Thank you. You've been making quite a few videos lately and I've been enjoying them, so hopefully you've enjoyed making them. A belated Happy New Year to you.
Thanks for watching! I was laden with work last year and family issues prior to that. I hope to be "freer" this year and hope to make more videos, but sadly, I know more family issues may be on the horizon.
I've been looking at the manufacturer's specs as well as yours and this seems like excellent value for money at $140/£110! In reality, I bet this holds up well against a certain £25K preamp that's been making the news lately.
I bet it won't. Not even close. Yes, it's a scandal asking 25k, but you're talking a very niche market where it's hard to find I really good phono preamp. This Fosi is nice for beginners. But I won't recommend it. I bought a Ayima T3 Pro for $80 and it's a much better deal, with upgraded opamps it can rival any us made phono up to $1k. This Fosi will not be used for (m)any MC cartridges I bet. No one will buy this $120 phonostage for a $500+ MC on a $1k+ turntable. Even though it works.
The OPA1656 is a ultra low nose device from the Burr Brown stables and has FET inputs the OPA1612 is nearly identical but has BIpolar inputs. Both are dual so take your pick ! but I think I understand why the 1612 was used as BIpolar inputs marginally less noisy at such low signal levels. I have used the 1656 in a headphone amplifier and sounds awesome to me (but it would !) I think I agree with you that the venerable NE5532 is just used as a buffer and the socket allows for the people who wanna play musical opamp's (pun intended) personally I wouldn't bother. Great work John.....cheers. EDIT:- that other chip you showed is a dual bipolar transistor package NPN and PNP side by side the two pins that are connected are the collector pins. Happy New Year !
Aren't distortion figures like 0.1% and 01% at those very low frequencies kinda academic? Seems practically 'margin of error' of the entire chain at those frequencies and voltages.
I wouldn't say so. There's no reason why THD would go up "just because"; there's always a source for those overtones. You can however profile your *measurement margin of error* by running a few runs and averaging it. Then take the median of the top half and the median of the bottom half of numbers, and you have an approximate distribution window to aim for.
Why didn't you just measure the capacitance at the input jacks? Or on the caps in circuit? (you might have to switch it on to get a reading, also due to the relays). Could you still do that and report back on that? It would be greatly appreciated. I don't know if you have this preamp still in your posession? There's really no real standard for input capacitance. Some have as low as 47pF (the Hagerman bugle even has none, but that's very rare), and some have as high as 220pF or even over 300pF. And a 200pF difference is absolutely audible. For the rest I think this is a good review. Certainly the best review I've seen so far, not the usual youtube influencer moronic advertizing. Subsonic filters often do just as much good as wrong. Most subsonic filters are often just too steep, reach too much in the audible realm etc. If it does -4dB at 15Hz or so, it's usually already plenty to prevent large woofer cone exursions at warps etc. These are typically sub 5Hz or so. So not having a subsonic filter is far better than a badly implemented one imho. Save a capacitance loading choice and not even any specifications given about it, it looks very decent for a very good price. Looks like they used good quality components. Although I would've rather seen a socketed 1st opamp on the input stage. For instance if that opa1612 could be swapped for a opa1656, you migth get away with removing that input capacitor all together. The opa1656 is fet based and has a very low input bias current, so hardly any current flows back into the cartridge. I speculate a bit that cap is also partly there to minimize that, apart from cartridge loading and the reason virtually no phono preamp is made without loading cap. Even though lots of popular cartridges would benefit of no capacitive loading where the interlink cable already provides enough capacitance. Most cartridges that have a Microline stylus in combination with a laminated coil core for example. They tend to get extremely much HF detail out of the grooves but probably therefore might be interpreted as 'bright' sometimes. Another way to compensate for that is to lower the resistance loading (from say 47k to 35k), but phono preamps never have that option sadly...
I tried to measure across the cap (same as input) but the meter refused the measurement probably because of the loading resistor. Other ways to measure but didn't bother. Wouldn't be too hard to remove them and add your own. The Rod Elliot phono preamp kit has none. I'd have to A B test the preamps to hear the difference. Both sound good but the Elliot amp has a bit too much gain for my system.
Nice review. Looks like it was responsibly designed but I do wonder about the final output stage op-amp. I've seen the first stage have the select-able gain for MM/MC modes followed by the RIAA stage so the final stage is just more stuff and in audio, less is usually more better.
They wanted to buffer the filter stage from the output. Looks like the final stage might have switchable gain. It is better to spread gain across amplifiers especially since it has such high gain. Good engineering to me.
@@JohnAudioTech your hospital must be nicer than our hospital is. Although I was just there last week and I guess it wasn't so bad really. At least not in the outpatient waiting room where I was. I was in preop too for a bit. It's the room wings where it gets bad.
One of the issues with actually measuring total harmonic distortion and similar distortion measurements is that it assumes that you divide the frequency spectrum you are measuring into infinitely small bandwidth bins (and so you have an infinite number of measurement bins) and the bins start at 0hz and extend to infinity hz. To measure THD you inject a reference frequency of a known amplitude and measure the amplitude of its bin and then collect and measure the amplitude of all the bins which contain an (integer) harmonic of that reference frequency. (So no bins are examined which have a lower frequency than the reference frequency). Every bin which does not contain the reference frequency or a integer harmonic of it is considered a noise bin (including those bins who's frequencies is bellow the reference frequency). This is of course impossible in real measurement gear. Real measurement gear has a bandwidth it can measure over, and harmonics higher than that are ignored, the individual bins are often 20-, 50-hz or even wider and there is a limit to the number of bins due to memory or processing limitations. Another real world problem is that the amplitude of the reference frequency matters. If the amplitude is too low then the signal to noise ratio will be low and ANY noise in a bin which contains a harmonic is considered as harmonic distortion and NOT noise. So as the signal to noise ratio falls the harmonic distortion measured appears to rise when the real distortion is not rising and may actually be falling. The reality is ONLY noise is rising. Conversely if the reference frequency is too high an amplitude it can overload the device being measured resulting in far more distortion than would be encountered it normal use, and because of the RIAA curve the maximum input level to a photo stage is far less at low frequencies than higher frequencies since the gain is more at lower frequencies. On some devices like the keithley 2015THD for example, the width of a bin, number of bins, maximum number of harmonics and bandwidth ALL vary according the the reference frequency. None of this is clearly or fully documented in the user or programming manual and it took me lot of measuring and experimentation to work out how the device was *actually* measuring. For example at low reference frequencies e.g. 40hz the keithley bin size is so wide each harmonic falls into the next available bin. So all bins contain harmonics and therefore no bins are considered to contain noise at all! In other words the bin width is so wide i.e. so low resolution ALL noise in the entire measurement bandwidth is considered harmonic distortion! - this of course inflates the very low reference frequency distortion figures. If you are measuring at 22hz the bin width needs to be 11hz ,5.5hz or better 2.75hz to avoid measuring *all* the noise as distortion. This would require 10s of thousands of bins. In the case of a phono stage, there is far more gain at lower frequencies and so there will be more noise and distortion and more noise will be interpreted as distortion and depending upon the bin size, if the bins are too wide the distortion analyser will take that extra noise and interpret it as extra distortion too. In addition unless the number of harmonics is kept the same, the lower the reference frequency the more measurement bins will contain a harmonic and therefore automatically the higher the measured distortion. For example the keithley maximum bandwidth is ~50Khz so if your reference frequency is 20Khz you will only get one bin with distortion in it. At 40hz all 1000+ bins are distortion, at 80hz half the bins are harmonics. THD + noise is often more accurate than just THD or just noise because it does not matter if harmonics and noise are conflated because EVERY bin except the bin containing the reference frequency is treated equally and included in THD + noise, reducing the effect of bin width and including all bins bellow the reference frequency in the calculation and so advantaging high reference frequencies so much. So to cut a long story short, you can get some "interesting" measurement results when you don't know the compromises the measurement tool is making (and by definition they must all break every assumption the THD maths says, i.e. each bin must have frequency width, there will be a limited number of bins and a limited bandwidth) when it measures and calculates. It would be nice if the user interface of measurement devices actually said the bin width, number of bins and device bandwidth.
Awesome thanks ! I have often wondered about those Keithley meters, but I don't know what i'm talking about (I know a bit more now !) I see people plug their amps in and it displays 0.005% and that's it, I think well that's a lot of trust in the instrument. How do you know if its even real ? I would imagine the QA kit that John is dribbling over (just kidding John if you read this !) is leaps and bounds ahead of the old Keithley ? or am I wrong ? Anyhow Happy New Year.
Thanks for the reply. As I understand it, the bin width = the sample rate / FFT size. In the video I think I used 192K rate and 256K size for bin widths of .75 Hz. The QA403 sample rate can be set down to 48K and a FFT size of 1024K for .04Hz bin widths. I used a 15dB attenuator on the output of the QA403 so I could use a larger output signal from it so further from its noise floor. Using a much higher value attenuator would be better given the preamp's gain at low frequencies. What I need to do is baseline the QA403 by looping back its output to its input and running numerous tests at various frequencies and amplitudes to determine the performance of the instrument. Some opamps have datasheet stated distortion levels below what it can measure.
@@JohnAudioTech the problem with bin width = sample rate / fft size, is that ideally want each harmonic to be in the middle of a bin so the bin width should be an integer divisor of the reference frequency. For example bin width = reference frequency / 4 ensures that every 4th bin is a harmonic and the harmonic is always in the same place within the bin i.e. middle of the bin. If the bin width is not done this way a harmonic can fall right on the edge between two bins and either not get counted in either or get counted in both bins. Making sure the divisor is > 1 also means that its impossible for two harmonics to be in the same bin. With sample rate / fft size if you go high sample rate and low fft size and the reference frequency is really low then the bin width can be so large that it will contain two or more harmonics. You also want/need all the bins to be the same size so the measurement bandwidth is also partly decided by the frequency of the right most edge of the last bin and the lowest frequency by the left edge of the first bin. So the actual measurement bandwidth also varies depending upon the bin width which depends on the reference frequency. Naturally the narrower the bin width the less of an effect this has. There is also an issue with the number of bins. Imagine there are 53 bins between harmonics. If you run out of bins just before the next harmonic then that harmonic will not be counted towards thd or thd + noise, but all the bins between the last harmonic that did get counted and just before the harmonic that did not will count as noise... now imagine you have one extra bin. suddenly the extra harmonic gets counted in the THD and THD + noise. now imagine you have 52 less bins. Now the last harmonic measured is also the last bin so all the noise past that is ignored so in this case the thd has not changed but the thd + noise has. So its not clear if the last bin should hold a harmonic, be mid way between two harmonics or just before or just after a harmonic. All these possibilities will effect the measured results and effect the thd + noise and thd differently. another interaction is between bin width and the absolute bandwidth of the measuring device. It is possible for the first or last bin to not contain the entire bin width of frequencies because the absolute bandwidth of the device falls within that bin. So the last bin may only be partly filled because the higher frequencies it should contain are outside the bandwidth of the measuring device and also the first bin because the lowest frequencies are outside the bandwidth. This can be an issue if a harmonic falls inside the last bin but just outside the measurement bandwidth. The bin will have an artificially low value and so not contribute as much to the distortion measurement as it should. Whereas a reference frequency of just one hz lower would move the harmonic to within the measurement bandwidth and get counted in the distortion figure. So in general to avoid this you throw away the first and last bins since they are only partially filled or reduce the number of bins so no bin straggles the measurement bandwidth of the device. In other words understanding how your device actually measures and understanding how a very small change in measurement setup can disproportionally effect the result requires a fair bit of experimentation (or really good documentation). It is easily possible to buy or build a passive reverse RIAA equalizer network and add it to your attenuation resistors. That way that you can much more accurately simulate what a phono cartridge would actually output including the low frequency attenuation. That would give you more realistic noise and THD measurements. The Keithley also has configurable low and high cut filters, only configurable in software via rs232 or gpib. However it allows you to set a cut off frequency inside a bin rather than just between bins. The keithley also does not display or provide a way to program it to get either the current bin width OR the starting or ending frequency of a bin. This also causes issues because the manuals don't specify the behaviour if a cut off filter is within a bin. Does it throw away the entire bin? keep the entire bin or remove the cut off frequencies before the FFT so they are not present in the bin... After some experimentation it turns out if a cut off filter is within a bin the entire contents of the bin is kept. It would really be nice if the manufacturers of distortion meters actually provided accurate and comprehensive documentation for the behaviour and interactions between measurement options ...
Thank you. You've been making quite a few videos lately and I've been enjoying them, so hopefully you've enjoyed making them. A belated Happy New Year to you.
Thanks for watching! I was laden with work last year and family issues prior to that. I hope to be "freer" this year and hope to make more videos, but sadly, I know more family issues may be on the horizon.
I've been looking at the manufacturer's specs as well as yours and this seems like excellent value for money at $140/£110! In reality, I bet this holds up well against a certain £25K preamp that's been making the news lately.
Indeed !
Build quality can't be worse.
I bet it won't. Not even close. Yes, it's a scandal asking 25k, but you're talking a very niche market where it's hard to find I really good phono preamp. This Fosi is nice for beginners. But I won't recommend it. I bought a Ayima T3 Pro for $80 and it's a much better deal, with upgraded opamps it can rival any us made phono up to $1k.
This Fosi will not be used for (m)any MC cartridges I bet. No one will buy this $120 phonostage for a $500+ MC on a $1k+ turntable. Even though it works.
I wonder how it compares to that 25000 quids Tom Evans amplifier 😂😂😂
If I were a billionaire, I'd still buy this Fosi preamp. Heck, I'd probably buy another one and mail it to Tom Evans.
@@ohmbug10 LOL!
Owch !kick a man when he's down LMFAO !
Copernicus you beat me to it.😂
Copyright strike! 😂😂
The OPA1656 is a ultra low nose device from the Burr Brown stables and has FET inputs the OPA1612 is nearly identical but has BIpolar inputs. Both are dual so take your pick ! but I think I understand why the 1612 was used as BIpolar inputs marginally less noisy at such low signal levels. I have used the 1656 in a headphone amplifier and sounds awesome to me (but it would !) I think I agree with you that the venerable NE5532 is just used as a buffer and the socket allows for the people who wanna play musical opamp's (pun intended) personally I wouldn't bother. Great work John.....cheers. EDIT:- that other chip you showed is a dual bipolar transistor package NPN and PNP side by side the two pins that are connected are the collector pins. Happy New Year !
I'll have to try the 1612 since popcorn noise seems to be endemic with the 4562.
Nice review. Haven't heard that song for ages. Nice.
The output impedance is specified as10K ohm.
If correct, it's far too high.
I'll have to check, but I doubt that specification is correct.
Home Audio Expert unit for The Home Audio Expert: JAT ! ! !
Aren't distortion figures like 0.1% and 01% at those very low frequencies kinda academic? Seems practically 'margin of error' of the entire chain at those frequencies and voltages.
I wouldn't say so. There's no reason why THD would go up "just because"; there's always a source for those overtones.
You can however profile your *measurement margin of error* by running a few runs and averaging it. Then take the median of the top half and the median of the bottom half of numbers, and you have an approximate distribution window to aim for.
Why didn't you just measure the capacitance at the input jacks? Or on the caps in circuit? (you might have to switch it on to get a reading, also due to the relays).
Could you still do that and report back on that? It would be greatly appreciated. I don't know if you have this preamp still in your posession?
There's really no real standard for input capacitance. Some have as low as 47pF (the Hagerman bugle even has none, but that's very rare), and some have as high as 220pF or even over 300pF. And a 200pF difference is absolutely audible.
For the rest I think this is a good review. Certainly the best review I've seen so far, not the usual youtube influencer moronic advertizing.
Subsonic filters often do just as much good as wrong. Most subsonic filters are often just too steep, reach too much in the audible realm etc. If it does -4dB at 15Hz or so, it's usually already plenty to prevent large woofer cone exursions at warps etc. These are typically sub 5Hz or so. So not having a subsonic filter is far better than a badly implemented one imho.
Save a capacitance loading choice and not even any specifications given about it, it looks very decent for a very good price. Looks like they used good quality components. Although I would've rather seen a socketed 1st opamp on the input stage. For instance if that opa1612 could be swapped for a opa1656, you migth get away with removing that input capacitor all together. The opa1656 is fet based and has a very low input bias current, so hardly any current flows back into the cartridge. I speculate a bit that cap is also partly there to minimize that, apart from cartridge loading and the reason virtually no phono preamp is made without loading cap.
Even though lots of popular cartridges would benefit of no capacitive loading where the interlink cable already provides enough capacitance. Most cartridges that have a Microline stylus in combination with a laminated coil core for example. They tend to get extremely much HF detail out of the grooves but probably therefore might be interpreted as 'bright' sometimes. Another way to compensate for that is to lower the resistance loading (from say 47k to 35k), but phono preamps never have that option sadly...
I tried to measure across the cap (same as input) but the meter refused the measurement probably because of the loading resistor. Other ways to measure but didn't bother. Wouldn't be too hard to remove them and add your own. The Rod Elliot phono preamp kit has none. I'd have to A B test the preamps to hear the difference. Both sound good but the Elliot amp has a bit too much gain for my system.
A lot smaller than Tom Evan's Mastergroove preamp
Nice review. Looks like it was responsibly designed but I do wonder about the final output stage op-amp. I've seen the first stage have the select-able gain for MM/MC modes followed by the RIAA stage so the final stage is just more stuff and in audio, less is usually more better.
They wanted to buffer the filter stage from the output. Looks like the final stage might have switchable gain. It is better to spread gain across amplifiers especially since it has such high gain. Good engineering to me.
@@JohnAudioTech Judging by the socket, seems more like the flavor of the month. Odd that they chose such an old op-amp for the output.
It sounded warm to me.
I noticed a dissonant harmonic above 16Kh.😂
My amplifiers are clinical. They smell like a hospital.
@@JohnAudioTech your hospital must be nicer than our hospital is. Although I was just there last week and I guess it wasn't so bad really. At least not in the outpatient waiting room where I was. I was in preop too for a bit. It's the room wings where it gets bad.
The soundstage was so deep, lush and engaging. I felt I was sitting next to Jimmy Page.
@@Margarinetaylorgrease soundstage height was not that impressive either 🙂
One of the issues with actually measuring total harmonic distortion and similar distortion measurements is that it assumes that you divide the frequency spectrum you are measuring into infinitely small bandwidth bins (and so you have an infinite number of measurement bins) and the bins start at 0hz and extend to infinity hz. To measure THD you inject a reference frequency of a known amplitude and measure the amplitude of its bin and then collect and measure the amplitude of all the bins which contain an (integer) harmonic of that reference frequency. (So no bins are examined which have a lower frequency than the reference frequency). Every bin which does not contain the reference frequency or a integer harmonic of it is considered a noise bin (including those bins who's frequencies is bellow the reference frequency).
This is of course impossible in real measurement gear. Real measurement gear has a bandwidth it can measure over, and harmonics higher than that are ignored, the individual bins are often 20-, 50-hz or even wider and there is a limit to the number of bins due to memory or processing limitations. Another real world problem is that the amplitude of the reference frequency matters. If the amplitude is too low then the signal to noise ratio will be low and ANY noise in a bin which contains a harmonic is considered as harmonic distortion and NOT noise. So as the signal to noise ratio falls the harmonic distortion measured appears to rise when the real distortion is not rising and may actually be falling. The reality is ONLY noise is rising. Conversely if the reference frequency is too high an amplitude it can overload the device being measured resulting in far more distortion than would be encountered it normal use, and because of the RIAA curve the maximum input level to a photo stage is far less at low frequencies than higher frequencies since the gain is more at lower frequencies.
On some devices like the keithley 2015THD for example, the width of a bin, number of bins, maximum number of harmonics and bandwidth ALL vary according the the reference frequency. None of this is clearly or fully documented in the user or programming manual and it took me lot of measuring and experimentation to work out how the device was *actually* measuring. For example at low reference frequencies e.g. 40hz the keithley bin size is so wide each harmonic falls into the next available bin. So all bins contain harmonics and therefore no bins are considered to contain noise at all! In other words the bin width is so wide i.e. so low resolution ALL noise in the entire measurement bandwidth is considered harmonic distortion! - this of course inflates the very low reference frequency distortion figures.
If you are measuring at 22hz the bin width needs to be 11hz ,5.5hz or better 2.75hz to avoid measuring *all* the noise as distortion. This would require 10s of thousands of bins.
In the case of a phono stage, there is far more gain at lower frequencies and so there will be more noise and distortion and more noise will be interpreted as distortion and depending upon the bin size, if the bins are too wide the distortion analyser will take that extra noise and interpret it as extra distortion too. In addition unless the number of harmonics is kept the same, the lower the reference frequency the more measurement bins will contain a harmonic and therefore automatically the higher the measured distortion. For example the keithley maximum bandwidth is ~50Khz so if your reference frequency is 20Khz you will only get one bin with distortion in it. At 40hz all 1000+ bins are distortion, at 80hz half the bins are harmonics. THD + noise is often more accurate than just THD or just noise because it does not matter if harmonics and noise are conflated because EVERY bin except the bin containing the reference frequency is treated equally and included in THD + noise, reducing the effect of bin width and including all bins bellow the reference frequency in the calculation and so advantaging high reference frequencies so much.
So to cut a long story short, you can get some "interesting" measurement results when you don't know the compromises the measurement tool is making (and by definition they must all break every assumption the THD maths says, i.e. each bin must have frequency width, there will be a limited number of bins and a limited bandwidth) when it measures and calculates. It would be nice if the user interface of measurement devices actually said the bin width, number of bins and device bandwidth.
Awesome thanks ! I have often wondered about those Keithley meters, but I don't know what i'm talking about (I know a bit more now !) I see people plug their amps in and it displays 0.005% and that's it, I think well that's a lot of trust in the instrument. How do you know if its even real ? I would imagine the QA kit that John is dribbling over (just kidding John if you read this !) is leaps and bounds ahead of the old Keithley ? or am I wrong ? Anyhow Happy New Year.
Thanks for the reply. As I understand it, the bin width = the sample rate / FFT size. In the video I think I used 192K rate and 256K size for bin widths of .75 Hz. The QA403 sample rate can be set down to 48K and a FFT size of 1024K for .04Hz bin widths.
I used a 15dB attenuator on the output of the QA403 so I could use a larger output signal from it so further from its noise floor. Using a much higher value attenuator would be better given the preamp's gain at low frequencies.
What I need to do is baseline the QA403 by looping back its output to its input and running numerous tests at various frequencies and amplitudes to determine the performance of the instrument. Some opamps have datasheet stated distortion levels below what it can measure.
@@JohnAudioTech the problem with bin width = sample rate / fft size, is that ideally want each harmonic to be in the middle of a bin so the bin width should be an integer divisor of the reference frequency. For example bin width = reference frequency / 4 ensures that every 4th bin is a harmonic and the harmonic is always in the same place within the bin i.e. middle of the bin. If the bin width is not done this way a harmonic can fall right on the edge between two bins and either not get counted in either or get counted in both bins. Making sure the divisor is > 1 also means that its impossible for two harmonics to be in the same bin. With sample rate / fft size if you go high sample rate and low fft size and the reference frequency is really low then the bin width can be so large that it will contain two or more harmonics. You also want/need all the bins to be the same size so the measurement bandwidth is also partly decided by the frequency of the right most edge of the last bin and the lowest frequency by the left edge of the first bin. So the actual measurement bandwidth also varies depending upon the bin width which depends on the reference frequency. Naturally the narrower the bin width the less of an effect this has.
There is also an issue with the number of bins. Imagine there are 53 bins between harmonics. If you run out of bins just before the next harmonic then that harmonic will not be counted towards thd or thd + noise, but all the bins between the last harmonic that did get counted and just before the harmonic that did not will count as noise... now imagine you have one extra bin. suddenly the extra harmonic gets counted in the THD and THD + noise. now imagine you have 52 less bins. Now the last harmonic measured is also the last bin so all the noise past that is ignored so in this case the thd has not changed but the thd + noise has. So its not clear if the last bin should hold a harmonic, be mid way between two harmonics or just before or just after a harmonic. All these possibilities will effect the measured results and effect the thd + noise and thd differently.
another interaction is between bin width and the absolute bandwidth of the measuring device. It is possible for the first or last bin to not contain the entire bin width of frequencies because the absolute bandwidth of the device falls within that bin. So the last bin may only be partly filled because the higher frequencies it should contain are outside the bandwidth of the measuring device and also the first bin because the lowest frequencies are outside the bandwidth. This can be an issue if a harmonic falls inside the last bin but just outside the measurement bandwidth. The bin will have an artificially low value and so not contribute as much to the distortion measurement as it should. Whereas a reference frequency of just one hz lower would move the harmonic to within the measurement bandwidth and get counted in the distortion figure. So in general to avoid this you throw away the first and last bins since they are only partially filled or reduce the number of bins so no bin straggles the measurement bandwidth of the device.
In other words understanding how your device actually measures and understanding how a very small change in measurement setup can disproportionally effect the result requires a fair bit of experimentation (or really good documentation).
It is easily possible to buy or build a passive reverse RIAA equalizer network and add it to your attenuation resistors. That way that you can much more accurately simulate what a phono cartridge would actually output including the low frequency attenuation. That would give you more realistic noise and THD measurements.
The Keithley also has configurable low and high cut filters, only configurable in software via rs232 or gpib. However it allows you to set a cut off frequency inside a bin rather than just between bins. The keithley also does not display or provide a way to program it to get either the current bin width OR the starting or ending frequency of a bin. This also causes issues because the manuals don't specify the behaviour if a cut off filter is within a bin. Does it throw away the entire bin? keep the entire bin or remove the cut off frequencies before the FFT so they are not present in the bin... After some experimentation it turns out if a cut off filter is within a bin the entire contents of the bin is kept. It would really be nice if the manufacturers of distortion meters actually provided accurate and comprehensive documentation for the behaviour and interactions between measurement options ...