Hi all,
I'm designing a flat pre-amp to interface with my turntable (Pioneer PLX-1000 with an Audio-Technica AT-XP5 ). This is going to feed a TI PCM4222EVM which will feed a miniDSP openDRC-DI. RIAA and subsonic filtering will be done inside the DSP. My line level "preamp" is a Topping DX7pro feeding DIY active speakers.
The flat pre-amp will be based on the AD8429 chip, so setting gain is just a matter of putting in the right resistor value.
The AT-XP5 has an output of 5.5 mV (at 1 kHz, 5 cm/sec). So about 55mV @ 20KHz. If I set the gain of the AD8429 to 40dB that gives a maximum output of 5.5Vrms or 3.89Vpk. The AD8429 can ouput a maximum of about 13Vpk on +/-15VDC rails giving me a headroom of 9.12V. This is about 7.4dB headroom.
Is this enough or should I lower the gain of the AD8429 to increase the headroom and avoid clipping on pops / scratches? I would like as much gain in the first amp as possible to get the best possible noise performance.
I'm designing a flat pre-amp to interface with my turntable (Pioneer PLX-1000 with an Audio-Technica AT-XP5 ). This is going to feed a TI PCM4222EVM which will feed a miniDSP openDRC-DI. RIAA and subsonic filtering will be done inside the DSP. My line level "preamp" is a Topping DX7pro feeding DIY active speakers.
The flat pre-amp will be based on the AD8429 chip, so setting gain is just a matter of putting in the right resistor value.
The AT-XP5 has an output of 5.5 mV (at 1 kHz, 5 cm/sec). So about 55mV @ 20KHz. If I set the gain of the AD8429 to 40dB that gives a maximum output of 5.5Vrms or 3.89Vpk. The AD8429 can ouput a maximum of about 13Vpk on +/-15VDC rails giving me a headroom of 9.12V. This is about 7.4dB headroom.
Is this enough or should I lower the gain of the AD8429 to increase the headroom and avoid clipping on pops / scratches? I would like as much gain in the first amp as possible to get the best possible noise performance.
Hi KatieandDad,
Thanks for the reply.
70dB dynamic range is not something I'm really worried about with this ADC. It's going to run @ 48KHz/24bit. 12 bits should be enough to capture the full dynamic range of 70dB. The ADC is capable of much more. Even when running @ a maximum input to the ADC of -20dBFS for best THD performance there are 8 bits or 48dB dynamic range to spare.
So I think I`m good in that regard. It's really the analog front end that I need some guidance on.
Thanks for the reply.
70dB dynamic range is not something I'm really worried about with this ADC. It's going to run @ 48KHz/24bit. 12 bits should be enough to capture the full dynamic range of 70dB. The ADC is capable of much more. Even when running @ a maximum input to the ADC of -20dBFS for best THD performance there are 8 bits or 48dB dynamic range to spare.
So I think I`m good in that regard. It's really the analog front end that I need some guidance on.
That doesn't sound right. 5.5 volts rms is 7.8 volts peak.
Fwiw, a Pioneer A80 I used to own had the (discrete) phono stages running on -/+45 volts.
I'm a great believer in empirical testing... have you looked at the output on a scope when playing something 'noisy and scratchy'? Sometimes you can research stuff and still not be convinced or come up with a conclusive answer wheras a quick real test can help put things in perspective.
Hi Mark,
That is veyr similar to the setup I will be ending up with. I need to check the TI EVM specs but ISTR 4V for FS. So you work back from that. A good rule of thumb is to set 5cm/s to be IIRO -23dBFS on the ADC. That will deal with all but the most extreme clicks. There is a school of thought that says you can clip on scratches as long as the amplifier recovers quickly.
There are more mathematical ways to calculate this, but there will always be a bit of suck it and see due to variability in records.
That is veyr similar to the setup I will be ending up with. I need to check the TI EVM specs but ISTR 4V for FS. So you work back from that. A good rule of thumb is to set 5cm/s to be IIRO -23dBFS on the ADC. That will deal with all but the most extreme clicks. There is a school of thought that says you can clip on scratches as long as the amplifier recovers quickly.
There are more mathematical ways to calculate this, but there will always be a bit of suck it and see due to variability in records.
Hi Mooly,
You are right, I miscalculated. So the headroom with 40dB gain would only be 4.4dB.
Right now I'm in the design phase of this little project (actually I got my Basket filled with parts @ Mouser, not ordered yet). So no measurements possible yet.
I will of course order a couple of 0.1% resistor values to play around with.
+/-45V rails will give a lot of extra headroom but that's not possible with this project. I hope to keep it simple (I'm hooking up a some evaluation boards and HW I have laying around).
Hi MarsBravo,
Slew rate @ 22KHz /13Vpk is about 1.8V/uS. This is the worst case situation. The AD8429 can do 22V/uS so plenty of headroom in that regard I think.
The Opendrc DI noise floor (from miniDSP forum)
digital I/O noise floor........................... –126.07dBFS due to asynchronous S.R.C. dither noise
You are right, I miscalculated. So the headroom with 40dB gain would only be 4.4dB.
Right now I'm in the design phase of this little project (actually I got my Basket filled with parts @ Mouser, not ordered yet). So no measurements possible yet.
I will of course order a couple of 0.1% resistor values to play around with.
+/-45V rails will give a lot of extra headroom but that's not possible with this project. I hope to keep it simple (I'm hooking up a some evaluation boards and HW I have laying around).
Hi MarsBravo,
Slew rate @ 22KHz /13Vpk is about 1.8V/uS. This is the worst case situation. The AD8429 can do 22V/uS so plenty of headroom in that regard I think.
The Opendrc DI noise floor (from miniDSP forum)
digital I/O noise floor........................... –126.07dBFS due to asynchronous S.R.C. dither noise
Hi billshurv,
Good point, so to get the best THD performance the maximum input signal should be about 0.4Vrms to the EVM.
With 55mV maximum from the cart the gain should only be 7.3X or 17.3dB.
That will insure plenty of headroom in the pre-amp stage. Noise performance will suffer a bit, probably around 4nV/SQRT Hz.
Good point, so to get the best THD performance the maximum input signal should be about 0.4Vrms to the EVM.
With 55mV maximum from the cart the gain should only be 7.3X or 17.3dB.
That will insure plenty of headroom in the pre-amp stage. Noise performance will suffer a bit, probably around 4nV/SQRT Hz.
The current noise of an AD8429 is a bit high for a cartridge with an inductance of 1 H, especially when you are interested in the noise floor in between records (so without the record surface noise). As a rule of thumb, noise optimizing the RIAA- and A-weighted noise is equivalent to optimizing the spot noise at 3852 Hz. 1 H at 3852 Hz is 24.2 j kohm, so 1.5 pA/sqrt(Hz) has the same effect as 36.3 nV/sqrt(Hz).
Hi MarcelvdG,
So in total the noise would be about 40nV/sqrt(hz) in the pre-amp stage.
Now my knowledge starts to run out a bit, that would be 280mV over a bandwith of 20KHz, right? That seems very bad with a signal level of 400mV.
I'm sure I`m doing something wrong here, can't really compare these numbers like that?
Edit, did a bit of Googling (haven't done these type of calculations in ages), seems I did make a mistake. 40nV/sqrt(hz) is equal to about 5.6uV 20Hz - 20KHz.
Noise from the cart itself would be around 2.8uV (thermal, with 25Kohm impedance), before the needle hits the record. So about 8.5uV noise to be expected. That is about 93dB S/N w.r.t. a signal of 400mV.
So in total the noise would be about 40nV/sqrt(hz) in the pre-amp stage.
Now my knowledge starts to run out a bit, that would be 280mV over a bandwith of 20KHz, right? That seems very bad with a signal level of 400mV.
I'm sure I`m doing something wrong here, can't really compare these numbers like that?
Edit, did a bit of Googling (haven't done these type of calculations in ages), seems I did make a mistake. 40nV/sqrt(hz) is equal to about 5.6uV 20Hz - 20KHz.
Noise from the cart itself would be around 2.8uV (thermal, with 25Kohm impedance), before the needle hits the record. So about 8.5uV noise to be expected. That is about 93dB S/N w.r.t. a signal of 400mV.
Last edited:
Hi billshurv,
For sure, the turntable / record will be the limiting factor in all of this. Still I'm having fun already figuring all this out.
It will probably be a while before I have this up and running (3 small children running around). But should be easy enough when using of the shelf evaluation boards.
DSP part is actually the easiest to figure out, just a couple of clicks in RePhase and let the PC crunch the numbers.
Out of curiosity, what will your setup be? Custom ADC and DSP board or similar setup to what I'm up to?
For sure, the turntable / record will be the limiting factor in all of this. Still I'm having fun already figuring all this out.
It will probably be a while before I have this up and running (3 small children running around). But should be easy enough when using of the shelf evaluation boards.
DSP part is actually the easiest to figure out, just a couple of clicks in RePhase and let the PC crunch the numbers.
Out of curiosity, what will your setup be? Custom ADC and DSP board or similar setup to what I'm up to?
Oh it is fun. FWIW I have 4 phono stages currently in preparation. All delayed due to my second brood. I fear grandchildren may be only a few years away as well 
I'll have various phono stages into the TI EVM feeding miniDSP with a second feed to a laptop for if I want to go more advanced with signal grunging. Its worth getting the linear audio article on digital RIAA Linear Audio | your tech audio resource as it gives the biquads for miniDSP. There is also a thread here discussing that.
For a balanced flat pre that is suited to MMs here are a couple of options.
Phono Transfer System Bare Flat Preamp PC Boards
Retro - Balanced RIAA Stage (although this needs a couple of mods as RIAA needs removing and gain adjusting)
Note the TI EVM will do SE to bal conversion so you only need to be balanced all the way through for giggles.
I'll have various phono stages into the TI EVM feeding miniDSP with a second feed to a laptop for if I want to go more advanced with signal grunging. Its worth getting the linear audio article on digital RIAA Linear Audio | your tech audio resource as it gives the biquads for miniDSP. There is also a thread here discussing that.
For a balanced flat pre that is suited to MMs here are a couple of options.
Phono Transfer System Bare Flat Preamp PC Boards
Retro - Balanced RIAA Stage (although this needs a couple of mods as RIAA needs removing and gain adjusting)
Note the TI EVM will do SE to bal conversion so you only need to be balanced all the way through for giggles.
Hi billshurv,
Yes a came across the phono transfer system when I was doing a bit of online research. Looks nice, but a bit of DIY is half the fun. Soldering I did a lot already, also professionally (did reference designs for Digital TV demods before my previous employer went bankrupt), so that's not much of a challenge.
Going balanced all the way is good for bragging rights And actually it makes a lot of sense, at least for the cart - preamp interface. My whole system is now running balanced, no nasty ground loops or excessive noise. Just a better way of connecting stuff if you ask me.
I'll take a look at that Linear Audio publication. For now I simply put in the minimum phase parameters for RIAA in rephase:
NORMAL --- LOW-PASS --- 1st order --- 2122 Hz
COMPENSATE --- HIGH-PASS --- 1st order --- 500 Hz
NORMAL --- HIGH-PASS --- 1st order --- 50 Hz
And added a linear phase Horbach-Keele highpass filter (8Hz, 2.3ratio), Hann windowing. That looks pretty good with 6144 taps / channel available. About -40dB @ 1.5Hz and just -0.5dB @ 20Hz (compared to no HPF).
Yes a came across the phono transfer system when I was doing a bit of online research. Looks nice, but a bit of DIY is half the fun. Soldering I did a lot already, also professionally (did reference designs for Digital TV demods before my previous employer went bankrupt), so that's not much of a challenge.
Going balanced all the way is good for bragging rights
And actually it makes a lot of sense, at least for the cart - preamp interface. My whole system is now running balanced, no nasty ground loops or excessive noise. Just a better way of connecting stuff if you ask me.I'll take a look at that Linear Audio publication. For now I simply put in the minimum phase parameters for RIAA in rephase:
NORMAL --- LOW-PASS --- 1st order --- 2122 Hz
COMPENSATE --- HIGH-PASS --- 1st order --- 500 Hz
NORMAL --- HIGH-PASS --- 1st order --- 50 Hz
And added a linear phase Horbach-Keele highpass filter (8Hz, 2.3ratio), Hann windowing. That looks pretty good with 6144 taps / channel available. About -40dB @ 1.5Hz and just -0.5dB @ 20Hz (compared to no HPF).
For just the RIAA correction 6144 taps is really a lot. But the high pass filter does need a lot of taps.
The miniDSP is not doing anything else for the moment, so why not use them, right?
Maybe I can also do some EQ in the miniDSP to correct the frequency response of the cart, but that opens up a whole new can of worms (white or pink noise records don't seem the right way to go from the bit of Googling I did).
The miniDSP is not doing anything else for the moment, so why not use them, right?
Maybe I can also do some EQ in the miniDSP to correct the frequency response of the cart, but that opens up a whole new can of worms (white or pink noise records don't seem the right way to go from the bit of Googling I did).
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.