MPP

Joachim

I took a closer look to your circuit and I'm not shure if it works correctly. When you inject about 40ma through T6, T3 must source the same amount of current. This current has to be delivered by R1/T7 from the positive power supply. As you are using a current mirror for your jfet the current is limited to the LTP current source from T4/T5. With 180 ohms on the emitter this current source will deliver 3.5ma max. In this way I think T6 will saturate. In addition (when the current for T6 is less than T4) the jfet diff amp is never balanced as the additional current for T6 has always delivered by T7.

Just my 2 ct.
 
Last edited:
Yes, I think this op-amp should work fine. You don't need a lot of open loop compensation. My folded cascodes work fine with as little as 47pF for C3, but without this cap the amp oscillates. You should use a little more current for T3/T10, maybe 5ma in this case (R1/R4 = 220R). In this way both transistors are always class-A biased.

In addition you can drive the LSK389C much harder. When you set the current to 20ma (R1 = 33R) or so and increase the current for T3/T10 to 20 - 30ma (R1/R4 = 39R) I think you can omit the output buffer. Although the JFETs are slightly forward biased under worst condition in this way, I wouldn't expect any problems. Maybe John Curl knows more about this.

When doing so I'd also change T3, T10, T12, T13 and T14 to BD139 / BD140 or similar, as the power dissipation rises to about 0.3 - 0.5w for each transistor. A little bit too much for the BC550 / BC560. This is also the reason for T12. When you omit this transistor T3 dissipates twice the power of T10 so slight imbalances can occur when warmed up.
 
Equivalent input noise current

Hi Bocka
I'm working on something similar as I/V stage for a DAC.
Did you consider the noise generated by R1, R2, R17 & R18 in your I/V stage?

Since the cascode and folded cascode stage have a current gain of 1x, these resistors effectively appear parallel to the input signal of the amplifier. The combined parallel resistance is only 34 Ohm [1/(2/100+2/220)], resulting in 40nV/sqrtHz at the output of the amplifier (R_IV=1.8K). From my own simulation experiments I found it is beneficial to have huge voltage degeneration for those current sources (>1kOhm), using higher rail voltages (e.g. 15V to 20V).

PS: this is a great thread, a source of inspiration.
 
Last edited:
I asume you mean R2 = 33 Ohm and not R1 = 33 Ohm ?
Is there enough voltage gain in the Fet stage left when R1/R4 are as low as 39 Ohm ?
Thanks again for good sugestions. This way the project goes forward very fast.
I am very happy about that. Another Diy fellow is doing a Microcap simulation until tomorrow. This is a great forum with a lot of knowledgeable people here.
Could i also use 2N4401 / 4403 instead of BC550C/BC560C. They are a bit more robust and i find the BD139/140 kind of big in terms of size. Of cause you already recommended similar devices in smaller packages.
 
I asume you mean R2 = 33 Ohm and not R1 = 33 Ohm ?
Yes, that's a typo.
Is there enough voltage gain in the Fet stage left when R1/R4 are as low as 39 Ohm ?
As the jfet drives a common base transistor the input inpedance of this cascode is only 2 ohms or so. But your right, 39 ohms looks very low on a first view. Maybe a current source is better here. I don't know, maybe it's worth trying both variants. More LEDs are worth trying, too.

The 2N4401/4403 looks fine especially when driving currents in the range of 30ma. Seems to be a good choice.

@jeepee
No, I did't make a noise analysis yet, as my posted circuit is currently conceptional only. I did it some days ago and it's not tested. It has some other weakness, too. PSRR is not optimal. But I think this can be solved. But you're right, 40nV/sqrtHz is a little bit high.
 
Yes, that's a typo.
As the jfet drives a common base transistor the input inpedance of this cascode is only 2 ohms or so. But your right, 39 ohms looks very low on a first view. Maybe a current source is better here. I don't know, maybe it's worth trying both variants. More LEDs are worth trying, too.

The 2N4401/4403 looks fine especially when driving currents in the range of 30ma. Seems to be a good choice.

@jeepee
No, I did't make a noise analysis yet, as my posted circuit is currently conceptional only. I did it some days ago and it's not tested. It has some other weakness, too. PSRR is not optimal. But I think this can be solved. But you're right, 40nV/sqrtHz is a little bit high.

Thanks again Bocka. I will start to build something and try to find out the best solution.
 
As my circuit is not phono related I've started a new dac i/v thread here. This should it make easier to seperate the discussion.

@Joachim
I think you will also have some noise issues when you lower the values as I've suggested. The more voltage you have across the emitter resistors of the current sources or the current mirror the better. jeepee is right here. In this way I'd choose the LED string.
 
I am building a preamp for a friend. For many years he was listening to CD and never wanted back to Vinyl. Until the day i brought my turntable and phonostage to his home.
Now he gets a preamp with High Z Capacitive MPP, Tentlabs volume control ( he wanted remote) and PSG buffer. That buffer improved the sound of his AudioAero CD player a lot.
Eather someting is wrong with the digital volume control on that player or the PSG adds some extremely pleasing distortion. We cranked the volume control on the CD player full and then added the buffer with a conventinal TKD pot as an experiment and he was blown away. Anyway for his Pre i choose the Placid PSU from Twisted Pear. Actually this is my first experience of a Shunt PSU on my equipment and i tested it out on my Headphone Amp. Compared to the LC Audio with inductor i heard an improvement but it was not night and day. The Headphone Amp has superb PSU rejection and even with simple 9V blocs there is no hiss or humm at all. The placid had a slightly softer sound i prefered because most CDs are a bit hot and compressed to say the least. The Placid made CDs more pleasant to listen too. Quiet pasages where more quiet so to say and i heard some micro details that escaped me with the LC Audio PSU. Bass was on the other hand was more clear and detailed and even a bit tighter and extended.
See also a picture of my frieds Turntable that i have around for adjustment. It´s a Clearaudio with magnetic bearing and for 2000,-€ you get it with a good arm and cartridge. I substituted the cartridge with my Lyra Helicon and will that replace again when the new Delos comes. It sound really great and i was surprised how little the difference to my 35.000 € monster is. Well the little Clearaudio does not reach the mastertape precission of my setup but it is very dynamic and alive sounding and makes lots of fun. Rumble is extremely low courtesy of the magnetic bearing and speed stabilty is good. I have it sitting on a simple Ikea table but touching the surface with my hand, vibration was very low. That simple silicon feet that are fitted work exceptionally well.
 

Attachments

  • Placid PSU.jpg
    Placid PSU.jpg
    97.9 KB · Views: 609
  • Clearaudio Turntable.jpg
    Clearaudio Turntable.jpg
    97.2 KB · Views: 615
With the help of Sigurd we developed the discrete Opamp further. The first version is ready and the results are very good. Open loop bandwidth is 40kHz (-3B), gain is 54dB,
distortion is under 0.01% up to 10kHz and around 0.02% up to 100kHz (open loop).
We call it JG OPamp 2.21. I can be usefull for many things. Linestage, filter etc.
I will try it as the second stage in my High Z Capacitive MPP.
We are working on a second version, working title 3.1, and will decide later what to use.
 

Attachments

  • JG_Discrete_OPamp_ver_2[1].21_REV_A.pdf
    13.9 KB · Views: 198
  • JG_Discrete_OPamp_ver_2[1].21_THD_REV_A.pdf
    19 KB · Views: 148
Here is a simulation of the High Z Capacitive MPP and the Ina.
Values are mathematically correct. If you take the closest standard values and 1% parts
precission "detorientates" to +- 0.1dB.
 

Attachments

  • Claudio_with_Extremely_Accurate_Freq_Response__REV_B.pdf
    76.4 KB · Views: 133
  • High_Z_MPP_FET_DC_Offset_Fine_Adjust_Capacitor_RIAA-2.pdf
    69.6 KB · Views: 169
Today the stuffed adaptor boards with the LME49990 arived from Stephan Horwege so i was able to build the Balanced INA input for the JG Self. I wish Holger whould be here.
The slightly rough and ready sound of the simple Phonoclone is gone replaced by a much more smooth and liquid sound with great bass and very good micro detail resolution.
This is not the lowest noise phonostage i ever designed. The discrete circuits i have shown here have the edje but up to medium high volume noise was not audible at the listening seat with my 96dB efficient MPL. At very high volume it was slightly audible but had a very "benign" quality with very little agressive treble content. This latest generation Opamps come fritening close in sound to discrete circuits so i am even more motivated to come up with a superiour discrete Opamp. One advantage of the INA topology is DC precission and i measured only 2.5mV at the output of the INA. A Servo is not necessary i think. Before the subsonic filter i am measuring 600mV now instead of 270mV and i think that is still fine. i decoupled all three Opamps in the INA with one pair of ELNA Silmic 2 330uF 25V and put only one 0.1uF Roederstein MKP from PIN 4 to Pin 7 of each Opamp. I talked about that in the "Best sounding OPamp" thread but got so much critique that i sopped to post on that thread for a while. Both the LME49990 and the ADA4627 are very fast but i have no troube with oscilation at all.
See the picture. You see the 3 0.1uF MKP´s ( blue rectangles) directly soldered over
the OPs at the input. I left the Phonoclone in for the ocasion that somebody wants to here the difference.
One other notewothy thing. Allen Wright of Vacuum State will visit me on sunday. He brings some tube equipment, so this will be interseting to compare with what i have.
 

Attachments

  • JG Self INA.jpg
    JG Self INA.jpg
    99.8 KB · Views: 562
I am very happy with the sound of the Balanced INA i added to the JG-Self. It sounds much more natural then the simple Phonocone i used before. The Phonoclone is very dynamic and direct sounding but lacks liquidity and is a little stessfull sounding over the long run. I will convert that stage into a simple balanced input and see if that sounds better.
I also redesigned the linestage. I raised the current in the buffer after Sigurd made some simulations to gain linearity and i will add a Mono-Stereo switch. I have separate pot meters for left and right , so this will help me to find the center position more easily. Sigurd and i have more or less decided on the topology for the discrete OPamp but do not expect something new until next week. I really goes into the details now.
 

Attachments

  • DSJG Simple Balanced Input.TSC - TINA.pdf
    39.6 KB · Views: 172
  • JG Self Linestage Mono-Stereo.TSC - TINA.pdf
    61.2 KB · Views: 171
OPAmp's

Hi Joachim,


I ordered the Opamp's and as soon as they arrive I'll start building your RIAA and compare it to my own passive phono amp.

I was wondering what the DC resistance is of the inductors after the LC Audio PSU you use. If I remember correctly it was a 27 mH inductor?

I use a J-fet as a MC head amp cause my phono is a MM type.
With tube's! I've used a SUN Audio transformer befor but this J-fet is faster than the SUN sound wise.

I'll read your posts tomorrow, it's late again :eek:

All the best.


Audiofanatic ;)
 
Riaa

Hi Joachim,


I meant the DSJG.TSC The one with OPA1642 and LSK 389 and LME 49713.
The reason I want to try it is because I never liked OPAmp's in the past.
But looking at the specs of the new models I think it's time to be a bit more tolerant to those little creechurs ;-)

I'll order also a LC Audio PSU and start building your RIAA.

I'll report later when I'm done.

Thanks again for sharing.

Cheers,


José

Audiofanatic ;)