Sure very true.
For now it's just the quick and easy way. RePhase is a brilliant tool. And since the cost is 0 why not? It's also what I have been using, so kind of a natural thing to do for now.
I have no preference whatsoever of FIR vs. IIR. They are just tools to get the job done.
I originally bought the miniDSP as a FIR tool (to make my active speakers linear phase), but to be honest I could hear 0 difference between minimum and linear phase speakers.
So out with the miniDSP, just another piece of equipment not doing anything usefull, until now
For now it's just the quick and easy way. RePhase is a brilliant tool. And since the cost is 0 why not? It's also what I have been using, so kind of a natural thing to do for now.
I have no preference whatsoever of FIR vs. IIR. They are just tools to get the job done.
I originally bought the miniDSP as a FIR tool (to make my active speakers linear phase), but to be honest I could hear 0 difference between minimum and linear phase speakers.
So out with the miniDSP, just another piece of equipment not doing anything usefull, until now
It's a SU from the late seventies, I guess the SU-9600, but I'm not sure. I'll have a look up at the HiFiEngine... nope.
Last edited:
I was just pointing out that you have IIR input PEQs that do the job exactly and save the FIR for what it's good at.
True but the RIAA filter itself is an infinite impulse response filter by definition and has significant response amplitude upto and beyond 50ms - given its 2 poles and a single zero and low-Q its clearly a match for IIR implementation.
A rumble filter could be implemented efficiently as FIR as you can decimate the rate to get a suitable low-pass filter at a much lower sample rate, interpolate that up again and subtract from the original signal. Basically once you have a stopband that spans most of the frequency space you can make FIR much more efficient by multi-rate techniques. The RIAA doesn't really have a stopband, IIR will be as good as anything (particularly as no high-Q poles/zeros)
Hi billshurv, Mark Tillotson
O.K. I'm convinced, all good arguments. I'll take the IIR route.
I've seen plenty of Bi-quad calculators around, actually already downloaded one from the miniDSP forum that has RIAA, HPF, LPF and so on (All-digital-coefs v1.2.xls).
Not sure how accurate that is, but I'll find out.
Don't know if I can mix and match IIR and FIR with the openDRC-DI, and if it is possible how the use of biquads affects the number of FIR taps available.
Ahh, more (re)searching to do, just first getting the kids to bed and cleaning up ....
O.K. I'm convinced, all good arguments. I'll take the IIR route.
I've seen plenty of Bi-quad calculators around, actually already downloaded one from the miniDSP forum that has RIAA, HPF, LPF and so on (All-digital-coefs v1.2.xls).
Not sure how accurate that is, but I'll find out.
Don't know if I can mix and match IIR and FIR with the openDRC-DI, and if it is possible how the use of biquads affects the number of FIR taps available.
Ahh, more (re)searching to do, just first getting the kids to bed and cleaning up ....
Runs almost from this topic (be aware of moderators!), but to have things sorted out on this sideline.
It is the Technics SU9600 proper, with the third stage running on 136Vdc for maximum headroom. In my days working for Matshusita, no single unit ever was brought in for service.
Attachments
The equivalent input noise voltage and current are largely uncorrelated and uncorrelated noise adds in power, so the noise of the preamp will be something like 36.4 or 36.5 nV/sqrt(Hz) when you look at the combination of the current and voltage terms, excluding the 47 kohm termination resistor. That is, the current dominates entirely, so an instrumentation amp with a bit more voltage and a bit less current noise would be better.
The 47 kohm termination resistor produces a noise current of sqrt(4 k T/R) ~= 0.5869 pA/sqrt(Hz), which produces 14.2 nV/sqrt(Hz) of noise voltage density across the cartridge impedance at 3852 Hz. Together with what we already had, that's 39.2 nV/sqrt(Hz), getting close to your 40 nV/sqrt(Hz).
The thermal noise of the cartridge itself depends on the real part of its impedance. That would be 990 ohm, the DC resistance, if there were no iron losses or skin effect, but in reality iron losses are anything but negligible. All in all, it is a bit vague. For one specific type of Shure cartridge that an acquaintance of mine has measured, it was about the same as the termination resistor noise. Using that estimate, the total without any record playing would be 41.7 nV/sqrt(Hz) at 3852 Hz.
Thanks to the RIAA- and A-filtering, the noise bandwidth is only 3219 Hz relative to the gain at 1 kHz. Hence, the RIAA- and A-weighted noise is sqrt(3219 Hz)*41.7 nV/sqrt(Hz) ~= 2.37 uV when there is no record playing.
The sensitivity of the cartridge is 5.5 mV at 1 kHz, 5 cm/s. Hence, the SNR without record noise would be 67.3 dB(A) with respect to the nominal level of 5 cm/s. Assuming that the record noise is -60 dB(A) with respect to the nominal level of 5 cm/s at 1 kHz, the noise of the preamplifier including termination resistor and the cartridge would increase the total noise from -60 dB(A) to -59.26 dB(A), an SNR loss of 0.74 dB when the record is playing. That -60 dB(A) estimate for record surface noise is based on measurements from some people on this forum, I think it is for a dry record in good condition. (Mind you, 5 cm/s is really a nominal rather than a maximal level, so there is not necessarily any contradiction with the 70 dB dynamic range mentioned earlier in this thread.)
With no record playing, the noise of the preamplifier including termination resistor and cartridge together is 9.35 dB(A) higher than the noise of the cartridge alone, assuming that the estimate holds that the cartridge's thermal noise is about the same as the termination resistor noise. Hence, by redesigning the amplifier, you can improve the noise floor in between records much more than the noise floor with a record playing.
Last edited:
How accurate do you want? .001dB enough or do you want 0.00000001dB? Both are possible with IIR
Your setup is basically and old fashioned miniDSP with an FIR block in the middle, so you still get input and output biquads to play with. Cake can be had and eaten in this case!
Well I already have the DSP, and would really like a digital out (S/PDIF or AES/EBU). So yes, unless there are really good arguments against it.
That's a lot of info, thanks!
A couple of things are a given for now. The cart is almost new so I would like to keep that. Termination resistors are also a must so that's another one that's fixed. The instrumentation amp is of course a variable to play with. I am planning to order some EVAL-INAMP-82RZ boards. Those come with an assortment of amps included.
The AD8221 is one of them, could be a better option, 40fA/sqrt(Hz) and 8nV/sqrt/(Hz).
Another one is the AD8421, 200fA/sqrt(Hz) and 3nV/sqrt(Hz).
Since they are included I think I will start with the AD8221, slew rate is only 2.5V/uS, but that should still be more than enough considering max 22Khz and 0.56Vpk typical output.
Voltage noise from the amp itself will probably be around 10nV/sqrt(Hz) with a gain setting of 17.3dB. Current noise from the amp should be about 30x lower compared to the AD8429. I'll add some 6K8 resistors to my basket to set the gain.
How accurate? As good as possible without going overboard. 0.001dB would be more than enough. I'll take 0.1dB also.
Yeah been playing with the openDRC 2x2 app, indeed I can have my cake and eat it to, really like that
Now I know from my experiments with minimum vs. linear phase speakers I`m completely phase deaf, but still it has some appeal to implement the rumble HPF as a linear phase one. Since it's DSP I could always just program a LR4 @ 15Hz or something similar if I don't like it.
Yes, not the greatest cartridge for sure. But not a bad price/performance trade-off. Was also hoping it's a bit rugged.
Edit: Any suggestions for a better cart <E250, are welcome. Maybe the next upgrade. I'm not looking for the ultimate Hi-Fi setup, my PC running Kodi and the Topping DX7pro have that covered. Just like the whole experience of getting a record out, putting it on the TT and listening without the ability to grab the remote and select another song. Somehow that's just really relaxing. Good glas of Port and some snack to go with it and I'm a happy man...
Started with an old Akai AP-B21 turntable (got that one for free years ago). That had wow and flutter so bad I couldn't stand it anymore (Adele 25 album sounded really bad). So new belt went in, still bad. New cart, still bad. Modified it with XLR on the outputs, still bad. After a while I figured out it was the bearing that was no good anymore. No amount of polishing and lubrication fixed it.
By that point I had enough and just bought a Pioneer PLX-1000, major improvement!
So I transferred the cart from the Akai to the Pioneer, it's done maybe 25h in total, so really hoping I can enjoy it for a while still. Maybe the upgrade bugs bites me after this little project is done. Any upgrade should be reasonably priced (turntable was about E500,- this preamp will cost me about E400,-).
INA from TI on veroboard is also an option, but I can get the AD eval boards for about E35,- each, which includes the instrumentation amps. Think that is a pretty decent deal.
Last edited:
Well, I really want differential in, with good CMRR. I have no problem feeding the PCM4222EVM with a single ended signal. Since it's also diff in it will have no problem in suppressing anything picked up in the couple of cm of wire between the 2 boards. If I was running the signal from 1 box to the next then yes, but in this particular case I don't think it will bring much benefit. The ref output of the AD8221 should be connected to the GND I guess or maybe to the Vcom of the TI OPA1632...
Edit, thinking about it, that could be an option right? Would save me some coupling capacitors.
Last edited:
Of course if the unwanted signal is not CM to the diff input all is lost.
As far as I know the ground connection of the PLX1000 is not connected to the cart, so the outputs are fully floating. I'm a bit hesitant to put a DMM on the leads with a cart attached to verify this, those coils are fragile.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.