https://www.diyaudio.com/community/threads/chipomatic-balanced-input-riaa.331009/ Salas noticed that too).
Hi, yes he spotted the device, I went through his thread (I have great respect for Salas).
Nevertheless I did not see he connected it with Robert Demrow design from 60's and later actualized-upgraded by Wayne Kirkwood with help of Hans Polak......
After looking at many 1nV/sqrHz mic preamp data sheets, I will try THAT1580 as soon as time allows, difference from all others is that it allows gain adjustment even in trans-impedance mode (which is noise wise game changer) and very conveniently provides balanced output.
Small take on filter capacitors, particularly film and foil, and particularly on polystyrene, tab terminated vs extended foil and ESL
This is about film capacitors for filters,, like RIAA filter in this thread. As I'm finishing TT and coming back to preamp, choosing caps is back as an question.
First , couple of statements about my experience and learning:
- Filter caps are different story than coupling and decoupling caps, so only about filters.
- I have great preference for PS capacitors as it has been proven many times that they are closest to ideal. I disregard all snake oil audiophile capacitors.
In recent looking about others implementation of RIAA filters, I have seen many classic PS "cylinder bugs" installed, so I wonder how they perform comparing to my preferred RIFA PFE extended foil 1% caps...
Here is short take:
About measurements:
My input card ADC is 192/24 capable, which is still limiting frequency at 96 something kHz , I have no means to measure above so I can just catch corner of capacitor knee.
Simple 1st order filter is created, Input signal of 970 mV feed through 9k6 resistor in series with DUT capacitor. Scope input is across capacitor.
I run 20kHz to 100kHz sweep and recorded trace, 1/24 smoothed to remove digital artifacts...
DUT's are, in close but not exact nF values:
top : preffered RIFA extended foil PFE, 4n57 1% , green trace
next: classic "bug" tab terminated 5n6 10% yellow trace
next: PP ERO 33n +-5%, MKP 1845 , red trace
Bottom: PP ERO 3n3 MKP1846, orange trace
Now order of above picture is mixed up due capacitors values:
What I read is, polystyrene capacitors show perfect linear behavior ( green and yellow).
I cannot say that deviation of the curve close to 96kHz is because of impedance or my ADC limitations, anyway it is far above audio frequency.
In any case , both PS types seems to be perfect for filter.
On the other hand, by many tests next to perfect polypropilene , MKP (red and orange trace) , shows quite bit of deviant characteristic vs PS
This is about film capacitors for filters,, like RIAA filter in this thread. As I'm finishing TT and coming back to preamp, choosing caps is back as an question.
First , couple of statements about my experience and learning:
- Filter caps are different story than coupling and decoupling caps, so only about filters.
- I have great preference for PS capacitors as it has been proven many times that they are closest to ideal. I disregard all snake oil audiophile capacitors.
- I also have preference to extended foil capacitors vs tab terminated, since later should theoretically show inductance and extended foil shouldn't
- still I did not consider MLCC as NGO for filters, probably some ancient prejudice, in any case, this is only about film - foil
In recent looking about others implementation of RIAA filters, I have seen many classic PS "cylinder bugs" installed, so I wonder how they perform comparing to my preferred RIFA PFE extended foil 1% caps...
Here is short take:
About measurements:
My input card ADC is 192/24 capable, which is still limiting frequency at 96 something kHz , I have no means to measure above so I can just catch corner of capacitor knee.
Simple 1st order filter is created, Input signal of 970 mV feed through 9k6 resistor in series with DUT capacitor. Scope input is across capacitor.
I run 20kHz to 100kHz sweep and recorded trace, 1/24 smoothed to remove digital artifacts...
DUT's are, in close but not exact nF values:
top : preffered RIFA extended foil PFE, 4n57 1% , green trace
next: classic "bug" tab terminated 5n6 10% yellow trace
next: PP ERO 33n +-5%, MKP 1845 , red trace
Bottom: PP ERO 3n3 MKP1846, orange trace
Now order of above picture is mixed up due capacitors values:
What I read is, polystyrene capacitors show perfect linear behavior ( green and yellow).
I cannot say that deviation of the curve close to 96kHz is because of impedance or my ADC limitations, anyway it is far above audio frequency.
In any case , both PS types seems to be perfect for filter.
On the other hand, by many tests next to perfect polypropilene , MKP (red and orange trace) , shows quite bit of deviant characteristic vs PS
Hi Mark, correct, something was wrong with orange (MKP), here again, 5 sweeps to ensure all computer garbage is removed, tightened contacts. The step down at 56k is gone, but still not very linear cap. 2 flat Polystyrene cap traces I got from first and only go.
Mark, I admit you are right!
To increase resolution and kind of avoid my computer slowness, I increased sweep time to maximum 60sec. Than took 2 of each measurements of PS and MKP capacitors.
Here on graph overlapping traces are from same capacitor, with irregularities coming just from too slow computer. Now both show straight lines, except from measurement computer issues... green and orange are PS, blue and brick are PP.
Still over 90khz it is ADC limitations so don't bother to look...
It seems it is science on its own how to measure audio things with ADC, supposedly cheap and cheerful.... Yeah, right
Cheers!
To increase resolution and kind of avoid my computer slowness, I increased sweep time to maximum 60sec. Than took 2 of each measurements of PS and MKP capacitors.
Here on graph overlapping traces are from same capacitor, with irregularities coming just from too slow computer. Now both show straight lines, except from measurement computer issues... green and orange are PS, blue and brick are PP.
Still over 90khz it is ADC limitations so don't bother to look...
It seems it is science on its own how to measure audio things with ADC, supposedly cheap and cheerful.... Yeah, right
Cheers!
Been a while and I revisited capacitors briefly, at least figured how to measure FR without computer interrupting with it 🙂. In REW, measurement window (not RTA) and longest possible sweep time. Now results make sense, and all behave similar up to 80kHz measurement limit.
What I discovered, well , almost nothing. All usually good capacitors behave the same !
What I discovered, well , almost nothing. All usually good capacitors behave the same !
- all capacitors except top trace are 3n3 or 3n9, PS and PP.
- Top trace is different animal for comparison, ceramic disc for RF suppression, and here we can see it is visibly more hairy trace than recommended caps.
- Bottom trace is actually 2 traces, but so precisely overlapping that its impossible to separate them. One is 3n9 10% usual PS little barrel of rolled foil, other is 3n92 1% RIFA extended foil cap (my preferred for filters, until now).
Been a while again, but my TT is now done except for some fine tuning that will probably go on forever anyway 🤣 https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-8023636 ,
So I have some time to spend on preamp and build it.
Contrary to title of this thread, my first iteration is for MM cart. I want to have second TT with MM and recently got Revox B751 for that, it will play every day and 2000€ MC with not replaceable stylus is just for special occasions.
The concept I want to achieve is 3 interchangeable stages;
For this purpose I designed 2 variations (A and B below) IMHO it is practical and no nonsense approach to good preamp:
Variation B is technically more advanced as it:
Despise I have chosen variant A because it gives me flexibility to play with old LP pre RIAA curves and different cartridges.
About headroom pre RIAA or overload margin, so many words were said and differing opinion. I never had this type of issue with 66db flat MC gain stage, and recently I decided to trust only in old Shure max velocities chart . If anyone has better more recent information, please shout
So it goes up to 70 cm/sec. Lets take it that it is 80 cm/s. That is 16 times more than nominal 5 cm/s.
I would say that any amp that can take 16 times nominal output of cart measured at 5 cm/s has no overload issues at all.
In case A above 5mV cart x 40db flat front end gain X 16 = 8 VRMS, this is nothing that differential amp capable of 18 VRMS should worry about. IMHO almost any needle will jump out of groove before producing this output.
RIAA capacitors for completely passive filter I selected already, so no issue for me at least.
As for remaining loop gain at 20 kHz, according to datasheet for OP27 and NE5534 with gain of 40db we have some or about 16 db left for feedback. In next post I will show measurements, I think the distortion is many many times less than any cart will achieve with LP.
During the testing I figured that adding double balancing with differential line receivers (SSM2141) did not bring any benefits so I removed them. Nevertheless keeping at least one of them can be essential if following preamp is single ended.
PS, this idea I borrowed from Cohen - Kirkwood design.
This is final diagram used for testing that I will show:
Nevertheless, if someone wants technical perfection , possibility to use low V PS , and needs no flexibility as I wish, circuit B is probably way to go, only to take care that opam is stable at unity gain.
So I have some time to spend on preamp and build it.
Contrary to title of this thread, my first iteration is for MM cart. I want to have second TT with MM and recently got Revox B751 for that, it will play every day and 2000€ MC with not replaceable stylus is just for special occasions.
The concept I want to achieve is 3 interchangeable stages;
- front end (changeable for MM or different MC's) This should bring flat signal to line level and will always be installed inside the TT
- anti-RIAA stage where I want flexibility to adjust also for pre RIAA curves. This will be in preamp
- final gain and output stage
For this purpose I designed 2 variations (A and B below) IMHO it is practical and no nonsense approach to good preamp:
Variation B is technically more advanced as it:
- allows more HF headroom due to reduced HF gain
- allows more loop gain for input opamps
- makes RIAA capacitors selection easier
Despise I have chosen variant A because it gives me flexibility to play with old LP pre RIAA curves and different cartridges.
About headroom pre RIAA or overload margin, so many words were said and differing opinion. I never had this type of issue with 66db flat MC gain stage, and recently I decided to trust only in old Shure max velocities chart . If anyone has better more recent information, please shout
So it goes up to 70 cm/sec. Lets take it that it is 80 cm/s. That is 16 times more than nominal 5 cm/s.
I would say that any amp that can take 16 times nominal output of cart measured at 5 cm/s has no overload issues at all.
In case A above 5mV cart x 40db flat front end gain X 16 = 8 VRMS, this is nothing that differential amp capable of 18 VRMS should worry about. IMHO almost any needle will jump out of groove before producing this output.
RIAA capacitors for completely passive filter I selected already, so no issue for me at least.
As for remaining loop gain at 20 kHz, according to datasheet for OP27 and NE5534 with gain of 40db we have some or about 16 db left for feedback. In next post I will show measurements, I think the distortion is many many times less than any cart will achieve with LP.
During the testing I figured that adding double balancing with differential line receivers (SSM2141) did not bring any benefits so I removed them. Nevertheless keeping at least one of them can be essential if following preamp is single ended.
PS, this idea I borrowed from Cohen - Kirkwood design.
This is final diagram used for testing that I will show:
Nevertheless, if someone wants technical perfection , possibility to use low V PS , and needs no flexibility as I wish, circuit B is probably way to go, only to take care that opam is stable at unity gain.
Last edited:
And the test results of amp as shown above.
To feed amp I installed fake cartridge as test signal in diagram above. It gives 5mV signal over 800 ohm. All measurements are in open air, no chassis, and exposed to any RFI or whatever. Please note absence of typical 50 Hz (or 60 in US) peaks with harmonics always observed in single ended amps. By the way that 50HZ I find most annoying thing.
So first, typical 1kHz:
At 10 Khz THD is still low enough not to worry:
To feed amp I installed fake cartridge as test signal in diagram above. It gives 5mV signal over 800 ohm. All measurements are in open air, no chassis, and exposed to any RFI or whatever. Please note absence of typical 50 Hz (or 60 in US) peaks with harmonics always observed in single ended amps. By the way that 50HZ I find most annoying thing.
So first, typical 1kHz:
At 10 Khz THD is still low enough not to worry:
Further on measurements, this is anti RIAA curve , recorded frequency sweep.
Please note these are 2 traces of L and R channel, but you cant see the difference
This is very close up of above curve, channel and RIAA matching at within about 0.3db
Please note these are 2 traces of L and R channel, but you cant see the difference
This is very close up of above curve, channel and RIAA matching at within about 0.3db
Last edited:
On opamp choice, all components on my PCB are from late 80's and early 90's, so this could have been built 35 years ago.
In comparison to OP27 I tested NE5534 , this is the shot taken somewhere in testing still with SSM2141 outputs. It is just to show difference between opamps. NE5534 is respectable, but OP27 is measurably better in this circuit.
I was swapping only first stage opamps. Second stage remained OP27.
OP27 is OP27GP from PMI (for younger folks PMI is Presision Monolitics Inc)
NE5534 is NE5534P from TI , green trace
And these are the numbers, OP27 on left, NE5534 on right
NE5534 is good, but measurably worst than OP27 particularly noise wise in given circuit. Than again, probably no one will hear the difference
In comparison to OP27 I tested NE5534 , this is the shot taken somewhere in testing still with SSM2141 outputs. It is just to show difference between opamps. NE5534 is respectable, but OP27 is measurably better in this circuit.
I was swapping only first stage opamps. Second stage remained OP27.
OP27 is OP27GP from PMI (for younger folks PMI is Presision Monolitics Inc)
NE5534 is NE5534P from TI , green trace
And these are the numbers, OP27 on left, NE5534 on right
NE5534 is good, but measurably worst than OP27 particularly noise wise in given circuit. Than again, probably no one will hear the difference
And about the build, as I mentioned, all components are ancient, but most still available:
PCB is home brew, as I do since ever. It brings issues, I had to fix one broken trace, 2 shorts to gnd plane in second layer and you name it, lots of work. I am seriously considering ordering pcb from China next time, but I have issue with those subsidized manufacturers killing our local production.... another subject..
Anyway, this is PCB layout:
It is now made as one PCB, but you can see it can easily be cut in 3 stages.
About power supply:
I use 18V LiIon batteries for power tools. They have voltage between 17 and 21 V depending on charge. OP27 and many other opamps take +-22 V DC so I could feed them directly from batteries, but SSM2014 and any other differential line reciver take max +-18V, so I built regulator.
Now when I see that SSM2014 is redundant anyway, I will test with direct PS from batteries... that's pending work. This is simple PS that is used:
Battery is remote on cable, and it will always be as it doesn't fit in TT and it is not practical to recharge them inside. Because of that cable I put 2 small chokes (but with very thick wire = minimal R) in series. Most likely this is not needed anyway as internal impedance of 2.5 Ah battery is so low that no RFI will interrupt there.
For V ref I tried to measure various LED-s but measurement results always ended below noise floor of my RME soundcard + 60db mic amp gain... Anyway, I cant measure noise from this PS, it is low enough I guess.
For decoupling capacitors at opamps legs X7R MLCC is used as per diagram above in right corner. I don't hesitate to use these as it is analog circuit that ends at 20khz, I do understand people insisting on COG but for digital circuits with much higher frequencies of interest.
- All capacitors that signal sees are Polystyrene
- In RIAA circuit they are 1% extended foil polystyrene ... well, these are not easy to get today but are the best filter capacitors that I can think off.
- Wherever possible resistors are 0.05% 25ppm mil grades from cold war era. For riaa R2 part I deliberately used 2% 4420R so I could select them for 4412R that was needed. Where I run out of mil 0.05% resistors, I used hand matched 1% pieces. Particulary precise should be 10k feedback resistors.
PCB is home brew, as I do since ever. It brings issues, I had to fix one broken trace, 2 shorts to gnd plane in second layer and you name it, lots of work. I am seriously considering ordering pcb from China next time, but I have issue with those subsidized manufacturers killing our local production.... another subject..
Anyway, this is PCB layout:
It is now made as one PCB, but you can see it can easily be cut in 3 stages.
About power supply:
I use 18V LiIon batteries for power tools. They have voltage between 17 and 21 V depending on charge. OP27 and many other opamps take +-22 V DC so I could feed them directly from batteries, but SSM2014 and any other differential line reciver take max +-18V, so I built regulator.
Now when I see that SSM2014 is redundant anyway, I will test with direct PS from batteries... that's pending work. This is simple PS that is used:
Battery is remote on cable, and it will always be as it doesn't fit in TT and it is not practical to recharge them inside. Because of that cable I put 2 small chokes (but with very thick wire = minimal R) in series. Most likely this is not needed anyway as internal impedance of 2.5 Ah battery is so low that no RFI will interrupt there.
For V ref I tried to measure various LED-s but measurement results always ended below noise floor of my RME soundcard + 60db mic amp gain... Anyway, I cant measure noise from this PS, it is low enough I guess.
For decoupling capacitors at opamps legs X7R MLCC is used as per diagram above in right corner. I don't hesitate to use these as it is analog circuit that ends at 20khz, I do understand people insisting on COG but for digital circuits with much higher frequencies of interest.
Further on build and some photos:
For every pair of opamps I used DIL 16 socket. It allows to change single opamps or to easily make adapter board from any SMT dual opamp to DIL 16. But using DIL8 dual opamps is not easy, I have no dual dil 8 opamps in mind anyway.
Top side with removed_ bypassed output SSm2141:
Bottom with decouplers and Input PS capacitors, after some repairs:
Test setup assembled at motherboard:
It goes into TT, TT chassis connected to big brass bolt at main pcb. Cart cable plugged in amp 4 pin socket:
And finally, back side of TT closed:
For every pair of opamps I used DIL 16 socket. It allows to change single opamps or to easily make adapter board from any SMT dual opamp to DIL 16. But using DIL8 dual opamps is not easy, I have no dual dil 8 opamps in mind anyway.
Top side with removed_ bypassed output SSm2141:
Bottom with decouplers and Input PS capacitors, after some repairs:
Test setup assembled at motherboard:
It goes into TT, TT chassis connected to big brass bolt at main pcb. Cart cable plugged in amp 4 pin socket:
And finally, back side of TT closed: