Hi!
The Turbo can play starting à 58V. Works fine here with 63V caps but the toroid should be well calculated to stay around 60V.
Or, you can use 80V caps if you are afraid.
For C15/C16: Mouser 594-MAL225632222E3
The Mini can also be placed horizontally as in the photo but I'm not sure I understood your explanation..
Stef.
The Turbo can play starting à 58V. Works fine here with 63V caps but the toroid should be well calculated to stay around 60V.
Or, you can use 80V caps if you are afraid.
For C15/C16: Mouser 594-MAL225632222E3
The Mini can also be placed horizontally as in the photo but I'm not sure I understood your explanation..
Stef.
Attachments
Hi Stef, how "strict" is this statement? I thought I could use my 4x36 VAC toroid + Saligny rectifiers (approx. +/-51V) also for the Q17 Turbo in my modular amplifier design.
Hello everyone,
I've spent a significant amount of time this summer delving into the circuitry of the Q17, reevaluating all the components thoroughly. Building on that, I've constructed several versions, tweaking numerous values until achieving an excellent sound.
The current version exhibits a slight sonic difference compared to my earlier builds, with the acoustic stage gaining precision. Overall, the new setup offers advantages in reproducing complex music, particularly noticeable in drum solos where individual beats are portrayed with greater precision and less softness.
Another noteworthy aspect is the implementation of an amplifier whose power stage doesn't require a downstream L-C filter. This enhances speaker control, proving quite impressive. Significant differences emerge, especially with highly audiophilic speakers possessing substantial acoustic potential but presenting a complex load (due to crossovers without impedance corrections).
I can now appraise Tibi's recent ideas regarding the Q17, meticulously implemented by Stef, with reserved enthusiasm. The circuit requires considerable adjustments to play with the same ease and naturalness. While it might appeal to enthusiasts of exotic SMD components, it doesn't necessarily outshine the old 2SC1845. I entertain the idea that the cascode needs fundamental reworking, possibly in combination with BJT.
The changes I made, such as altering the R-C filter behind the operational amplifier, along with other details, all serve the purpose of stabilizing the amplifier and minimizing the risk of oscillation. I cannot publish measurements representative of the amplifier, as the results indicate that, in the low-frequency range, the amplifier has no measurable impact on my measurement chain.
Best regards, Tim
I've spent a significant amount of time this summer delving into the circuitry of the Q17, reevaluating all the components thoroughly. Building on that, I've constructed several versions, tweaking numerous values until achieving an excellent sound.
The current version exhibits a slight sonic difference compared to my earlier builds, with the acoustic stage gaining precision. Overall, the new setup offers advantages in reproducing complex music, particularly noticeable in drum solos where individual beats are portrayed with greater precision and less softness.
Another noteworthy aspect is the implementation of an amplifier whose power stage doesn't require a downstream L-C filter. This enhances speaker control, proving quite impressive. Significant differences emerge, especially with highly audiophilic speakers possessing substantial acoustic potential but presenting a complex load (due to crossovers without impedance corrections).
I can now appraise Tibi's recent ideas regarding the Q17, meticulously implemented by Stef, with reserved enthusiasm. The circuit requires considerable adjustments to play with the same ease and naturalness. While it might appeal to enthusiasts of exotic SMD components, it doesn't necessarily outshine the old 2SC1845. I entertain the idea that the cascode needs fundamental reworking, possibly in combination with BJT.
The changes I made, such as altering the R-C filter behind the operational amplifier, along with other details, all serve the purpose of stabilizing the amplifier and minimizing the risk of oscillation. I cannot publish measurements representative of the amplifier, as the results indicate that, in the low-frequency range, the amplifier has no measurable impact on my measurement chain.
Best regards, Tim
Attachments
Hi,
You can put just about anything like opamp. I even tried the old OPA627 and it works. I did not notice any oscillation of the circuit. On the other hand, when you compare an OPA1611 with an OPA134 for example, you can really hear the difference. We lose definition and the grain of the OPA134 is really quite unpleasant.
Regards,
Stef.
You can put just about anything like opamp. I even tried the old OPA627 and it works. I did not notice any oscillation of the circuit. On the other hand, when you compare an OPA1611 with an OPA134 for example, you can really hear the difference. We lose definition and the grain of the OPA134 is really quite unpleasant.
Regards,
Stef.
Electrically evaluated, the TL071 is still the most suitable for the Q17 from your selection. With this operational amplifier you should not have any problems with the Q17Mini. I think you can think about building a dedicated op-amp, it might be cheaper than buying an OPA1611 in Turkey.
here is a guide for the building of dedicated operational amplifiers for the hifi nerd:
http://diy.ucborgmann.de/index.php/...te/opb3-diskreter-audio-operationsverstaerker
Last edited:
Hello everyone,
I would like to discuss the recent changes to the Q17 and derive ideas for optimization. Originally, a current mirror was formed with Q9 and Q11, with its input current depending on Q12 and R24+R25, regulating the current for the cascode. Thus, it was a feedback correction. The variant from this summer (by Tibi), accurately reproduced by Stef, changed this by replacing the simple current mirror with a Wilson current mirror.
So far, nothing new.
Now, this Wilson current mirror has an impact on the current for the cascode and the currents in the circuit of Q5+Q6. Therefore, it is understandable that Tibi reduced the resistors R10 and R13 to 9.1 ohms. However, the adjustment of R3 is missing. In the original concept with a simple current mirror, R3 was chosen quite ideally at 150 ohms. In the concept with the Wilson current mirror, this value needs to be reduced to the order of 100 ohms for the circuit to work well-balanced again.
Since the circuit is a Class A feedback correction with Q5 and Q6, generating higher impulses when driving two pairs of Mosfets (Q15, Q16, Q17, and Q18 - in Stef's Turbo), there is increased instability in terms of oscillation behavior. I've dampened the oscillation behavior a bit by adding a resistor (R34) to C10 – this suggests that the mica capacitor without a resistor is not perfect (I can demonstrate this in the Spice simulation) and that a high-frequency capacitor with high ESR – like the cheap ceramic capacitors – would be more suitable in this position.
Additionally, I'm damping high-frequency oscillations of the feedback path (Q12 – R24+R25) by adding an additional damping element (R35 + C20) compared to the central feedback (through R27).
Now, in this context, I have determined that R3 is still too high in my latest version (#2,204) and should be adjusted.
During these investigations, I explored the question of which alternatives to Q12 can have a positive influence. In the simulation, I picked up the possibility of 2N5551 and found a solution with a high base resistor (R9). Since I want to keep the quality of the components as high as possible, I chose the ZTX458 as an alternative to 2N5551 (as Q12). As the simulation of the Wilson current mirror behaves little positively compared to the simple current mirror, I also simulated a double current mirror with P-MosFET. Now there is the idea of building all components of the Wilson current mirror, including Q10, with ZTX558. This results in a simulation result that is even more precise than with P-Mosfets.
This adjustment allows a lowering of resistors R11 and R12, making the power stage in Class B almost error-free at the zero crossing.
The reason for this investigation is provided by the Hafer interpretation: https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7493521
In this amplifier, the Class B power stage also works very precisely, which the Q17 in its current form does not. However, the described circuit works with extremely small signals, which reduces my enthusiasm. Therefore, I consider it more promising to further optimize the Q17 circuit.
What do you think?
Best regards, Tim
I would like to discuss the recent changes to the Q17 and derive ideas for optimization. Originally, a current mirror was formed with Q9 and Q11, with its input current depending on Q12 and R24+R25, regulating the current for the cascode. Thus, it was a feedback correction. The variant from this summer (by Tibi), accurately reproduced by Stef, changed this by replacing the simple current mirror with a Wilson current mirror.
So far, nothing new.
Now, this Wilson current mirror has an impact on the current for the cascode and the currents in the circuit of Q5+Q6. Therefore, it is understandable that Tibi reduced the resistors R10 and R13 to 9.1 ohms. However, the adjustment of R3 is missing. In the original concept with a simple current mirror, R3 was chosen quite ideally at 150 ohms. In the concept with the Wilson current mirror, this value needs to be reduced to the order of 100 ohms for the circuit to work well-balanced again.
Since the circuit is a Class A feedback correction with Q5 and Q6, generating higher impulses when driving two pairs of Mosfets (Q15, Q16, Q17, and Q18 - in Stef's Turbo), there is increased instability in terms of oscillation behavior. I've dampened the oscillation behavior a bit by adding a resistor (R34) to C10 – this suggests that the mica capacitor without a resistor is not perfect (I can demonstrate this in the Spice simulation) and that a high-frequency capacitor with high ESR – like the cheap ceramic capacitors – would be more suitable in this position.
Additionally, I'm damping high-frequency oscillations of the feedback path (Q12 – R24+R25) by adding an additional damping element (R35 + C20) compared to the central feedback (through R27).
Now, in this context, I have determined that R3 is still too high in my latest version (#2,204) and should be adjusted.
During these investigations, I explored the question of which alternatives to Q12 can have a positive influence. In the simulation, I picked up the possibility of 2N5551 and found a solution with a high base resistor (R9). Since I want to keep the quality of the components as high as possible, I chose the ZTX458 as an alternative to 2N5551 (as Q12). As the simulation of the Wilson current mirror behaves little positively compared to the simple current mirror, I also simulated a double current mirror with P-MosFET. Now there is the idea of building all components of the Wilson current mirror, including Q10, with ZTX558. This results in a simulation result that is even more precise than with P-Mosfets.
This adjustment allows a lowering of resistors R11 and R12, making the power stage in Class B almost error-free at the zero crossing.
The reason for this investigation is provided by the Hafer interpretation: https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7493521
In this amplifier, the Class B power stage also works very precisely, which the Q17 in its current form does not. However, the described circuit works with extremely small signals, which reduces my enthusiasm. Therefore, I consider it more promising to further optimize the Q17 circuit.
What do you think?
Best regards, Tim
Attachments
Hello!
I need help...
I built a Q17 turbo 1.2.2.
If I connect the supply voltage (power supply +- 50V current limit set to 1A) then Q16 and Q18 quickly get hot. The LEDs don't light up either.
So I have a mistake somewhere...
Q16, Q18, Q6, Q14 tested.
Does anyone have an idea for me as to what I should check?
I need help...
I built a Q17 turbo 1.2.2.
If I connect the supply voltage (power supply +- 50V current limit set to 1A) then Q16 and Q18 quickly get hot. The LEDs don't light up either.
So I have a mistake somewhere...
Q16, Q18, Q6, Q14 tested.
Does anyone have an idea for me as to what I should check?
Hello!
I checked it.
Very puzzling. Because I have installed the same board twice and have the same symptom.
So there must be something fundamentally wrong...
I just can't find anything yet.
Q1, Q13,Q9, Q10,Q3,Q13,Q5,Q6 all unsoldered and measured, all good
U1 and U2 swapped...
I checked it.
Very puzzling. Because I have installed the same board twice and have the same symptom.
So there must be something fundamentally wrong...
I just can't find anything yet.
Q1, Q13,Q9, Q10,Q3,Q13,Q5,Q6 all unsoldered and measured, all good
U1 and U2 swapped...
