If anyone has the formulae for determining RIAA component values they may care to consider the values used in the gramophone stage. The chosen capacitor values are standard ones and wonder about the 27 nano-farad value. I think that value might err too much on the low side.
The formulae for determining component values in the particular RIAA feedback network are:
Theoretical
R129/R125 equals 12.38
C123 uF value by R129 in ohms equals 2937 uS
C117 uF value by R125 in ohms equals 81.1 uS
Implementation (component tolerances ignored)
R129/R125 equals 12.12
C123 (.027) by R129 (100000) equals 2700 uS
C117 (.01) by R125 (8200) equals 82.5 uS
The middle result stands out since it is 9 percent less than the ideal result whereas the others are close to the mark.
We have to consider that the situation is not that bad due to tolerances perhaps of 10% and selection of examples at the upper end of the range. The only way to know is to remove the part and measure the value.
An ad hoc approach would be to fit a 2n2 capacitor in parallel with C123
Re Post #22, The calculations for component RIAA component values were originally due to W.H.Livy purportedly in 1957 (I was never able to track down the article in Wireless World). The calculations were recast in a more usable form by Baxandall.
These can be accessed from "Radio TV and Audio Reference Book " by S.W.Amos, an Article in the English edition of ETI "RIAA Equalisation" by Wilfred Harms published in the edition for June 1987 p.p. 25-27, and "Valve and Transistor Audio Amplifiers" J.L. Linsley-Hood p.p. 199-203
In regard to the significance of the deviation of series feedback equalization networks Harms article sets out the equation for working out the signal level at a specified frequency. That may be of interest in regard to responses in the ultra-sonic region and the question of added corrective RC network on the output. I will post these if there is sufficient interest.
These can be accessed from "Radio TV and Audio Reference Book " by S.W.Amos, an Article in the English edition of ETI "RIAA Equalisation" by Wilfred Harms published in the edition for June 1987 p.p. 25-27, and "Valve and Transistor Audio Amplifiers" J.L. Linsley-Hood p.p. 199-203
In regard to the significance of the deviation of series feedback equalization networks Harms article sets out the equation for working out the signal level at a specified frequency. That may be of interest in regard to responses in the ultra-sonic region and the question of added corrective RC network on the output. I will post these if there is sufficient interest.
Last edited:
andy_c is technically very well-regarded and used to be frequent poster here.
he created this page of good info on riaa networks for single op amp designs: Andy's Active RIAA Phono Equalization Design Page.
i can also strongly recommend analog devices op amp handbook, section 6 starting at page 16 here
best regards,
mlloyd1
he created this page of good info on riaa networks for single op amp designs: Andy's Active RIAA Phono Equalization Design Page.
i can also strongly recommend analog devices op amp handbook, section 6 starting at page 16 here
best regards,
mlloyd1
Guys, thanks for all the input. First things first, I'm am going to try swapping opamps. Couple questions:
Should I stick to surface mount opamps? If not, how do I attach a socket/adapter to the tiny surface mount connections on the board?
I will try NE5532N or the premium version NE5532AN. Only source is TI and that would mean TI's equivalent, correct?
thanks,
Herman
Should I stick to surface mount opamps? If not, how do I attach a socket/adapter to the tiny surface mount connections on the board?
I will try NE5532N or the premium version NE5532AN. Only source is TI and that would mean TI's equivalent, correct?
thanks,
Herman
SOIC to DIL sockets are available so you could install one of these with less risk of damaging the IC.
Surface mount technology requires more patience and hand to eye precision. The job will be made easier if a fine tip temperature controlled soldering iron and ideally a de-solder station are available.
A pair of tweezers and sticky tape might help with lining up the adapter socket while you solder the first lead to the board. The same process could be used in fitting SOIC substitute chips.
TI has a good reputation you don't need to look elsewhere. You would perhaps use an adapter if chances were you might later want to try another chip.
Newark stocks both DIL8 and SM 5532's from TI this week. I bought four to check against ST33078 since I already put sockets in my mixer. Those old DIL16's they used to have were hot back in the day when people worried about offset voltage for electrometers hydrometers and such DC instruments, but offset voltage is not a problem for audio with capacitor coupling.
Thanks for that diy-audio-engineering link to the ideal circuit. Paralleling the existing 3300 pf ceramic in the RA88a with a rice grain 150 pf cap, and messing around with the resistors to get rid of the 1.5 meg resistor my RA88a has got, sounds useful and doable. 922k series 75k might hiss less than 1.5 meg. A bit more doable than stuffing the .033 uf 400 v in there I bought with ~150kohm, which might be possible with the extra drive current of the 33078 but doesn't fit well physically.
Note for long life without fooling around to scrape oxide off after five years, I like phosphor bronze DIP sockets instead of brass contacts.
For the ultrasonics my unit already has 100 pf RF filter caps to ground on the RCA input sockets. These help keep interferance out from the CBer driving by, which used to be a problem with this unit. (along with a choke on the DC power supply input) The 1k series 25 uf to ground parallel the network to minus op amp input my unit has instead of 1.78 kohm may also protect some against deadly ultrasonic Quadraphonic records. (I only own one, and that used)
Thanks for that diy-audio-engineering link to the ideal circuit. Paralleling the existing 3300 pf ceramic in the RA88a with a rice grain 150 pf cap, and messing around with the resistors to get rid of the 1.5 meg resistor my RA88a has got, sounds useful and doable. 922k series 75k might hiss less than 1.5 meg. A bit more doable than stuffing the .033 uf 400 v in there I bought with ~150kohm, which might be possible with the extra drive current of the 33078 but doesn't fit well physically.
Note for long life without fooling around to scrape oxide off after five years, I like phosphor bronze DIP sockets instead of brass contacts.
For the ultrasonics my unit already has 100 pf RF filter caps to ground on the RCA input sockets. These help keep interferance out from the CBer driving by, which used to be a problem with this unit. (along with a choke on the DC power supply input) The 1k series 25 uf to ground parallel the network to minus op amp input my unit has instead of 1.78 kohm may also protect some against deadly ultrasonic Quadraphonic records. (I only own one, and that used)
Last edited:
You would do well to replace the ceramic capacitors in the feedback equalization. Polystyrene or polypropylene are suggested. Only in tube equipment does one need a 400 volt rated part. That should take the pressure off in terms of finding components that fit the footprint of your circuit board.
The 1k in series with ground is the decoupling leg of the negative feedback and the ratio of these determines the low frequency roll-off and affects the overall voltage gain. Increasing the value from 1k to 1.78k will have reduced the gain of the gramophone circuit block and extended the bass roll off to a lower level.
It may have been due to a modification during production to reduce the sensitivity of the module and allow more headroom for signals. Disco music is highly compressed and there are jockeys who have made an art out of the deliberate scratching of old gramophone discs.
The component values used in the RA88a do not reconcile with the Baxandall equations. To implement these you would need to replace all the component values in the network between the module output and the inverting input terminal. I can be more specific on this point if you are interested.
I lifted the drawing of the feedback network from the diy-audio-engineering website. Rather than reproduce the diagram here, I read it as {[2.49kohm series (3300 pf parallel 150 pf parallel 922 kohm) series (1000 pf parallel 75 kohm) ]parallel 1.78 kohm to ground.}
In my RA88a mixer, pulling the 1.5 meg resistor and the 100k resistor the builder installed and replacing them with the values above seems doable with rice grain sized tenth watt resistors. I was suspicious of the 1.5 meg anyway, resistors over 100 k tend to hiss in my PAS 2 tube preamp and I like to lower the resistance if possible, or put in metal film instead of carbon comp . (metal film over 100 k helped the PAS2 hiss a lot ). I may have to stay with the ceramic caps in the RA88a already installed, as long plastic caps would have long leads looping around that might short on something. Vertical mount of an axial cap is not an option. The ceramic caps installed are square and green, I can hope they are cpo and not something cheap like Y5u.
What I dont know about is the 1 k series 25 uf to ground the RA88a has on the minus in pin. I suppose the capacitor to ground is important and can't be eliminated, but it has different impedance at different frequencies and I don't know what the best compromise is versus the drawing in the article. I'm not a disco DJ, I just like something with more accessible components than the PAS2 to play records on. I've tried 3 times to get a new PAS2 volume pot, and version 2 installed now requires vise grip pliers on the volume shaft to turn it. Version 3 pot is in the parts bag, is stereo 250 k and turns okay, but doesn't have the volume bass boost pot section on the back.
Really the RA88a makes a piano sound from LP more like my Steinway calibrator now, with the wrong feedback RIAA values, than the PAS2 with polyester MKP brand .022 uf caps replacing deteriorated GI .02 uf paper/wax caps. So the RIAA can't be that far off. But if I'm going in there to change the op amp from 33078 to 5532 or LM4562, I might as well change the resistors and put the 150 pf 10% CPO cap I have parallel the .0033. I don't have a cap meter to mesure what the cap values really are, so it is a **** shoot really. The PAS2 has a 1% ceramic cap from 1961 installed in it.
BTW, I already had to put 33 pf across the feedback network, and .1 uf between +15 and -15 near the two input op amps, to make the 33078 not oscillate ~ 1 mhz. That 4th pole the article pontificates about, to get rid of the cutting lathe rumble, the RA88a already has a low cut filter later to eliminate footsteps from making the woofer cones go in and out.
Thanks for listening. any help on what to replace the 1k series 25 uf parallel to ground would be useful.
In my RA88a mixer, pulling the 1.5 meg resistor and the 100k resistor the builder installed and replacing them with the values above seems doable with rice grain sized tenth watt resistors. I was suspicious of the 1.5 meg anyway, resistors over 100 k tend to hiss in my PAS 2 tube preamp and I like to lower the resistance if possible, or put in metal film instead of carbon comp . (metal film over 100 k helped the PAS2 hiss a lot ). I may have to stay with the ceramic caps in the RA88a already installed, as long plastic caps would have long leads looping around that might short on something. Vertical mount of an axial cap is not an option. The ceramic caps installed are square and green, I can hope they are cpo and not something cheap like Y5u.
What I dont know about is the 1 k series 25 uf to ground the RA88a has on the minus in pin. I suppose the capacitor to ground is important and can't be eliminated, but it has different impedance at different frequencies and I don't know what the best compromise is versus the drawing in the article. I'm not a disco DJ, I just like something with more accessible components than the PAS2 to play records on. I've tried 3 times to get a new PAS2 volume pot, and version 2 installed now requires vise grip pliers on the volume shaft to turn it. Version 3 pot is in the parts bag, is stereo 250 k and turns okay, but doesn't have the volume bass boost pot section on the back.
Really the RA88a makes a piano sound from LP more like my Steinway calibrator now, with the wrong feedback RIAA values, than the PAS2 with polyester MKP brand .022 uf caps replacing deteriorated GI .02 uf paper/wax caps. So the RIAA can't be that far off. But if I'm going in there to change the op amp from 33078 to 5532 or LM4562, I might as well change the resistors and put the 150 pf 10% CPO cap I have parallel the .0033. I don't have a cap meter to mesure what the cap values really are, so it is a **** shoot really. The PAS2 has a 1% ceramic cap from 1961 installed in it.
BTW, I already had to put 33 pf across the feedback network, and .1 uf between +15 and -15 near the two input op amps, to make the 33078 not oscillate ~ 1 mhz. That 4th pole the article pontificates about, to get rid of the cutting lathe rumble, the RA88a already has a low cut filter later to eliminate footsteps from making the woofer cones go in and out.
Thanks for listening. any help on what to replace the 1k series 25 uf parallel to ground would be useful.
Last edited:
a few weeks ago i modified a phono preamp - like this one - for a friend.
input electrolytic (1uF) changed to a plastik cap.
power supply cleaned (lm317 reg instead of a noisy smpsu), essential.
riaa network enhanced (mkp and polystyrene caps built in), more exact.
elcap at the output was 4,7uF (instead of 47uF here).
input electrolytic (1uF) changed to a plastik cap.
power supply cleaned (lm317 reg instead of a noisy smpsu), essential.
riaa network enhanced (mkp and polystyrene caps built in), more exact.
elcap at the output was 4,7uF (instead of 47uF here).
Last edited:
Possibly a stupid question:
in the riaa network, and signal path resistors are used. From what I've seen -- always. Can we use small transformers in place of these resistors? Sort of like they use autoformers for volume control in place of resistors or pots.
It seem that transformers have an advantage, they attenuate without throwing out signal??
Herman
in the riaa network, and signal path resistors are used. From what I've seen -- always. Can we use small transformers in place of these resistors? Sort of like they use autoformers for volume control in place of resistors or pots.
It seem that transformers have an advantage, they attenuate without throwing out signal??
Herman
These resistors work with the capacitors in the RIAA network to set three time constants that define the RIAA curve needed.
There are numerous ways of designing a circuit with the three time constants, but transformers are of no practical use here.
They are however useful at the input, to step up the voltage (usually by a factor of 10) without adding too much noise.
There are a few RIAA preamps that use LR (inductor-resistor) filters instead, but these have their own problems.
Last edited:
That phono output cap sees another similar series cap and finally a 10k input impedance (neglecting some larger resistors to ground in the circuit).
You could change these caps to a smaller value. If both series caps were 10uF, the rolloff would be at around 3Hz.
Reducing these as far as 1uF would roll off at 30Hz, too high.
Last edited:
Take care with the power supply.
If it is on a separate pcb then fine.
If not keep the power supply ground and audio ground separate otherwise you will have big hum problems. Just join the audio ground once at the power supply smoothing capacitor centre tap.
I got caught out seriously with a USB mixer when I just routed the pcb ground any way it came out. With shorted input I was getting 1 volt hum !
On the 2nd pcb revision I complete separated the power supply and the hum was negligible.
If it is on a separate pcb then fine.
If not keep the power supply ground and audio ground separate otherwise you will have big hum problems. Just join the audio ground once at the power supply smoothing capacitor centre tap.
I got caught out seriously with a USB mixer when I just routed the pcb ground any way it came out. With shorted input I was getting 1 volt hum !
On the 2nd pcb revision I complete separated the power supply and the hum was negligible.
Q........I want to hear what happens when I change out two electrolytics in the signal path with something better......
.........you should listen yourself what happens...........🙂
i would say the sound is a little bit more transparent, better fine details with the plastic caps......personally preferred.
-----------
the transformer should be far away from phono preamp inputs - if not it will beginn to humm........
.........you should listen yourself what happens...........🙂
i would say the sound is a little bit more transparent, better fine details with the plastic caps......personally preferred.
-----------
the transformer should be far away from phono preamp inputs - if not it will beginn to humm........
C105 1uf
C127 47uf
I want to replace both of these with polypropelene.
Can I reduce the value of the 47uf -- it's the output coupling cap? If so to what value?
C127 47uf
I want to replace both of these with polypropelene.
Can I reduce the value of the 47uf -- it's the output coupling cap? If so to what value?
........c127/128:
.....see post 34........
and i used 4,7uF/35V.....63V in my phono pre without problems in this place. a smaller output cap helps to filter out subsonic frequencies (rumble).
in a friend's phono pre i built in 2,2uF at the output - and he was happy with it.
if you think in the future: the cap was too small, you can parallel another one to the existing cap.......
.....see post 34........
and i used 4,7uF/35V.....63V in my phono pre without problems in this place. a smaller output cap helps to filter out subsonic frequencies (rumble).
in a friend's phono pre i built in 2,2uF at the output - and he was happy with it.
if you think in the future: the cap was too small, you can parallel another one to the existing cap.......
Rayma,
>>If both series caps were 10uF, the rolloff would be at around 3Hz.
By "both" do you mean the 1uf (input) and 47uf (output)? Or all the 47uf in the signal path?
there are about 10 of the these 47uf in the signal path for each channel! ...studio mixer, lots of stages.
Maybe that would be make noticeable improvement.
Now I am thinking to keep the 1uf value stock but replace all coupling 47uf with 4.7uf film. If not possible to replace than to bypass with film cap.
>>If both series caps were 10uF, the rolloff would be at around 3Hz.
By "both" do you mean the 1uf (input) and 47uf (output)? Or all the 47uf in the signal path?
there are about 10 of the these 47uf in the signal path for each channel! ...studio mixer, lots of stages.
Maybe that would be make noticeable improvement.
Now I am thinking to keep the 1uf value stock but replace all coupling 47uf with 4.7uf film. If not possible to replace than to bypass with film cap.
The formula for determining the frequency is f=1/(2pi.R.C) where f is in hertz, R is Ohms and C is Farads. You can interpolate R or C putting either one on the left hand side of the equation and putting f in their place on the right.
As far as C105 goes larger value capacitors in this position introduce less internal noise - typically and in the interests of not picking up any external radiated fields these would be compact in size - i.e.electrolytic.
If the I.C. dual op.amp you are using has low enough input bias current you can do without an input coupling capacitor entirely.
With a 100 ohm series resistor bridging in place of C105 it is possible to do this with an SE5532. The resistor could be dispensed with and a link put in place of C105 with a 33078.
You would need to compensate for the modification by downstream changes with coupling capacitor values - using the given formulas. An old strategy to attenuate subsonic noise without an active high pass filter is to arrange for successive stages to roll off at 7 hertz.
Electrolytic capacitors are O.K. for inter stage coupling in this subsonic region and cumulative roll-off is more practical to implement as in physical terms polypropylene capacitors are large.
In the last regard I have built several RIAA projects over the years. Two earlier ones of these have low value input coupling capacitors (Polypropylene 470n and 1uF) while the latter ones - successive steps forward - have direct coupling at the input.
Polypropylene and polystyrene have their place in equalization and filter networks where accuracy is much more critical and the values involved will require only small packages.
- Status
- Not open for further replies.