I played a little with P3, but did not go far away: max one turn from the center position. That hardly had any influence on the H2 in the distortion graph, as measured with an Akitika 1kHz source, a Focusrite sound card and REW software. Both channels, L and R, have the same harmonic structure within a few dB from each other, so matching seems to be OK.
I started doubting my measurements, so I measured a Schiit Aegir for comparison, and that showed very low H2 (< -90 dB) as specified by Schiit. That reassured me my measurements were not ********.
I guess I can play a little further from the center positions with P3. After all, it is a 200 Ohm pot in parallel to a 2*10 Ohm resistor, so effects should certanly start showing at a few turns from the end, so over 5 turns from the center (25 turn pot). Any experience with the number of turns of P3 to really have an impact on H2 magnitude?
I started doubting my measurements, so I measured a Schiit Aegir for comparison, and that showed very low H2 (< -90 dB) as specified by Schiit. That reassured me my measurements were not ********.
I guess I can play a little further from the center positions with P3. After all, it is a 200 Ohm pot in parallel to a 2*10 Ohm resistor, so effects should certanly start showing at a few turns from the end, so over 5 turns from the center (25 turn pot). Any experience with the number of turns of P3 to really have an impact on H2 magnitude?
You can get H2 down to -90dB quite easily using P3. Needs about 5-6 turns, and from memory the degeneration across the N-JFET has to be reduced. It will start rising a bit after a point, and my best results are achieved when matching the distortion profile of both channels rather than the lowest I can absoluitely get. I don't use matched JFETs so it's a little more complex for me than most. H3 is unaffected (it actually does go down by a dB or two, but not significantly).
Thanks you very much for the tip, Sangram! I just did the optimization you described, and indeed found the best settings around 5 turns to the left of the middle position, so around 1/3 of the total pot range. That implies that the degeneration/feedback resistors R3 and R4 now virtually are 9.4 and 8.7 Ohm. I do use matched JFETs, that could be the cause that I could go there so easily. In my case, H3 also went down. The attached graph gives the distortion spectrum of one channel (the other channel is almost identical). H2 is indeed around -80 dB, H3 is -105 dB, higher harmonics sink in the noise at -120 dB (probably determined more by my measurement tool than by the amp). 50 Hz is at -105 dB, harmonics of 50 Hz are at -100 dB. Not bad I think. The data are taken at around 1 W output power, and the distortion remains fairly low up to 50 Watt in 8 Ohm (the Zu Audio Omen Def Supreme).
i was considering entering another build, the F5X. But now I am not sure anymore it is really worth the effort. It would bring balanced operation (which I actually prefer over SE) and further suppression of amp-related even harmonics, but I suspect that the speaker actually could be a dominant source of H2, overshadowing the amp. Therefore my next excursion will be to measure the harmonics behavior of my speakers, using the Akitika 1 kHz source at low level and an amp without measurable harmonics the Aegir or a Hypex amp I have).
i was considering entering another build, the F5X. But now I am not sure anymore it is really worth the effort. It would bring balanced operation (which I actually prefer over SE) and further suppression of amp-related even harmonics, but I suspect that the speaker actually could be a dominant source of H2, overshadowing the amp. Therefore my next excursion will be to measure the harmonics behavior of my speakers, using the Akitika 1 kHz source at low level and an amp without measurable harmonics the Aegir or a Hypex amp I have).
Hello,
I am assembling a PASS F5-Turbo stereo 50watt class A, reading the list of components, I read that in the FRONT-END board, resistors R25 - R30, capacitors C5, C6 and transistors Q7 and Q8, are to be used only when a cascoding configuration is desired, however in many videos or photos I have noticed that these components are used, what do you recommend, to use them or not?
The boards I use are the ones I bought from DiYaudiostore.com, Gain-Stage board V2.2, P-Channel and N-Channel are V3.0.
Thanks to everyone who answers the question.
I am assembling a PASS F5-Turbo stereo 50watt class A, reading the list of components, I read that in the FRONT-END board, resistors R25 - R30, capacitors C5, C6 and transistors Q7 and Q8, are to be used only when a cascoding configuration is desired, however in many videos or photos I have noticed that these components are used, what do you recommend, to use them or not?
The boards I use are the ones I bought from DiYaudiostore.com, Gain-Stage board V2.2, P-Channel and N-Channel are V3.0.
Thanks to everyone who answers the question.
Thank you for your reply Sangramma,
considering that the DC current is +- 32V, I assume that I must adopt a cascoding configuration that if I then only assemble a 50WATT stereo configuration, or perhaps I must mean that the 25V you mentioned refers to the AC current generated by the transformer.
I don't think I understand your thinking correctly.🤔
Thanks
considering that the DC current is +- 32V, I assume that I must adopt a cascoding configuration that if I then only assemble a 50WATT stereo configuration, or perhaps I must mean that the 25V you mentioned refers to the AC current generated by the transformer.
I don't think I understand your thinking correctly.🤔
Thanks
Current is in amperes, not volts.
Is +- 32V the rail voltages that you're trying to achieve? If so, then it would be very strongly advised to cascode.
No.
Just a small misunderstanding, perhaps. However, you do seem to either be confusing two separate things and/or not understanding overall.
This may help tremendously. Have you read, and more importantly, do you understand some of the fundamentals of this article?
https://www.firstwatt.com/pdf/art_f5_turbo.pdf
The article should have most if not all your answers. Between that and 6L6s guide, it's fantastic. Please continue to ask questions as needed.
Thank you for your reply and clarification, I will try to be less clumsy in future questions.
Thanks.
Rail always refers to the supply voltage for the amplifier, and usually denotes DC.
For a standard F5/turbo, the JFET has a little less than the rail voltage across it, about 3-4V less. The P-channel JFET has a breakdown voltage of 25V, hence the maximum you can go to without cascode is about 29V if you leave zero safety margin.
These are now expensive, rare and difficult to find, so you should treat them with care. I prefer running them at 12V or below personally, and a cascode has a lot of important sonic and reliability benefits that I wouldn't ignore. This means that even for low rail voltages (20V onwards) I choose to cascode these devices.
For a standard F5/turbo, the JFET has a little less than the rail voltage across it, about 3-4V less. The P-channel JFET has a breakdown voltage of 25V, hence the maximum you can go to without cascode is about 29V if you leave zero safety margin.
These are now expensive, rare and difficult to find, so you should treat them with care. I prefer running them at 12V or below personally, and a cascode has a lot of important sonic and reliability benefits that I wouldn't ignore. This means that even for low rail voltages (20V onwards) I choose to cascode these devices.
Considering the algorithm given at the bottom of the BOM for F5 Turbo , cascode voltage = (rail voltage / (R26 + R28)) * R28 and using R26 and R28 as resistor values = 4.75Kohm
we obtain a cascode voltage = (32V /(4.75K +4.75K)) * 4.75K = 16.2V.
I ask another question, but should transistors Q7 = Tip31 and Q8=Tip32 be used?
If the answer is no, should I jumper C-E (2-3)?
Thanks for the technical support.
The ratio of the two cascode resistors determines the voltage drop across the cascode.
The current through the tail should be between 1 and 2mA, more is not necessary and simply wastes power.
Therefore the rail voltage is multiplied by 1000 to find the total of the two resistors, for a 32V rail this is 32k.
To achieve the <12V JFET D-S voltage, we arrive at 10k and 22k as the cascode values, this gives 1mA thru the tail and a very handy 8-9V across the JFET. This also meas you can run it at a nice high current in order to get better sonic results. You could also use two 15k resistors for 50% rail voltage, totally your call.
The current through the tail should be between 1 and 2mA, more is not necessary and simply wastes power.
Therefore the rail voltage is multiplied by 1000 to find the total of the two resistors, for a 32V rail this is 32k.
To achieve the <12V JFET D-S voltage, we arrive at 10k and 22k as the cascode values, this gives 1mA thru the tail and a very handy 8-9V across the JFET. This also meas you can run it at a nice high current in order to get better sonic results. You could also use two 15k resistors for 50% rail voltage, totally your call.
@Sangram,
Reading the last 10-20 pages of this thread, one question that came up for me is what would be the detriment to using a completely separate power supply that was a lower voltage for the FE board and leave the big CRC supply for just the output stage? The implementation of cascoding the JFETS was to insulate it from the higher voltage rails (>25V as many have mentioned) as well as any power supply perturbations that may be occurring. A separate power supply for the FE would solve the issue?
I have a feeling that the supply voltage to the JFET front end and MOSFET output stages needs to be same for the design itself but I am not sure why that is so. I am also thinking that this will involve breaking traces and such. Perhaps I don’t see it since I don’t have F5 Turbo boards in front of me (it’s time to start shopping! 😍 ).
Perhaps including the schematic to this post will help from a teaching standpoint:
Note, R26 is 10K ohm of course.
As always, thanks for the tutelage and guidance.
Best,
Anand.
Reading the last 10-20 pages of this thread, one question that came up for me is what would be the detriment to using a completely separate power supply that was a lower voltage for the FE board and leave the big CRC supply for just the output stage? The implementation of cascoding the JFETS was to insulate it from the higher voltage rails (>25V as many have mentioned) as well as any power supply perturbations that may be occurring. A separate power supply for the FE would solve the issue?
I have a feeling that the supply voltage to the JFET front end and MOSFET output stages needs to be same for the design itself but I am not sure why that is so. I am also thinking that this will involve breaking traces and such. Perhaps I don’t see it since I don’t have F5 Turbo boards in front of me (it’s time to start shopping! 😍 ).
Perhaps including the schematic to this post will help from a teaching standpoint:
Note, R26 is 10K ohm of course.
As always, thanks for the tutelage and guidance.
Best,
Anand.
I don't see any achievement from this connection except for the possibility of a quieter front end by using regulated supplies, but the cascode is already improving PSRR significantly.
Besides, the entire drive voltage for the output is generated by the drop across R5/6 and P1/2, which means you have to couple them somewhere even if the supplies start off separately.
Besides, the entire drive voltage for the output is generated by the drop across R5/6 and P1/2, which means you have to couple them somewhere even if the supplies start off separately.
Hello,
an 800VA transformer with 25V secondary outputs should not cause any problems for an F5 Turbo V 2.
I decided to do it in cascode configuration, with cascode voltages between 10V and 11V, I used 10K resistors for R25 and R26 and 4K75 for R28.
Assuming a track voltage of max 35V, I will have a cascode voltage of about 11V.
Can anyone confirm my assertions?
Thank you.
an 800VA transformer with 25V secondary outputs should not cause any problems for an F5 Turbo V 2.
I decided to do it in cascode configuration, with cascode voltages between 10V and 11V, I used 10K resistors for R25 and R26 and 4K75 for R28.
Assuming a track voltage of max 35V, I will have a cascode voltage of about 11V.
Can anyone confirm my assertions?
Thank you.
I can confirm, tricksters. In my F5T I also have 25 V transformer outputs, in my case at 1000 VA, and after the CRC my rail voltage is 30.5 V. I have also chosen the cascode, using TIP31 and TIP32. I used 10 k resistors for R25-28, so my cascode voltage is around 15 V. But cascode voltage 10-11 Volt will also work.
Hello! I am SO sorry if this has been asked a zillion times already...but we are on 81 daunting pages....
I will be building a stereo F5T which will be converted to monoblock when future funding permits a second set of parts. So, I'll go ahead and install the cascode in the FE. The boards and FETs have been ordered. Time for the power tranny. There are many options when it comes to rail VDC. My question is, is it sonically better to run a quite-high rail voltage and keep the bias current low? Or is it sonically better just to bump up the rail voltage slightly (36-40VDC) and keep the mosfets biased high? I want that classic full Class A sound and will be heatsinking appropriately either way. Thanks!
I will be building a stereo F5T which will be converted to monoblock when future funding permits a second set of parts. So, I'll go ahead and install the cascode in the FE. The boards and FETs have been ordered. Time for the power tranny. There are many options when it comes to rail VDC. My question is, is it sonically better to run a quite-high rail voltage and keep the bias current low? Or is it sonically better just to bump up the rail voltage slightly (36-40VDC) and keep the mosfets biased high? I want that classic full Class A sound and will be heatsinking appropriately either way. Thanks!