An example of socket for trying out resistors and jfets:
https://www.mouser.se/ProductDetail/?qs=ecHgFjcWJS/3lcAZPkoGYA==
https://www.mouser.se/ProductDetail/?qs=ecHgFjcWJS/3lcAZPkoGYA==
I don't know if this is the right place to ask this-- but if not, maybe someone can direct me. Somewhere in the diyAudio posts Zen Mod mentioned the importance of having a very good POWER SUPPLY for any kind of preamp/buffer. (Right now I have implemented the full circuit of the B1 Rev 2 (2019) and have the +-24V divider/filter circuit implemented for each rail. I'm feeding it with a +24V MeanWell power supply that can handle 1Amp. The circuit is supplying +-11.9x volts per rail). My buffer is going to be some naked vishay audio resistors with the matched transistor pairs from the diyAudio Store-sort of a stripped down version of the buffer. I have a very high bandwidth amp that has high gain and input sensitivity of 0.75V.
HERE's My QUESTION: will having a better power supply affect the performance of the buffer very much? It doesn't seem like it will draw much current.. but I'm hoping to supply those transistor pairs with as much current as needed for any instantaneous changes in waveform signals. (I'm using 192khz ultra HD music streamed and have no idea as to the capabilities of those transistors in terms of frequency range). Any comments, perspectives, or suggestions? Have any of you noticed audible changes in going from the supplied circuit design/24V power splitting circuitry to a fancier or more powerful power supply? Thank you all for your help!
HERE's My QUESTION: will having a better power supply affect the performance of the buffer very much? It doesn't seem like it will draw much current.. but I'm hoping to supply those transistor pairs with as much current as needed for any instantaneous changes in waveform signals. (I'm using 192khz ultra HD music streamed and have no idea as to the capabilities of those transistors in terms of frequency range). Any comments, perspectives, or suggestions? Have any of you noticed audible changes in going from the supplied circuit design/24V power splitting circuitry to a fancier or more powerful power supply? Thank you all for your help!
It is very sensitive to power supplies. 24 volts btw is double of what Nelson specifies for this circuit, and will also make the JFETs more noisy. I use 12v per rail. Also don’t expect extremely high PSRR in this no feedback circuit.
The very high bandwith also makes this circuit sensitive to power supply inductance.
You can achieve less voltage and also less PSU inductance through adding RC filtering as close to the circuit as possible. Preferably four RC’s, on pos and neg per chan. Also using high R will reduce psu noise.
The litterate will probably tell you tHat your SMPS is the worst alternative, then pure linear, then series regulated, then shunt to top it off
If I were you, I’d use what I have and consider upgrades later.
The very high bandwith also makes this circuit sensitive to power supply inductance.
You can achieve less voltage and also less PSU inductance through adding RC filtering as close to the circuit as possible. Preferably four RC’s, on pos and neg per chan. Also using high R will reduce psu noise.
The litterate will probably tell you tHat your SMPS is the worst alternative, then pure linear, then series regulated, then shunt to top it off
If I were you, I’d use what I have and consider upgrades later.
Last edited:
Thank you andynor and dennis--
I spoke incorrectly-- I do have +- 12 V (not +- 24v). I never thought of power supply inductance-new concept--thanks andynor. I do have the 4x 4.7K and the 3300uF caps and two 10 ohm 3 watt per channel. I'm going to build the thing with the best resistors I can find and check it out. Without any improvements and just using potentiometers -- it was not transparent enough for my system...so I'm giving it the best shot I can.
Dennis I bought from the store an SMPS DC Filter PO89ZB kit... is that the kind of thing you are speaking about? I've not tried it yet.
Thanks so much!
I spoke incorrectly-- I do have +- 12 V (not +- 24v). I never thought of power supply inductance-new concept--thanks andynor. I do have the 4x 4.7K and the 3300uF caps and two 10 ohm 3 watt per channel. I'm going to build the thing with the best resistors I can find and check it out. Without any improvements and just using potentiometers -- it was not transparent enough for my system...so I'm giving it the best shot I can.
Dennis I bought from the store an SMPS DC Filter PO89ZB kit... is that the kind of thing you are speaking about? I've not tried it yet.
Thanks so much!
That filter is perfect, I would try it out. It should be pretty transparent. The most striking quality in my opinion is the absense of hi fi mumbo jumboness to the sound: very natural.
Don’t worry about inductance, the 3300uF caps take care of that.
If you experience hf noise you can just increase the value of the 10R resistors.
Don’t worry about inductance, the 3300uF caps take care of that.
If you experience hf noise you can just increase the value of the 10R resistors.
Last edited:
That's the idea, but you need something for the negative rail as well.
As an illustration of the effect of inductor filtering on the high frequency noise, have a look here (Another project, but also using an SMPS):
https://www.diyaudio.com/community/threads/hagerman-bugle-2.237627/#post-3605491 (scroll down that post to the oscilloscope traces..)
As an illustration of the effect of inductor filtering on the high frequency noise, have a look here (Another project, but also using an SMPS):
https://www.diyaudio.com/community/threads/hagerman-bugle-2.237627/#post-3605491 (scroll down that post to the oscilloscope traces..)
I finished and posted my B1R2 with the bells and whistles elsewhere on the site but it did not get much notice with the B1R2 crowd. I obviously put the post in the wrong place. So here it is again for my fellow B1R2-o-philes.
This is for my office rig that also uses an ACA-Mini with boards laid out so it can be housed in a conventional chassis with heat sinks.
I'm working on balanced supply version of that as well. I populated the boards and power supply this weekend.
The output buffer is where the magic is and is pretty much as designed by Papa (Thanks Papa! 🙏). But I also added a few other things I wanted,
There is an adjustable LM317/337 based power supply w/RCRC front end, a volume pot and a configurable gain stage all on one PCB.
The AC source is a 24Vwall transformer using a standard 2.1 mm barrel connector.
This build has 12dB of gain and I really like that a lot.
But the gain stage can be de-populated and omitted or easily changed with socketed components.
Same goes for the PS. It's easy enough to de-pop and bypass,
The pot is a 20k Alps RK2.
Using a 20k Pot with 50k input impedance was pretty stupid. That will change.
The next build will use a wider chassis and and a front panel 1/8" input, an LED power indicator. load R's on the inputs and a 10k pot with a higher input Z. Maybe a mute switch, too. The next build will go in the main rig.
How does it sound? I think it's great but I am biased. It is very open, detailed and dynamic with great imaging and depth. My office stereo never sounded this good. BTW, dDoes anyone want an almost completed BA-3 gain stage? I see no reason to finish it. How much better could if be, if audibly better to me at all?
The board was created in KiCad. I'd be happy to share if anyone is interested in the design files or Gerbers
The attached schematic is as built. I have a new rev with tiny improvements not yet built and it probably never will be built. I ordered the five board minimum for the first build and expect to only build a few more.
This is for my office rig that also uses an ACA-Mini with boards laid out so it can be housed in a conventional chassis with heat sinks.
I'm working on balanced supply version of that as well. I populated the boards and power supply this weekend.
The output buffer is where the magic is and is pretty much as designed by Papa (Thanks Papa! 🙏). But I also added a few other things I wanted,
There is an adjustable LM317/337 based power supply w/RCRC front end, a volume pot and a configurable gain stage all on one PCB.
The AC source is a 24Vwall transformer using a standard 2.1 mm barrel connector.
This build has 12dB of gain and I really like that a lot.
But the gain stage can be de-populated and omitted or easily changed with socketed components.
Same goes for the PS. It's easy enough to de-pop and bypass,
The pot is a 20k Alps RK2.
Using a 20k Pot with 50k input impedance was pretty stupid. That will change.
The next build will use a wider chassis and and a front panel 1/8" input, an LED power indicator. load R's on the inputs and a 10k pot with a higher input Z. Maybe a mute switch, too. The next build will go in the main rig.
How does it sound? I think it's great but I am biased. It is very open, detailed and dynamic with great imaging and depth. My office stereo never sounded this good. BTW, dDoes anyone want an almost completed BA-3 gain stage? I see no reason to finish it. How much better could if be, if audibly better to me at all?
The board was created in KiCad. I'd be happy to share if anyone is interested in the design files or Gerbers
The attached schematic is as built. I have a new rev with tiny improvements not yet built and it probably never will be built. I ordered the five board minimum for the first build and expect to only build a few more.
Looks like you've done amazing work audiosteve-- I just thought I would acknowledge that, but my posting is not related to what you have just posted. I'm looking at going from "bells and whistles" to totally stripped down. What's the simplest the circuit can go and sound good? The goal of my buffer is impedance matching between DAC and AMP with utmost transparency and frequency range. I recall Andynor had very simple circuit of the B1 rev 2 without resistors...(I think post #389 and prior). I'm building something like diagram in post #454 (diagram 16B only without the extra two resistors in front for attenuation). I've removed the input selection and volume control.
REMOVING OUTPUT RESISTORS: I bought some fancy zfoil audio resistors (low inductance) that were pricey...but found out the best sound I have from my B1 rev 2 buffer (2019) is if I remove ALL the resistors from the output. The transistors from the store are so well matched that without a pot and even with removing output resistors the DC voltage at the src output of the transistors is only 0.7mV for one channel and 0.4mv for the other. My power supply with filtering is working great.
NO ATTENUATION: On the input I did some testing using those sockets suggested and found that the simplest was the best sounding for my setup-- by that I mean no -12dB or -6dB attenuation... just a 930 ohm resistor (replacing the 1K in the schematic) and having a 26K shunt resistor at the input to ground (instead of Pot+1Meg) and I'm using my DAC for volume control. The resistors I'm using at the front right now are audio resistors. I'm not selecting between inputs. It seems the more I reduce components the more transparent it gets (I can still hear slight effects of the buffer).
QUESTION about INPUT RESISTORS: I haven't tried stripping out all the resistors on the input but I'm wondering about that. All of my interconnects to and from this buffer are helix design, low noise, low capacitance, low mass KLE rca.. I'm trying to squeeze the last ounce of transparency that I can out of the buffer. (My amp is a very high bandwidth amp...sensitive to everything).
Is it safe to remove all resistors at the front end of the B1 rev 2? (will it cause problematic oscillations): do I need the 930ohm (1K in the diagram) at the gates to the transistors? If I try it without a resistor in there and it sounds okay will it perhaps later damage the transistors somehow?
Is the 26K resistor from input to ground... is it really necessary? Would it help to have the 26K resistor bumped up way higher (even if it's a traditional resistor and not an audio low inductance one)? Any suggestions?
Thank you to Andynor and R-K Ronningstad and Sov for helping me get this far! It's sounding better and better as I experiment with trying to squeeze out the essence of this simple design. (I'm hoping to try and avoid having to buy a fancy preamp).
REMOVING OUTPUT RESISTORS: I bought some fancy zfoil audio resistors (low inductance) that were pricey...but found out the best sound I have from my B1 rev 2 buffer (2019) is if I remove ALL the resistors from the output. The transistors from the store are so well matched that without a pot and even with removing output resistors the DC voltage at the src output of the transistors is only 0.7mV for one channel and 0.4mv for the other. My power supply with filtering is working great.
NO ATTENUATION: On the input I did some testing using those sockets suggested and found that the simplest was the best sounding for my setup-- by that I mean no -12dB or -6dB attenuation... just a 930 ohm resistor (replacing the 1K in the schematic) and having a 26K shunt resistor at the input to ground (instead of Pot+1Meg) and I'm using my DAC for volume control. The resistors I'm using at the front right now are audio resistors. I'm not selecting between inputs. It seems the more I reduce components the more transparent it gets (I can still hear slight effects of the buffer).
QUESTION about INPUT RESISTORS: I haven't tried stripping out all the resistors on the input but I'm wondering about that. All of my interconnects to and from this buffer are helix design, low noise, low capacitance, low mass KLE rca.. I'm trying to squeeze the last ounce of transparency that I can out of the buffer. (My amp is a very high bandwidth amp...sensitive to everything).
Is it safe to remove all resistors at the front end of the B1 rev 2? (will it cause problematic oscillations): do I need the 930ohm (1K in the diagram) at the gates to the transistors? If I try it without a resistor in there and it sounds okay will it perhaps later damage the transistors somehow?
Is the 26K resistor from input to ground... is it really necessary? Would it help to have the 26K resistor bumped up way higher (even if it's a traditional resistor and not an audio low inductance one)? Any suggestions?
Thank you to Andynor and R-K Ronningstad and Sov for helping me get this far! It's sounding better and better as I experiment with trying to squeeze out the essence of this simple design. (I'm hoping to try and avoid having to buy a fancy preamp).
Last edited:
I may be wrong but I would do the following since you run it without a pot:
Remove the shunt resistor, reduce the 930R to 400-500R, and leave the outputs open. You can also potentially try to remove the 1M resistor to gnd, the one that is originally attached to the wiper, but this will likely provide little or no audible benefits.
You could drop the gate resistors, but the gates can theoretically begin to draw current and create issues, so if you do this I would monitor the output with a scope and see what happens in the high freq reign. MZM sez approx 500R is fine in that position. Papa uses 1K also in his beast cells and the original F6 circuit, for safeguarding. But he also notes he did not observe oscillations without it. So maybe worth a shot testing with 500R and some lower values too
Remove the shunt resistor, reduce the 930R to 400-500R, and leave the outputs open. You can also potentially try to remove the 1M resistor to gnd, the one that is originally attached to the wiper, but this will likely provide little or no audible benefits.
You could drop the gate resistors, but the gates can theoretically begin to draw current and create issues, so if you do this I would monitor the output with a scope and see what happens in the high freq reign. MZM sez approx 500R is fine in that position. Papa uses 1K also in his beast cells and the original F6 circuit, for safeguarding. But he also notes he did not observe oscillations without it. So maybe worth a shot testing with 500R and some lower values too
Thanks GndSense and I largely agree with What Andynor says .....
But you could need a path from the FET gates to ground if they are to auto bias. I would not remove the 1M R before the gate unless there is another R to ground right in front of it.
You want the gate at 0V. Hence the R from gate to ground and, in my case above, a coupling cap to prevent a DC level at the gate.
You could drop that value to 100k and it would still be fine. But you should have something and, considering it's not in the signal path, I'm not sure the specific value is of huge importance or audible.
.
But you could need a path from the FET gates to ground if they are to auto bias. I would not remove the 1M R before the gate unless there is another R to ground right in front of it.
You want the gate at 0V. Hence the R from gate to ground and, in my case above, a coupling cap to prevent a DC level at the gate.
You could drop that value to 100k and it would still be fine. But you should have something and, considering it's not in the signal path, I'm not sure the specific value is of huge importance or audible.
.
Great work! Regarding the BA3, it may sound better or it may not sound better. But it will most certainly sound different if you are into that sort of thing.
I have built a variety of these pass designs and find myself switching back and forth for fun.
Andynor,
I think the R to ground (i.e. DC ref to ground), is needed. I made a mistake on my first layout and put the coupling cap between the 1M and the 1k R to the gate. So the gate had no DC ground reference. I got no music from the buffer. I patched the board to move the 1M to ground between the cap and 1k and it started to work. That gave a 1M DC path to ground instead of an open. It seems odd that, even through it is 1M, it worked. But with no DC ref to ground at all it did not work. Maybe something else was happening as well. I'm just a hobbyist, not an expert. I'm just relating my experience.
I think the R to ground (i.e. DC ref to ground), is needed. I made a mistake on my first layout and put the coupling cap between the 1M and the 1k R to the gate. So the gate had no DC ground reference. I got no music from the buffer. I patched the board to move the 1M to ground between the cap and 1k and it started to work. That gave a 1M DC path to ground instead of an open. It seems odd that, even through it is 1M, it worked. But with no DC ref to ground at all it did not work. Maybe something else was happening as well. I'm just a hobbyist, not an expert. I'm just relating my experience.