Thanks for the encouragement! 
I think you may misunderstand however. I have no idea what I'm doing either. I know that we run these devices with certain voltages and current through them and to look at the SOA - safe operating area curve to make sure we are comfortably within the safe zone but even there the SOA may not be defined for DC operation or even implied safe continuous operation as many or most of these SiC devices are meant for use in a switching circuit where it's not 'on' all the time as we see with class A amps.
The next thing I learned is that you may have to play with some values to get the appropriate Vgs to the gate and then sometimes that also affects how much current goes through the device (bias) (looking at the Vgs(th) value)
Input capacitance seems to have a direct connection to THD levels, lower capacitance gives you lower distortion and output capacitance is important as well if it's too low, it may force you to limit the other components in the chain ... I think that's written in the F2J paper but also referenced here and there in the forum.
I'm still in the very early days of experimentation. I am doing "swap and turn it on to see if it still works" stage. I can then do a bit of adjusting for THD and have a used distortion analyzer to look at the residual harmonics but you can use an FFT from a soundcard to look at the H2 and H3 levels individually to see if it's dominant one way or the other. What it won't necessarily show you on an FFT however is the phase so maybe you'll just have to grow golden ears to determine if it's +phase or -phase H2 by swapping speaker leads.
I've watched Mike's videos on load lines but actually designing for these operating points is well beyond me so I'm content for now with swapping, listening and seeing if it melts down. I don't have anything built with 2 channels yet so listening is mono and I have no idea what the rest of the spectrum looks like as I haven't done any frequency sweeps either. Definitely no shortage of new and fun activities still left.
What got me started on this to begin with was because I had built an F2 12 years ago and it was my favorite amp for a very very long time simply based on listening pleasure. I do have the appropriate speakers for it and my room and choice of music just also happens to work well with those speakers and the F2 so I've always wondered what an F2J would sound like.
Many things have changed since then. I had forgotten entirely how to do board layouts as I started using Eagle back in 2008 and did my F2 on a laser printed (Gootee method) self-etched board. This time I (inspired by fellow member Ripson - thank you P.) locked myself in my office and picked up KiCad over a weekend and was able to get boards made overseas. Once I had these I was able to start experimenting with different devices. What makes the F2 a great candidate for this is that there are no special or expensive parts outside of the gain device if you choose to use Semisouth. All I have to do is keep a fire extinguisher handy and if a device self destructs, I've gathered a new data point 🙂 - see my signature btw.
I originally joined diyaudio.com because I wanted to build an amp that sounded just like something expensive without the price tag but Papa has spoiled us with so many excellent sounding projects, it's no longer about building something that 'sounds the best' because that label is subjective and relatively short lived. To me, now, diy is something that I made that sounds special to my ears, whether it's exactly like how Papa or Wayne or Zen Mod or Mike R or Mark J or any of the other unmentioned contributors to this forum designed or a variation that I prefer and modified myself.
Papa always says - it's entertainment
... but I have a FW J2 just in case my cooking is actually horrible
This is what DIY looked like in 2008
I think you may misunderstand however. I have no idea what I'm doing either. I know that we run these devices with certain voltages and current through them and to look at the SOA - safe operating area curve to make sure we are comfortably within the safe zone but even there the SOA may not be defined for DC operation or even implied safe continuous operation as many or most of these SiC devices are meant for use in a switching circuit where it's not 'on' all the time as we see with class A amps.
The next thing I learned is that you may have to play with some values to get the appropriate Vgs to the gate and then sometimes that also affects how much current goes through the device (bias) (looking at the Vgs(th) value)
Input capacitance seems to have a direct connection to THD levels, lower capacitance gives you lower distortion and output capacitance is important as well if it's too low, it may force you to limit the other components in the chain ... I think that's written in the F2J paper but also referenced here and there in the forum.
I'm still in the very early days of experimentation. I am doing "swap and turn it on to see if it still works" stage. I can then do a bit of adjusting for THD and have a used distortion analyzer to look at the residual harmonics but you can use an FFT from a soundcard to look at the H2 and H3 levels individually to see if it's dominant one way or the other. What it won't necessarily show you on an FFT however is the phase so maybe you'll just have to grow golden ears to determine if it's +phase or -phase H2 by swapping speaker leads.
I've watched Mike's videos on load lines but actually designing for these operating points is well beyond me so I'm content for now with swapping, listening and seeing if it melts down. I don't have anything built with 2 channels yet so listening is mono and I have no idea what the rest of the spectrum looks like as I haven't done any frequency sweeps either. Definitely no shortage of new and fun activities still left.
What got me started on this to begin with was because I had built an F2 12 years ago and it was my favorite amp for a very very long time simply based on listening pleasure. I do have the appropriate speakers for it and my room and choice of music just also happens to work well with those speakers and the F2 so I've always wondered what an F2J would sound like.
Many things have changed since then. I had forgotten entirely how to do board layouts as I started using Eagle back in 2008 and did my F2 on a laser printed (Gootee method) self-etched board. This time I (inspired by fellow member Ripson - thank you P.) locked myself in my office and picked up KiCad over a weekend and was able to get boards made overseas. Once I had these I was able to start experimenting with different devices. What makes the F2 a great candidate for this is that there are no special or expensive parts outside of the gain device if you choose to use Semisouth. All I have to do is keep a fire extinguisher handy and if a device self destructs, I've gathered a new data point 🙂 - see my signature btw.
I originally joined diyaudio.com because I wanted to build an amp that sounded just like something expensive without the price tag but Papa has spoiled us with so many excellent sounding projects, it's no longer about building something that 'sounds the best' because that label is subjective and relatively short lived. To me, now, diy is something that I made that sounds special to my ears, whether it's exactly like how Papa or Wayne or Zen Mod or Mike R or Mark J or any of the other unmentioned contributors to this forum designed or a variation that I prefer and modified myself.
Papa always says - it's entertainment
... but I have a FW J2 just in case my cooking is actually horrible
This is what DIY looked like in 2008
I appreciate the time taken to tell an awesome story and provide some background, Twitchie.
Still emerging as a DIYer and learning bits as I go. Digging through posts both old and new saves me a ton of time and provides a lot of encouragement.
Still emerging as a DIYer and learning bits as I go. Digging through posts both old and new saves me a ton of time and provides a lot of encouragement.
twitchie, you are an inspiration, along the likes of Dennis and Ben; though I don't recall either of them encouraging the sparks and smoke method of experimentation and learning you are engaged in. It's all good. (But "test" the SSs last!)
🙂
🙂
I think about this a lot. I love diy-ing all the great circuits on this forum, but I still aspire to actually own a genuine PL or FW amp. Partly to have a benchmark to compare with, part because I’ve always wanted one, and part because it would be nice to have an amp that I didn’t feel the urge to tinker with nonstop 😀
Thank you for the kind words everyone 😛 I am eternally grateful for the contributors to this site as well as the community that makes it so enjoyable to be a part of. I'm also flattered to be able to inspire others just as I am inspired by many others so right back at you
As far as the trial by fire and smoke testing goes, I do take safety precautions (using a variac) and observe electrical safety like all good greedy boyz should do, but never underestimate the benevolence of a guardian Zen Mod now and again to slap some sense into you if you veer too far into the Dodo zone 😀
The recent act of generosity by Papa should (hopefully) spur a harvest of new creations that folks can share the experience back with the rest of us.
@codyt - I can't speak to all of Papa's creations but I think it's probably safe to say that he has been very successful in designing things that not only sound excellent, but are very much reproducible by the diy'er. The common hurdle I've found myself has to do with execution, particularly for noise and ground loops. Tame that beast and be reassured that what you have is very close to what a FW version sounds like. It's almost as if he had intended to do this 🙄
Anyway, for anyone who hasn't read Ben's mini-blog on how to build quiet amps - https://www.diyaudio.com/forums/pass-labs/366312-25w-single-hammond-193v-choke-loaded-2sk180-lamp-2.html#post6529826
... back to regularly scheduled programming ... I'll get some time this weekend to try out the non-Cree devices and have more fun pics posted.
As far as the trial by fire and smoke testing goes, I do take safety precautions (using a variac) and observe electrical safety like all good greedy boyz should do, but never underestimate the benevolence of a guardian Zen Mod now and again to slap some sense into you if you veer too far into the Dodo zone 😀
The recent act of generosity by Papa should (hopefully) spur a harvest of new creations that folks can share the experience back with the rest of us.
@codyt - I can't speak to all of Papa's creations but I think it's probably safe to say that he has been very successful in designing things that not only sound excellent, but are very much reproducible by the diy'er. The common hurdle I've found myself has to do with execution, particularly for noise and ground loops. Tame that beast and be reassured that what you have is very close to what a FW version sounds like. It's almost as if he had intended to do this 🙄
Anyway, for anyone who hasn't read Ben's mini-blog on how to build quiet amps - https://www.diyaudio.com/forums/pass-labs/366312-25w-single-hammond-193v-choke-loaded-2sk180-lamp-2.html#post6529826
... back to regularly scheduled programming ... I'll get some time this weekend to try out the non-Cree devices and have more fun pics posted.
UJ3C120080K3S
Datasheet says:
This SiC FET device is based on a unique ‘cascode’ circuit
configuration, in which a normally-on SiC JFET is co-packaged with a Si
MOSFET to produce a normally-off SiC FET device. The device’s
standard gate-drive characteristics allows for a true “drop-in
replacement” to Si IGBTs, Si FETs, SiC MOSFETs or Si superjunction
devices.
Surely enough, my cheap device tester recognizes it as a cascoded device of some sort (incorrectly calling it an IGBT). While reading a thread on IGBT devices, none other than our own Papa notes some of the downsides at the time as the unavailability of complementary parts (when push-pull and supersymmetry were king of the H3 landscape after defeating H2).
Well here we are in 2022 with the last few remaining Semisouth SiC FETs being released into the wild but it turns out, everyone is distracted by the SIT show 🙂 and there seems to be waning interest in finding suitable alternatives. Seems like a fool's errand but I think it's safe to come forward and admit that if nothing else, I am definitely a fool
Back in the middle of covid boredom, I picked a few more SiC parts out of the digikey catalog from Genesic, Littlefuse, UnitedSIC, Cree/Wolfspeed and Microchip with the intention of testing them in F2 but life had other plans so these sat for over a year during my vacation from DIY. I've learned a few things since my last round of device testing and have gotten much better at taking FFTs so I decided to do some testing in a different design (current feedback amp) where you have square law OS and single common source jfet FE, using IRFP150 for the upper half (current source) and swapping out N-channel devices for SiC testing.
I was very happy to find that every device worked and worked well, with the exception of 1 which was likely just a bad choice that you wouldn't have selected if you knew what to look for in a datasheet, and found a family of devices which really caught my attention (title of this post).
Test results:
In pretty much every case, H3 was roughly 20dB down from H2. This design is all H2 loveliness (negative phase of course) so I won't bother reporting different H2/H3 numbers. Just assume what you see is 99% H2. All testing was with input signal at -6dBV which put the output just below 1W into 8R. I had a testing anomaly where none of the devices would go higher than about 15W before >1%THD set in but I suspect my input JFET has too high IDSS or the fault is in my lousy pcb layout skills
Requisite eye candy time
The winner!!!
Some specs
REW FFT
Most of you probably stopped reading at this point so the important information will go here
. I looked for SOA curves before ordering any of these and also based decisions on dissipation, input capacitance and transconductance information whenever it was available. I was easily able to trim offset to below 1mV and bias was approximately 1.3A in all cases. I only ran each device for several minutes and none of them died premature deaths, not making any statements on device longevity - you are reading this at your own risk.
Encouraged by the UnitedSIC result, I left the UJ3N120080K3S running for several hours over the course of 2 days and it's still healthy. I attempted a frequency response sweep with my QA402 and all I got was a mostly flat line so either it's great or I am not measuring it correctly. In any case, it's good enough for me to move on to building 2 channels to give it a listen. Come back in 2023 to hear how it goes perhaps ...
Cheers,
Stephen
Datasheet says:
This SiC FET device is based on a unique ‘cascode’ circuit
configuration, in which a normally-on SiC JFET is co-packaged with a Si
MOSFET to produce a normally-off SiC FET device. The device’s
standard gate-drive characteristics allows for a true “drop-in
replacement” to Si IGBTs, Si FETs, SiC MOSFETs or Si superjunction
devices.
Surely enough, my cheap device tester recognizes it as a cascoded device of some sort (incorrectly calling it an IGBT). While reading a thread on IGBT devices, none other than our own Papa notes some of the downsides at the time as the unavailability of complementary parts (when push-pull and supersymmetry were king of the H3 landscape after defeating H2).
Well here we are in 2022 with the last few remaining Semisouth SiC FETs being released into the wild but it turns out, everyone is distracted by the SIT show 🙂 and there seems to be waning interest in finding suitable alternatives. Seems like a fool's errand but I think it's safe to come forward and admit that if nothing else, I am definitely a fool
Back in the middle of covid boredom, I picked a few more SiC parts out of the digikey catalog from Genesic, Littlefuse, UnitedSIC, Cree/Wolfspeed and Microchip with the intention of testing them in F2 but life had other plans so these sat for over a year during my vacation from DIY. I've learned a few things since my last round of device testing and have gotten much better at taking FFTs so I decided to do some testing in a different design (current feedback amp) where you have square law OS and single common source jfet FE, using IRFP150 for the upper half (current source) and swapping out N-channel devices for SiC testing.
I was very happy to find that every device worked and worked well, with the exception of 1 which was likely just a bad choice that you wouldn't have selected if you knew what to look for in a datasheet, and found a family of devices which really caught my attention (title of this post).
Test results:
In pretty much every case, H3 was roughly 20dB down from H2. This design is all H2 loveliness (negative phase of course) so I won't bother reporting different H2/H3 numbers. Just assume what you see is 99% H2. All testing was with input signal at -6dBV which put the output just below 1W into 8R. I had a testing anomaly where none of the devices would go higher than about 15W before >1%THD set in but I suspect my input JFET has too high IDSS or the fault is in my lousy pcb layout skills
Requisite eye candy time
The winner!!!
Some specs
REW FFT
Most of you probably stopped reading at this point so the important information will go here
. I looked for SOA curves before ordering any of these and also based decisions on dissipation, input capacitance and transconductance information whenever it was available. I was easily able to trim offset to below 1mV and bias was approximately 1.3A in all cases. I only ran each device for several minutes and none of them died premature deaths, not making any statements on device longevity - you are reading this at your own risk.Encouraged by the UnitedSIC result, I left the UJ3N120080K3S running for several hours over the course of 2 days and it's still healthy. I attempted a frequency response sweep with my QA402 and all I got was a mostly flat line so either it's great or I am not measuring it correctly. In any case, it's good enough for me to move on to building 2 channels to give it a listen. Come back in 2023 to hear how it goes perhaps ...
Cheers,
Stephen
Well here we are in 2022 with the last few remaining Semisouth SiC FETs being released into the wild but it turns out, everyone is distracted by the SIT show 🙂 and there seems to be waning interest in finding suitable alternatives. Seems like a fool's errand but I think it's safe to come forward and admit that if nothing else, I am definitely a fool
I'm still stuck on "SIT show". Someone PLEASE design that amp
Invasion of the SIT has swiftly corrected this… 🙂
On another note, someone going through a similar exercise years ago mentioned UF3C120400K3S as a device that measured close to SJEP performance-wise. Might be worth taking a look, though it’s possible some of the devices on your list have surpassed it.
This is just reinventing the wheel from tube amps.
Much better just to go SE tubes, unless it is more fun to try to source unobtanium silicon.
Even tubes has fully differential circuits with low distortion and even harmonics cancellation.
Just look at a well-built circlotron.
Cheers,
Patrick
It is more fun to find simple solid state alternatives being that the possibilities with tubes has been well explored.
I guess reinventing the wheel is also fun for some.
Especially when one runs out of other things to invent.
😊
Cheers,
Patrick
PS BTW I really like the M2 output stage. Shame it has not been used more extensively elsewhere.
.
Especially when one runs out of other things to invent.
😊
Cheers,
Patrick
PS BTW I really like the M2 output stage. Shame it has not been used more extensively elsewhere.
.
I wouldn’t say that. There are things you can do with toobz u cannot with fets, but also the other way around. Take the SIT-1 for example. I am no expert, but I doubt such a simple circuit would be achievable with toobz.