Not sharing for obvious reasons 
It was a hodge-podge of some of the ideas presented in the first thirty pages of the thread, plus our own guesswork for the PFB addition. We had to prove the design in simulation, as there's no proper schematic. The best way to describe what I got finally was a mix of the F5 with the NFB from the Plantfeve design. The PFB was a tough nut to crack and I hope to compare it against Nelson's if/when he releases his. I don't think it's 100% there but it works in both the sim and real life, and that's what matters.
Nelson said it would be more fun this way, and it was. Of course, the PCBs had a big boo-boo that needed manual correction with a rotary engraver - I added the cascode and took the gate drive from the emitter instead of the collector, which resulted in total destruction of most of the parts on the PCB. We were using hand-match Renesas MOSFETs, but after multiple issues with consistency switched to the Exicon dual-die devices.
It was a hodge-podge of some of the ideas presented in the first thirty pages of the thread, plus our own guesswork for the PFB addition. We had to prove the design in simulation, as there's no proper schematic. The best way to describe what I got finally was a mix of the F5 with the NFB from the Plantfeve design. The PFB was a tough nut to crack and I hope to compare it against Nelson's if/when he releases his. I don't think it's 100% there but it works in both the sim and real life, and that's what matters.
Nelson said it would be more fun this way, and it was. Of course, the PCBs had a big boo-boo that needed manual correction with a rotary engraver - I added the cascode and took the gate drive from the emitter instead of the collector, which resulted in total destruction of most of the parts on the PCB. We were using hand-match Renesas MOSFETs, but after multiple issues with consistency switched to the Exicon dual-die devices.
I have done some preliminary search on the F7 topic and found a fair bit of info on the web, especially from a few Japanese sites. I'm looking for something that could potentially keep me occupied for a while, and this amp looked promising with feedback combinations/mucking around with input impedance values... etc. Also, I like the simplistic approach... Everything Should Be Made as Simple as Possible, But Not Simpler
Thanks for the pointers you provided
Thanks for the pointers you provided
Last edited:
After three years of lack of music due to a very unpleasant deafness after an explosion occurred in the rustic house, my ear deformed the musical notes from the high D, third octave of my flute, it was heard as an E flat out of tune. I was slightly hypoacusic but without deformation. I've gone back to Bruckner and Tony Levin and am bringing back the amps we had built, the F7 really does sound very balanced and distortion free. Little by little I have submerged myself in DIY and above all to investigate again in the column of the teacher Nelson Pass. Discovering that so much knowledge was warming up and that the people and the teacher continued with their work, slowly but surely.
From Spice:
With 33k / 62k I measured an increase of 1.46db
With 16k5 / 31k I measured an increase of 1.66db
But of course adjusting the size of the Pos FB Resistor could produce greater or lesser values
Actually, in adjusting these resistor values what we are aiming at is to create a reasonably high damping factor. Nelson's F7 has a rated DF of 100.
The values needed to achieve this depend upon other factors:
1) The closed loop gain of the amp
2) The size of the current sense resistor.
So to arrive at a finished design it might be more valuable to test for this kind of damping factor in spice.
The simplest, not technical way to do this is to measure without & then with a load resistor.
when adding the load resistor the o/p voltage should only drop by a very small amount.
e.g.
Without a load o/p = 5.00V
With an 8R load o/p = 4.96V
This would give a DF of about 120
Hands up admission here - if you set this up correctly and then halve both resistor values as I mentioned above, you will probably end up with a negative damping factor which might be considered to be a bit "over the top". Negative damping factor means that when you add in the load, the o/p voltage actually rises !
The values needed to achieve this depend upon other factors:
1) The closed loop gain of the amp
2) The size of the current sense resistor.
So to arrive at a finished design it might be more valuable to test for this kind of damping factor in spice.
The simplest, not technical way to do this is to measure without & then with a load resistor.
when adding the load resistor the o/p voltage should only drop by a very small amount.
e.g.
Without a load o/p = 5.00V
With an 8R load o/p = 4.96V
This would give a DF of about 120
Hands up admission here - if you set this up correctly and then halve both resistor values as I mentioned above, you will probably end up with a negative damping factor which might be considered to be a bit "over the top". Negative damping factor means that when you add in the load, the o/p voltage actually rises !
Last edited:
No no - I think these are different measurements - bit busy now - will get back later
generg
Regarding the above posts I suggest that you choose a value for the series input resistor and current sense resistor and then adjust the pos FB resistor until that damping factor is about 100 ( into 8 ohms ).
Practically speaking you can measure damping factor by measuring the AC voltage @ 50 hz at the o/p of the amp with a digital multi meter first with the load disconnected and then with an 8 ohm resistor connected in as the load.
Adjust the input sinwave level until the amp o/p measures 5V RMS on the multimeter. Then add the load resistor into the circuit and see how much the voltage drops.
Adjust the Pos feedback resistor value until the RMS voltage under load condition is about 4.96V. ( You will need to reset the off load voltage to 5V every time you change the Pos FB resistor value )
Then for your interest you can see how much the output level increases when you add the Pos FB functionality . . . . but it's best to set up the damping factor first and then check o/p level increase afterwards
hope that helps
mike
Regarding the above posts I suggest that you choose a value for the series input resistor and current sense resistor and then adjust the pos FB resistor until that damping factor is about 100 ( into 8 ohms ).
Practically speaking you can measure damping factor by measuring the AC voltage @ 50 hz at the o/p of the amp with a digital multi meter first with the load disconnected and then with an 8 ohm resistor connected in as the load.
Adjust the input sinwave level until the amp o/p measures 5V RMS on the multimeter. Then add the load resistor into the circuit and see how much the voltage drops.
Adjust the Pos feedback resistor value until the RMS voltage under load condition is about 4.96V. ( You will need to reset the off load voltage to 5V every time you change the Pos FB resistor value )
Then for your interest you can see how much the output level increases when you add the Pos FB functionality . . . . but it's best to set up the damping factor first and then check o/p level increase afterwards
hope that helps
mike
Hi All,
I seem to remember that Nelson said that he'd be interested to see with what people came up with when "having a go" with the F7 type designs so after playing around with the F7 for a while here's a little summary of my journey through this design. Just to be sure I checked this with Nelson yesterday and he encouraged me to share it here.
From the first time I saw the F5 circuit diagram I was kind of blown away by the elegant simplicity of the design and I thought it would be worth trying to spend some time seeing how good an amp like this could get. So I've been playing around with F5/7 type circuits on and off for about three years and this is a very short summary of my findings and proposed next step:
My first attempt that this was this:
https://photos.app.goo.gl/nXSQURf2CJhR2ENXA
It shows:
This is what I'm listening to at present and it's clearly my best effort so far:
The treble & midrange are much improved. it sounds more natural and imaging is clearer.
However, I think there is still room for improvement:
So my new proposed circuit to address these points is this:
https://photos.app.goo.gl/M2ncG1P4GLZX4Tfg8
I'm wondering if anyone else has tried or seen this kind of approach with this or other designs.
One concern about this idea is that the transistioning between the two OLG plateaus might subjectively affect the midband.
Any Feedback / Ideas / Suggestions would be welcome
I am planning to get a prototyping PCB made that supports all of these ideas and also supports an option for using BJTs in the I/P stage.
cheers
mike
I seem to remember that Nelson said that he'd be interested to see with what people came up with when "having a go" with the F7 type designs so after playing around with the F7 for a while here's a little summary of my journey through this design. Just to be sure I checked this with Nelson yesterday and he encouraged me to share it here.
From the first time I saw the F5 circuit diagram I was kind of blown away by the elegant simplicity of the design and I thought it would be worth trying to spend some time seeing how good an amp like this could get. So I've been playing around with F5/7 type circuits on and off for about three years and this is a very short summary of my findings and proposed next step:
- In several of my versions of the F5 & F7 I had noticed that the midband could sound really very good with a ( natural ) damping factor of 10-12.
- I tried adding the postive feedback to several different versions but my personal take on it was that I marginally preferred the sound without it.
- I also noticed that with the DF set to around 5 or 6 the treble become very clean & natural sounding. Higher levels of FB caused the treble to sound harsh & rough and tracks with strong sibilance really did not sound good.
My first attempt that this was this:
https://photos.app.goo.gl/nXSQURf2CJhR2ENXA
It shows:
- The circuit
- 10Khz 5V square waves
- AC sweep showing Open & Closed Loop Gains
This is what I'm listening to at present and it's clearly my best effort so far:
The treble & midrange are much improved. it sounds more natural and imaging is clearer.
However, I think there is still room for improvement:
- I think the mids in the 5-10khz region could be a little cleaner so I would like this region to have slightly lower FB
- I think the high impedence of the I/P series resistor and the high impedence current sources are contributing to some lack of dynamics in the bass
- You can also see that this method gives, shall we say, rather "relaxed" square waves
So my new proposed circuit to address these points is this:
https://photos.app.goo.gl/M2ncG1P4GLZX4Tfg8
- Now impedances are kept low
- The OLG has two plateaus so that the lower FB extends into the critical 5-10khz region.
- The i/p stage current is only about 3mA but at high frequencies, the majority of the current comes though the R C "OLG reducer shunts" The charge stored in the C1 & C3 appears to be responsible for surprisingly fast square waves considering that the FB and I/P stage current is so low.
I'm wondering if anyone else has tried or seen this kind of approach with this or other designs.
One concern about this idea is that the transistioning between the two OLG plateaus might subjectively affect the midband.
Any Feedback / Ideas / Suggestions would be welcome
I am planning to get a prototyping PCB made that supports all of these ideas and also supports an option for using BJTs in the I/P stage.
cheers
mike
Last edited: