Excellent program, you have to buy a license for 79,- Euro which is a fair price, but in a world where so much freeware exists ......
Hans
Yes, S. No one. Not one single person on Earth. Ever.
Do you have "Yet no-one can hear a difference in a blind test." as text in a keyboard shortcut? Because you should. You know, in case someone ever posts on a forum and hasn't heard you inform them of this apparent "law".
Last edited:
Oh wow...it was a sarcastic remark, but as I don't post too often here it might have been hard to realize.
Sarcastic because I find it ironic that people who asks for DBT's to prove Sound difference and who claim that differences of Sound are mainly "imagination" when spotted in a sighted test, do claim that 0.00x% distortion should be achieved. Try to spot a 0.01% vs 0.1% H2 difference in an ABX....
That said, I'm no fan of ABX for spotting small Sound differences anyway.
But we are far OT.
What do you think about ARTA and RIAA testing, btw?
You can't load files and you can't save files.
So what exactly can you do when it is not even possible to load files ?
Hans
Some interesting discussion and some programs suggested seem like they could be what I'm looking for. I initially thought surely it would be easy to find free Linux software for doing the kind of FFT analysis and giving me the kind of graphs I'm after, but it seems I was wrong or haven't found that software yet... however RMAA looks like it does just the things I'm after, and seems to run OK under WINE though I get some occasional crashes. I have the RMAA test signal generated and applied inverse RIAA curve to it with Audacity (I'm assuming it's preset RIAA eq is accurate...) now all that is left is pass that file through the phono preamp and analyze the results in RMAA. Haven't got around to that yet but maybe later today.
There were some interesting points brought up about the importance (or lack of) of RIAA accuracy, since commonly even a good cartridge might be only accurate to within -1/+2dB from 20Hz-20kHz. That's certainly a valid point and for real world listening I agree - as long as the resulting sound from the cartridge+phono preamp combo is to your liking and nothing sticks out the wrong way to your ears, then surely the preamp has good enough accuracy - it's way more accurate than the cartridge anyway.
But if I'm trying to evaluate different phono preamps or different type of designs, I want to make sure they are both flat, otherwise the slight tonal balance differences become a very deciding factor in perceived sound. Say I modify the preamp so that it's flat at 100Hz and then change RIAA network values just a bit so that it's +0.3dB at 100Hz - while the change is not drastic, it's clearly heard and it's not just a case of "this one has slightly more bass than the other one", it also affects the midrange and even how we perceive the high frequencies. I was thinking of making this preamp into a "reference" unit for myself, as far as RIAA accuracy goes. Btw. I'm using MC-cartridges, so capacitive loading etc. are not issues for cartridge behaviour. Whether the eventual SUT+phono preamp combo maintains same accuracy of RIAA response between different cartridges is interesting point though. MC cartridge impedances might vary from just a couple ohms all the way up to 40, I guess the interaction between cart+pre could affect things to a noticeable degree.
Another reason I want these measurements is this idea I've been toying with, to make the RIAA network adjustable on the fly with a couple of switches, for example making it possible to cut or boost the poles +0.3 / flat / -0.3. While that kind of compensation is not enough to make the response of a typical cartidge flat, it would be enough to at least gently tilt the response towards flatness. Also it would be enough adjustment to make the RIAA response mimic that of another preamp, to make a level playing field for other things. Not sure if I'll implement this, it's just a fun idea I had and would go along with the "reference" thing nicely.
Btw. for what is worth the preamp in question is TCC TC-750, this far I've changed the DC-coupling caps at the input and output to good PP-film types, adjusted the RIAA network ever so slightly and got rid of the stock switching power supply.I plan to change the mica caps in the RIAA network to PP as well, but haven't done that yet. I've also messed around with different power supplies since the design is very sensitive to it. This far I've tried 9V batteries, lead gel batteries a linear PSU and better switching supply than the stock one, with some additional filtering. Interesting on that front as well. Maybe it's a case of trying to make a silk purse out of a sow's ear, but it's about learning things first and foremost anyway. Also after the mods it sounds very nice actually, excellent imaging and instrument separation, does some things way better than any op-amp based commercial design I've owned. Haven't owned anything fancy though, but as it is I'd take it over a typical <$300 unit.
There were some interesting points brought up about the importance (or lack of) of RIAA accuracy, since commonly even a good cartridge might be only accurate to within -1/+2dB from 20Hz-20kHz. That's certainly a valid point and for real world listening I agree - as long as the resulting sound from the cartridge+phono preamp combo is to your liking and nothing sticks out the wrong way to your ears, then surely the preamp has good enough accuracy - it's way more accurate than the cartridge anyway.
But if I'm trying to evaluate different phono preamps or different type of designs, I want to make sure they are both flat, otherwise the slight tonal balance differences become a very deciding factor in perceived sound. Say I modify the preamp so that it's flat at 100Hz and then change RIAA network values just a bit so that it's +0.3dB at 100Hz - while the change is not drastic, it's clearly heard and it's not just a case of "this one has slightly more bass than the other one", it also affects the midrange and even how we perceive the high frequencies. I was thinking of making this preamp into a "reference" unit for myself, as far as RIAA accuracy goes. Btw. I'm using MC-cartridges, so capacitive loading etc. are not issues for cartridge behaviour. Whether the eventual SUT+phono preamp combo maintains same accuracy of RIAA response between different cartridges is interesting point though. MC cartridge impedances might vary from just a couple ohms all the way up to 40, I guess the interaction between cart+pre could affect things to a noticeable degree.
Another reason I want these measurements is this idea I've been toying with, to make the RIAA network adjustable on the fly with a couple of switches, for example making it possible to cut or boost the poles +0.3 / flat / -0.3. While that kind of compensation is not enough to make the response of a typical cartidge flat, it would be enough to at least gently tilt the response towards flatness. Also it would be enough adjustment to make the RIAA response mimic that of another preamp, to make a level playing field for other things. Not sure if I'll implement this, it's just a fun idea I had and would go along with the "reference" thing nicely.
Btw. for what is worth the preamp in question is TCC TC-750, this far I've changed the DC-coupling caps at the input and output to good PP-film types, adjusted the RIAA network ever so slightly and got rid of the stock switching power supply.I plan to change the mica caps in the RIAA network to PP as well, but haven't done that yet. I've also messed around with different power supplies since the design is very sensitive to it. This far I've tried 9V batteries, lead gel batteries a linear PSU and better switching supply than the stock one, with some additional filtering. Interesting on that front as well. Maybe it's a case of trying to make a silk purse out of a sow's ear, but it's about learning things first and foremost anyway. Also after the mods it sounds very nice actually, excellent imaging and instrument separation, does some things way better than any op-amp based commercial design I've owned. Haven't owned anything fancy though, but as it is I'd take it over a typical <$300 unit.
Room EQ Wizard does pretty much everything ARTA does, with very few exceptions.
It's free. There's a similar learning curve. I mention it -- yet again -- because I'm an audio professional, and would easily justify purchase of any software, and have spent thousands on software for my business. Yet, I find myself using REW! There are a few things I wish it did, and since it's always a work in progress, those features might show up at some point.
Re: the title of the thread...you can absolutely use Room EQ Wizard to perform high resolution test on your RIAA preamp. Any response issues with your computer sound card are compensated by the setup and calibration routine, so a perfectly flat computer sound card isn't even necessary (I use a good external sound card for noise reasons). You can manually load the RIAA curve as a target, and your resulting response plots will be normalized against that, so no inverse RIAA net is required.
I really don't know why anyone would bother with anything else.
It's free. There's a similar learning curve. I mention it -- yet again -- because I'm an audio professional, and would easily justify purchase of any software, and have spent thousands on software for my business. Yet, I find myself using REW! There are a few things I wish it did, and since it's always a work in progress, those features might show up at some point.
Re: the title of the thread...you can absolutely use Room EQ Wizard to perform high resolution test on your RIAA preamp. Any response issues with your computer sound card are compensated by the setup and calibration routine, so a perfectly flat computer sound card isn't even necessary (I use a good external sound card for noise reasons). You can manually load the RIAA curve as a target, and your resulting response plots will be normalized against that, so no inverse RIAA net is required.
I really don't know why anyone would bother with anything else.
It seems there are two noise camps, camp white and camp pink.
For measuring a speaker, it is much better to use pink noise because FR goes wildly up and down which when using white noise / FFT is not giving any useful information.
That's why speakers are always measured with pink noise, thereby reflecting the perceived sound of our auditory system, which is also logarithmic and perceives pink noise as white noise.
A system like Room EQ Wizzard with RTA is perfectly suited for this job.
An electronic device like a Riaa preamp does not have the wild fluctuations that loudspeakers have and can very well be evaluated with white noise with the advantage that much more energy in used at the high end of the spectrum, keeping a sound distance from signal to the amp's own noise.
Hans
For measuring a speaker, it is much better to use pink noise because FR goes wildly up and down which when using white noise / FFT is not giving any useful information.
That's why speakers are always measured with pink noise, thereby reflecting the perceived sound of our auditory system, which is also logarithmic and perceives pink noise as white noise.
A system like Room EQ Wizzard with RTA is perfectly suited for this job.
An electronic device like a Riaa preamp does not have the wild fluctuations that loudspeakers have and can very well be evaluated with white noise with the advantage that much more energy in used at the high end of the spectrum, keeping a sound distance from signal to the amp's own noise.
Hans
You can also use HOLMImpulse, and I have done so. If you're clever you can load in the RIAA curve to compare. Does pretty much what REW does, without the need for a reference channel. Both will do a nice job, just pick the one you like.
My impression was that pink noise is equal energy per octave so before computers real analog filters had to be used and fractional octave analysis is the only thing that makes sense (which then outputs as flat with frequency).
You can use Audacity for that matter, it has arbitrary IIR filters built in so you can model RIAA or IRIAA in whichever order you choose.
I was looking at the exact same thing about 2 months back and found this.
Daqarta - RIAA Phono Equalization Testing:
I have used this. It was a little frustrating going through the learning curve but was finally able to get useful measurements to compare different phono preamps. What I found was that though I was hearing considerable differences in frequency responses the measurements were closer than I would have guessed.
Daqarta - RIAA Phono Equalization Testing:
I have used this. It was a little frustrating going through the learning curve but was finally able to get useful measurements to compare different phono preamps. What I found was that though I was hearing considerable differences in frequency responses the measurements were closer than I would have guessed.
Somehow I have the strong impression that we are not understanding each other.
Pink noise IS equal energy per octave, and before computers could do FFT's, fractional analogue (or digital) octave filters were giving a flat FR response.
But nowadays although we have computers doing FFT's of any length, pink noise is still the one to use for measuring speakers with (partial) octave filters because results are more in line with what we hear, i.e. also equal energy per octave and not per Hz. That's why we also perceive Pink noise as a flat spectrum like those fractional octave filters.
It is no coincidence that the program that Jaddie mentions is called Room EQ Wizzard.
Hans
Last edited:
Adding to my previous posting, a bit more detail what benefit White noise can bring.
Amplifiers generate noise. Above a certain point in frequency where 1/f noise stops , this knee point depending on technology used, noise is white, meaning constant energy per Hz.
Pink noise falling with 10dB / decade has 30dB less energy per Hz at 20 Khz as it has at 20 Hz.
That means that the distance from the Pink noise signal to the amp's noise is 30dB less at 20 Khz or S/N is 30 dB less at that point and both can interfere when coming too close resulting in a false reading.
When using white noise as a signal, the S/N remains constant above the knee point making interference with amp noise much less likely to happen.
Hans
Amplifiers generate noise. Above a certain point in frequency where 1/f noise stops , this knee point depending on technology used, noise is white, meaning constant energy per Hz.
Pink noise falling with 10dB / decade has 30dB less energy per Hz at 20 Khz as it has at 20 Hz.
That means that the distance from the Pink noise signal to the amp's noise is 30dB less at 20 Khz or S/N is 30 dB less at that point and both can interfere when coming too close resulting in a false reading.
When using white noise as a signal, the S/N remains constant above the knee point making interference with amp noise much less likely to happen.
Hans
Last edited:
No, that's not why pink is used for speaker and acoustic measurements. FFT data can be smoothed and processed, and measurement bandwidth adjusted to provide reasonable results. Pink noise/RTA hasn't been used for measuring speakers for quite some time.
Pink noise was developed as a test signal that analog real-time analyzers with log bandwidth (octave or fractional octave) filters would read as "flat". Noise used as a test source for acoustic measurements was an important improvement over using single tone or swept tone signals because it avoids excitation of resonances to some degree, and more importantly smooths single-frequency acoustic problems caused by resonances and reflections. That's it, that's the only reason pink noise even exists at all. And no, pink noise and RTA is NOT what's used for acoustic measurements today, other than for quick checks and basic EQ setting.
This seems to imply REW is just an RTA. Nothing could be less accurate! The RTA function is but one of many, including the primary swept-sine/FFT measurement system that provides full energy/time/frequency data. Depending on how the FFT data is windowed, it can be used to display frequency response, FR vs time, THD vs frequency, THD vs time, ETC (energy/time curve), pretty much anything you'd ever need, and some you've never thought of.
An RIAA preamp many not have wild response fluctuations, but that's not what testing with noise is all about anyway. Pink noise testing is fast, accurate and easy to interpret. With high resolution RTA filters, like 1/24th octave (never practical in the analog world), you get all the resolution you need to respond to these gentle variations. With a simple 10 second or so average applied, you get highly accurate and smoothed FR data quickly and easily. By pre-loading the RIAA curve as a "target", you don't need an r-RIAA network, you get a normalized display of deviation from that curve. A flat curve would then be one with zero deviation from perfect RIAA.
With REW, you can use an RIAA target curve with either the RTA function or the swept-sine/fft function. You can set up the test frequency range arbitrarily, or go 20-20K. Typically 4 sweeps are done and integrated. You can then choose to display FR, THD vs frequency, phase vs frequency, etc. The test takes seconds to perform. Multiple tests can be displayed together, color coded, so you can easily compare before/after as you make changes. If you're concerned about residual noise affecting an RTA FR result, that's your solution. You can also run an FFT of the residual noise without running the generator and see where the signal and noise are in relation to each other on a frequency plot.
White noise is also available as a "generator" choice, but frankly, in all the years of testing audio having white as an option, I've never used it because the other tools are so much better, faster, and easy to interpret. Using white with a spectrum display is fine, but then requires smoothing and measurement bandwidth tweaking. It's just busy-work, which gets in the way of getting accurate answers and responding to them.
It looks like HOLMImpulse may do much of this too, I haven't tried it. I like REW because it's kept up to date with respect to OS updates, and it's cross-platform, so I run it on my Windows10 pc and my MacBook Pro. Works exactly the same on both.
I feel like an REW salesman here! I'm not, I have nothing to gain or loose, and other than a few email exchanges, I don't know the developer. I will say, when I send him a feature request, he gets and reads my email and considers it for a future version. I kind of like that. And I thought it might be interesting to hear what someone who makes a portion of his living measuring stuff likes to use.
I find the name an unfortunate choice. Room EQ Wizard was initially developed for acoustic testing and EQ, that's true. But the capabilities that came along make it far more useful that just for that. I actually use it more for electronics than for room EQ. Though I have to say, with it integrated with certain DSP devices, it's pretty cool to enter a room target curve, run a few tests in a few mic positions, and have REQ calculate and upload the filters! Wow, really fast!
Using analog test gear for decades, primarily Tek and HP, I have Champaign tastes when it comes to testing systems, and lean toward Audio Precisions and the likes. It's out of budget for most of my stuff, though. I like the Quant Assylum stuff, but find the application to acoustic work limited, and unfortunately married to their hardware. REW hits an interesting balance of cost and functions that it's hard to beat until you hit the d-Scope stuff and up to AP.
I managed to get the job done running RightMark Audio Analyzer under WINE, so big thanks to whoever it was that brought that software to my attention.
The good news is my LTspice model seems quite accurate from 200Hz-20kHz and the measured graph looked like it's supposed to, within +/- 0.1dB. From there on it starts to gently rise until reaching a peak of +0.5dB at 40Hz, after which it falls to -0.2dB at 20Hz - basically this means the 2SC945 transistor model I have is not really a good fit for the STC945 transistors the preamp has. No big deal and I'm not sure that bump is even worth fixing, though it would take only a few minutes to lower the value of the resistor in the RIAA network which should fix that. I might even have a suitable resistor ready. The original is 348k, I previously upped it to 416k because simulation showed that to be better value, but in listening I noticed a bump in the low bass, so went back to original.
I had a bit of a noise issue which I wasn't able to fix, which creates some artefacts at 50Hz and harmonics, seems ground related but adding a ground wire from preamp to computer only made it worse. Anyway I tested two preamps and it doesn't seem to affect the measurements other than making the multitone frequency response graph look a bit weird with sharp triangular dips at the noise frequencies.
EDIT: oh and thumbs up for whoever manufactures the TC-750, either I got lucky or they've got some QC going - channel balance was within 0.05dB through the bass and midrange, but then in the treble it got to around 0.15dB. Still, I find this to be very good and even from an unit which costs about 40 euros retail.
The good news is my LTspice model seems quite accurate from 200Hz-20kHz and the measured graph looked like it's supposed to, within +/- 0.1dB. From there on it starts to gently rise until reaching a peak of +0.5dB at 40Hz, after which it falls to -0.2dB at 20Hz - basically this means the 2SC945 transistor model I have is not really a good fit for the STC945 transistors the preamp has. No big deal and I'm not sure that bump is even worth fixing, though it would take only a few minutes to lower the value of the resistor in the RIAA network which should fix that. I might even have a suitable resistor ready. The original is 348k, I previously upped it to 416k because simulation showed that to be better value, but in listening I noticed a bump in the low bass, so went back to original.
I had a bit of a noise issue which I wasn't able to fix, which creates some artefacts at 50Hz and harmonics, seems ground related but adding a ground wire from preamp to computer only made it worse. Anyway I tested two preamps and it doesn't seem to affect the measurements other than making the multitone frequency response graph look a bit weird with sharp triangular dips at the noise frequencies.
EDIT: oh and thumbs up for whoever manufactures the TC-750, either I got lucky or they've got some QC going - channel balance was within 0.05dB through the bass and midrange, but then in the treble it got to around 0.15dB. Still, I find this to be very good and even from an unit which costs about 40 euros retail.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.