Resistor swapping is hard work and i do not want to earn the reputation as the new Hiraga. So when you do not think that loading resistors make an audible difference ones they are of good quality like any 1% metalfilm ignore the next sentences and focus on the end of my post.
Today it did more listen to the High Z Fet MPP after i had DC coupled it and rebuild the servo. The trimmimg of offset and distortion is not in place yet. I will try that over the weekend because i have to buy some parts.
The rebuild circuit sounds much more open and dynamic then the one with the industrial 2uF cap so i had to do a reaccessment of the loading resistors. In the AC coupled version the Caddock MK132 came out first because it brought a very wellcome opening up of the sound. The latest DC coupled version sounds much closer to what i am used to so it was much more difficult this time to make a decission. I had the following resistors at hand : Dale RS60 antimagnetic 1kOhm, Tantal antimagnetic 1kOhm, Caddock MK132 1kOhm, YAM2 3kOhm. I put them on the front part of an Eichmann connector to avoid as much metal as i could. I can plug in those jumpers into a parallel teflon insulated RCA jack. As solder i used Mundorf silver-gold not because i think it is uperiour souding but because i had it awailable. Sofar i have only limited experience with the sound of different solder so i have not made up my mind if that is super critical. They all behave different and i usually prefer thoose that make a shiny soldering spot if ony for the sake of aesthetics.
First i listened to the Caddock and found the same openness and transparency that they are known for. This time it was a mixed blessing and very programm dependend. On very dense Pop material like "Stringimimi" of the italien band Matia Bazar i liked the high amoint of sorting and diferentiating of the densly recorded song. Lot´s of analog synth, drums, percussion, voice etc. Kind of effect then Alice in wonderland. On "Split", Alto Edition with Patricia Barber it was clearly the non magnetic tantal that made the nicest noise very closely followed by the YAM2. On that track the caddock was a very little spitty on the silbilance. If you love light Jazz grab the YAM2 and the Tantals when you can get them. The are quite obscure now. The YAM2 and the Tantal have an elegegance that is mesmerising. The Tanatal more rose coloured and the YAM2 more neutral. The Tantal more light and spacious and the YAM2 more close to earth. This was quite reasuring because it mimiked quite closely the results i got when i tested passive componetns in the 90th. I can say now with a bit of confidence that the High Z Fet Headamp is honest and natural sounding for my ears. What about the Dale RS60 ? Not bad at all but slighly less informative then any of the other. Extremely fine detail like studio acoustics and flaws in the recording (like tape splitting) escape that resistor. I whould describe the sound as "normal" like a good piece of bread with butter. Nothing wrong there and i use the ones with less wattage in volume without hassitation in a lot of less critical position in my commercial products.
Now a question to Scott Wurzer:
I have drawn the circuit you send me and can not see how the input pair is biased. Also
3 of the Opamp stages are drawn like indistinguished Gain stages. Are they invertiong or non inverting and what are the missing resistor value ?
Well, i am getting a bit lazy.
Today it did more listen to the High Z Fet MPP after i had DC coupled it and rebuild the servo. The trimmimg of offset and distortion is not in place yet. I will try that over the weekend because i have to buy some parts.
The rebuild circuit sounds much more open and dynamic then the one with the industrial 2uF cap so i had to do a reaccessment of the loading resistors. In the AC coupled version the Caddock MK132 came out first because it brought a very wellcome opening up of the sound. The latest DC coupled version sounds much closer to what i am used to so it was much more difficult this time to make a decission. I had the following resistors at hand : Dale RS60 antimagnetic 1kOhm, Tantal antimagnetic 1kOhm, Caddock MK132 1kOhm, YAM2 3kOhm. I put them on the front part of an Eichmann connector to avoid as much metal as i could. I can plug in those jumpers into a parallel teflon insulated RCA jack. As solder i used Mundorf silver-gold not because i think it is uperiour souding but because i had it awailable. Sofar i have only limited experience with the sound of different solder so i have not made up my mind if that is super critical. They all behave different and i usually prefer thoose that make a shiny soldering spot if ony for the sake of aesthetics.
First i listened to the Caddock and found the same openness and transparency that they are known for. This time it was a mixed blessing and very programm dependend. On very dense Pop material like "Stringimimi" of the italien band Matia Bazar i liked the high amoint of sorting and diferentiating of the densly recorded song. Lot´s of analog synth, drums, percussion, voice etc. Kind of effect then Alice in wonderland. On "Split", Alto Edition with Patricia Barber it was clearly the non magnetic tantal that made the nicest noise very closely followed by the YAM2. On that track the caddock was a very little spitty on the silbilance. If you love light Jazz grab the YAM2 and the Tantals when you can get them. The are quite obscure now. The YAM2 and the Tantal have an elegegance that is mesmerising. The Tanatal more rose coloured and the YAM2 more neutral. The Tantal more light and spacious and the YAM2 more close to earth. This was quite reasuring because it mimiked quite closely the results i got when i tested passive componetns in the 90th. I can say now with a bit of confidence that the High Z Fet Headamp is honest and natural sounding for my ears. What about the Dale RS60 ? Not bad at all but slighly less informative then any of the other. Extremely fine detail like studio acoustics and flaws in the recording (like tape splitting) escape that resistor. I whould describe the sound as "normal" like a good piece of bread with butter. Nothing wrong there and i use the ones with less wattage in volume without hassitation in a lot of less critical position in my commercial products.
Now a question to Scott Wurzer:
I have drawn the circuit you send me and can not see how the input pair is biased. Also
3 of the Opamp stages are drawn like indistinguished Gain stages. Are they invertiong or non inverting and what are the missing resistor value ?
Well, i am getting a bit lazy.
I decided to listen more with the High Z Fet MPP DC with 1kOHm Taltal loading.
More Jazz (Helen Merill in Tokio) and classical (Die Kluge, Orf; Suite Espaniol etc.) and got such a good sound that i think now that the High Z version is as good as the Low Z version if not totally similiar. I then measured the current over the 10 Ohm source resistors and found that it was very similiar. 64.2mV on the N types and 64.1mV on the P types. I am not sure any more that an asymmetric trim is of any advantage but i will try to measure distortion and see if i can get a better null. My circuit seems to force the Fet´s into balance without trimming. That is particular surprising because the P types had around 25% less IDss in their "natural" state.
That result made me think if a Fet- Bipolar cascode whould be the best way to start my inamp survay. Let´s see with what i come up. One real advantage is that i do not have to worry about biasing in that case.
More Jazz (Helen Merill in Tokio) and classical (Die Kluge, Orf; Suite Espaniol etc.) and got such a good sound that i think now that the High Z version is as good as the Low Z version if not totally similiar. I then measured the current over the 10 Ohm source resistors and found that it was very similiar. 64.2mV on the N types and 64.1mV on the P types. I am not sure any more that an asymmetric trim is of any advantage but i will try to measure distortion and see if i can get a better null. My circuit seems to force the Fet´s into balance without trimming. That is particular surprising because the P types had around 25% less IDss in their "natural" state.
That result made me think if a Fet- Bipolar cascode whould be the best way to start my inamp survay. Let´s see with what i come up. One real advantage is that i do not have to worry about biasing in that case.
Did I send that schematic? I thought I sent a link to the Seventh Circle preamp schematic. I will send a figure tomorrow that should make it clear.
Thank you Scott !
Hi Rüdiger ! Triple groundplane is a bit hard to explain. I am not sure if i inverted it but i have known double groundplane for a long time in HF circuits. In priciple it is two double metalised PCB boards, one for each channel. Top and bottom copper are connected at strategic places by experiment. The turntable ground is connected in between. By coincidence i found when i put a third (or if you will fifth) plane under thoose and connect it with one wire the hum got down even more. In fact i found that out when one of the upper groundplanes got in contact with a gold plated frontplate of an amp that sat under it. I made a small scetch. Unfortunately the colour was lost during transfer to PDF but i hope you see the princple. Soon i will make some photos. I am still not back in the factory where all my better measurement equipment and digital camera is because i suffer since over 2 weeks from a bad case of "Hexenschuss".
Fortunately when i sit the pain is ok. When i walk or stand for more the 30 secomds i could cry tears.
When i have the camera availlable i make a photo and then it is clear.
Hi Rüdiger ! Triple groundplane is a bit hard to explain. I am not sure if i inverted it but i have known double groundplane for a long time in HF circuits. In priciple it is two double metalised PCB boards, one for each channel. Top and bottom copper are connected at strategic places by experiment. The turntable ground is connected in between. By coincidence i found when i put a third (or if you will fifth) plane under thoose and connect it with one wire the hum got down even more. In fact i found that out when one of the upper groundplanes got in contact with a gold plated frontplate of an amp that sat under it. I made a small scetch. Unfortunately the colour was lost during transfer to PDF but i hope you see the princple. Soon i will make some photos. I am still not back in the factory where all my better measurement equipment and digital camera is because i suffer since over 2 weeks from a bad case of "Hexenschuss".
Fortunately when i sit the pain is ok. When i walk or stand for more the 30 secomds i could cry tears.
When i have the camera availlable i make a photo and then it is clear.
Most of the time i also connect the upper and middle plate where the turntable ground is.
When you have worked with that style of construction for a while usually you get very good at finding the right contact points. One option is to try with aligator clips first. I like this open style of contruction now even better then my little metal boxes because the parts are better accesable and aesier to change in type and topology.
When you have worked with that style of construction for a while usually you get very good at finding the right contact points. One option is to try with aligator clips first. I like this open style of contruction now even better then my little metal boxes because the parts are better accesable and aesier to change in type and topology.
Today was interesting but also a bit frustrating.
I had some comunication with Peter Suchi of Clearaudio concerning my strobe lamp.
The Led was glowing but i could not see the lines stop on the stobe mat. Peter sugested to change the batteries and after doing that everything was fine. The Led strobe flashes with 300Hz so is more accurate then a 50Hz strobe. I found that my turntable was a bit fast, something that escaped my 50Hz strobe. After adjusting the speed i listened to some known tracks. The sound is a bit fuller now but not much. I very slightly turned down my subwoofer. A have to do more listen but the general impression is fine. No big deal.
I also started to put some thought into the discreet balanced input stage of my Intrumentation RIAA. The latest version of the High Z Fet MPP sounds fine and i am geting better friends with Fet´s. I decided on a Fet-Bipolar cascode with adjustable current sources, an often used recipy for good results This time i build the circuit with run of the mill parts to isolate the topology from effords to tune the circuit with passive components. Resistos are standart 1% metalfilm, elcaps simple Yamicon TK 105° and i used HS10 leadfree silver solder. I got now quite good to use it. Temperature has to be adjusted for different jobs but i am getting better looking joints the more i use it. I did not select the Fets and used 2N4401/4402 bipolars also without selecting. The adjustable current sources are quite convenient and after i had connected two 9V alkaline cells i got a good balance with the trimmers. currents and voltages seemed ok so i tried to listen to that stage without any further electronics by AC coupling the output. I also implemented the 75usec time constant in my usual transimpedance way. Unfortunately i got no output so there must be a topologigal problem or i had bad contacts because i use really the ceapest gold plated jacks i could fine. Anyway i gave up for today not without testing if the source resistors connected to ground gave a signal and i also tried the input stage unbalanced. No succes ether. I will post the circuit as it is today to give you an idea and maybe one of you has a solution. Tomorrow i will look further if i can lokate the problem
I had some comunication with Peter Suchi of Clearaudio concerning my strobe lamp.
The Led was glowing but i could not see the lines stop on the stobe mat. Peter sugested to change the batteries and after doing that everything was fine. The Led strobe flashes with 300Hz so is more accurate then a 50Hz strobe. I found that my turntable was a bit fast, something that escaped my 50Hz strobe. After adjusting the speed i listened to some known tracks. The sound is a bit fuller now but not much. I very slightly turned down my subwoofer. A have to do more listen but the general impression is fine. No big deal.
I also started to put some thought into the discreet balanced input stage of my Intrumentation RIAA. The latest version of the High Z Fet MPP sounds fine and i am geting better friends with Fet´s. I decided on a Fet-Bipolar cascode with adjustable current sources, an often used recipy for good results This time i build the circuit with run of the mill parts to isolate the topology from effords to tune the circuit with passive components. Resistos are standart 1% metalfilm, elcaps simple Yamicon TK 105° and i used HS10 leadfree silver solder. I got now quite good to use it. Temperature has to be adjusted for different jobs but i am getting better looking joints the more i use it. I did not select the Fets and used 2N4401/4402 bipolars also without selecting. The adjustable current sources are quite convenient and after i had connected two 9V alkaline cells i got a good balance with the trimmers. currents and voltages seemed ok so i tried to listen to that stage without any further electronics by AC coupling the output. I also implemented the 75usec time constant in my usual transimpedance way. Unfortunately i got no output so there must be a topologigal problem or i had bad contacts because i use really the ceapest gold plated jacks i could fine. Anyway i gave up for today not without testing if the source resistors connected to ground gave a signal and i also tried the input stage unbalanced. No succes ether. I will post the circuit as it is today to give you an idea and maybe one of you has a solution. Tomorrow i will look further if i can lokate the problem
Momentary i do not post on the cable thread but i found this to be interesting:
"A case for subjectivity" by Jens Blauert.
You can download the file from the Linkwitz Lab - Loudspeaker Design webpage
"A case for subjectivity" by Jens Blauert.
You can download the file from the Linkwitz Lab - Loudspeaker Design webpage
Colour me blind but i found one problem. I was trying to listen to the common mode signal ! The N channel reverses the phase so the signal comes out with negative polarity. The P channel inverses the negative signal so it comes out at positive polarity. + and - are zerro so no miracle that i could hear vitually nothing. On the other hand what a simple and great way to test common mode suppresion ! Under certain circumstances i could get a small output signal. For example when i tied the source degeneration resistors to ground without adjusting the current sources. Coincidence is the teacher sometimes.
One solution seen for example in the La Boheme phonostage whould be to use N channels for both the noninverted and the inverted output but that needs coupling caps i try to avoid.
One solution seen for example in the La Boheme phonostage whould be to use N channels for both the noninverted and the inverted output but that needs coupling caps i try to avoid.
I played a little Mikado with my resistors and got the inamp input stage working. Distortion is low in this stage and noise is superiour low too. I supply the circuit with 2 12V Lead Gel cells and hum is simply not existing.
Tomorrow i will try the full inamp version (circuit2). Unfortunately i have no Opamps around that can drive more then 600 Ohms so the gain resistor has to be quite high at 20 Ohm and disturbs the noise a little. With an OP that can drive more current this resistor can be lowered to insignificant values of cause. Doubling the input transistors again whould bring 3dB less noise and then the law of diminishing returns sets in. 0.5nV/qHz is posible and that is very good for an inamp. Around 7dB better then the INA163. I improved the current generators a little with a Fet. Set the current to around 3mA. The Led is very low noise then.
3rd i show a posibilty to drive the input as transimpedance by gate grounding the input transitors. If that works well i have to try out.
How does the unbalanced input stage sound ? Not bad at all but when you now what output caps i use you will kill me. On the positive side i use an ages old green Röderstein MKT and on the positive side i use a Wima MKC4 that is also obsolete now.
Allen Wright will be happy because this time i start with 47kOhm loading.
Tomorrow i will try the full inamp version (circuit2). Unfortunately i have no Opamps around that can drive more then 600 Ohms so the gain resistor has to be quite high at 20 Ohm and disturbs the noise a little. With an OP that can drive more current this resistor can be lowered to insignificant values of cause. Doubling the input transistors again whould bring 3dB less noise and then the law of diminishing returns sets in. 0.5nV/qHz is posible and that is very good for an inamp. Around 7dB better then the INA163. I improved the current generators a little with a Fet. Set the current to around 3mA. The Led is very low noise then.
3rd i show a posibilty to drive the input as transimpedance by gate grounding the input transitors. If that works well i have to try out.
How does the unbalanced input stage sound ? Not bad at all but when you now what output caps i use you will kill me. On the positive side i use an ages old green Röderstein MKT and on the positive side i use a Wima MKC4 that is also obsolete now.
Allen Wright will be happy because this time i start with 47kOhm loading.
They will go tomorrow when i build the DC couple Inamp version.
Anyway, the circuit sounds great. The best combination of raw power and elegance so far.
Yes, the Rifa PHE450 for example whould the uppermost treble pass through with less coloration and loss but this circuit proves that topology matters more then exotic parts. Please do not understand me wrong. I can imagine better sound with carefully coosen components.
If it is the distortion canceling mechanism of the adjustable current sources, the super low noise or the light 47kOhm loading i do not know yet but this shows the way that this tread is not a dead end.
Anyway, the circuit sounds great. The best combination of raw power and elegance so far.
Yes, the Rifa PHE450 for example whould the uppermost treble pass through with less coloration and loss but this circuit proves that topology matters more then exotic parts. Please do not understand me wrong. I can imagine better sound with carefully coosen components.
If it is the distortion canceling mechanism of the adjustable current sources, the super low noise or the light 47kOhm loading i do not know yet but this shows the way that this tread is not a dead end.
Small mistake in post 834. I use the Röderstein on the positive side and the Wima on the NEGATIVE side. I thought that whould be better then only Röderstein on one channel and Wimas on the other. Kind of mixed dielectric idea i experimented a lot with in the early 80th. Pollin had great bagages of all kind of caps for very little money at that time and still does i think. The first 200 pairs of Audio Phyic speakers where all a little different because i used all cocktails of caps. I learned a lot too.
I did the DC and distortion trim on the High Z Fet MPP and it was successful.
I disconnected the servo and got 80mV in one channel and 130mV DC offset in the other channel initially. I could trim the DC down to +-10mV and that proved good enough to play it directly into the capacitive RIAA that has a servo inbuild.
The sound got better still and is now very fresh and natural. The resistors that supply the Leds are still in place and i will put in the Fet again to try if the bias is even more stabe then. This litte change has a lot of benefits and is a nice simplification. No low level instabilty any more and no concern about the "sound" of different servo ops.
Can that circuit be improved ? I think yes, everything can be improved. One thing is to supply the Leds with less currnt, say 3mA instead of 9mA to lower the noise from that source. The base resistors in the cascodes can be raised to 1kOHm for even better regulation. The Fets can be doubed ( 4 N channel ,4 P channel) for a theretical improvement of 3dB now close to the physical limit of 0.3nV/qHz. The cascode ransistos have to be doubled or quadrupled now, when you use the BC550C/BC560C combination because the drain resistors have to be lowered to acomodate the Gm rise.
They whould otherwise add some noise due to their 140 Ohm Rbb. You can use 2N4401/2N4403 without problem instead. The circuit i will show has all the changes with the ecception of the transistor doubling.
One small thing: the bias trim is very sensitive. I could reduce the DC of 130mV in one channel with less of a turn of a 1kOhm 10 gang trimmer. That ergonomic problem can be reduced if you use a resistor with a lower value, say 330 Ohm and use a pot with a higher value ( something between 2 and 5kOhm.
I trimmed the offset only on the positive side because usually N channel fets have higher transconductance then P channel and it is easier to adjust only one pot instead of two.
Make sure that the trimmer is aproximately in the middle position before you insert it or otherwise a big current could flow when the trimmer is near zerro.
I disconnected the servo and got 80mV in one channel and 130mV DC offset in the other channel initially. I could trim the DC down to +-10mV and that proved good enough to play it directly into the capacitive RIAA that has a servo inbuild.
The sound got better still and is now very fresh and natural. The resistors that supply the Leds are still in place and i will put in the Fet again to try if the bias is even more stabe then. This litte change has a lot of benefits and is a nice simplification. No low level instabilty any more and no concern about the "sound" of different servo ops.
Can that circuit be improved ? I think yes, everything can be improved. One thing is to supply the Leds with less currnt, say 3mA instead of 9mA to lower the noise from that source. The base resistors in the cascodes can be raised to 1kOHm for even better regulation. The Fets can be doubed ( 4 N channel ,4 P channel) for a theretical improvement of 3dB now close to the physical limit of 0.3nV/qHz. The cascode ransistos have to be doubled or quadrupled now, when you use the BC550C/BC560C combination because the drain resistors have to be lowered to acomodate the Gm rise.
They whould otherwise add some noise due to their 140 Ohm Rbb. You can use 2N4401/2N4403 without problem instead. The circuit i will show has all the changes with the ecception of the transistor doubling.
One small thing: the bias trim is very sensitive. I could reduce the DC of 130mV in one channel with less of a turn of a 1kOhm 10 gang trimmer. That ergonomic problem can be reduced if you use a resistor with a lower value, say 330 Ohm and use a pot with a higher value ( something between 2 and 5kOhm.
I trimmed the offset only on the positive side because usually N channel fets have higher transconductance then P channel and it is easier to adjust only one pot instead of two.
Make sure that the trimmer is aproximately in the middle position before you insert it or otherwise a big current could flow when the trimmer is near zerro.
I started to build the discreet INA stage but today is sunday so i stopped for today and dicided to listen a bit to the trimmed High Z MPP.
Soundwise it´s now as good as the Low Z version. A lot of energy and resolution. I had no desire to change the Tantal 1kOhm loading resistor.
My son and i will look Avatar tonight so i see you tomorrow.
Soundwise it´s now as good as the Low Z version. A lot of energy and resolution. I had no desire to change the Tantal 1kOhm loading resistor.
My son and i will look Avatar tonight so i see you tomorrow.