Totally possible. I've experimented with up to 20% ratio (Schade feedback). At a 20% ratio, say 2k and 10k Schade resistors, the gain is about 6X, which is about 15 db, instead of 20 db with 1k/10k.
Here's some rough numbers for various values of Rg and assuming 10k for Rsch:
Rg, gainX, db gain
1k, 10X, 20 db
2k, 6X, 15 db
3k, 4.5X, 13 db
4k, 3.5X, 11 db
5k, 3X, 9.5 db
I have not tested these values.
(If we keep going, we'll reach 10k/10k, and then it becomes like a regular source follower with unity gain. The curves look very straight. For example, see the figure in Mike's MoFo write-up.)
This whole thing began with me building an F4. Then, the usual question, "Can it drive an F4?" To drive the F4, I started with a BA-3 preamp, which is also quite phenomenal BTW, and then onto the usual #26->01a->10Y tube pre journey. But then it got ridiculous searching for 100 year valves and that's when I landed on this topology. Triodes on the cheap with great performance AND sound. The intention with the SCG was to drive an F4. Now, I've ended up with a VFET follower, which is also a zero gain power buffer, but that's where the SCG shines in my opinion and will be able to contribute the majority of its character. Doesn't mean its gain cannot be reduced for different purposes.
Feel free to experiment, it won't break anything. Test first with a cheap amp and speakers and proceed if stable.
Last edited:
wow thanks for the details. Alternatively, maintaining the value of Rg at 1k, and varying the Rsch may just work as well, am i correct? i will order some resistors to be ready to try the gain adjustments. i am targeting maybe around 10db of gain, a reduction from the default gain of your design of 20db.
The first generation of the Tubelab SE (TSE) board had provisions for a volume pot. When the second generation (TSE-II) board was in development I asked for inputs, suggestions and complaints via the forum and email. One of the top ten change requests was to ditch the volume pot. Virtually nobody ever populated it as it limits the board mounting possibilities, and there are just too many different volume adjustment scenarios and pots to get consensus on any given footprint. The space was wisely used for heat management and extra mounting / ventilation holes.
This circuit was a spin-off from my UNSET design which used pentode vacuum tubes for the gain device. The "Schade" feedback resistors control the "triodness" of the gain device, which influences its gain, output impedance and distortion profile. All three interact, but there is usually a happy medium somewhere that will satisfy a given use case. In this case the circuit is used as a gain stage preamp with minimal loading, so one can optimize for gain and distortion. It will also function as an output stage where output impedance is important, so gain is usually traded for damping factor.
I have collected all the parts for an all solid state version of the UNSET which will use some large UJ3N065080K3S SiC fets running on a couple hundred volts into a 600 ohm vacuum tube OPT for the output stage and something similar to this for the gain stage. It is all waiting for space on the bench. I need room for the 20 pound heat sink!
This circuit was a spin-off from my UNSET design which used pentode vacuum tubes for the gain device. The "Schade" feedback resistors control the "triodness" of the gain device, which influences its gain, output impedance and distortion profile. All three interact, but there is usually a happy medium somewhere that will satisfy a given use case. In this case the circuit is used as a gain stage preamp with minimal loading, so one can optimize for gain and distortion. It will also function as an output stage where output impedance is important, so gain is usually traded for damping factor.
I have collected all the parts for an all solid state version of the UNSET which will use some large UJ3N065080K3S SiC fets running on a couple hundred volts into a 600 ohm vacuum tube OPT for the output stage and something similar to this for the gain stage. It is all waiting for space on the bench. I need room for the 20 pound heat sink!
Yes, you could change Rsch. Lower values of Rsch won't affect the trimmer pot, but it would mean more current through the Schade network. For example, if you reduced Rsch to 5k, then,
gainX = (Rsch + Rg)/Rg = (5+1)/1= 6X, or 15 db.
The network now draws about 60V/(6k+trimpot). Let's say trimpot is at middle setting of 5k. So, 60V/11k = 5.4 mA. You would have to subtract this value from the CCS. Not a big difference from where we are, but if you go lower, check the current and also make sure the dissipation in the network and trimpot is okay.
It would also impact the output impedance, which is usually dominated by the internal resistance of the device at the given operating point.
Good to know, George! Thank you!
Awesome! Really looking forward to what you come up with. I have been thinking about it also and I was wondering if a high voltage low current version is better than a low voltage high current version. Here's a thought:
The CCS means double the B+ than with a transformer, but this is not such a problem with SS devices because you can go to a higher current. But if you go to a higher current, then the transformer has to be bigger, and this is the reason for the above circuit. Something like 200V B+ and 500 mA might very well be possible. It neatly sidesteps the problems of this design with tubes. You are not hit with a very high B+, nor do you have to get a really large transformer. But of course, there must be trade-offs. I'm just throwing out ideas that I'd like to pursue.
I bet your 600 ohm OPT is really big!
Attachments
Last edited:
I'm doing a tube version first because I have all the parts.
The circuit you show can be done with a suitable choke instead of the CCS. Tube amp builders call this "parafeed." as in parallel feed. They don't realize that the two inductances are effectively in parallel thus reducing the value of the total inductance. I have experimented with this design using a big sweep tube for the CCS and a KT88 for the output tube. An old experiment is here, new experiments will use bigger output tubes:
http://tubelab.com/articles/circuits/active-loaded-se-output-stages/
The 600 ohm OPT's that I have are these, they weigh 3kg each and will probably do 20 to 25 watts if they work like the similar 1500 ohm Toroiry's that I have been using.
https://www.tme.com/us/en-us/details/ttg-6c33cse/toroidal-transformers/toroidy/
At one time I had three 5 Hy @ 5 Amp chokes. They WERE huge and therefore did not make the trip when I moved everything I owned and was willing to keep 1200 miles. The thought was for a choke loaded SS or multi regulator triode source or cathode follower output stage directly driving a speaker. I still have a bunch of 6336A tubes. Each of the two sections is rated for 400 mA of continuous current.
Member
Joined 2006
A gyrator will work, but much like the CCS circuit, the power supply feeding the CCS or gyrator must be about twice the idle plate voltage to emulate the voltage rise caused by the energy stored in the transformer or chokes magnetic field being turned back int EMF as it is released.
Amazing circuit. Thank you for that!
Yes, the transformer is in parallel with the CCS for AC frequencies. I know very little about transformer design, but I read that SE transformers have to big to keep the core from saturating. With this "parafeed" construction, there is no DC current through the transformer.
Anyway, just throwing stuff on the wall to see what sticks.
Ah, I did not the reason for the double voltage was so that it can "emulate" the transformers stored energy. I was thinking about it in terms of voltage swing. As the input signal swings negative, the top of the plate/drain will swing down, and there should be enough voltage above the CCS to make sure it keeps conducting. Am I thinking about this right?
In a choke or transformer loaded common cathode / source / emitter circuit the drain voltage will be close to the B+ voltage at idle, limited only by the DCR of the choke / transformer. It will be pulled toward ground as the output device is driven into conduction. How close it gets to ground depends on the saturation characteristics of the output device. A typical audio tube like a 6L6GC saturates at around 50 volts unless it's driven into A2. A TV sweep tube can get down into the under 20 volt region without A2, and a semiconductor can go below 1 volt. During this time a large magnetic field builds in the choke or transformer. As the output device is being driven toward cutoff this magnetic field will collapse creating a voltage in the inductance of the winding. This voltage will drive the drain of the output device more positive than the B+ voltage. In a theoretical lossless circuit operated at its maximum undistorted output voltage, the drain voltage will rise to twice the B+ voltage. In a loaded (speaker on the secondary) lossy circuit the voltage will be less than 2X the B+ voltage. If the circuit is driven to cutoff the path for current flow in the transformer primary is temporarily broken and the voltage can rise above twice the B+ value. I have measured over 2500 volts in a guitar amp with a B+ of 430 volts, when overdriven about 10 dB beyond clipping while operating into a speaker at or slightly above its resonant frequency.
An artificial inductor, gyrator or CCS can emulate the impedance VS frequency curve of a real inductor, but it does not have any magnetic storage capability so the headroom for the expected voltage swing must be provided by the power supply. Twice the typical B+ is usually sufficient for a circuit that does not normally operate in clipping.
Mu-Follower Output
One of the things to try was a mu-follower output instead of the output connected connected to the bottom of the CCS. I tried this today and it sounds quite a bit better. This will definitely be on the next PCB. If the folks with PCBs want to try it out, I can post here. It involves desoldering one pin of the output cap.
One of the things to try was a mu-follower output instead of the output connected connected to the bottom of the CCS. I tried this today and it sounds quite a bit better. This will definitely be on the next PCB. If the folks with PCBs want to try it out, I can post here. It involves desoldering one pin of the output cap.
Sure Ra!
Always into tweaking for the betterment
. Details of this mod would be appreciated. TY
Always into tweaking for the betterment
. Details of this mod would be appreciated. TY
Wow, that was fast. Got a yes from both you and Alfi! I didn't want to disrupt the builds but sounds like it is a go. Okay, here it is:
Mu-Follower Mod
To achieve this on the existing PCB, the output cap pin (C103/203) closer to the outside should be wired to the inside leg of R104/204 (and not wired into the C103/203 solder pad/hole).
Pretty simple really. You can even hook it up with alligator clips, but be sure to desolder the C103/203 pin from its hole.
All this does is that it takes the output from the source of the CCS FET instead of from the drain of the gain FET. There is no change to the gain or any setting required. It has produced quite a bit of sonic ecstasy for me.
The other change to make is to take the current to 25-30 mA when using the TO-220 parts. For the STP30N10F7, the operating point is about 35-40V above the Vs voltage and 25-30 mA current. If you do this, make sure you have good heatsinking on the reg, top CCS, and gain FETs. Also, this much current will be marginal at best with the TO-92 P-channel FET (bottom FET). I am currently breadboarding it with a TO-92 part (VP0106), but it will run much happier with a TO-220 part (like the FQP ones).
Let us know what you hear.
Mu-Follower Mod
To achieve this on the existing PCB, the output cap pin (C103/203) closer to the outside should be wired to the inside leg of R104/204 (and not wired into the C103/203 solder pad/hole).
Pretty simple really. You can even hook it up with alligator clips, but be sure to desolder the C103/203 pin from its hole.
All this does is that it takes the output from the source of the CCS FET instead of from the drain of the gain FET. There is no change to the gain or any setting required. It has produced quite a bit of sonic ecstasy for me.
The other change to make is to take the current to 25-30 mA when using the TO-220 parts. For the STP30N10F7, the operating point is about 35-40V above the Vs voltage and 25-30 mA current. If you do this, make sure you have good heatsinking on the reg, top CCS, and gain FETs. Also, this much current will be marginal at best with the TO-92 P-channel FET (bottom FET). I am currently breadboarding it with a TO-92 part (VP0106), but it will run much happier with a TO-220 part (like the FQP ones).
Let us know what you hear.
Perfect!
Thanks for the details, should be a quick modification. Will try to do this in the morning. I’m really enjoying this pre as is, it gets better?…😀
Thanks for the details, should be a quick modification. Will try to do this in the morning. I’m really enjoying this pre as is, it gets better?…😀
Member
Joined 2006
I have Ale’s gyrator board that was in use for the 10Y pre. With that board, I tried both the mu out and CCS out and preferred the CCS out. The gyrator just provides some curvature on the load to match the triode curves. In Merlin’s preamp book, he says the mu out from the CCS circuit here has the same impact. I haven’t analyzed the two deeply enough but I could throw the gyrator in to see how it sounds.