RIAA and iRIAA accuraccy in the real and simmed world.

the road not taken.
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

RIAA and iRIAA accuraccy in the real and simmed world.

Post by izzy wizzy »

daves edit.... split his off from Naz's thread


Hi Lars,

I'm not so sure about the measuring thing when it comes to phonos. Look at the accuracy required of instrumentation to ensure that what you are measuring is accurate. There are 3 pieces required to get it right. A frequency generator, an AC voltmeter and a Frequency Counter. All of these have be an order of magnitude better than that being measured otherwise, IRL, what is the point?

So say we are aiming for our EQ to be in a 1dB window which is about 1%. To measure that accurately will require test equipment with 0.1% accuracy. That is very expensive test equipment. Each test point will have to be beasured at its input and output because I doubt the Generator will maintain its output accuracy over such a wide range.

I reckon that the best that can be achieved is to sim it and bung the parts in. Go for 1% or better. Personally I aim for a 0.1dB window in the sim. With part tolerances at 1% or better, the result seems to be OK. What happens with an LCR and the choke values, I have no idea. My methods could be all wrong for that type of thing :oops:

If you want to measure anything, in RC EQs, the worst deviation will be 100Hz down where for the most part, quite large deviations don't matter that much. Measure that and forget the rest :wink:
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

I measure with testrecord/PU/arm/turntable/phonostage. Think this should be the most accurate. Have earlier used an old Decca-record but must have a new one. Any suggestions?
Brgds
Lars
Naz
Posts: 71
Joined: Wed Aug 06, 2008 4:27 am
Location: Sydney Australia

Post by Naz »

Interesting points. I've been using a Hagerman Rev RIAA filter along with some pretty capable (and expensive) test equipment. But I've found that the theoretical never produces the optimal sound in a given system without some minor tweaking, and a perfect response doesn't necessarily sound best.

I originally incorporated adjustable controls in my Phono for a couple of the elements in the RC RIAA. I then found them useful for balancing badly Equ'd recordings. It makes enough difference to make some records playable (and perhaps even sound good) that you'd normally totally avoid.

I'd like a more accurate Rev RIAA but I've never got around to doing anything about it. Any ideas?

Naz
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Nothing against the Hagtech iRIAA (I own one) but with all due respect, it isn't accurate enough IMHO to decide if the EQ in a phono stage is to spec or not as the parts are not high enough tolerance and it is a compromise at best.

The best inverse RIAA I've seen was in a Glass Audio article which I still have. The basics are on my sim page http://www.izzy-wizzy.com/audio/preampneweq.html

The SPICE Hagerman might even be better for sims but I haven't checked it against the RIAA standard.
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Nothing against the Hagtech iRIAA (I own one) but with all due respect, it isn't accurate enough IMHO to decide if the EQ in a phono stage is to spec or not as the parts are not high enough tolerance and it is a compromise at best.

The best inverse RIAA I've seen was in a Glass Audio article which I still have. The basics are on my sim page http://www.izzy-wizzy.com/audio/preampneweq.html

The SPICE Hagerman might even be better for sims but I haven't checked it against the RIAA standard.
I noticed with the Hagerman RC version versus the Spice version you guys were playing with, that I get more bass out of the Hagerman, in other words the iRIAA seems to cut the bass a tad too much so the designers is led to believe they need filter the bass less. The end result seems to be stages that have a bit too much bass emphasis.

-- Andrew
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

The SPICE Hagerman might even be better for sims but I haven't checked it against the RIAA standard.
The "SPICE Hagerman" IS the theoretical RIAA standard! If one changes the 322,08ohm to 56 ohm we get the original Lipshitz RIAA standard.

I have done a few sims feeding the Pultec/Tango directly from the SPICE-RIAA and it has to be tweaked quite a bit to confirm. Actually the Tango should be terminated with 598ohms to work best!

Maybe the 322,08ohm conversion isn´t exactly right as the 4th constant is said to be a 2nd order filter. Anyone who knows if it is a Butterworth, Bessel or something else?

Stephen think I saw the 2nd order filter in your site....
Attachments
LCRRIAAdirekt.asc
(17 KiB) Downloaded 1243 times
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Had to check what happened with a 2nd order 50kHz Bessel-filter added at the revRIAA.

When adjusting the LCR to get the best possible frequencyplot 20-20kHz and than adopt this to the Hagerman/AllenWright something interesting happens!

Hagerman and Allen Wright have done it all wrong if the 50k is 2nd order!!!!

To get the best out of Hagermans SPICE RIAA the 322,08ohm has to be adjusted to 278ohm but still it is not exact.

So use the Lipshitz followed by the ideal-opamp filter instead. The Lipshitz has a 374kHz filter, maybe that should be removed?

Yet to find out what filter-characteristic to use.....
Attachments
LCRRIAAdirekt2nd.asc
(8.96 KiB) Downloaded 1026 times
Last edited by reVintage on Sat Mar 28, 2009 1:03 pm, edited 1 time in total.
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I've yet to look at your sims but have saved them for a look later.

I'll have to go look at my notes but the inverse RIAA I use I thought was to the RIAA standard. IIRC, it is within 0.05dB at worst and a lot closer over most of its range. Now you bring up the notion that there is the RIAA standard and Lipshitz RIAA :?

I incorporated the 50kHz filter as a seperate filter after a discussion with Morgan Jones. If it was used, then it would have been a seperate filter to roll off the feed to the cutter head after the inv RIAA EQ applied at cutting boosting the signal too much (which I guess we all know) Morgan suggested it has way more affect than it looks to have and so you can't just calculate it and modify one part of the EQ. And to that effect, most will only put in a resistance in an RC EQ at about 80% of the calculated value for it to sound right. That was what I was doing until I resimmed with the 50kHz filter.

The re-sim showed just how much it affects the upper mids and so when I altered the EQ in my phono, it sounded more together and whole. You'd also think that it would sound more brilliant with the 50kHz point added but it doesn't really; it just sounds a more right. Overall that is. I'd love to have this switchable but I don't think it's just a matter of switching one thing as it seems to alter many of the variables.

This whole area of discussion makes you wonder just how accurate many phonos are and to what standard are people making their claims? Never mind what is in those all-in-1 cans.

BTW guys, I've been waiting years to have some dicussion with like minds on this subject :D
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Stephen,
Sorry, found some errors in my spicemodels. Not as big as the errors found in the works by the so called diygurus.
Hagermans RIAA was off, the 50kHz filter that you refered to Morgan Jones was off.

Attached is the original RIAA followed by the Neumann compensated RIAA. They drive LCRs adjusted for best frequency response 20-20kHz.
Attachments
LCRRIAAdirekt2ndnew.asc
(8.83 KiB) Downloaded 1084 times
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I better clarify, Morgan suggested how I should model it. The filter design has nothing to do with him. I just checked it and at 50kHz, it's -3dB with the values I show on my site.

Isn't that what it should be?
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Stephen,

1009 and 757p when I calculate a Bessel. But my slideruler can be worn :lol: !

Also used Bessel in my sims but I doubt they used that. Probably Butterworth.

If you use Butterworth you get completely different figures :( . Anybody got a Neumann manual lying around?
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I think we might be getting into areas where the accuracy cannot be obtained or verified IRL with real world components, variations, instruments and so on. One of the side issues of using sims - lots of numbers. Best to get the "standard" as close as possible though.

Just to add to the complications, I have read that some records were cut with a different cutter rolloff freq :?
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hi Stephen,

In a Spice sim we do exactly the same mathematics and calculations that would have taken us hours to do manually. So this is the tool to use. To get to the best real life situation, the rev-RIAA model we start with must be correct. From there we can then make compensations and changes.

Have done some research(Google:wink:) and found that the Bessel-filter used in your simulations isn´t correct, as Neumann uses Sallen-Key. The Bessel introduces a 0,3dB error at 20kHz when compared to a LCR optimised for Neumann or Pultec/Tango:cry: ! See the sims below: First Bessel, second Sallen-Key, third RIAA Pultec/Tango and fourth Hagerman/Wright. At 50kHz we will still have an error of ca 2dB wether using Bessel or Sallen-Key, due to the fact that we compensate a 12dB/oct filtering with a 6dB/oct.

The rev RIAA proposed by Hagerman and Allen Wrigth has an unacceptable error of 0,5dB if my sims are right! There might be a flaw in my sims as the error is opposite to the assumed :oops: !

As you pointed out there are also two frequencies used. SX-66 used 33kHz and the more commonly used SX74 used 50kHz.

What we also should look at when compensating is the phase-errors introduced but we can leave that for now.........


Great this subject came up, as it shows we all used incorrect assumptions about revRIAA.
Attachments
LCRRIAAdirekt3ndnew.asc
(17.97 KiB) Downloaded 1067 times
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I think we may have had a slight misunderstanding. I'm all for getting this accurate and correct. I was just wondering about just how accurate we need it for real life. So essentially we're saying the same IMO :)

My reference is Horowitz and Hill: The Art of Electronics. But going to the net, I found this http://www.daycounter.com/Filters/Salle ... ator.phtml From my reference, my understanding is Sallen and Key are a type of filter block that can be configured for Bessel, Butterworth and Chebyshev style filter responses.

I think I had assumed Butterworth even though my values are a bit off. The Bessel shows approx -5dB @ 50kHz while the Butterworth with correct values shows -3.01dB and the one on my site -2.94dB. IIRC, I probably thought this was close enough for a phono stage.

So I'm a bit surprised you thought I used a Bessel for the 50kHz pole?

I still haven't had time to look at the RIAA EQ variations.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

You used Bessel, nothing to argue about :wink: ! The error in the filter in your site is that it is a close to 64k Bessel with 3dB loss at 50khz.

When using Butterworth (in daily use wrongly called Sallen-Key) one cap has double the value of the other.

Sallen-Key is the filtertopology where the Q-factor is set to the right charasteristic ie Linkwitz-Riley Q=0,5 , Butterworth Q=0,707, Bessel 0,866 etc.

You have all the answers in the previously attached sim, why don´t you check it :P ? Please analyze them, as I am anxious to know your opinion of the four examples. Also, as I mentioned earlier , I am little troubled by the the results of the Hagerman.
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

At last I found it! These are the words of legendary Swedish cutter Goran Finnberg about the SX74/VG74:

"very slightly peaked Sallen & Key unity gain
active two pole butterworth filter set to - 3 dB @ 50 kHz as followed by
a single pole at 96 kHz"

Below is the revRIAA I made in accordance to Gorans findings. Running a RIAA compensated for this one leaves Allen Wrights with an error of ca
-0,3dB at 20kHz.
Attachments
VG74revRIAA.asc
(3.38 KiB) Downloaded 1039 times
Brgds
Lars
Naz
Posts: 71
Joined: Wed Aug 06, 2008 4:27 am
Location: Sydney Australia

Post by Naz »

Good and Interesting stuff guys. I haven't been able to check this thread for a few days but I like where it's going. Lars, I added your latest Rev RIAA to my optimised Phono schematic to see the comparison. Does the result meet with your expectation?

Cheers,
Naz
Attachments
LCR_GC_30-3-09_RevRIAAComparison.asc
(12.96 KiB) Downloaded 942 times
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I got a bit confused when you said in LCRRIAAdirekt3ndnew.asc that the last two circuits represented RIAA/Pultec/Tango and the last one was Wright/Hagerman. I don't see how they represent what you say so what have I missed/misunderstood? IOW, did Wright propose a SPICE invRIAA? Neither have done LCR EQs AFAIK.

The one you identify as Pultec/Tango has a fantastic response. OK it doesn't have the extra 90kHz cutter head rolloff but the EQ curve is wonderfully smooth. It may be able to be tweaked to be better in a sim but with or without the 90k point, it looks great IMO.

I tried the revised HF cutter rolloff sim on the sim of my phono and mine can be made better so will be keen to try to see how audible this is.

cheers,

Stephen
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Haven´t refered neither to Wright proposing a SPICE invRIAA or him having done a LCR.

What I wanted to show was that he used wrong assumptions, making his proposed RIAA correction totally wrong. It isn´t better than using the original RIAA.

Also the Tango isn´t fantastic :wink: ! It is just OK when fed with a theoretical standard RIAA.

Below is a plot showing how Wrights correction(AW) is more at fault than a std RIAA(P/T) when fed with the SX74 revRIAA with its 5th constant. The LS is mine adjusted to the SX74.

As you can see the Wright-compensated RIAA is as much at fault as the std RIAA, + or - 0,3dB, when fed with a Nemann compensated revRIAA.

IMPORTANT 1:
Phase response isn´t impreesing as we try to make our corrections with 1st order filters.

IMPORTANT 2:
Adjustment of response can only be done with testrecord/cartridge/arm/turntable/RIAA-stage

IMPORTANT 3:
Do we know for sure that the Neumann corrections aren´t compensated for, to make the record having the std RIAA response?

Also Neumann isn´t the only manufacturer on earth. What about Westrex and all the others.

Probably my fantastic Deccas from the mid-sixties aren´t made using SX-74 as it was introduced in the beginning of the seventies!

Will have a chat with my friend Jan-Erik Persson at Opus3 records to see if he can make any clarification. Opus3 uses SX74 when cutting their present 180gr vinyls 8) ! When he was at CuttingRoom he was coworker with Finnberg.
Attachments
revRIAAjfr.PNG
revRIAAjfr.PNG (60.45 KiB) Viewed 38147 times
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

This is what happens when we feed the different LCRs with a signal that is according to the RIAA standard.

P/T as is
AW optimized to Wright/Hagerman revRIAA
LS optimized to the SX74 5th constant revRIAA

This time AW is off by 0,6dB at 20kHz.
Attachments
revstdRIAAjfr.PNG
revstdRIAAjfr.PNG (61.47 KiB) Viewed 38145 times
Brgds
Lars
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Lars,

I'm deeply impressed with your efforts here! Really nice work. I may have missed this, but if I were to rework my 1k5 what curve/Spice/Circuit should I be using for the iRIAA? Recall out most excellent host provides taps either side on the inductors, so I have the ability to tweak up and down if needs be...

thanks,

-- Andrew
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

There are only three components that normally needs tweaking in a LCR.

1 The series/terminating resistor to fix low end, so it has nothing to do with the Neumann thing. But it needs trimming due the less than ideal surroundings :wink: !

2 The 126n cap from the Tango, 50,4 in yours.

3 The inserted resistor above the 126n

My suggestion is, either you do something that is right in the middle of all or you make it switchable. Think two will be enough. Anyway Hagerman that I earlier believed in, is as it looks right now a NONO as it seems to boost treble to much!

But I´d still like to see the final adjustment done with a testrecord.

But I don´t think we are ready with this issue yet, more research is needed.
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

I am still getting confused by some of your references and terminology.
Haven´t refered neither to Wright proposing a SPICE invRIAA or him having done a LCR.

What I wanted to show was that he used wrong assumptions, making his proposed RIAA correction totally wrong. It isn´t better than using the original RIAA.
And yet you say ....
P/T as is
AW optimized to Wright/Hagerman revRIAA
and another reference elsewhere to Hagerman/Wright when an invRIAA is feeding a LCR.

I know Jim H proposed a SPICE invRIAA which I remember you finding an error in but what has Allen W got to do with this? Is it that he proposed a 50kHz point and that is represented in the invRIAA of Jim H and so you have lumped them together?

Part of my problem with this may stem from my lack of understanding of the SPICE invRIAA which is why up until now, I used a circuit that worked rather than the block you have been using.

IIRC, the only place I've seen Allen W mention the 50kHz node is in his phono stage "white paper" and in the odd post on AA. How do you derive this into an invRIAA that is in error? And a SPICE one at that?

I'm not asking these questions to pick holes in your work Lars but trying to remain in the discussion and understand what for me has never been discussed before and as I also said before, I've been waiting years for this and don't want to be left behind so soon :)

It's no good looking at numbers in a sim if you don't know where the source or numbers come from/represent.
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

Can you please include the .asc for the last 2 posts you show graphs for?

And what does P/T stand for?

Thanks!
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hello Stephen,
Will not publish or answer anything more until I have had the chance to digest all what have come forward today. Both in favor and not for the Hagerman approximation :? .

The "Hagerman/Wright Connection" is that Hagerman refered to only Wright when spicing his iRIAA. Didn´t you read hagermans pdf :wink: ?
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,
Didn´t you read hagermans pdf ?
Well to be honest, I skim read it and then forgot about it. When the RIAA curve ends up being "derived" into some kind of formula to be represented in SPICE and I don't fully understand how it came to be, I lose trust in my ability to use it properly. I also end up being reliant that they got it right and so again, I don't have the hands on trust I need to use it. Remember, I don't measure at all so have to trust the sim. Then you find there's an error in it someplace :wink:

My preference to date has been my invRIAA sim circuit referenced point by point to the published RIAA standard. That up until now has been the best I can do. We might be able to do better than that now :D

Can you tell me what does P/T stand for?
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

My preference to date has been my invRIAA sim circuit referenced point by point to the published RIAA standard. That up until now has been the best I can do.
I´m lost :oops: ? Checked your website and the one you publish there is based on Wright, ie far from the STD RIAA. So how do you mean by point by point? The RIAA std is as I understand based on only three poles: 3180 318 and 75uS, not on points on a frequency-table.

As I want to sim your invRIAA, do you have the models for the various Op-amps involved?
Brgds
Lars
izzy wizzy
Posts: 64
Joined: Sun Jun 26, 2005 8:38 pm
Location: Folkestone, UK

Post by izzy wizzy »

Hi Lars,

See below for my invRIAA. I have it seperate from the 50k pole so I guess I confused you when I said that I'd checked my invRIAA circuit. I removed the blocks and have reverted to a raw .asc. You should have all the models as part of a standard LT SPICE install as IIRC, that's what I did.

I've also included my manual record of that check in .xls format.

Running my invRIAA and the SPICE invRIAA (forgetting any other poles for now) into a Tango LCR shows a spread of 0.04dB for the SPICE invRIAA vs 0.06dB spread for my invRIAA. If I could place an offset on one of the traces, they could be seen on the same graph but I don't know how to apply an offset.

But the subject of phase has been coming up slowly in this discussion and there my method may be lacking as a reference. I'm not wedded to this thing; It's only ny best shot to date. A bit of peer review is always a good thing :)
Attachments
invriaaeqreva.asc
Inv RIAA Circuit
(2.87 KiB) Downloaded 1057 times
riaa-comparison to LTSpice.xls
RIAA to LT SPICE comparison
(36.5 KiB) Downloaded 1041 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Thanks Stephen,
I am at the moment going through all different invRIAAs to make a comparision against the "correct" Neumann and the ideal std RIAA.
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Checked your circuit and unfortunately the filter isn´t capable of driving a 600ohm RIAA. The figures you get are off. To make it work, a buffer has to be inserted. As is below shows your filters response whatever load you add.

Compare with the ideal RIAA at the bottom. No need to use your overly complicated circuit to get less accuracy.

One good thing though. Probably your circuit is more like an IRL invRIAA. The same can be done with better control in the circuit at the bottom by just altering one resistor value.

Still waiting for the schematic of the SAL74 to arrive by E-mail......

EDIT: Added the Lipshitz/Jung passive at the bottom, this one a tad better than the top one but they show similarity.....
Attachments
invriaaIZZY].asc
(11.91 KiB) Downloaded 996 times
Brgds
Lars
Post Reply