
Home  Forums  Rules  Articles  diyAudio Store  Blogs  Gallery  Wiki  Register  Donations  FAQ  Calendar  Search  Today's Posts  Mark Forums Read  Search 
Analogue Source Turntables, Tonearms, Cartridges, Phono Stages, Tuners, Tape Recorders, etc. 

Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving 

Thread Tools  Search this Thread 
4th October 2017, 08:13 PM  #11  
diyAudio Member
Join Date: Apr 2011

Quote:
with their output capacitors between the buffers and U1/U2. A very direct supply connection is necessary. Then route these power supply connections first, before adding more parts. Something like this: U4 C4 U1U2U3 C3 U5 Last edited by rayma; 4th October 2017 at 08:24 PM. 

4th October 2017, 09:58 PM  #12 
diyAudio Member
Join Date: May 2016
Location: Bournemouth UK

Thanks rayma, I'll have a go at rejigging the layout. I find it difficult to know what to trade off in the design: keeping the feedback loop short; separating the supply voltage from the signal etc. When I was doing the layout for the LM3886 it was (eventually) possible to separate the DC from the signal, decouple right up against the IC and keep the feedback loop short... with the five op amps in this design it's not so straight forward.

4th October 2017, 10:03 PM  #13  
diyAudio Member
Join Date: Apr 2011

Quote:
i.e. the power supply and minimizing the op amp's inverting input connection lengths. 

5th October 2017, 12:43 PM  #14 
diyAudio Member
Join Date: May 2016
Location: Bournemouth UK

That sounds like good advice. I've had another go, I think I could still move U1,2&3 over to the left a bit more to get U2 closer to the supply caps but I'm a bit concerned about putting more distance between U1, the input and the RIAA network. Plus I'm not totally happy with the pours but it seems like a step forward.

5th October 2017, 03:00 PM  #15 
diyAudio Member
Join Date: Apr 2011


6th October 2017, 03:39 PM  #16 
diyAudio Member
Join Date: Mar 2003
Location: Haarlem, the Netherlands

I guess U4 and U5 are meant to filter the supplies. I'm not sure whether they will remain stable with C4 and C3 as a load. If C4 and C3 are ordinary runofthemill electrolytic capacitors, their effective series resistance will probably help to keep U4 and U5 stable, otherwise you may have to insert small resistors between U4/U5 and C4/C3.
By the way, the LM317 has a highvoltage version, the LM317HV, that can handle 60 V of input voltage (even with shorted output). The LM337 also has an HV version, but the LM337HV is only meant for 50 V. (The LM317 also needs the ESR of the decoupling capacitors to remain stable, so don't use it with only film or ceramic decoupling.) I don't understand your remark about 47 ohm in the first post, was that simply a typo for 47 kohm? Noisewise the LME49710 is not the best choice for phono amplifiers. Its noise is specified as 2.5 nV/sqrt(Hz) of equivalent input noise voltage density and 1.6 pA/sqrt(Hz) of equivalent input noise current density. The noise voltage is a bit high for moving coil and the noise current is a bit high for moving magnet. 
6th October 2017, 04:17 PM  #17 
diyAudio Member
Join Date: May 2016
Location: Bournemouth UK

Hi MarcelvdG,
I think that the following extract from application note is reference to U4 & U5: "the power supply rails must present a low and more importantly flat impedance across the audio bandwidth to preserve the audible spectral balance and overall integrity of the input signal. The PSRR of the LME49710/LM4562 is outstanding across the full audio bandwidth and therefore is highly immune to power supply impedance discontinuities. However, given the low level signals provided from moving coil phono cartridges, the effect is more pronounced and therefore the impedance of the supply rails is addressed in this application." For the caps I plan to use Wima MKS2 Thanks for pointing out the LM317HV, I'll take a look. I think I'll experiment with the PSU, I can't decide whether the inefficiencies of regulating down from 37V outweigh the space / cost of including an extra trafo. I did mean 47k (there does appear to be an error in the application note too as it refers to 47 Ohm in section 3.3) I think I'll carry on with this chip for now and use it to experiment... I'm learning as I go (if that's not obvious ). I am however interested in other options... do you have a suggestion for a better performing op amp that would require a circuit of similar simplicity/complexity? 
6th October 2017, 10:02 PM  #18 
diyAudio Member
Join Date: Mar 2003
Location: Haarlem, the Netherlands

OK, if you got that supply circuit from an application note, whoever wrote the note must have checked the stability.
When you look only at the noise matching to a movingmagnet cartridge, the good old NE5534A does very well, as do some modern JFETinput opamps, like the OPA1641 (or 1642 or 1644), or the OPA1652 (or 1654). To compare opamps with different equivalent input noise current and noise voltage densities, you need to know the exchange rate between the two. If the source impedance were constant, that would simply be the source impedance: multiply the equivalent input noise current density by the source impedance and you have calculated what equivalent input noise voltage density would give the same amount of output noise. The impedance of a movingmagnet cartridge is, however, far from constant over the audio band. As a rule of thumb, you can take the cartridge impedance at 3852 Hz. With a typical 500 mH, 1 kohm cartridge, the impedance at 3852 Hz is about 12 kohm. So each pA/sqrt(Hz) roughly counts as 12 nV/sqrt(Hz). LME49710: current 1.6 pA/sqrt(Hz) > has as much impact as 19.2 nV/sqrt(Hz) voltage 2.5 nV/sqrt(Hz) Total is the root of the sum of the squares: 19.36 nV/sqrt(Hz) NE5534A: current 0.4 pA/sqrt(Hz) > has as much impact as 4.8 nV/sqrt(Hz) voltage 3.5 nV/sqrt(Hz) Total is the root of the sum of the squares: 5.94 nV/sqrt(Hz) OPA164X: current 0.0008 pA/sqrt(Hz) > has as much impact as 0.0096 nV/sqrt(Hz) voltage 5.1 nV/sqrt(Hz) Total is the root of the sum of the squares: 5.1 nV/sqrt(Hz) OPA165X: current 0.003 pA/sqrt(Hz) > has as much impact as 0.036 nV/sqrt(Hz) voltage 4.5 nV/sqrt(Hz) Total is the root of the sum of the squares: 4.5 nV/sqrt(Hz) To put these figures into perspective, the 47 kohm termination resistor injects 0.5869 pA/sqrt(Hz) of noise current into the cartridge. According to my rule of thumb that corresponds to 7.04 nV/sqrt(Hz). The thermal noise of the cartridge itself increases this to around 10 nV/sqrt(Hz), so whenever the result of the calculation of the opamp noise is below 10 nV/sqrt(Hz), it contributes less to the total noise than the termination resistor and the cartridge do, even when there is no record playing. Please keep in mind that I assume a cartridge inductance of about 500 mH. Many cartridges have inductances of this order of magnitude, but there are also models that have far less inductance and are therefore less sensitive to noise current. Last edited by MarcelvdG; 6th October 2017 at 10:11 PM. 
7th October 2017, 06:44 PM  #19 
diyAudio Member
Join Date: May 2016
Location: Bournemouth UK

Thanks, that gives me a bit of research to do... I just looked up the OPA165 and can only find it in a SMD package. I've not tried surface mount soldering before  maybe I should give it a go.

7th October 2017, 08:02 PM  #20 
diyAudio Member
Join Date: Apr 2011


Thread Tools  Search this Thread 


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Passive Phono Preamp using National LME49710  binspaul  Analogue Source  3  3rd September 2016 02:25 PM 
FS: NAIM phono cards, good for diy phono stage.  piecor  Swap Meet  11  29th May 2015 11:04 AM 
DIY phono to replace roksan phono stage  seroxatmad  Analog Line Level  6  14th December 2008 05:46 PM 
Should I replace Scott 222C phono stage with Claret phono stage?  Bing Yang  Analogue Source  0  22nd August 2005 06:41 AM 
New To Site?  Need Help? 