Dears,
This is sub-thread of my turntable post: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7687982
It seems appropriate to open new one just for the preamplifier.
Since I'm waiting for motor parts to start significant upgrade of this 22 y.o. thing, I will revisit the preamp too.
So far I was using this design with 2 x SSM2017 in first stage and 2 different second stages, description is here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7688045
For the new version, I have 3 personal preferences that I would like to respect:
SSm2017 specifies 0.012% THD+N at gain of 1000, I probably get it bit better as 2 amps in differential configuration will cancel some distortion. However not the best in class THD, nevertheless doubt if it is hear-able. Nevertheless SSM2017 as ancient as it is, is still unbeatable on noise performance.
Here is PCB work in progress:
Alternatives (in my thoughts) for the first stage :
2nd option is to keep configuration but to upgrade SSM2017 with better INA like AD8429 for instance, as used here https://www.diyaudio.com/community/threads/chipomatic-balanced-input-riaa.331009/post-5627780 (but this is single ended output)
3rd option is classic differential receiver (same as my second stage now) , but that would call for 4 x AD797 and cost of over 70 Euro, I see no other opamp that would fit this purpose.
4th option is to use this ancient schematics and replace each AD797 with it in above schematics. Probably that is as far as low noise is possible.
Matched low noise dual transistors are rare today, but there are many dual JFET's that would do just fine here. Of course some more modern opamp would be used instead of OP27. However, this design for stereo would call for 12 dual matched low noise transistors, quite heavy on wallet.
Or, 5th idea, this very fine simple circuit from John Broskie's site. Looks very tempting even the gain will be max 40db (x100), so second stage will need to deal with another 40db, but that is not major issue if below circuit is quiet enough.
I'm sure that each of these options will work just nicely (when fully developed), but what would be the best?
Any comments, thoughts, ideas?
This is sub-thread of my turntable post: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7687982
It seems appropriate to open new one just for the preamplifier.
Since I'm waiting for motor parts to start significant upgrade of this 22 y.o. thing, I will revisit the preamp too.
So far I was using this design with 2 x SSM2017 in first stage and 2 different second stages, description is here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7688045
For the new version, I have 3 personal preferences that I would like to respect:
- Fully differential, I see great benefit and improvement from dropping ground of the signal path. Even still I'm using GND for referencing output voltage, but in new design I'm considering bootstrapping that with very low noise amplifier.
- Passive RIAA EQ
- As high as possible first gain stage, 60db (1000 V/V) seems just fine with my 0,34 mV nominal output cartridge. I do use higher PS voltage to allow for more output possibility; now +-18, next +-20V. This leaves second stage to deal with only about 20db gain.
SSm2017 specifies 0.012% THD+N at gain of 1000, I probably get it bit better as 2 amps in differential configuration will cancel some distortion. However not the best in class THD, nevertheless doubt if it is hear-able. Nevertheless SSM2017 as ancient as it is, is still unbeatable on noise performance.
Here is PCB work in progress:
Alternatives (in my thoughts) for the first stage :
2nd option is to keep configuration but to upgrade SSM2017 with better INA like AD8429 for instance, as used here https://www.diyaudio.com/community/threads/chipomatic-balanced-input-riaa.331009/post-5627780 (but this is single ended output)
3rd option is classic differential receiver (same as my second stage now) , but that would call for 4 x AD797 and cost of over 70 Euro, I see no other opamp that would fit this purpose.
4th option is to use this ancient schematics and replace each AD797 with it in above schematics. Probably that is as far as low noise is possible.
Matched low noise dual transistors are rare today, but there are many dual JFET's that would do just fine here. Of course some more modern opamp would be used instead of OP27. However, this design for stereo would call for 12 dual matched low noise transistors, quite heavy on wallet.
Or, 5th idea, this very fine simple circuit from John Broskie's site. Looks very tempting even the gain will be max 40db (x100), so second stage will need to deal with another 40db, but that is not major issue if below circuit is quiet enough.
I'm sure that each of these options will work just nicely (when fully developed), but what would be the best?
Any comments, thoughts, ideas?
Hi,
"...I see great benefit and improvement from dropping ground of the signal path. Even still I'm using GND for referencing ...."
Why?? What possible benefits, which possible improvements?
If You design Your gnd layout carefully there will be not enough 'improvement' left to really justify the increased complexity and effort of fully balanced circuitry.
Input INAs like ADIs SSM2017, TIs INA163/103 ot THATs 15xx used as INAs (bal-in, SE-out) are perfectly fine for the task and can accommodate from low-output MC pickups (>=150uV) to high-output MMs with very low noise.
I'm building phono stages using this input stage topology for 30 years now and I can just recommend it.
Add a dc-servo/subsonic to the INA, passively filter the 2120Hz pole and use a second amplifying stage with 50/500Hz filter in the feedback loop.
If wished for You can add either a SE-to-balanced output stage or an audio transformer (gnd-loop-free).
Circuitry-wise then everything has been done for the built of an excellent very-lownoise and highly flexible phono stage.
A major and much more decisive point You seemingly missed is the design of the gnd layout, resp. the whole PCB layout in general.
You can easily flush more than 20dB of THD+noise (and money) down the drain with a bad layout.
I'm sorry to say but the layout shown above leaves a lot room for improvement.
You'd be better of putting more thought and effort into the layout if You want to exploit and experience the full potential of the above described circuit topology.
jauu
Calvin
"...I see great benefit and improvement from dropping ground of the signal path. Even still I'm using GND for referencing ...."
Why?? What possible benefits, which possible improvements?
If You design Your gnd layout carefully there will be not enough 'improvement' left to really justify the increased complexity and effort of fully balanced circuitry.
Input INAs like ADIs SSM2017, TIs INA163/103 ot THATs 15xx used as INAs (bal-in, SE-out) are perfectly fine for the task and can accommodate from low-output MC pickups (>=150uV) to high-output MMs with very low noise.
I'm building phono stages using this input stage topology for 30 years now and I can just recommend it.
Add a dc-servo/subsonic to the INA, passively filter the 2120Hz pole and use a second amplifying stage with 50/500Hz filter in the feedback loop.
If wished for You can add either a SE-to-balanced output stage or an audio transformer (gnd-loop-free).
Circuitry-wise then everything has been done for the built of an excellent very-lownoise and highly flexible phono stage.
A major and much more decisive point You seemingly missed is the design of the gnd layout, resp. the whole PCB layout in general.
You can easily flush more than 20dB of THD+noise (and money) down the drain with a bad layout.
I'm sorry to say but the layout shown above leaves a lot room for improvement.
You'd be better of putting more thought and effort into the layout if You want to exploit and experience the full potential of the above described circuit topology.
jauu
Calvin
I had no success with op27. At high gain it sounded fuzzy. 5532 should be better.
I eventually used npn transistor and Lf353 for cassette tape head duty that worked very well.
I eventually used npn transistor and Lf353 for cassette tape head duty that worked very well.
Hi,
OP27 is in that old schematics from 1993 only, its not bad but today there are much better opamps, if NE than 5534 is better than 5532 for this, LM4562 even better, AD797 probably best.
BCM53DS might be what you need. It is supposed to have two dies inside that are the same as the one in a BC640, and I measured a base resistance of only 3.5 ohm on a BC640 a couple of months ago. The NPN equivalent (BC639, supposed to be the same as one half BCM56DS) had a 6.1 ohm base resistance.
@Calvin , thanks for your comments, appreciate.
Still I stick to fully differential, with years of use I found it much more solid and particularly immune on external interference. It doesn't cost me much as I have enough SSm2017 (if I end that route) in drawer, just 3 extra opamps for second stage .
Regarding layout, cant agree more it is fundamental. This board is work in progress, so only one channel is there right now. Here are different views with explanation if you will look at it and comment:
1. Top assy view:
2. Bottom assy view:
The complete board view is as in previous post, here I just extended 2 PS GND lines to emphasize this will connect later when PS is there....
How would you improve it?
Still I stick to fully differential, with years of use I found it much more solid and particularly immune on external interference. It doesn't cost me much as I have enough SSm2017 (if I end that route) in drawer, just 3 extra opamps for second stage .
Regarding layout, cant agree more it is fundamental. This board is work in progress, so only one channel is there right now. Here are different views with explanation if you will look at it and comment:
1. Top assy view:
- big blocks in middle are RIAA capacitors, very large as I use 1% PS for this. 3pcs for C2 to get 75.8000 nF and one 221 nF.
- INA amps are on this side; each PS leg decoupled with 1µF Murata MLCC (same on OPAMS in 2nd stage)
- here you can clearly see AGDN line that is separate from PSGDN line (as I always do) and connected only to a) headshell ground b) ref pins of INA's c) ref resistors infront of OPA's and d) input resistors. AGDN and PSGND will later be connected at PS
- For RIAA resistors I left space for 2 in parallel so I can get exact values needed
2. Bottom assy view:
- here you can clearly see PS GND going to decoupling capacitors. To reach INA decouplers there are vials going to front side.
- Second stage with 2 OPA I placed under large RIAA capacitors as there is lot of space. PS ground plane will cover free space later.
- Did not use SMT resistors because I have stock of 0,05% 25PPM mil resistors that are through hole
The complete board view is as in previous post, here I just extended 2 PS GND lines to emphasize this will connect later when PS is there....
How would you improve it?
Buy a bunch of ZTX851 and select for matching pairs - on each side the matching doesn't have to be great, just between the two sides. Should get to 0.223nV/√Hz - however with a single ended preamp you'd get 0.158nV/√Hz with half the number of low noise transistors, or 0.112nV/√Hz with the same number - balanced circuitry is inherently sqrt(2) noisier than single-ended all else being equal... Anyway its possible to get very low voltage noise with ZTX851 / 951 or similar devices, far better than the AD797 alone for instance. See this link: http://www.dicks-website.eu/low_noise_amp_part3/part3.html
Hi, thanks.
Just checked datasheet, here noise is not even specified and transfer is not particularly linear? Except low input Z, why? Is anything made with these already?
Well, I built a low-noise bandgap reference with them...
The lowest voltage noise transistors are usually medium power types that aren't meant for low noise applications and have no noise specifications. Silly, but true.
The lowest voltage noise transistors are usually medium power types that aren't meant for low noise applications and have no noise specifications. Silly, but true.
Interesting, I need to study bit more on the subject.
Where did you use bandgap reference, for MC preamp?
No, for an audio DAC, this one: https://www.diyaudio.com/community/threads/return-to-zero-shift-register-firdac.379406/
Hi,
You might consider a dc-servo for the first stage.
Output offsets could be high due to the high gain of this stage.
For the second stage You might consider a fully differential TI OPA1632, or a Dual (J)FET-input OPAmp.
jauu
Calvin
You might consider a dc-servo for the first stage.
Output offsets could be high due to the high gain of this stage.
For the second stage You might consider a fully differential TI OPA1632, or a Dual (J)FET-input OPAmp.
jauu
Calvin
Hi @Calvin ,
DC servo is generally good idea, but with existing SSM2017 pairs I had no issues . That doesn't mean that when I build new one I will not have issue.
Also I don't have problems using DC coupling caps at the end, or really large electrolytic or very fine film (too small cap will act as primitive subsonic filter too .-). Even in my TT combination never has subsonic resonances that I noticed.
Ill see if I have space on pcb to provide for one, I will not use it most likely, but for capacitor dislike-rs 🙂.
Dual JFET opamp is good idea, I started with OPA2828 BUT, Mouser has it only in HVSSOP , figured that's to small for my oldish hands and eyes, OPA828 are also in SOIC, thats (still) ok. I never listened to OPA828, but it looks perfectly specified for this application.
Idea with FDA however I don't understand? In second stage we need hi input impedance amp because of RIAA resistors. In my case they are 4947,50 Ohm, and that is quite low as I used capacitors on high side. In other designs this can be in 20kohm range.
What Rg would you use then and keep noise down while not attenuating signal a lot, normally around 1k is used here......
DC servo is generally good idea, but with existing SSM2017 pairs I had no issues . That doesn't mean that when I build new one I will not have issue.
Also I don't have problems using DC coupling caps at the end, or really large electrolytic or very fine film (too small cap will act as primitive subsonic filter too .-). Even in my TT combination never has subsonic resonances that I noticed.
Ill see if I have space on pcb to provide for one, I will not use it most likely, but for capacitor dislike-rs 🙂.
Dual JFET opamp is good idea, I started with OPA2828 BUT, Mouser has it only in HVSSOP , figured that's to small for my oldish hands and eyes, OPA828 are also in SOIC, thats (still) ok. I never listened to OPA828, but it looks perfectly specified for this application.
Idea with FDA however I don't understand? In second stage we need hi input impedance amp because of RIAA resistors. In my case they are 4947,50 Ohm, and that is quite low as I used capacitors on high side. In other designs this can be in 20kohm range.
What Rg would you use then and keep noise down while not attenuating signal a lot, normally around 1k is used here......
Hi, same as for @marcel, I never considered such mid power transistors in LTP dealing with microvolts, need to study and digest that a bit.
Is there anyone's else work using them in front end MC preamp?
Regarding DC servo loops and AC coupling capacitors:
If a DC servo loop should include the RIAA correction filter, it will shift the RIAA correction poles, particularly the first one. In this case, this doesn't apply if the loop is only around the first stage. See https://www.diyaudio.com/community/...o-precorrecting-riaa-correction-poles.406227/
AC coupling capacitors to passive RIAA correction networks can have a rather similar effect. See https://www.diyaudio.com/community/...eamp-phono-section-slp-70.407297/post-7557077
If a DC servo loop should include the RIAA correction filter, it will shift the RIAA correction poles, particularly the first one. In this case, this doesn't apply if the loop is only around the first stage. See https://www.diyaudio.com/community/...o-precorrecting-riaa-correction-poles.406227/
AC coupling capacitors to passive RIAA correction networks can have a rather similar effect. See https://www.diyaudio.com/community/...eamp-phono-section-slp-70.407297/post-7557077