I verified the FR is nearly identical out of the preamp without the output FETs.
But I have no doubt the output buffer improved load handling despite the higher level of distortion it introduced.
That’s good to know. At low enough distortion levels there seems to be no detectable difference in sound.
That is a big difference. See if it is coming from the SCG. If yes, then check the drain resistor, Rsch, and the gate to ground resistor, Rg. Those control the gain. At a 10x ratio, there is 20 db gain. So if the other is 8 db less, it would have to be the Rsch/Rg ratio.
Hopefully not a problem.
No, that doesn’t surprise me. The Vgs difference is less than a volt between gate and source. So, preamp output and buffer output and close to each other.
Here is one solution to the buffer. Keep the J113 in and replace J175 with 3k3-4k7. More on that tomorrow. Meanwhile, here is a nice measurement of the circuit with buffer swinging 40 Vpp (14.44 Vrms) at less than 1% THD. It sounds quite good. Some more measurements tomorrow. We might have to give up on the PP buffer with this performance.
Also note the nice low noise floor. There is a bit of 600 Hz but it is below -100db. Good enough. This is with Jan's Autoranger. Makes measuring different levels a breeze and I don't have to use my clunky attenuator.
I am liking the sound of the new power supply. For those of you who have tried other supplies, can you comment on the differences?
Also note the nice low noise floor. There is a bit of 600 Hz but it is below -100db. Good enough. This is with Jan's Autoranger. Makes measuring different levels a breeze and I don't have to use my clunky attenuator.
I am liking the sound of the new power supply. For those of you who have tried other supplies, can you comment on the differences?
I need to switch the 10 dB gain FE2022s to hear if that is the problem. Ten dB is the sensitivity difference between woofers and CD - thought I was asking too much of the digital stuff to make up this difference OR on the other hand running the CD with too much attenuation, again digital attenuation.
I am about to open up the left speaker to install a phase plug on the Celestion CD and to fill the cabinet with reticulated foam which made a worthwhile difference on the right side. I cannot imagine the foam adding 8 dB of sensitivity! But this needs to be done to keep both the same as much as possible.
Started hearing the thickness of a bit too much second order distortion on the right channel. Hope to unbuffer those boards over the weekend.
All in all: what I thought was a great line stage is even greater now.
Rahul has made all of us very lucky music lovers to have this.
Thanks, Rahul
Your post showed up AFTER I originally posted the above. Somehow I worry the problem is more likely the FE2022s - I measured the resistors I used for Rschade and Rg before installing. NOT that that precludes checking again!!!
Thanks, again,
I am about to open up the left speaker to install a phase plug on the Celestion CD and to fill the cabinet with reticulated foam which made a worthwhile difference on the right side. I cannot imagine the foam adding 8 dB of sensitivity! But this needs to be done to keep both the same as much as possible.
Started hearing the thickness of a bit too much second order distortion on the right channel. Hope to unbuffer those boards over the weekend.
All in all: what I thought was a great line stage is even greater now.
Rahul has made all of us very lucky music lovers to have this.
Thanks, Rahul
Your post showed up AFTER I originally posted the above. Somehow I worry the problem is more likely the FE2022s - I measured the resistors I used for Rschade and Rg before installing. NOT that that precludes checking again!!!
Thanks, again,
On this latest buffer thing.. as one of my scatterbrained ideas.. would it be possible to run double J113's in parallel and replace Q106 with a halved resistor value in order to double this buffer stage current? Might make for a more robust drive capability without ruining the identity.. (?)
Output impedance with knob at 12’o’clock:
Output impedance with knob at max:
That’s with output buffer desoldered and a 50K pot.
Output impedance with knob at max:
That’s with output buffer desoldered and a 50K pot.
I continue to wonder about this UJ3N065080K35 as a buffer.
I know nothing of what is needed for a buffer nor how would would use it as a buffer - but it certainly is rugged and take lots of voltage and current.
Certainly would not "fall into place" with the new PCB but could be finagled into place especially if only a resistor or two would be needed.
Could such a device make for a good buffer? It could easily accept the full 80 volts. Any advice? We could use a small choke to mimic TDV - 100uH?
By the way, Rahul, I love your power supply boards. I am using one for each SCG and for the FE2022s. I am placing a choke in the path between the remote transformers and the inputs to the boards since I have them. The SCGs get choke input - the FE2022s are using LT4320 with a choke after the input cap. - in fact sitting next to the board. Reg boards as close to each SCG as I can get them.
My system is quiet - never has been so quiet.
I know nothing of what is needed for a buffer nor how would would use it as a buffer - but it certainly is rugged and take lots of voltage and current.
Certainly would not "fall into place" with the new PCB but could be finagled into place especially if only a resistor or two would be needed.
Could such a device make for a good buffer? It could easily accept the full 80 volts. Any advice? We could use a small choke to mimic TDV - 100uH?
By the way, Rahul, I love your power supply boards. I am using one for each SCG and for the FE2022s. I am placing a choke in the path between the remote transformers and the inputs to the boards since I have them. The SCGs get choke input - the FE2022s are using LT4320 with a choke after the input cap. - in fact sitting next to the board. Reg boards as close to each SCG as I can get them.
My system is quiet - never has been so quiet.
Last edited:
This is absolutely a possibility. Double or quadruple the JFETs to get more current capability. Doubling is possible on the current board if you are adventurous. But even with a single J113 there is a benefit to showing a high load to the gain device and isolating it from the world. It sounds powerful with the buffer driving my Elmo.
This looks about right in terms of output impedance. But I’m not sure why a change in the pot position at the input changes the output impedance.
That’s a power JFET, which can drive a speaker directly. It can work I’m sure but might be better suited in a MoFo style follower. You can give it a try.
Edit: I’ll post the edits to the buffer tonight.
Great! Good to know. I’m getting good sound from the power supply. Very happy with it and was curious about other impressions.
I have measured output impedance using a high value resistor, say 10k, then measuring output, then starting with 1k and going lower until I hit a value where the Vout is half the value with the high value resistor. That is the output impedance.
Working on the buffer still. Checking a couple of things before finalizing.
Working on the buffer still. Checking a couple of things before finalizing.
I know it is a power FET but it seems to have many appealing qualities. It can take plenty of voltage, has minimal capacitance.
I have no idea how much current is required for it to perform well. Might require its own regulator board.
I was thinking of those who use 300Bs as line stages and Eric Barbour's 572 output stage of decades past.
I would not know how to begin to use it but I would be glad to give it a try if you could tell me values of components to use with it.
I have no idea how much current is required for it to perform well. Might require its own regulator board.
I was thinking of those who use 300Bs as line stages and Eric Barbour's 572 output stage of decades past.
I would not know how to begin to use it but I would be glad to give it a try if you could tell me values of components to use with it.
Looking at the datasheet, figure 6, it seems you would need more than -8V Vgs for currents in the range of 1-2A. Preamp levels would be similar. It doesn’t quite fit the Mofo circuit because of the bias requirements, but you could easily add a negative supply like is needed in some VFET designs.
In the SCG, you could try it in the gain position, maybe with a J74 instead of a J175 at the input to get the Vd in the right range. As a preamp buffer, again it will work but the large Vgs will waste a bit of voltage headroom. Not a problem if you have enough supply voltage.
In the SCG, you could try it in the gain position, maybe with a J74 instead of a J175 at the input to get the Vd in the right range. As a preamp buffer, again it will work but the large Vgs will waste a bit of voltage headroom. Not a problem if you have enough supply voltage.
Rahul, I appreciate you explaining to me why this would not work. Thanks. Now I can put it out of mind, if I can assume it came from there!
Was hoping there might be a higher voltage FET we could use for the buffer.
I have no desire to change the CIRCUIT. I am very pleased with what you have wrought and I would never be the one to find something to better it.
I remain,
Yr Obd Svt
Was hoping there might be a higher voltage FET we could use for the buffer.
I have no desire to change the CIRCUIT. I am very pleased with what you have wrought and I would never be the one to find something to better it.
I remain,
Yr Obd Svt
Well, that was quite a trek! But I figured out where the noise came from in the preamp build.
And....drum roll, it came from the power supply board...well, that is not quite clear, actually...
What I know for certain is that I had a PSU board with a 18V+18V+18V+20V Zener string in there, and, when I swapped in a PSU board with a 18V+18V+18V+10V Zener string, the noise went away.
All other parts on these boards are identical.
Of course, this also means the regulated output went from ~68V to ~58V...so it could be that the preamp board "prefers" 58V when the output JFETs are not on it.
I am glad it wasn't a ground loop or some wiring error, because I really tried everything in that area.
The question now becomes why is there noise when a 18V+18V+18V+20V Zener string is used.
I am starting to suspect the use of the STF3LN80K5 (https://www.mouser.com/ProductDetail/STMicroelectronics/STF3LN80K5) in place of the FQP19N20 (https://www.mouser.com/datasheet/2/149/fqp19n20-298362.pdf), but I do not know what's going on yet.
And....drum roll, it came from the power supply board...well, that is not quite clear, actually...
What I know for certain is that I had a PSU board with a 18V+18V+18V+20V Zener string in there, and, when I swapped in a PSU board with a 18V+18V+18V+10V Zener string, the noise went away.
All other parts on these boards are identical.
Of course, this also means the regulated output went from ~68V to ~58V...so it could be that the preamp board "prefers" 58V when the output JFETs are not on it.
I am glad it wasn't a ground loop or some wiring error, because I really tried everything in that area.
The question now becomes why is there noise when a 18V+18V+18V+20V Zener string is used.
I am starting to suspect the use of the STF3LN80K5 (https://www.mouser.com/ProductDetail/STMicroelectronics/STF3LN80K5) in place of the FQP19N20 (https://www.mouser.com/datasheet/2/149/fqp19n20-298362.pdf), but I do not know what's going on yet.
ElArte - I have used both of those MOSFETs and I do not detect a difference but my reg boards are only supplying one SCG. Maybe the STI has some kind of limitation? I cannot imagine it being stressed supplying SCG. I love to speculate so take it as that.
The pair without buffer has 79 volts. It is being fed something like 89 volts. The buffered pair are lower since I installed those before the buffer doubt and the advice to go to 80 volts or so. This is the voltage i had used on my perfboard versions.
What (forgive me for not going back to look) is the input voltage you are using?
I would go for no noise over "ideal" voltage, too.
The pair without buffer has 79 volts. It is being fed something like 89 volts. The buffered pair are lower since I installed those before the buffer doubt and the advice to go to 80 volts or so. This is the voltage i had used on my perfboard versions.
What (forgive me for not going back to look) is the input voltage you are using?
I would go for no noise over "ideal" voltage, too.
I was just wondering about that myself.. delta of V-in to V-out. You'll certainly want some headroom there. I would measure the V-in at pin 2 of Q1, since the R's in the raw DC filter are going to steal a little voltage.
Now that I have that problem in a corner, I am taking a break.
The donut is that 2 X 32V Antek which I wired in series and there is more than 64V of AC at the input because those Antek do not sag with such low load.
With a 18V+18V+18V+20V Zener string, I do not quite understand why the regulated output would be 68V only actually. Something somewhere is dropping more voltage than it should.
Granted, the STF3LN80K5 has Rds On = 2.75 ohm rather than 150 mOhm, but even that difference cannot account for that.
It's in a corner and I will get to the bottom of it.
The donut is that 2 X 32V Antek which I wired in series and there is more than 64V of AC at the input because those Antek do not sag with such low load.
With a 18V+18V+18V+20V Zener string, I do not quite understand why the regulated output would be 68V only actually. Something somewhere is dropping more voltage than it should.
Granted, the STF3LN80K5 has Rds On = 2.75 ohm rather than 150 mOhm, but even that difference cannot account for that.
It's in a corner and I will get to the bottom of it.