I listened to the revised headphone amp build tonight without the buffer at Vds = 33V. Spectacular.
This album blew my mind — what a recording masterpiece!
Rob van Bavel – Time for Ballads: The Studio Sessions
Highly recommended!
I think @metaphile's results match my findings, albeit at 10W into 8R. It's hard to fully assess because the headphone buffer masks some details.
As Rahul mentioned previously, 70V of input voltage with 35V of Vds is probably where the preamp wants to be.
In my setup, the headphone buffer provides more than enough buffering, and the output buffer on the preamp board doesn't add much value at this time.
This album blew my mind — what a recording masterpiece!
Rob van Bavel – Time for Ballads: The Studio Sessions
Highly recommended!
I think @metaphile's results match my findings, albeit at 10W into 8R. It's hard to fully assess because the headphone buffer masks some details.
- Jumpering the buffer off reduces THD.
- Increasing Vds reduces THD, but increasing Vds without a corresponding increase in supply voltage causes asymmetric clipping and reduces Vpp at clipping — unless you remove the buffer entirely, apparently.
- Removing the buffer from the board reduces THD. (I didn't do that yet.)
As Rahul mentioned previously, 70V of input voltage with 35V of Vds is probably where the preamp wants to be.
In my setup, the headphone buffer provides more than enough buffering, and the output buffer on the preamp board doesn't add much value at this time.
Fantastic report @metaphile! Great info in there. Your observations match mine. I’ll add that you can go up to 45V and the distortion drops further. The J111 will only have about 32-35V across it at 45V because the J175 has more than 10V across it and so the J111 doesn’t go beyond its dissipation limits. But 40V Vd is enough. For symmetrical swing and a bit of margin on the CCS, you can up the supply to 75-80V. Looking back at post#1321, 80V was where I had it setup.
@ElArte Most excellent! Glad you found an operating point you like. Love the feedback. I have not heard that song but will be cueing it up tomorrow. Hope you can try it without the JFETs. I am certain it will blow your mind!
You guys are playing with the circuit. I had a lot of fun tweaking it. Hope you are enjoying it also!
For lowest distortion and highest output swing, remove buffer JFETs. Use buffer to drive difficult loads, otherwise use preamp only.
@ElArte Most excellent! Glad you found an operating point you like. Love the feedback. I have not heard that song but will be cueing it up tomorrow. Hope you can try it without the JFETs. I am certain it will blow your mind!
You guys are playing with the circuit. I had a lot of fun tweaking it. Hope you are enjoying it also!
For lowest distortion and highest output swing, remove buffer JFETs. Use buffer to drive difficult loads, otherwise use preamp only.
I certainly am enjoying it. Next step is to hook it up to the TDV and listen to at least one channel, I haven't listened to the preamp or the amp yet. Looking forward to it.
My inane suggestion occurred after a full day of trying to install the magnificent camillaDSP to on an X86 motherboard I had laying around.
The same lack of sense was responsible for both.
The same lack of sense was responsible for both.
All good Rick. You can’t experiment without blowing a few fuses. Once I went through about a dozen before I found the problem. Now I buy them in packs of ten and keep several values handy.
On the preamp build, I will aim to not have the output JFETs. Desoldering them will be the first thing I do, so if I screw that up, it will be the end of that story for now. I don't want to make it sound like I have not desoldered parts before, I have plenty of time, but my odds of failure are definitely not zero and I hate doing it.
On the headphones amp build, not so sure, it will depend on what the preamp build reveals with that change in place. I have something that works well right now, and I don't need more voltage swing or getting H2 closer to H3.
I am not looking for a low-THD headphones amp with this build. This has always been about being able to drive my most stupid headphones (that, in truth, should not exist) with a slice of H2 on top for harmonic fun.
Does removing the buffer improve other technicalities beyond THD and potential voltage swing?
I go through dn2540s like you go through fuses.
Getting ready to install the left side boards - luckily I had not soldered in one of the boards buffer FETs.
The other I just clipped - I had left enough lead on the back that i can push some of that to make a better join when you find the solution. No big deal at all. This will be the case for the right when I get around to those.
I am looking forward to having camilladsp one of these days.
As always, thanks for your work, kindness, generosity and all-around goodness.
Getting ready to install the left side boards - luckily I had not soldered in one of the boards buffer FETs.
The other I just clipped - I had left enough lead on the back that i can push some of that to make a better join when you find the solution. No big deal at all. This will be the case for the right when I get around to those.
I am looking forward to having camilladsp one of these days.
As always, thanks for your work, kindness, generosity and all-around goodness.
I read @ElArte's post again this morning with the second round of measurements. The call out that distortion wasn't ruler flat at all Vd levels prompted me to look again at my distortion measurements, and I noticed that in mine, too, when I went from 35Vd to 40Vd, despite a lower absolute THD figure and lower H2, the higher harmonics were more prominent in the 40Vd measurement. 5th, 6th, and 7th harmonics, not present at 35Vd, started to appear at 40Vd.
I might follow @ra7 's try a higher input voltage or I may just stick with what I have.
If I stick with the 60.4V supply, I might back off from the 40Vd operating point a bit to get a more H2 dominant distortion profile with fewer higher harmonics rather than optimizing for the absolute lowest THD regardless of distortion character.
Of course once I have actually * listened * to it I can make more informed choices.
Just musing.
I might follow @ra7 's try a higher input voltage or I may just stick with what I have.
If I stick with the 60.4V supply, I might back off from the 40Vd operating point a bit to get a more H2 dominant distortion profile with fewer higher harmonics rather than optimizing for the absolute lowest THD regardless of distortion character.
Of course once I have actually * listened * to it I can make more informed choices.
Just musing.
ElArte - For successfully removing parts you have a few options... if we don't care about saving the component, (but we definitely don't want any PCB damage) it's easiest to just cut the part out and toss away. If nothing is going back into those holes, you're done. If a part is going back in, just pluck each leg remnant out with the soldering iron tip and fine tweezers. Then suck the remaining solder out of the holes with the desoldering handpiece. If we're saving the part... On a JFET like this I would mount the PCB in my board holder upside down (solder source side up), then cut a short piece of small wire that spans the width of the three legs at the same time. Solder this piece so that all three legs are soldered / shorted together. With a wide tip in the iron and this technique it's very easy to transfer the heat quickly to all three pads without damaging anything. The idea is that the part will basically fall out when all three solder joints reach the melting point at the same time. It's quick and the part will be fine with the brief heating. Clean everything back up and you're done.
I intentionally mount my precious Toshiba K170's and J74's without trimming the legs too short. Makes it easy to just cut at board level and transfer over to the "next big thing" without stressing the parts in any way.
I intentionally mount my precious Toshiba K170's and J74's without trimming the legs too short. Makes it easy to just cut at board level and transfer over to the "next big thing" without stressing the parts in any way.
I've never tried this but this sounds like a great technique thanks for sharing.
Yes. Obviously, I am not against measurements...they are tremendous to validate a build and finetune parameters, but your ears must agree or I don't know why we would not all just buy mainstream gear.
Thank you for the tips. I have seen Mr. SolderFix on YouTube use some of these with great ease. But, he is master at that and makes everything look trivial. I should practice those technique on a dummy board.
After sorting to find those FETs I doubt anyone wants to throw them away. Best to have an easy way to restore them to the circuit.
Please correct me if I have it wrong AGAIN - but leaving them in with legs cut at pin ones would be much less stressful than removing them and then replacing them.
Are there any untoward effects from leaving them in with cut legs?
Please correct me if I have it wrong AGAIN - but leaving them in with legs cut at pin ones would be much less stressful than removing them and then replacing them.
Are there any untoward effects from leaving them in with cut legs?
Something I forgot to mention: With the single supply and (10R || 100nF) ground lift from preamp board to earth ground, there is no noise whatsoever with even the 8-ohm headphones.
New donuts came in today...
I had prepared the rest for installation...
So, this is the pure preamp build...with output buffer JFETs desoldered! It went well.
After putting the secondaries in series, I get about 68.5V out of 1 PSU board. I share that between left and right channel.
34V Vds
Vin 0.5V
Vout 3.86 Vrms, 10.97 Vpp
THD+N: 0.29%
THD: 0.23%
H2: 0.23%
H3: 0.03%
Rounding errors apply.
40 Vds
Vin: 0.5V
Vout: 3.88Vrms, 11.02 Vpp
THD+N: 0.16%
THD: 0.12%
H2: 0.12%
H3: 0.02%
Rounding errors apply.
Now, as far as swings go...
Vds 40V
Vin: 2V (a full 2V)
Vout: 15.34Vrms, 43.46Vpp
This is quite a clean sine.
Clipping doesn't happen until around 2.2V of input.
That's the corresponding spectrum. There are more harmonics in this party, but still quite decent.
So, my personal conclusions:
The combination of these changes:
a) Increasing supply voltage to near 70V
b) Increasing Vds to near 40V
c) Desoldering the output buffer JFETs
Brings these benefits:
1) Decrease in THD
2) Increase in output Vpp
But, for my headphones amp build, desoldering the output buffer JFETs would not change much at all. (0.22% vs. 0.23% THD) [That is granted I cannot muster the supply I would need when preamp board and headphones buffer board draw 0.25 to 0.3A.]
All in all, I think that looks very good. It's still all H2 and it can swing big.
I'll leave this one on the bench if you guys would like me to test other things.
I had prepared the rest for installation...
So, this is the pure preamp build...with output buffer JFETs desoldered! It went well.
After putting the secondaries in series, I get about 68.5V out of 1 PSU board. I share that between left and right channel.
34V Vds
Vin 0.5V
Vout 3.86 Vrms, 10.97 Vpp
THD+N: 0.29%
THD: 0.23%
H2: 0.23%
H3: 0.03%
Rounding errors apply.
40 Vds
Vin: 0.5V
Vout: 3.88Vrms, 11.02 Vpp
THD+N: 0.16%
THD: 0.12%
H2: 0.12%
H3: 0.02%
Rounding errors apply.
Now, as far as swings go...
Vds 40V
Vin: 2V (a full 2V)
Vout: 15.34Vrms, 43.46Vpp
This is quite a clean sine.
Clipping doesn't happen until around 2.2V of input.
That's the corresponding spectrum. There are more harmonics in this party, but still quite decent.
So, my personal conclusions:
The combination of these changes:
a) Increasing supply voltage to near 70V
b) Increasing Vds to near 40V
c) Desoldering the output buffer JFETs
Brings these benefits:
1) Decrease in THD
2) Increase in output Vpp
But, for my headphones amp build, desoldering the output buffer JFETs would not change much at all. (0.22% vs. 0.23% THD) [That is granted I cannot muster the supply I would need when preamp board and headphones buffer board draw 0.25 to 0.3A.]
All in all, I think that looks very good. It's still all H2 and it can swing big.
I'll leave this one on the bench if you guys would like me to test other things.
Okay, that’s looks great! Thanks for doing the measurements. I have all I need in terms of measurements and yours match mine. Maybe you can do some listening next and see if this version sounds different.
I am continuing to troubleshoot the buffer. Will report when I find something.
I am continuing to troubleshoot the buffer. Will report when I find something.