10H seems big but you could ask Torsten himself, ID here http://www.diyaudio.com/forums/member.php?s=&action=getinfo&userid=3984
One thing for sure you will need very quite supply voltage for the first JFET stage. You could without any harm increase the 2200 µF to 10 mF or more and exclude the inductor. You could also increase the resistance but the amp will get a longer start time. I can imagine up to at least 220 ohms would work.
One thing for sure you will need very quite supply voltage for the first JFET stage. You could without any harm increase the 2200 µF to 10 mF or more and exclude the inductor. You could also increase the resistance but the amp will get a longer start time. I can imagine up to at least 220 ohms would work.
Re: Power supply inductor
Koinichiwa,
You are reading the Circuit right.
Yes, the Inductor is 10 H / 30 Ohm / 10mA.
There is only one thing that you are wrong about. The inductor is not HUGE, it is quite small.
It is an item from Steinmusic in Germany.
Steinmusic Chokes
Does that help?
Sayonara
Koinichiwa,
rookster said:
You are reading the Circuit right.
Yes, the Inductor is 10 H / 30 Ohm / 10mA.
There is only one thing that you are wrong about. The inductor is not HUGE, it is quite small.
It is an item from Steinmusic in Germany.
Steinmusic Chokes
An externally hosted image should be here but it was not working when we last tested it.
Does that help?
Sayonara
Rookster
I have been using a similar front end for MC. Tried both 2sk147 and 2sk170 and settled for two parallel 2sk170GR as a compromise between resolution, noise and bass. For power i use the contents of 4 Nokia NiMeH rechargable batteries connected in series, i doubt if a much better sounding solution exists and of course it will sound better than sharing a common supply with the riaa section. Plus you won't need the inductor.
What really bothers me about El-cheapo are the humongous output coupling caps, surely they needn't be that big? If i had to build this exclusively for MC i'd check what's the output offset is like and if need be would periodically readjust it just to get rid of the elcos.
cheers
peter
I have been using a similar front end for MC. Tried both 2sk147 and 2sk170 and settled for two parallel 2sk170GR as a compromise between resolution, noise and bass. For power i use the contents of 4 Nokia NiMeH rechargable batteries connected in series, i doubt if a much better sounding solution exists and of course it will sound better than sharing a common supply with the riaa section. Plus you won't need the inductor.
What really bothers me about El-cheapo are the humongous output coupling caps, surely they needn't be that big? If i had to build this exclusively for MC i'd check what's the output offset is like and if need be would periodically readjust it just to get rid of the elcos.
cheers
peter
Koinichiwa,
There is good sense in all of this. First, I have removed the DC blocking capacitor inside the negative feedback loop. This causes a lot of offset even with low offset, J-Fet input Op-Amp's. That is due to the very high DC gain of the circuit. Up to 1V offset is not unusual even with OPA637's.
Of course, you could use some for of external Offset adjustment or a DC servo. I never got either of these options to sound as good as an output capacitor with a suitable bypass combo as shown.
You CAN leave off the big electrolytic capacitors completely if your load will not be any lower than around 40kOhm, as I wrote in the text, which you clearly not bothered to read. By using very oversized audio grade (Elna Starget or Silmic) the influence of these on the sound is minimal as they in effect "bypass" the 2u2 Film Coupling Capacitor at low frequencies. To have a low sonic foot print you need these capacitors back-2-back and biased from the negative rail, plus very large values.
I have tested this whole circuit extensively and tried pretty much all options. For reliability (no DC on output) and smallest sonic footprint I found exactly the solution you dislike so much to be sonically most transparent.
I will still say that this circuit is not the final word for Phonostages, but in the context of invested materiels and costs vs. performance it is very good, IF BUILD AS DIRECTED, including observing the general passive parts quality.
Sayonara
analog_sa said:
There is good sense in all of this. First, I have removed the DC blocking capacitor inside the negative feedback loop. This causes a lot of offset even with low offset, J-Fet input Op-Amp's. That is due to the very high DC gain of the circuit. Up to 1V offset is not unusual even with OPA637's.
Of course, you could use some for of external Offset adjustment or a DC servo. I never got either of these options to sound as good as an output capacitor with a suitable bypass combo as shown.
You CAN leave off the big electrolytic capacitors completely if your load will not be any lower than around 40kOhm, as I wrote in the text, which you clearly not bothered to read. By using very oversized audio grade (Elna Starget or Silmic) the influence of these on the sound is minimal as they in effect "bypass" the 2u2 Film Coupling Capacitor at low frequencies. To have a low sonic foot print you need these capacitors back-2-back and biased from the negative rail, plus very large values.
I have tested this whole circuit extensively and tried pretty much all options. For reliability (no DC on output) and smallest sonic footprint I found exactly the solution you dislike so much to be sonically most transparent.
I will still say that this circuit is not the final word for Phonostages, but in the context of invested materiels and costs vs. performance it is very good, IF BUILD AS DIRECTED, including observing the general passive parts quality.
Sayonara
Thorsten, have you checked my DC servo in my QSXM3 phono amp?Kuei Yang Wang said:
The servo can be seen in the datasheet of LT1115. See page 10 in the datasheet. If this circuit is to be used you have to investigate how OPA637 is built inside. When you have got everything right this is a cool way to connect a DC-servo.
The servo in my amp works really nicely.
Koinichiwa,
No, it is the first time I noticed that. But I tried a good quality cermet offset correction pot and even that made the sound worse by more than the output capacitors do.
In the end you are still injecting a signal into the circuit. You have chosen a different junction but it is still the same story. Also, it is I think not readily applicable to the OPA637. I'd have to see the LT1028 schematic.
All first order servos I have tried where clearly audible, even when the "pull" range was minimised and audibly worse than a simple output coupling capacitor combo. I had little luck with 2nd or 3rd order servos, they tend to be, shall we say, tempramental (predictably so, look at the bode plot).
Hence the output coupling combo as it stands. These days I use a MKP of a few uF, a KP of 0.47uF (usually largerst value avialable) plus a few nF each of Styrene and Silver Mica.
Sayonara
peranders said:
No, it is the first time I noticed that. But I tried a good quality cermet offset correction pot and even that made the sound worse by more than the output capacitors do.
In the end you are still injecting a signal into the circuit. You have chosen a different junction but it is still the same story. Also, it is I think not readily applicable to the OPA637. I'd have to see the LT1028 schematic.
All first order servos I have tried where clearly audible, even when the "pull" range was minimised and audibly worse than a simple output coupling capacitor combo. I had little luck with 2nd or 3rd order servos, they tend to be, shall we say, tempramental (predictably so, look at the bode plot).
Hence the output coupling combo as it stands. These days I use a MKP of a few uF, a KP of 0.47uF (usually largerst value avialable) plus a few nF each of Styrene and Silver Mica.
Sayonara
Servo's
Hi Kuei,
As time goes by, we seem to agree on more and more things!
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50096&highlight=elso+servo&r=&session=
Hi Kuei,
As time goes by, we seem to agree on more and more things!
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=50096&highlight=elso+servo&r=&session=
Re: Servo's
Elso,
Hardly as time goes by. Most of the work on the AA Phono was done in 1996/97, so is 6 - 7 Years old. The file creation date on the Graphics for this article on my Disk puts them at Feb 1998. This is anchient stuff for me for me at least.
So we have agreed for a very long time, except you did not know...
Sayonara
Elso,
Elso Kwak said:
Hardly as time goes by. Most of the work on the AA Phono was done in 1996/97, so is 6 - 7 Years old. The file creation date on the Graphics for this article on my Disk puts them at Feb 1998. This is anchient stuff for me for me at least.
So we have agreed for a very long time, except you did not know...
Sayonara
It remains to be seen if it's possible to fiddle with the offset trim pins of the OPA637 but I think it's impossible to use trimpots with so high DC gain.Kuei Yang Wang said:
If you check closely I have used a low noise OP07, made it slow, together with an extra filter so not very much of the signal passes through the servo backwards.
Koinichiwa,
Well, I suspect you CAN find a way, if neccesary with added current mirrors and stuff for level shifting. My original point remains that this does not remove the servo any more from the "signal path" (I hate that word) as feeding via a voltage divider into the "foot" of the feedback loop does.
Trimpots do work, but you need a low value, high quality pot and large value range limiting resistors. Use low tempco and the thing is even fairly stable with temperature. But with as high a DC gain I figured "Belts & Braces".
Very poor AC performance. I found the best DC servo for the OPA637 is an OPA627 but it is more expensive and sounds worse than the output capacitor combo.
I wish people would loose their paranoid fear of Capacitors or Transformers "in the signalpath". By avoiding a component in one place you only bring two or three in through the backdoor.
Like those "DC Coupled" Valve Amplifiers that proudly proclaim "no capacitors in the signal path", excepting of course the cathode bypass electrolytics and PSU electrolytics that could have been kept out of the SIGNAL CURRENT LOOP by using one good quality coupling capacitor, choke anode loading and fixed bias....
Anyway, you may wish to actually compare the two solution. Not in theory but in application. You may be surprised....
I started with Self's circuit from EWW with a DC blocker in the NFB Loop. Awful. Changed Op-Amp's, still awful. Increased the feedback loop impedance and used a film DC blocker, much better. Shifted the Cap from the feedbacl loop to the output, MUCH better, tried servos for about halve a year but always put the coupling caps back.
Go figger. Maybe I like plain straightforward capacitor distortion better than having amplified by the servo and reamplified by the Amplifier. Maybe I'm deaf.
Sayonara
peranders said:
Well, I suspect you CAN find a way, if neccesary with added current mirrors and stuff for level shifting. My original point remains that this does not remove the servo any more from the "signal path" (I hate that word) as feeding via a voltage divider into the "foot" of the feedback loop does.
Trimpots do work, but you need a low value, high quality pot and large value range limiting resistors. Use low tempco and the thing is even fairly stable with temperature. But with as high a DC gain I figured "Belts & Braces".
peranders said:
Very poor AC performance. I found the best DC servo for the OPA637 is an OPA627 but it is more expensive and sounds worse than the output capacitor combo.
I wish people would loose their paranoid fear of Capacitors or Transformers "in the signalpath". By avoiding a component in one place you only bring two or three in through the backdoor.
Like those "DC Coupled" Valve Amplifiers that proudly proclaim "no capacitors in the signal path", excepting of course the cathode bypass electrolytics and PSU electrolytics that could have been kept out of the SIGNAL CURRENT LOOP by using one good quality coupling capacitor, choke anode loading and fixed bias....
Anyway, you may wish to actually compare the two solution. Not in theory but in application. You may be surprised....
I started with Self's circuit from EWW with a DC blocker in the NFB Loop. Awful. Changed Op-Amp's, still awful. Increased the feedback loop impedance and used a film DC blocker, much better. Shifted the Cap from the feedbacl loop to the output, MUCH better, tried servos for about halve a year but always put the coupling caps back.
Go figger. Maybe I like plain straightforward capacitor distortion better than having amplified by the servo and reamplified by the Amplifier. Maybe I'm deaf.
Sayonara
I fully agree about the servos. The offset pots otoh have worked much better for me than couling caps. Of course one needs to be sufficiently neurotic to periodically (daily?) adjust the offset but for me that's part of the fun. It makes the listening experience that more special. And you can forget connecting carts directly, as their impedance will influence the offset too The topology i used was significantly different using split passive riaa and therefore 3 offset pots.
Paranoid fears about caps? We all have different sound idiosyncrasies (oops, meant priorities) and that's why a single solution doesn't fit all.
cheers
peter
The topology i used was significantly different using split passive riaa and therefore 3 offset pots.Paranoid fears about caps? We all have different sound idiosyncrasies (oops, meant priorities) and that's why a single solution doesn't fit all.
cheers
peter
Kuei Yang Wang said:
OP07 is slow and made even slower (intergrator) and the purpose of that is just that I want DC-signals to pass or only low freuency signals. There is absolute no need for a high speed opamp in a 1-10 Hz DC servo. You have lots and lots of feedback left in the opamp.
One another advantage of a slow opamp is that you can be more sure of no oscillations.
At least for discrete semiconductor circuits, I find that a DC servo is _by_far_ a better-sounding solution than an output capacitor.
But this is with the caveat that the servo circuit is intelligently designed and well-implemented. A poorly designed and implemented DC servo will definitely have an adverse effect on the sound.
If anyone decides to go with a DC servo, they should be prepared to invest a major effort in designing the circuit and board layout. If this is not feasible, or you do not have confidence in your design abilities, I recommend staying with an output capacitor. Simpler and easier - much less to go wrong with it.
For further study, I suggest reading AN232 from Analog Devices, authored by Walter Jung. Even if you don't use any of the methods proposed in the application note, it should give you some food for thought.
hth, jonathan carr
But this is with the caveat that the servo circuit is intelligently designed and well-implemented. A poorly designed and implemented DC servo will definitely have an adverse effect on the sound.
If anyone decides to go with a DC servo, they should be prepared to invest a major effort in designing the circuit and board layout. If this is not feasible, or you do not have confidence in your design abilities, I recommend staying with an output capacitor. Simpler and easier - much less to go wrong with it.
For further study, I suggest reading AN232 from Analog Devices, authored by Walter Jung. Even if you don't use any of the methods proposed in the application note, it should give you some food for thought.
hth, jonathan carr
MC stage
Thank you Kuei Yang Wang/Thorsten and analog sa and others for your responses and lively discussions.
analog sa/Peter,
apart from the slight change to the 2sk170GR and the power supply, have you altered much in the MC circuit? Have you had any issues with noise? I will be utilising an AT OC9 cartridge with an RB300 arm on a Rega Planar 3 (about 10 years old).
Kuei Yang Wang,
I am very inexperienced in the area of electrical circuits, but why is the 10H inductor/choke needed? I thought it was a filter, but the circuit is already running of a DC supply anyway. Thankyou for the details on the inductors, but I feel the cost and logistics of ordering 2 of these from Germay to Australia, would make it quite a hassle. Can i substitue something else? Any ideas? If this is the best option, then I will implement it.
Thank you Kuei Yang Wang/Thorsten and analog sa and others for your responses and lively discussions.
analog sa/Peter,
apart from the slight change to the 2sk170GR and the power supply, have you altered much in the MC circuit? Have you had any issues with noise? I will be utilising an AT OC9 cartridge with an RB300 arm on a Rega Planar 3 (about 10 years old).
Kuei Yang Wang,
I am very inexperienced in the area of electrical circuits, but why is the 10H inductor/choke needed? I thought it was a filter, but the circuit is already running of a DC supply anyway. Thankyou for the details on the inductors, but I feel the cost and logistics of ordering 2 of these from Germay to Australia, would make it quite a hassle. Can i substitue something else? Any ideas? If this is the best option, then I will implement it.
Rookster
As i mentioned previously the number of parallel FETs is a compromise between noise, resolution and bass. A single device sounds the most resolved to my ears but the noise with a 100uV source is intolerable. At 400uV i find two parallel 170s to sound just fine. If i may chime for Kuei the 10H is to separate the MC section from PS interference coming from the RIAA. Even batteries are not perfect and wire has impedance too. It probably works very well but for me it's easier and likely better to use a separate battery.
cheers
peter
As i mentioned previously the number of parallel FETs is a compromise between noise, resolution and bass. A single device sounds the most resolved to my ears but the noise with a 100uV source is intolerable. At 400uV i find two parallel 170s to sound just fine. If i may chime for Kuei the 10H is to separate the MC section from PS interference coming from the RIAA. Even batteries are not perfect and wire has impedance too. It probably works very well but for me it's easier and likely better to use a separate battery.
cheers
peter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.