Looks like this thread has run its course. I hope that the information has been useful.

cheers

steve

Title: Re: RIAA Inaccuracies
Post by: tmazz on January 24, 2022, 04:39:41 PM
Perhaps my editing screwed up the message. My point was that the rating of a resistor is within that % over the entire lifetime of the resistor. So theoretically if a resistor has an x% tolerance it can come off the assembly line at +x% of its rated value and shift to -x% overtime and still be with specs.  My experience with this has been with more run of the mill resistors. It seems from your later posts that you have not experienced this with the  ones you have used.  If the is the case I certainly yield to your expertise in that area and will modify my comment to say that unless you use carefully selected high quality parts, value shifts can cause inaccuracies over time. (And that does not have to mean super expensive tight tolerance resistors,  just high quality)

And on I certainly hope you did not take any of what I said as a negative comment on your posts. Quite the opposite I was trying to highlight  what I saw as at least one of the reasons why it is so difficult to get dead on right as opposed to just throwing something together that is just close enough to get out the door and on the market. Guys like you and Dave really sweat the details and are few and far between. One of my biggest career regrets is that although I started out as an EE, I quickly got tracked into the management and financial end of the business and although I used my engineering knowledge through the day I retired I actually got to spend very little time doing actual design work.  As a result I know
W enough to be dangerous.  :lol:

Even though I don’t have the level of detailed that you guys do, I know enough to have a deep appreciation for your expertise and what it took for you to get there.  :thumb:

Title: Re: RIAA Inaccuracies
Post by: steve on January 24, 2022, 06:17:40 PM
Perhaps my editing screwed up the message. My point was that the rating of a resistor is within that % over the entire lifetime of the resistor. So theoretically if a resistor has an x% tolerance it can come off the assembly line at +x% of its rated value and shift to -x% overtime and still be with specs.  My experience with this has been with more run of the mill resistors. It seems from your later posts that you have not experienced this with the  ones you have used.  If the is the case I certainly yield to your expertise in that area and will modify my comment to say that unless you use carefully selected high quality parts, value shifts can cause inaccuracies over time. (And that does not have to mean super expensive tight tolerance resistors,  just high quality)

And on I certainly hope you did not take any of what I said as a negative comment on your posts. Quite the opposite I was trying to highlight  what I saw as at least one of the reasons why it is so difficult to get dead on right as opposed to just throwing something together that is just close enough to get out the door and on the market. Guys like you and Dave really sweat the details and are few and far between. One of my biggest career regrets is that although I started out as an EE, I quickly got tracked into the management and financial end of the business and although I used my engineering knowledge through the day I retired I actually got to spend very little time doing actual design work.  As a result I know
W enough to be dangerous.  :lol:

Even though I don’t have the level of detailed that you guys do, I know enough to have a deep appreciation for your expertise and what it took for you to get there.  :thumb:


No sweat on this end Tom. All is good here. I appreciate the clarification and your findings and I have also noticed variations of a questionable brand or two; but been too long to remember. I have not had any problems with mills except a very rare noisy 5 watters. All seems to be well with copper leaded Resista, Holcos, Vishay, but I did have a little problem with the new PRPs with initial tolerance variations. Haven't had enough time with them for a longevity study either. I stay away from Caddock 132s, a metal oxide type, and a bland sound.

For general consumption, A + to - 1% variance will alter the RIAA in my circuit in the less than 0,1 dbs range. Three resistors are crucial in the RIAA network, plate, R1, and R2, as listed in my link based on Stanley's calculations. The good resistors seem to always measure within 0,25% or so using a Textronix meter. Resista, Holco, vishay resistors seem to measure less than 0,25% between dozens of resistors, pretty reliable. I see Michael Percy has Vishays for $12.00.

Thank you for the appreciation Tom. I cannot calculate all the R&D, sophisticated, proprietary listening test since 1980 or sooner. And for no pay; out of my own pocket. I have even tested if wires transmit in the audio realm and at what frequencies.

Cheers and all the best Tom.

steve
Title: Re: RIAA Inaccuracies
Post by: tmazz on January 24, 2022, 09:03:04 PM
Just so you know where I'm coming from, here is the story behind the resistor shifts. Back in the first generation of CD players there was a resistor in the transport across which a voltage was read that corresponded to the position of the laser. This resistor has to be an exact value to properly report the position but the first generation machines were selling in the $1000+ range so they could afford to put a high priced precision resistor in that spot. Fast forward a year or two and CD players had already dropped into the $150 price range. At that retail price the manufacturers could not afford to use those expensive resistors so instead they put in a cheap resistor and used a trim pot to dial the combination to the exact number of ohms they needed. At the time I had a frind who was a repair tech at one of the local brick and motor stereo shops. He told me that about a year after this new generation of CD player came out the started to get flooded with repairs of units that had begun to mistrack. The advice he got from the manufacturers on the problem was that the feedback resistor had shifted in value during break-in and all he needed top do was to adjust the trim pot to bring the combo back into spec. His advice to me when I bought a CD player during that time frame was to put in a disc when I first got it, set the machine for continuous repeat and let it play for a week or two strait the break in that resistor so if it shifted (and they didn't all shift out of spec) it did so before the player was out of warranty.

So I just wanted you to know that I didn't pull the resistor shift idea out of my left ear, it came from real world experience, not only of myself, but of a technician who probably spent 100s of hour making repairs on CD player that had the very problem I was talking about.

I am however glad to hear that you have not experience the same kind of thing. Maybe it is a thing of the past and there was some kind of factory flaw in resistors back then and also I think you would be much more picky about who you source things from than a mass market manufacturer like Sony or Sharp who would be planning production runs of literally millions of units.
Title: Re: RIAA Inaccuracies
Post by: steve on January 24, 2022, 10:34:33 PM
Just so you know where I'm coming from, here is the story behind the resistor shifts. Back in the first generation of CD players there was a resistor in the transport across which a voltage was read that corresponded to the position of the laser. This resistor has to be an exact value to properly report the position but the first generation machines were selling in the $1000+ range so they could afford to put a high priced precision resistor in that spot. Fast forward a year or two and CD players had already dropped into the $150 price range. At that retail price the manufacturers could not afford to use those expensive resistors so instead they put in a cheap resistor and used a trim pot to dial the combination to the exact number of ohms they needed. At the time I had a frind who was a repair tech at one of the local brick and motor stereo shops. He told me that about a year after this new generation of CD player came out the started to get flooded with repairs of units that had begun to mistrack. The advice he got from the manufacturers on the problem was that the feedback resistor had shifted in value during break-in and all he needed top do was to adjust the trim pot to bring the combo back into spec. His advice to me when I bought a CD player during that time frame was to put in a disc when I first got it, set the machine for continuous repeat and let it play for a week or two strait the break in that resistor so if it shifted (and they didn't all shift out of spec) it did so before the player was out of warranty.

So I just wanted you to know that I didn't pull the resistor shift idea out of my left ear, it came from real world experience, not only of myself, but of a technician who probably spent 100s of hour making repairs on CD player that had the very problem I was talking about.

I am however glad to hear that you have not experience the same kind of thing. Maybe it is a thing of the past and there was some kind of factory flaw in resistors back then and also I think you would be much more picky about who you source things from than a mass market manufacturer like Sony or Sharp who would be planning production runs of literally millions of units.

I don't doubt your experience with resistors. I have had two particular cd players, both the same model, that had the problem you described. The third one, same model, is still working, but I don't use it often. Will see if time causes the same drifting and subsequent problem.

Cheers and thanks Tom.

steve