I worked on the Balanced Discreet INA today. First the input stage has to be AC coupled to the opamp inversters. Then i got an expected stability problem. The OPs got hot ears more then ones. I use LT1115 in that position and OPA134 in the diferential stage. I then added the HF compensation that is used in the Collin Phono and also in some of syn08´s designs. Sofar no success. The op´s get warm and also the DC ouput runs wild especially when i raise the supply voltage. I have them generously decoupled and by
passed so i have no idea momentarily what makes the problem. Did i already fry the op´s, do i have a layout or topology problem ? I do not know. This gets quite complex and needs more work then i thought.
I am going back to the High Z Fet MPP and look if i can get even more stable bias when i insert back the constant current FET. I will also try to run the Leds on less current and listen if that sounds any better.
The circuits today show the balanced stage in it´s momentary condition and the reworked High Z MPP Head Amp.
passed so i have no idea momentarily what makes the problem. Did i already fry the op´s, do i have a layout or topology problem ? I do not know. This gets quite complex and needs more work then i thought.
I am going back to the High Z Fet MPP and look if i can get even more stable bias when i insert back the constant current FET. I will also try to run the Leds on less current and listen if that sounds any better.
The circuits today show the balanced stage in it´s momentary condition and the reworked High Z MPP Head Amp.
Well, i think i can get a better noise performance in the end. This is only the beginning and i sometimes think i should not post to early. On the other hand i thought it could be fascinating to folow a design process from beginning to end, mistakes, dead ends and all.
I have a balanced stage with INA163 working and although not bad at all my latest discreet circuits sound better. Simply more involving and moving,
I have a balanced stage with INA163 working and although not bad at all my latest discreet circuits sound better. Simply more involving and moving,
There are a lot of very quite, low distortion and stable state of the art MC step-up devices on the market. But there are very few with these qualities which preserve the living integrity of the music adequately; they are too often just sterile and do not pass the emotion of music with enough true intensity. Also there are very few experienced designers out there who allow their ears to be the final arbiters. That you post the full story of your design progress here in truthful detail is all but unique. It is a very healthy and transparent approach which some others ought to follow.
Last edited:
Joachim,
I suspect your stability problems come in three parts:
1/ You have a huge gain before you even get to the opamps, a CCS loaded jfet/bipolar cascode must have immense gain.
2/ You are running the LT1115 opamps with largish Pin 6 to Pin 2 feedback caps, which will drop their gain to unity at some HF frequency. BUT... the LT1115 data sheets on Page 7 says that unity gain operation is NOT recommended.
Then you feedback the whole thing with no sign of any frequency compensation...
It looks really spooky to me, but then again I'm not a huge OLG type of person. I'd just passive eq it after the jfet/bipolar cascode, and add a little extra gain (discretely without NFB) after the RIAA network if it's needed.
Regards, Allen
I suspect your stability problems come in three parts:
1/ You have a huge gain before you even get to the opamps, a CCS loaded jfet/bipolar cascode must have immense gain.
2/ You are running the LT1115 opamps with largish Pin 6 to Pin 2 feedback caps, which will drop their gain to unity at some HF frequency. BUT... the LT1115 data sheets on Page 7 says that unity gain operation is NOT recommended.
Then you feedback the whole thing with no sign of any frequency compensation...
It looks really spooky to me, but then again I'm not a huge OLG type of person. I'd just passive eq it after the jfet/bipolar cascode, and add a little extra gain (discretely without NFB) after the RIAA network if it's needed.
Regards, Allen
There is frequency compensation, but it's not good enough.
See guys, that's probably why you so passionately dislike negative feedback; it requires a little more than good ears
the heat goes up, i like that but not the LT1115´s !
First again to Onvinyl : I could not do this topology with the INA163 (for example) subtituting the three opamps. As you see the first stage is inverted and i would not have a choice of feedback resistors. The internal resistors in the INA163 are set to 3kOhm and it is not ment to be driven inverted.
To Allen: R11 and R12 set the gain of the input stage to 26dB so i tried not to put the full gain at the colectors to use. The ground referenced cascode is a bit unusual and this is my first atempt to mate that with a shunt feedback stage no wonder i do not understand that circuit in all it´s details yet. Using the compensating capacitor from Pin2 to Pin6 of the LT1115 should help to lower oszilation tendency. You have to look at the LT1115 as part of the whole feedback chain and not in isolation. Circuits like that have proven to work. Syn08, i know that i am stupid. In earliers posts i told everybody that i have not mastered modern circuit simultion yet. I was formally trained in the late 70th at the University of Padeborn in telecomunications. That is also the home of Nixdorf/Wincor computers. My main task was to calculate 9 dimensional matrixes by hand. No calculator allowed. In the 90th i worked with Prof.Hawksford in Essex but training there was informal, inbetween the lines, so to say. I have a basic understanding of feedback systems and stability margins but whould not say that i am an expert on feedback theory like you. i use gobs of feedback in my commercial designs but they all use finished building blocks like intergrated Operational amplifiers that of cause behave much more predictable then an atempt to visionary discreet design.
I simply could call up a friend and he whould put the circuit through it´s paces in no time. I have a lot of friends that are professional engineers that master that art perfectly well so i never have seen a need to look more into that myself. In fact i have founded world wide operating companies like Audio Physic on a beer felt. Well that time is come now and i should look for a circuit simulator that would fit my purpose. Any sugestion ? Microcap?
To blueskynis: R7C11, R8C12 and C13,C14 belong to a high frequency compensation that rolls of the open loop gain at a certain frequency in the quest for stability after feedback has been set. In my case the circuit is still oscilating. In the words of Syn08 i whould say "it´s not good enough". In my words "it´s not ready yet". That sounds better to my stupid but wise ears.
P.S. originally i wanted to post new adventures with the High Z MPP but now i need a Tequila beer.
First again to Onvinyl : I could not do this topology with the INA163 (for example) subtituting the three opamps. As you see the first stage is inverted and i would not have a choice of feedback resistors. The internal resistors in the INA163 are set to 3kOhm and it is not ment to be driven inverted.
To Allen: R11 and R12 set the gain of the input stage to 26dB so i tried not to put the full gain at the colectors to use. The ground referenced cascode is a bit unusual and this is my first atempt to mate that with a shunt feedback stage no wonder i do not understand that circuit in all it´s details yet. Using the compensating capacitor from Pin2 to Pin6 of the LT1115 should help to lower oszilation tendency. You have to look at the LT1115 as part of the whole feedback chain and not in isolation. Circuits like that have proven to work. Syn08, i know that i am stupid. In earliers posts i told everybody that i have not mastered modern circuit simultion yet. I was formally trained in the late 70th at the University of Padeborn in telecomunications. That is also the home of Nixdorf/Wincor computers. My main task was to calculate 9 dimensional matrixes by hand. No calculator allowed. In the 90th i worked with Prof.Hawksford in Essex but training there was informal, inbetween the lines, so to say. I have a basic understanding of feedback systems and stability margins but whould not say that i am an expert on feedback theory like you. i use gobs of feedback in my commercial designs but they all use finished building blocks like intergrated Operational amplifiers that of cause behave much more predictable then an atempt to visionary discreet design.
I simply could call up a friend and he whould put the circuit through it´s paces in no time. I have a lot of friends that are professional engineers that master that art perfectly well so i never have seen a need to look more into that myself. In fact i have founded world wide operating companies like Audio Physic on a beer felt. Well that time is come now and i should look for a circuit simulator that would fit my purpose. Any sugestion ? Microcap?
To blueskynis: R7C11, R8C12 and C13,C14 belong to a high frequency compensation that rolls of the open loop gain at a certain frequency in the quest for stability after feedback has been set. In my case the circuit is still oscilating. In the words of Syn08 i whould say "it´s not good enough". In my words "it´s not ready yet". That sounds better to my stupid but wise ears.
P.S. originally i wanted to post new adventures with the High Z MPP but now i need a Tequila beer.
The MPP High Z Headamp is ready ! I inserted the Fet current source back and now offset is stable to + - 3mV after trimming without DC servo. I run the Leds now on 4mA. No other changes. The circuits sounds to my satisfaction and performs well in the lab. It is fast ( -3dB at 5 MHz), low noise ( 0.5nV/qHz) and low distortion (-100dB at 2V out). This is the best i can do without global feedback and better then most commercial stages on the marked especially in noise performance. 1nV/qHz is advertised as very low noise and is a knind of industry standart. This circuit is 6dB better. In my system ( low sensitivity speakers and 0.5mV cartridge, the noise, both hum and hiss is unaudible at the listening position at insane volume and i hear loud ( over 100dB peak). The sound combines elements of raw energy with elegant handling of tonality. The output is current so transimpedance RIAAs are posible, even all in one go but i did not try that out. If you plan to win the next noise contest use 4 Jfets on each side. The drain resistors have to be lowered and the cascode transistors have to be doubled or replaced by something like 2N4401/4403 in that case just like in the Ultra Low Noise Low Z MPP. Any better performance is only posibe with global feedback or error cancelation and the second circuit gives an example of a posible topology. HF compensation has to be added and a DC servo can be used to trim the offset of that very high open loop gain design. Maybe i will try that variation too but momentaryly i am working on stabilizing the Discreet Intrumentation Amp.
Today i got Z foil resistors from Sigurt in sweden. Many thanks to suporting my effort !
I will now design a PVCB bord that will alow you to try that circuit if you wish. I will also look now more closely into the shunt regulators that Salas designed in the backround. Sorry Salas that i am not ready yet. I build parts of the circuit but i have to order more parts. This thread is not my only ocupation as you might gess.
Thanks for contributing to all. It´s an enyoyable and exiting ride.
P.S. I will try the DC and distortion trim on the Low Z MPP too.
Today i got Z foil resistors from Sigurt in sweden. Many thanks to suporting my effort !
I will now design a PVCB bord that will alow you to try that circuit if you wish. I will also look now more closely into the shunt regulators that Salas designed in the backround. Sorry Salas that i am not ready yet. I build parts of the circuit but i have to order more parts. This thread is not my only ocupation as you might gess.
Thanks for contributing to all. It´s an enyoyable and exiting ride.
P.S. I will try the DC and distortion trim on the Low Z MPP too.
Microcap is indeed a superb simulator with easy-to-use GUI! Highly recommended.
Spectrum software also has a very helpful support. Micropcap is
developed continiously and bugs are fixed quickly.
You can DL it here: Micro-Cap Evaluation Program From Request
If yo need any help with Microcap, I am willing to help out as much as I can.
I have designed a few RIAA amplifier with it.
Sigurd
Spectrum software also has a very helpful support. Micropcap is
developed continiously and bugs are fixed quickly.
You can DL it here: Micro-Cap Evaluation Program From Request
If yo need any help with Microcap, I am willing to help out as much as I can.
I have designed a few RIAA amplifier with it.
Sigurd
Hi John! My e-mail is : xxxxx@xxxxx.com
send me a mail and i respond.
Hi Sigurd, thanks for the link.
Edited out after enough time so to have been seen by the interested party, due to spam security reasons. Salas.
send me a mail and i respond.
Hi Sigurd, thanks for the link.
Edited out after enough time so to have been seen by the interested party, due to spam security reasons. Salas.
Today i tried the 1kOhm loading resistors. I compared the nonmagnetic Tantal (my reference) to the Texas Components Z Foil. First the resistive element looks eyactly like in the Charcroft but the resistor is put on something that looks like aluminim oxide. whereas the charcroft has a kind of bakalite material. I measured both resistors, one cold and the other still hot from soldering. They both measured exactly 1kOhm so the tolerance is better then 0.1% and the temperature coefficient seems to be very low.
Soundwise they are the sons of the Tantals. Again i could hear a slight improvement in microdetail but not with the light sound of the Caddock MK132. Something other was wrong and i found that the current through the Leds of the High Z MPP had quite a drastic effect to the sound. I had reduced the current in the Leds to 4mA but that seems to be to low for best sound. I adjusted the Fet back to 8-9mA and changed the base resistors back to 235 Ohm from 1kOhm. Now the sound gained a lot in size, power and presence. I like that better and this change makes a much bigger effect then swapping the resistors i must admit. I did the resistor comparison again and heard the same tendency. The Tantal with a very elegant vintage sound without much loos of information and the Z Foil with a more open and clear presentation. Ultimately i think the Z foil is more honest but i could happy life with both of them. So when you can grab an NOS Tantal for a good price, buy it. The Z foil is a fantastic modern replacement. It´s awaillable in small quantities with no problem. It´s expensive so one of my projects is to find an afordable resistor that sounds well too.
See the latest circuit diagram where i have replaced the fixed resistor on the Fet with a potmeter if you want to experiment. I used a value of 220 Ohm in the end.
Soundwise they are the sons of the Tantals. Again i could hear a slight improvement in microdetail but not with the light sound of the Caddock MK132. Something other was wrong and i found that the current through the Leds of the High Z MPP had quite a drastic effect to the sound. I had reduced the current in the Leds to 4mA but that seems to be to low for best sound. I adjusted the Fet back to 8-9mA and changed the base resistors back to 235 Ohm from 1kOhm. Now the sound gained a lot in size, power and presence. I like that better and this change makes a much bigger effect then swapping the resistors i must admit. I did the resistor comparison again and heard the same tendency. The Tantal with a very elegant vintage sound without much loos of information and the Z Foil with a more open and clear presentation. Ultimately i think the Z foil is more honest but i could happy life with both of them. So when you can grab an NOS Tantal for a good price, buy it. The Z foil is a fantastic modern replacement. It´s awaillable in small quantities with no problem. It´s expensive so one of my projects is to find an afordable resistor that sounds well too.
See the latest circuit diagram where i have replaced the fixed resistor on the Fet with a potmeter if you want to experiment. I used a value of 220 Ohm in the end.
