I have some ECX10N20's around here somewhere and most of the parts and boards for the MoFo. But I don't think I want to sweeten things up anymore than what I have now, it seems just about right. With my previous SissySIT R.3, I always used a gain stage in front, either internal or external, to get the overall gain where I needed it for my system. And those gain stages were always "clean" to maintain that sweetness in the.... sweet spot..since the SS R.3 has it's own pleasing character. I listen to a wide variety of music and if the total distortion gets too high, I think things just start sounding too sloppy, particularly on complex material.
I really like this square law HGF. It just thrives into the lower impedance loads and sounds great. My speakers are essentially 5 ohms and not super efficient, so the amp has some nice synergy there..
I really like this square law HGF. It just thrives into the lower impedance loads and sounds great. My speakers are essentially 5 ohms and not super efficient, so the amp has some nice synergy there..
Breaking the “no building in the Spring or Summer” over here.
I want to say it’s an entry level DAC/streamer, but I cannot tell when I listen to it. My Lord, what a rabbit hole!
I want to say it’s an entry level DAC/streamer, but I cannot tell when I listen to it. My Lord, what a rabbit hole!
Looks fantastic! Don’t tease us with those pics… tell us more. Which DAC is it? I recently ordered an R2R dac as well, though it hasn’t arrived yet.
@william2001 I totally get what you are saying with too much H2. I don’t need a lot of power, so the single ended follower is just right for me. But a push-pull lateral follower would cancel most of the H2. Now that would be interesting. Of course, there are tons of amps with that config already. Would be fun to try that with SCG driving.
It's a ProtoDAC. TD1837 X 8. A NOS DAC HAT on top of a Raspberry Pi 5.
In its current incarnation, the PSU is shared between the Pi and the DAC, which is not the best scenario. But it sounds pretty great already, probably because the Pi only has 1 job: Running the RAAT bridge.
At some point in the future, I'll add a low noise linear supply for the HAT, but for now, I just need to clean up the build and put a lid on it.
That should be my last update on this.
I replaced all JFETs and used some old J175s made by Siliconix. These rate higher than anything else in terms of Vp far away from zero.
Put the board back in and the puttering was higher still. 🙄😳🤦
It was a setback.
But clearly all fingers were pointed at the PSU board once again.
I built a new PSU with higher rated parts, 3W 33R resistors, higher rated caps, 2W R6, no snubber parts, 1.5W 20V and 6.2V Zener diodes, etc. Everything overrated.
Put that PSU and…total silence.
Victory achieved!
The preamp makes 0.2% of H2, flat across the FR, at the output level that matters.
The preamp can also swing a little more than 40 Vpp before clipping.
I let that run 8 hours straight to ensure durable stability and it passed the test.
Subjectively, the sound is rich (H2), detailed and open. You can certainly hear all the layers in a mix.
What impresses me the most is the quality of timbre reproduction.
Bass rolloff is a tad above 2 dBr at 20 Hz, but nothing to be too concerned about.
Fantastic preamp! I can’t wait to find a way to put it in my main system.
Thanks again, Rahul!
Awesome! Great work! So good to hear. Chasing down the noise gremlins is always tough 
You could play with the Vd a bit and make the H2 go down if so desired. It acts like a triode and you can play with the Vd to adjust the harmonics.
A side note: I have been thinking about a couple more variations on the SCG theme. Just some ideas I want to try. Will post them here.
You could play with the Vd a bit and make the H2 go down if so desired. It acts like a triode and you can play with the Vd to adjust the harmonics.
A side note: I have been thinking about a couple more variations on the SCG theme. Just some ideas I want to try. Will post them here.
I did play with that. I dialed the supply voltage down to 62V from the original 70V. Then I played around with Vd from 25V to 30V. I settled on 28V, as I like the sound of that.
It’s really great that you can do some rolling like that by just turning the trimpots.
You could come full circle and drive some 300Bs. 😂 That could be too much H2 stacking, though.
I have upgraded my build of the original SCG design (FQP12P20/STP30N10F7) to JFET's (J111 and J175) according to ra7's notes in post #1348. I would like to use the SCG to drive to a SIT Mu Follower (Ben Mah's design) which requires a preamp with low output impedance and significant voltage swing to achieve full output. Would the JFET upgrade be suitable for this, or should I look at moving to the new design with buffer?
Thoughts and comments would be appreciated.
Thoughts and comments would be appreciated.
Driving Ben's SIT Mu Follower directly without the buffer would be fine, and since you built it already, just go for it. I would try it with the buffer also, it will probably drive it better. Some amps, like F4, already have a buffer on the input and with those you definitely don't need the buffer on the SCG.
Thanks for your comments. As you suggest, I'll go with the JFET upgrade version at present and build the new SE buffer version as well, since you have so kindly provided the Gerbers, BOM and a wealth of information regarding construction, tuning and mods. Many thanks for making these great designs available to the diyAudio community.
Rahul,
After listening without the buffers for many months I am curious to try the NEW buffer.
Would the selection process for the J113s be the same - using the 47R resistor?
I am thinking of doubling, maybe even quadrupling them. Would there be additional things to "match" in this case? Or would finding four that measured the same in your jig be enough?
Thanks and take care,
PS have to say what ACOURATE can do is close to astonishing.
After listening without the buffers for many months I am curious to try the NEW buffer.
Would the selection process for the J113s be the same - using the 47R resistor?
I am thinking of doubling, maybe even quadrupling them. Would there be additional things to "match" in this case? Or would finding four that measured the same in your jig be enough?
Thanks and take care,
PS have to say what ACOURATE can do is close to astonishing.
Beefy Buffer
Hi Rick,
If you are looking for a beefier buffer, there are a few options. Yes, you can double, triple, or quadruple the JFET in the buffer and get more current. But there is another way.
First, with the JFETs. When paralleling JFETs, I'd highly recommend using a separate source resistor per JFET and connecting the other end of the source resistor to ground. In this post, we are setting it to ~10 mA per JFET. So, you would have one 4k7 resistor per JFET when paralleling.
The JFETs to be paralleled should have as close Vgs as possible at the operating point. So, yes, use a 47R resistor in the source with a 9V battery and pick JFETs within < 0.1 Vgs (at worst) of each other when producing the same current. This might take a few parts to get good matches.
Another way, and what I would do, is to use a bigger part. I know you have some KSC3503 in your stash because we went through them during development. That part can take more heat and you can put a small heatsink on it. See the schematic below. There may be other small part changes on the schematic--ignore them--the things to focus on are Q1 and R12. If you adjust Vd for 40V, you will have about 40V - 0.7V (one diode drop), or about 40V, on the emitter pin of the KSC3503. So, then if you put a 1k resistor from the emitter to ground, you will get 40 mA through it. Plenty of juice. Be sure to put a heatsink on the KSC3503 and the 1k resistor will need to be properly sized for heat dissipation as well. If you put a 2k resistor there, you will get 20 mA through it--plenty for most applications.
In fact, this is how I have it configured. I have it at a little less than 2k for about 25-30 mA and it now has enough grunt to drive the Elmo. The sound is very close to the JFET buffer, very transparent.
I'll post an updated PCB with the footprint for a KSC3503.
Hi Rick,
If you are looking for a beefier buffer, there are a few options. Yes, you can double, triple, or quadruple the JFET in the buffer and get more current. But there is another way.
First, with the JFETs. When paralleling JFETs, I'd highly recommend using a separate source resistor per JFET and connecting the other end of the source resistor to ground. In this post, we are setting it to ~10 mA per JFET. So, you would have one 4k7 resistor per JFET when paralleling.
The JFETs to be paralleled should have as close Vgs as possible at the operating point. So, yes, use a 47R resistor in the source with a 9V battery and pick JFETs within < 0.1 Vgs (at worst) of each other when producing the same current. This might take a few parts to get good matches.
Another way, and what I would do, is to use a bigger part. I know you have some KSC3503 in your stash because we went through them during development. That part can take more heat and you can put a small heatsink on it. See the schematic below. There may be other small part changes on the schematic--ignore them--the things to focus on are Q1 and R12. If you adjust Vd for 40V, you will have about 40V - 0.7V (one diode drop), or about 40V, on the emitter pin of the KSC3503. So, then if you put a 1k resistor from the emitter to ground, you will get 40 mA through it. Plenty of juice. Be sure to put a heatsink on the KSC3503 and the 1k resistor will need to be properly sized for heat dissipation as well. If you put a 2k resistor there, you will get 20 mA through it--plenty for most applications.
In fact, this is how I have it configured. I have it at a little less than 2k for about 25-30 mA and it now has enough grunt to drive the Elmo. The sound is very close to the JFET buffer, very transparent.
I'll post an updated PCB with the footprint for a KSC3503.
ra7
Thanks for posting your beefy driver, three questions if I may:
1). In post 1863 you had a J113 below a DN2535 but in post 1913 you got rid of the J113 and replaced it with a resistor, is this for better sound?
2). In post 1913, I think R5 should be on the other side of the 10uF output cap, as drawn it is in parallel to the 1k resistor, lowering the resistance hence increasing the output current beyond your 40mA.
3). If I was to build your Elmo, how do I eliminate one capacitor by mating the SCG which has an output cap to Elmo which has an input cap and biasing arrangement without stuffing up the lateral's bias and input resistance etc?
Thanks for posting your beefy driver, three questions if I may:
1). In post 1863 you had a J113 below a DN2535 but in post 1913 you got rid of the J113 and replaced it with a resistor, is this for better sound?
2). In post 1913, I think R5 should be on the other side of the 10uF output cap, as drawn it is in parallel to the 1k resistor, lowering the resistance hence increasing the output current beyond your 40mA.
3). If I was to build your Elmo, how do I eliminate one capacitor by mating the SCG which has an output cap to Elmo which has an input cap and biasing arrangement without stuffing up the lateral's bias and input resistance etc?
1) That's just how my trials in Spice went. Result of multiple edits to the schematic in Spice. The J113 should be in the SCG CCS.
2) Yes, that's right. It's a remnant from a previous trial in Spice.
3) You can eliminate the cap in the SCG or the Elmo. If putting SCG and Elmo into the same chassis, this makes a lot of sense.
Here's where my standalone Elmo (when not driving with beefy SCG) development stands. This is for illustration purpose only. It WILL definitely change before calling it final. The point here is that you could direct couple the lateral FET driver (KSC3503) with the output stage, but the input will need a cap. This is easier to do when building the power amp section separate from the gain stage (standalone Elmo). Alternatively, you can direct couple the driver to SCG (as in post 1913) and put a cap between the driver and the lateral FET, thereby saving one choke (L2 below). The choke in the schematic below (L2) is simply so we don't need two power supplies for Elmo, otherwise to get symmetrical swing, we'd need a bigger supply on Q1 or a bipolar supply.
I just listened to Beefy SCG driving Elmo. It makes wonderful music!
2) Yes, that's right. It's a remnant from a previous trial in Spice.
3) You can eliminate the cap in the SCG or the Elmo. If putting SCG and Elmo into the same chassis, this makes a lot of sense.
Here's where my standalone Elmo (when not driving with beefy SCG) development stands. This is for illustration purpose only. It WILL definitely change before calling it final. The point here is that you could direct couple the lateral FET driver (KSC3503) with the output stage, but the input will need a cap. This is easier to do when building the power amp section separate from the gain stage (standalone Elmo). Alternatively, you can direct couple the driver to SCG (as in post 1913) and put a cap between the driver and the lateral FET, thereby saving one choke (L2 below). The choke in the schematic below (L2) is simply so we don't need two power supplies for Elmo, otherwise to get symmetrical swing, we'd need a bigger supply on Q1 or a bipolar supply.
I just listened to Beefy SCG driving Elmo. It makes wonderful music!
Dear Rahul,
You have confused the easily confused one!
Would L2 be a useful addition when driving the THF51, also? Would a higher Henry choke be OK? Since the current will be quite small there are not any 2 Hy chokes that I am aware of that are not rather large. But plenty of smaller ones with more Henrys. The resistance was not specified - that could easily make the difference here!
Getting ready to start over on my SCG - trying to remember what you have said in the past and cannot find what I think I was looking for.
Is J175 the most critical part for noise? I think you said that in the past.
If that is true do you think there could be any advantage in using those expensive "low noise" variants at MOUSER? I remember I bought some of the expensive J113s and they were not usable with SCG in the CCS. At that point I had no idea how to adjust the current so that thought is not to be taken seriously. I see there are old new stock variants on EBAY - I was motivated to look by ElArte who said they measured further away from zero than any he had.
Is there a "best" J175?
Since I am using Fe2022s before SCG if there is a loss of gain I can regain it with Fe2022 - but I wonder how much gain can be lost between a J175 that measures far away from zero versus one that is close to zero? An estimate?
Sorry to be so inquisitive.
THANKS and take care,
You have confused the easily confused one!
Would L2 be a useful addition when driving the THF51, also? Would a higher Henry choke be OK? Since the current will be quite small there are not any 2 Hy chokes that I am aware of that are not rather large. But plenty of smaller ones with more Henrys. The resistance was not specified - that could easily make the difference here!
Getting ready to start over on my SCG - trying to remember what you have said in the past and cannot find what I think I was looking for.
Is J175 the most critical part for noise? I think you said that in the past.
If that is true do you think there could be any advantage in using those expensive "low noise" variants at MOUSER? I remember I bought some of the expensive J113s and they were not usable with SCG in the CCS. At that point I had no idea how to adjust the current so that thought is not to be taken seriously. I see there are old new stock variants on EBAY - I was motivated to look by ElArte who said they measured further away from zero than any he had.
Is there a "best" J175?
Since I am using Fe2022s before SCG if there is a loss of gain I can regain it with Fe2022 - but I wonder how much gain can be lost between a J175 that measures far away from zero versus one that is close to zero? An estimate?
Sorry to be so inquisitive.
THANKS and take care,
Not at all, these are great questions and make me think.
I didn't look at the specifics of the J175 with respect to noise, thinking it is mostly in the same group as other JFETs in the J1XX series. I haven't yet explored the relationship between voltages and currents of the J175 and noise and the SCG operating point possibilities. That might be fun to look into.
All my J175s were from Digikey or Mouser, the cheapest ones only, and measure about the same in the Vg range. Gain for a JFET is highest near 0V Vgs. ElArte had some other kinds of J175 and they did measure farther away. Yes, that would be a fun experiment to see what that does to the sound. Gain might be higher or lower, will have to be looked at with the actual samples. If being further negative from 0V sounds better, even at the expense of some gain. It does to me with the J111 and J175, so it might be worth doing the experiment. Maybe El Arte can chime in with his experience with the J175 variants.
The ElMo is in the same chassis as the TDV was, so I haven't had a chance to test the beefy SCG with an SIT yet. It will not hurt to have a beefier driver. Based on Papa's published articles though, it does not appear that the SIT needs beefy driver. Indeed in the SIT 4, a single 2sk170 is driving the SIT. So, the original JFET buffer should work just fine.
L2 should be about 70-100 ohms, ideally around 85 ohms. I've just received the Hammond 156M, which fits the bill.
Will follow-up further.
I didn't look at the specifics of the J175 with respect to noise, thinking it is mostly in the same group as other JFETs in the J1XX series. I haven't yet explored the relationship between voltages and currents of the J175 and noise and the SCG operating point possibilities. That might be fun to look into.
All my J175s were from Digikey or Mouser, the cheapest ones only, and measure about the same in the Vg range. Gain for a JFET is highest near 0V Vgs. ElArte had some other kinds of J175 and they did measure farther away. Yes, that would be a fun experiment to see what that does to the sound. Gain might be higher or lower, will have to be looked at with the actual samples. If being further negative from 0V sounds better, even at the expense of some gain. It does to me with the J111 and J175, so it might be worth doing the experiment. Maybe El Arte can chime in with his experience with the J175 variants.
The ElMo is in the same chassis as the TDV was, so I haven't had a chance to test the beefy SCG with an SIT yet. It will not hurt to have a beefier driver. Based on Papa's published articles though, it does not appear that the SIT needs beefy driver. Indeed in the SIT 4, a single 2sk170 is driving the SIT. So, the original JFET buffer should work just fine.
L2 should be about 70-100 ohms, ideally around 85 ohms. I've just received the Hammond 156M, which fits the bill.
Will follow-up further.
Hi Rahul, I'm following your road through the Elmo, as I'm a big fan of lateral mosfets (I've been using them in Juma's Cubie 1, 2 and 3, as well in Svaja https://www.diyaudio.com/community/threads/25w-class-a-amp-with-lateral-mosfets.183482/post-2474936 and in the Transnova-style with complementary folded cascode f.e. https://www.diyaudio.com/community/threads/transnova-schade-os-amp.359000/post-6648312 and even in a beefy headphones amp (Sony VFET FE + LatFET output stage). Could you elaborate more on "The problem is it needs a beefy driver": do you mean output impedance of the buffer? and/or current? In my experience, a pair of complementary latfets are happy with 7,5 mA and a 2SK170-based buffer is absolutely fine to drive them.
As I'm current working on a slighty modified version of Zenductor 2, I'd like to play with output Latfet (I'm using the old good Hitachi/Renesas) still maintaining low supply voltage (16V) and idle current (my speakers are in the 100dB/W league, thus I don't need more than, say, 10W). The current configuration is then jfet buffer - autoformer - output stage, I wonder if a second buffer stage (after the autoformer) would be used (if output impedance of the buffer, multiplied by the autoformer ratio isn't enough). How much is the minimum inductance for L2? Would a (cheap) Triad 1H 240mA https://www.mouser.it/datasheet/2/410/C_24X-1892581.pdf be sufficient for the task?
Lots of questions, sorry...😉
Guido
The Siliconix J175 I tried, I bought from a surplus store in NJ. These rated high, as @ra7 mentioned.
They sound similar to others I tried and the noise and distortion is also quite the same.
I have not gone as far as to check the gain, but I also got near 40 Vpp of swing before clipping at the output.
They sound similar to others I tried and the noise and distortion is also quite the same.
I have not gone as far as to check the gain, but I also got near 40 Vpp of swing before clipping at the output.