did have one of the FETs explode during testing with the 9 volts battery, though.
That happened to me too, but I suspect I did something wrong when placing it in the jig.
I would never say that was not possible!
It did take me by surprise.
What is the "midway" voltage on your SCG?
It did take me by surprise.
What is the "midway" voltage on your SCG?
I would never say that was not possible!
It did take me by surprise.
What is the "midway" voltage on your SCG?
About 25V.
But Rahul promises more bliss if I dial it up to 35V.
I am building another version with 60V of supply.
I have not lost one to high voltage yet. As I mentioned a few pages ago, they survived even 50+ volts. Of course we don't want to exceed 40V at most across one device.From my experience with higher voltages I did worry you were cutting it a bit close. I have never had one of "the little guys" go away.
I did have one of the FETs explode during testing with the 9 volts battery, though.
Wima is good for that spot also.Another one! I see you have changed the compensation capacitor to 100 pF.
Another place to use the WIMA or have you found something you like better?
I am trying to recreate the setup in post #1337. Still working on it.About 25V.
But Rahul promises more bliss if I dial it up to 35V.
I am building another version with 60V of supply.
I have not done the switch to 35V yet. I am giving the build as it is some good listening time, so I can compare better later.
Sounds fantastic with all my difficult-to-drive headphones.
I tried it with one of my problem children: A 9 ohm, 95 dB/mW set
With that one I get some audible noise when nothing is playing.
The headphones buffer has 1-ohm source resistors, so its output impedance is 0.5 ohm.
0.5 ohm into 9 ohm should work, but...that strange noise, it's not a hum.
The same source (DAC) into a Topping amp is totally quiet with those.
It can't be a ground loop because the preamp board is left floating right now. In fact, I have nothing connected to PE in that chassis (except the chassis itself, of course). RCA-in brings the ground reference, which is established at the source (the DAC).
Maybe these headphones are too sensitive for the SCG with so much gain.
Sounds fantastic with all my difficult-to-drive headphones.
I tried it with one of my problem children: A 9 ohm, 95 dB/mW set
With that one I get some audible noise when nothing is playing.
The headphones buffer has 1-ohm source resistors, so its output impedance is 0.5 ohm.
0.5 ohm into 9 ohm should work, but...that strange noise, it's not a hum.
The same source (DAC) into a Topping amp is totally quiet with those.
It can't be a ground loop because the preamp board is left floating right now. In fact, I have nothing connected to PE in that chassis (except the chassis itself, of course). RCA-in brings the ground reference, which is established at the source (the DAC).
Maybe these headphones are too sensitive for the SCG with so much gain.
Next time I put it on the bench, I’ll populate TH_Chassis with a 10R resistor and connect that to chassis/PE ground. Maybe that’ll do it.
Yes, try that, though 95 db/mW is lower than usual, so I am not sure it is the sensitivity. It must the 9 ohm load. I am not sure yet of the cause of the noise, but it is a tough load. The buffer probably needs more bias at that low load. It was designed for 32 ohms. Though again, not sure if that will remove the noise.
It must the 9 ohm load.
It is. I did an experiment where I placed a 22 ohm resistor in line with both channels and the noise went away.
Easy fix, I can make an adapter cable with 30 ohm built in.
I will still improve noise management, but that will do in any case.
OK, if you are sensitive to the immense audiofoolery one can wander in, you may want to close your eyes and skip this post.
This box hosts 2 resistors and 2 6.35mm jacks. A bit of 20-AWG wire too.
The output of SCG with headphones buffer goes in on the left, headphones with ridiculously low input impedance go in on the right. It adds 33 ohm of impedance via 2W metal film resistors.
It's not a miracle cure, but that pair of 9-ohm headphones I did this for has ruler flat input impedance across the frequency range; so it works.
This box hosts 2 resistors and 2 6.35mm jacks. A bit of 20-AWG wire too.
The output of SCG with headphones buffer goes in on the left, headphones with ridiculously low input impedance go in on the right. It adds 33 ohm of impedance via 2W metal film resistors.
It's not a miracle cure, but that pair of 9-ohm headphones I did this for has ruler flat input impedance across the frequency range; so it works.
What!? Another!?
Yes!
Slightly different configuration.
One will be a full-time preamp, the other a full-time headphones amp.
Yes!
Slightly different configuration.
One will be a full-time preamp, the other a full-time headphones amp.
Amazing builds, Elarte! How does it sound?
Past few days I have been investigating the discrepancy in the measurements I made some time ago and those made by ElArte and others using the new design. There is a considerable difference and I still haven’t figured out the exact cause. I have tried various operating points and undid other changes in the circuit, eliminating variables. I have it narrowed down to the J175. It appears that the first batch of J175 had different parameters to the new batch. I need to verify this and will do so over the next few days.
It is still a good performer, delivering 30V peak to peak, which is about 15W rms into 8 ohms, at about 2% THD, mostly 2nd harmonic, which is pleasing. Plenty of power and better than a lot of tube amps. More importantly, it manages to stay clean at loud volumes, which I have found is not easy to do.
Past few days I have been investigating the discrepancy in the measurements I made some time ago and those made by ElArte and others using the new design. There is a considerable difference and I still haven’t figured out the exact cause. I have tried various operating points and undid other changes in the circuit, eliminating variables. I have it narrowed down to the J175. It appears that the first batch of J175 had different parameters to the new batch. I need to verify this and will do so over the next few days.
It is still a good performer, delivering 30V peak to peak, which is about 15W rms into 8 ohms, at about 2% THD, mostly 2nd harmonic, which is pleasing. Plenty of power and better than a lot of tube amps. More importantly, it manages to stay clean at loud volumes, which I have found is not easy to do.
Amazing builds, Elarte! How does it sound?
As with the previous build, it sounds fantastic! I love it when more than 90% of the distortion is H2. The SCG I built in 2023 had much more H3; it still sounded good, though.
I am still making arrangements because my situation is non-standard; i.e., I have the headphones buffer in my builds, and it turns out it requires careful supply planning.
This morning, I changed the wiring to give one PSU to the preamp board and one PSU to the headphones buffer. This simple change allows the preamp board to have a ~29V midpoint, while my goal was 30V. I know you recommended 35V, but I don't think there is a good way to get there in my situation, unless I have a custom transformer made, or I upgrade various parts on the PSU to drop all the way down from a 115V secondary.
Past few days I have been investigating the discrepancy in the measurements I made some time ago and those made by ElArte and others using the new design. There is a considerable difference and I still haven’t figured out the exact cause. I have tried various operating points and undid other changes in the circuit, eliminating variables. I have it narrowed down to the J175. It appears that the first batch of J175 had different parameters to the new batch. I need to verify this and will do so over the next few days.
OnSemi's latest J175s are definitely on the tame side. I bought and sorted about 150 by now... they are clearly more predictable in terms of their parameters, but they may not be ideal for audio. I am not sure.
On the other hand, I have Siliconix J175s that measure more erratically but may be better for audio, if you can match them.
There are also low-noise InterFET J175s, but sampling 100 of those would cost $1,000. Not going there.
It is still a good performer, delivering 30V peak to peak, which is about 15W rms into 8 ohms, at about 2% THD, mostly 2nd harmonic, which is pleasing. Plenty of power and better than a lot of tube amps. More importantly, it manages to stay clean at loud volumes, which I have found is not easy to do.
It's an excellent performer. Remember, the power you need is a lot more important than the power you can achieve.
Eheh, alas the new supply configuration gives lots of noise when RCA cables are not connected and some audible noise when RCA cables are connected.
The tinkering lives on.
The tinkering lives on.
With dual supplies in my projects I normally tie those two grounds together with a short fat wire and connect the client grounds at the center... "T" style..
Thanks, William.
Yes. I should have thought things through before I made that wiring change.
I created a situation where the two supplies satisfy very different loads (and therefore have almost 15V of regulated difference between them) and the only ground path between them is through signal ground.
So, I went back to the previous wiring scheme, where the two supplies output nearly the same regulated voltage and silence is back.
Then, I figured I can use the biasing of the headphones buffer to even out the output of both supplies exactly... LOL. Obvious, I have to keep it reasonable not to create bad channel matching.
Using a fat wire to connect the grounds of both supplies should also work, but the load on each supply would still be very different, so I wonder if that won't still create current to flow in undesirable ways.
A 2 X 70V transformer is probably the only way to get where Rahul wants me to be in my setup.
Alas, I couldn't find one of those.
Granted, I am chasing a theoretical nirvana here and what I already have is tremendously good.
Alas, I couldn't find one of those.
Granted, I am chasing a theoretical nirvana here and what I already have is tremendously good.
- Home
- Amplifiers
- Pass Labs
- Schade Common Gate (SCG) Preamp