Hi Hans
Thanks for visiting the post.
I confirm that the formula as I posted it is correct. I re checked it by entering it in a spreadsheet. It produces the proper results.
There maybe a glitch in the copy-paste you have done.
Also the formula you provided gives strange results with or without these two red zeros. Please check again.
George
Thanks for visiting the post.
I confirm that the formula as I posted it is correct. I re checked it by entering it in a spreadsheet. It produces the proper results.
There maybe a glitch in the copy-paste you have done.
Also the formula you provided gives strange results with or without these two red zeros. Please check again.
George
George we are both right, I overlooked one of the many of the brackets that you used.
But (100*(0,001*A1)^2+2.505) is the same as 100*((0.001*A1)^2+0.002505).
Then join 444.18 with 100 and you have 4.442 (I made a silly typo with 444.18/100=4.412 instead of 4.442)
It should be now with typo's and brackets corrected:
=10*LOG10(4.442*((0.001*A1)^2+0.2505)/(((0.001*A1)^2+0.002505)*((0.001*A1)^2+4.503)))
Hans
Last edited:
Hans
Why didn’t you just blind copy/paste the g d thing like I did?
I started worried about your future when I read your post where you wrote that you want to know how things work (see, curiosity killed the cat).
Now you make me worry even more!
Be well and be productive
George
Just out of curiosityHans
Why didn’t you just blind copy/paste the g d thing like I did?
George
Hans
Hi George,Your corrected formula is OK.
It is 0.030dB higher that the other two RIAA formulae.
Well done.
(I don't ask how you derived it, I will not understand the explanation)
George
I won't explain, it was just because your formula looked so different from the usual exact and more universal version,
10log(97,97*((2*pi*t2*A1)^2+1)/(((2*pi*t1*A1)^2+1)*((2*pi*t3*A1)^2+1))).
with t1=3180, t2=318 and t3=75 usec.
that it made me go into details, and of course at the end one could be translated into the other.
So see it as a puzzle that I tried to solve.
Unfortunately I overlooked a single bracket in your formula leading to some unwanted irritation on your side.
And may I mention without any hard feelings that when the IRiaa .wav was presented in #3, there was no mentioning that it concerned pseudo random noise, leading me to even suspect RMAA for a short while.
But then it became clear how Scott had created the file.
So what I want to say, it is never wrong trying to understand things.
Hans
Hans
I hope you understand I was teasing you.
I have the greatest respect for the people who are motivated by a healthy curiosity, who never stop searching and asking.
George
>Edit. Yes, I had not mentioned that the IRIAA file was a pseudo random noise.
Scott has appropriately named it psrnoise_IRIAA_48.wav, so it’s not Scott to be blamed for concealment but me.
I exported a test signal file from RMAA, then applied reverse RIAA to that using audacity, played the file through the phono stage and recorded it, then analyzed that file with RMAA. The only problem with this approach was I ran into a noise issue, since I had to adjust playback gain to -20dB or something on a file that had already had an inverse RIAA applied to it - but as it didn't seem to affect the actual results besides reating little anomalies at 50Hz and some harmonics, I didn't worry about it. Next time I'll keep the playback gain at max and use an attenuator between soundcard and phono pre, that should help keep the noise floor sufficiently low.
Finally had time to play...
These were made with Room EQ Wizard. The sound card is a Motu 4Pre, sampled at 96kHz. The phono preamp is a lowly ART USB Phono Plus. The input is driven from the 4Pre line out through a pad, approx 40dB. The Phono Plus analog output was taken into the 4Pre line input. Levels adjusted for no clipping of the Phono Plus.
Unfortunately, I kind of blew it with the vertical scale. But a dB is a dB, so it still shows what's important.
The first graph is with no RIAA compensation, just the playback curve of the Phono Plus.
Next, the RIAA curve in tabular data posted earlier was loaded into REW as a .cal file. The results are the difference between the ideal RIAA curve and what the Phono Plus does, shown in graphs 2 (swept) and 3 (swept and RTA/pink). There are some errors in the RTA response at the low end, likely due to low FFT resolution in the lower bands. The RTA plot goes above 20kHz, but the cal file stops at 20kHz, hence the roll off above 20kHz. There is some noise in the HF area, but you can see the RTA and swept results track pretty well.
The advantage of the RTA/Pink test is you don't have to sync the generator, pink could be recorded. The advantage of the swept test is better resolution, lower noise.
These were made with Room EQ Wizard. The sound card is a Motu 4Pre, sampled at 96kHz. The phono preamp is a lowly ART USB Phono Plus. The input is driven from the 4Pre line out through a pad, approx 40dB. The Phono Plus analog output was taken into the 4Pre line input. Levels adjusted for no clipping of the Phono Plus.
Unfortunately, I kind of blew it with the vertical scale. But a dB is a dB, so it still shows what's important.
The first graph is with no RIAA compensation, just the playback curve of the Phono Plus.
Next, the RIAA curve in tabular data posted earlier was loaded into REW as a .cal file. The results are the difference between the ideal RIAA curve and what the Phono Plus does, shown in graphs 2 (swept) and 3 (swept and RTA/pink). There are some errors in the RTA response at the low end, likely due to low FFT resolution in the lower bands. The RTA plot goes above 20kHz, but the cal file stops at 20kHz, hence the roll off above 20kHz. There is some noise in the HF area, but you can see the RTA and swept results track pretty well.
The advantage of the RTA/Pink test is you don't have to sync the generator, pink could be recorded. The advantage of the swept test is better resolution, lower noise.
Here's the result I got with RMAA - ignore the triangular weirdness, those happen because of noise like I explained in my previous post and don't seem to affect accuracy of measurement apart from those very narrow bands (I tested couple other phono stages as well).
The phono preamp is a slightly modified TC-750, the things I've done which affect response are adjust the RIAA circuit a bit - 100pF cap in parallel with the 3.3nF cap and 22k resistor lowered to 21.5k - the left channel shows behaviour pretty much identical to simulated from 200Hz upwards. Oh and I also added 3300pF to create a low pass filter with the output resistor, this evens out the response from 10-20kHz slightly (like 0.2dB difference at 20kHz...) - the -3dB point for that filter is 100kHz or something. Probably pretty pointless, but I thought it did sound slightly better with those caps in place, at least didn't sound noticeably worse, so I let it be.
According to simulations, power supply voltage does affect the lowend response somewhat - more voltage -> slightly more bass. The stock supply was 12V but I'm using about 21V now. I'm not yet sure about the final PS, but once I've got that figured out, evening out the below 200Hz response should be no big deal, though for subjective purposes I might leave it there. 0.5dB in that area is next to nothing BUT it does sound better with the higher voltage and I'm not sure how much of that is about the slightly different bass response - so if I even it out, it might be it actually sounds slight bit worse to my ears in my system, but would go well with the reference RIAA curve idea...
The phono preamp is a slightly modified TC-750, the things I've done which affect response are adjust the RIAA circuit a bit - 100pF cap in parallel with the 3.3nF cap and 22k resistor lowered to 21.5k - the left channel shows behaviour pretty much identical to simulated from 200Hz upwards. Oh and I also added 3300pF to create a low pass filter with the output resistor, this evens out the response from 10-20kHz slightly (like 0.2dB difference at 20kHz...) - the -3dB point for that filter is 100kHz or something. Probably pretty pointless, but I thought it did sound slightly better with those caps in place, at least didn't sound noticeably worse, so I let it be.
According to simulations, power supply voltage does affect the lowend response somewhat - more voltage -> slightly more bass. The stock supply was 12V but I'm using about 21V now. I'm not yet sure about the final PS, but once I've got that figured out, evening out the below 200Hz response should be no big deal, though for subjective purposes I might leave it there. 0.5dB in that area is next to nothing BUT it does sound better with the higher voltage and I'm not sure how much of that is about the slightly different bass response - so if I even it out, it might be it actually sounds slight bit worse to my ears in my system, but would go well with the reference RIAA curve idea...
I'm actually kind of surprised the TC-750 doesn't match the RIAA curve better, even according to their specs, much less post-tweak. I know it's not an expensive box, but the little ART thing is $80 with a decent ADC in it, and an SPDIF and Optical input too. I was surprised the ART tracked the RIAA so well, and it's noise is down 80dB unweighted. I never gave it much credit, I guess it's ok after all.
You have to be really cautious of making subjective judgements after simulations and minor tweaks, though. It's all highly biased, and sighted, so anything sounding slightly worse or better to your ears...I'd just ignore that, and measure very, very carefully.
The job of the RIAA preamp is to supply gain and accurate RIAA EQ, not to trivialize a rather difficult job. However, tone control is best done elsewhere. There are huge variations in actual program EQ, far greater than a .5dB bump in the RIAA match. Tweak-able equalizers and tone controls are not of the Devil, they're part of life and in the chain upstream of the record anyway (not to mention the gigantic "EQ" sitting just past the amp - your speakers and room). It's no sin to adjust a knob to suit. But attempting to "voice" a phono pre seems a bit silly, if you'll pardon my opinion.
Actually I was rather impressed with that curve and it's the best of the phono preamps I measured, so I'm not quite sure what you're getting at - the ART curve you posted certainly didn't look any better, though it looks good as well, especially considering the price.
On the TC-750, the right channel isn't that great as it is, though it's actually still pretty good compared to the competition, but clearly one of the stock caps or resistors doesn't match it value as closely as the left channel.
The left channel is actually pretty much ideal, from 200Hz-7k it's within 0.05dB and even further up the treble it still stays within 0.1dB. I think there's absolutely no sense in trying to make it better than that. The graph in my previous post was of the multitone test, swept sine seems a bit more forgiving and it sums the channels in the plots - it looks like this using 0.5dB steps in the scale similar to the ART plots you posted:
I'll be replacing the RIAA caps with polypropylene (the stock caps are mica), matching all components to what's currently there in the left channel should provide really accurate results. I'll have to check what supply voltage does to bass in real life and then adjust accordingly, changing one resistor should be all it takes to even it out if need be.
I agree with this. I did make little recordings between tweaks, frankly I don't care enough to blind test, but it makes it possible to switch sample on the fly. Frankly the RIAA adjustments were a bit... uhm, maybe I heard something, maybe I didn't, mostly they could be made out by focusing on some very specific difference, like before adjustment on that particular track there was a violin passage where the strings sounded slightly more metallic and harsh preadjustment, but really subtle difference. Changing the coupling caps from electrolytics to film was a clear improvement on the other hand.
Yeah I get that, and I'm not attempting to voice it really, accuracy is the goal but on the other hand listening to music is the end result, so if it sounds better with that bump, I don't think straightening it out is necessarily worth the effort. In all honesty I doubt how audible it's going to be anyway, all things considered, I don't think 0.5dB in 40-50Hz amounts to much. If it was in the midrange that would be different.
Btw. I don't currently have any sort of EQ in the chain, that's not out of some principle of being against them, just that the way things are setup currently. My other amp has tone controls and I certainly have no qualms about using them. Actually it's a bit sad tone controls have all but disappeared from 'serious' bits of hifi and the implementations on lower end gear are often of the sort that make you think all tone controls sound like **** and you can't find the "tone bypass" button fast enough. It seems DSP based room correction has brought back tone control, it's just not called that anymore and is way more invasive and effective too.
I admit to being an audiophile of sorts, though I'm also well aware of the things that go on in studios and the amount of "stuff" most recordings go through, which make it a bit silly pursuit, but it's a hobby. Frankly I get a bigger kick out of the silly things in audio which are hard or even impossible to measure, like tweaking turntables and all the little magical mystery bits that can go to it if you approach it that way.
The swept curve I posted and the initial graphs you posted did look quite a bit different. Once you used the swept-sine technique, the differences vanished. That's one reason I don't use RMAA.
I wouldn't say "more forgiving", I'd say "more accurate".
Unless you're dealing with tolerance issues, I can't see why polyprops would help anything. But I'll be interested to see how it ends up.
All I can say is, "ABX" before you conclude.
Having calibrated lots of rooms, pro and home, I have yet to find a room/speaker combo that didn't benefit from some form of EQ. The new FIR stuff is darned good.
We differ. I'm of the "everything you hear is measurable" camp, but you do have to know how to measure it and more importantly to correlate with audibility. There's where the productive work is.
Helma's Riaa deviation curves where a factor 4 better than yours, 0.25 dB vs 1 dB from 30Hz to 20 Khz.
I have put them together in one image, both brought to exactly the same scale.
So maybe you should give RMAA a try after all to come with better results.
You may put yourself in as many camps as you prefer, right or wrong.
However stop trying to convince the rest of the world that only you have the true believe.
Hans
Attachments
The difference between the initial curves I posted and the swept sine are quite small really - the biggest difference is the initial curve I posted had the steps/lines at 0.1dB intervals while the swept curve used 0.5dB. Ok, in the bass range the multitone graphs show about 0.5dB bump while the swept shows more like 0.3dB. I don't know why the swept analysis doesn't differentiate between channels and I haven't found a way to make it plot the channels separately.
My guess is the disperancy between channels is a capacitor issue. I haven't measured them, but I plan to once I replace them with new caps (which is waiting for when I put together an order from Mouser) - but it would seem a 0.1nF difference in value between 3.3nF caps in right/left channel would create that difference. That's about 3%, so could happen even with 2% tolerance caps unless you match them by hand. Btw. I don't ofc expect PP caps to mystically "solve" anything, the biggest thing is changing the caps I can match them by hand between channels. I'm also curious to see whether there seems to be any audible difference between the cap types in RIAA circuit.
Btw. even as it is the channel balance is actually very good on the TC-750 and neither of the two other phono stages I measured performed better in this respect, actually they were worse. It stays well within 0.1dB for most of the range and even at 10kHz where the difference is highest it's only 0.15dB. This is not bad for a phono stage.
As for measuring things, I too believe that everything is measurable, at least in an ideal world but in practice some things are hard or impossible to measure at least for a hobbyist. I'll just leave it at that since I don't really have anything to add to this subject that hasn't been discussed to death in gazillion threads in pretty all audio forums in existence.
Excuse me, Hans, but look at my graphs again. My swept-sine graph shows a maximum deviation of .26dB. Helma's initial curves show about .5dB.
Here's a blow-up of my graph data with cursors to show the devition
My initial comparison was made with the above data and this graph from Helma (note the radical difference in vertical scale!):
Then, we moved on.
What we're seeing here is a difference in measurement methods, and without considering calibration. His multi-tone graph and my swept-sine do exhibit the differences differences I alluded to. His swept-sine shows better accuracy.
I explained when I posted the graphs why the RTA results were inaccurate, and that the swept sine was much better. I also explained with the pink/RTA method was still useful, and under what circumstances.
But all of this splitting .1dB hairs is a supreme exercise in futility. All efforts to match the RIAA curve with the preamp alone are more than somewhat pointless. The goal is to get the entire system, cartridge through preamp output, to hit the ideal curve match, and we aren't looking at that at all. Cartridge/preamp interface and interaction can be on the order of up to several dB. Anyone testing that? The only way is to stimulate the cartridge, typically with a test records, and measure the total system response. And, because test records with RIAA eq sweeps vary significantly between each other, it's best to use a test record with a sweep made without RIAA EQ...and that's the STR100. The expected result is not the RIAA curve, but is known, predictable, and can also be loaded into REW as a .cal file.
Regardless, all that matters is the result, RIAA fit, cartridge to preamp output. Swapping caps and resistors to try to nail the curve looking the preamp alone is literally wasted effort.
I find it amusing and slightly annoying that you think that's what I'm trying to do. If you don't find someone's lifetime of experience worth anything, you don't have to pay attention, now do you? But judging for everyone? Hmm... not sure that's a good thing either.
For my part, I'm only interested in learning and sharing knowledge, so long as it's based on science and fact. Sometimes that means separating fact from non-fact. I've spent my life doing that, pretty much not going to stop now.
Last edited:
That's certainly true, and in fact I've wondered why no one ever made (at least to my knowledge) a cartridge/preamp combination which was designed together to provide accurate RIAA.
I don't think anyone here is oblivious to the fact that cartridges typically exhibit far greater errors in freq response than any half decent preamp. But that doesn't mean playing around with this preamp has been an exercise in futility, I've learnt quite a bit doing it which is sort of the point.
Btw. I tested the preamp with different power supplies and like simulated, supply voltage does affect the low end response and that bump flattens away when used with 12V voltage like it was designed for. But to my ears higher voltage sounds better and it does lower THD. Now I'm a bit unsure which I'm hearing more, the difference in response or the lower THD. It's hard to tell, since the difference is subtle and I'd say more in the high frequencies, but playing around with this thing I've noticed just because you change response in some parts, doesn't mean that's where you perceive the change to be.
There are preamps that are specifically designed for one type of moving-magnet cartridge, such as this one:
Linear Audio | your tech audio resource
I don't remember how accurate the response was.
Linear Audio | your tech audio resource
I don't remember how accurate the response was.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.