This is my first "from scratch" design and I'm hoping to get some feedback before I make any serious mistakes. I put "from scratch" in quotes since I certainly drew inspiration from many other designs posted here and elsewhere (particularly from Pete Millett and John Broskie).
I built a pair of 300B P-SET monoblocks last year and have been thrilled with the sound, but I'd like to have a bit more power and try something different. The SET amps were based on the ANK Interstage kit with lots of embellishments (a couple photos attached), but using the basic circuit design and transformers from the kit.
So this will be my first tube amp build without starting with an existing design. The attached schematic shows the basic amp circuitry that I'm currently planning. My goal is to get 55-60W class A1. I don't need a lot of gain since my DAC and Preamp can easily source several volts. I'm planning to use a CLCLC supply for the 425V B+ (currently using 220uF caps and 4H chokes). The gain/driver stage will use a Neurochrome Maida regulator to supply regulated 300V. I'm planning regulated DC supplies for all the filaments.
The output stage bias will be controlled by a Tent Labs Negative Bias Supply, so each of the 100K resistors will connect separately to this controller.
I'm planning a power supply controller which will power up the filaments first, then bias controller, then the B+. This will run off a 12V power supply which will also provide a -12V reference (using a separate regulator) for the LTP current source.
My current choice for the OPT is a Toroidy 2K with CFB windings. I'm going to use these windings to provide symmetrical feedback to the input stage LTP cathodes. I will probably incorporate a couple switches to select global feedback level and triode vs ultra-linear output tube operation. I've read that toroidal transformers can work very well for tube OPTs as long as there is minimal DC imbalance, and from what I've read, the Tent Labs bias controller does a good job with this.
I've currently selected a 12AX7 for the gain stage. I'm not sure if this is the best choice, but I wanted something with enough gain that I could avoid two stages. If anyone has suggestions for other tube types to consider, I'd appreciate it.
Thanks.
Below are a couple photos of my 300B PSET monoblocks completed last fall.
I built a pair of 300B P-SET monoblocks last year and have been thrilled with the sound, but I'd like to have a bit more power and try something different. The SET amps were based on the ANK Interstage kit with lots of embellishments (a couple photos attached), but using the basic circuit design and transformers from the kit.
So this will be my first tube amp build without starting with an existing design. The attached schematic shows the basic amp circuitry that I'm currently planning. My goal is to get 55-60W class A1. I don't need a lot of gain since my DAC and Preamp can easily source several volts. I'm planning to use a CLCLC supply for the 425V B+ (currently using 220uF caps and 4H chokes). The gain/driver stage will use a Neurochrome Maida regulator to supply regulated 300V. I'm planning regulated DC supplies for all the filaments.
The output stage bias will be controlled by a Tent Labs Negative Bias Supply, so each of the 100K resistors will connect separately to this controller.
I'm planning a power supply controller which will power up the filaments first, then bias controller, then the B+. This will run off a 12V power supply which will also provide a -12V reference (using a separate regulator) for the LTP current source.
My current choice for the OPT is a Toroidy 2K with CFB windings. I'm going to use these windings to provide symmetrical feedback to the input stage LTP cathodes. I will probably incorporate a couple switches to select global feedback level and triode vs ultra-linear output tube operation. I've read that toroidal transformers can work very well for tube OPTs as long as there is minimal DC imbalance, and from what I've read, the Tent Labs bias controller does a good job with this.
I've currently selected a 12AX7 for the gain stage. I'm not sure if this is the best choice, but I wanted something with enough gain that I could avoid two stages. If anyone has suggestions for other tube types to consider, I'd appreciate it.
Thanks.
Below are a couple photos of my 300B PSET monoblocks completed last fall.
Last edited:
What I'd suggest you is to move the coupling caps before the power drive, and DC couple the CCS loaded mosfets to the output tubes' grids (one per tube would be better).
Given the fact that even so called matched output tubes are seldom exactly matched and the sensitivity of toroidal outputs for dc imbalance I'd add separate adjustable negative grid voltages for each of the KT 88's. Besides you then have a wider choice regarding output tubes (EL 34's, 6l6gc's) and they don't have to be exactly matched.
If you want to avoid magnetizing your outputs at all cost, adjust the output tubes bias currents with shorted primaries and when both halves are equal undo the shorts.
Regarding your schematic, as you use the input transformer as phase splitter the current source for the ecc 83 is unnecessary (unless you think it sounds better).
Concerning the fet followers you use, Alan Wright when discussing his pre amp comparing the two versions he designed, one with cathode followers (E88CC's) one with fet followers stated that the cathode follower version sounded much better. So may be you can try 12au7's instead. Or try both and see what you like best.
If you want to avoid magnetizing your outputs at all cost, adjust the output tubes bias currents with shorted primaries and when both halves are equal undo the shorts.
Regarding your schematic, as you use the input transformer as phase splitter the current source for the ecc 83 is unnecessary (unless you think it sounds better).
Concerning the fet followers you use, Alan Wright when discussing his pre amp comparing the two versions he designed, one with cathode followers (E88CC's) one with fet followers stated that the cathode follower version sounded much better. So may be you can try 12au7's instead. Or try both and see what you like best.
Of course ! , for best results each of four KT88 have to be biased separate ,
and just to correct myself , V3 can be Ecc82 , 6SN7 , or 12BH7 , or simmilar twin triode tube ...
Best Regards!
edit , this fully balanced front section end do not use any input transformer .
Last edited:
If you plan to have an amp working totally in A1, voltage shound be lower.
Around 300V B+, 4k Raa per pair of tubes and 110mA per tube at idle, you get around 30 Wrms in class A1 per pair of tubes.
With 425V you get 60 Wrms per pair of tubes, but this is AB1.
Just consider how you want to work, because Toroidy transformer cannot be used above 80 Wrms.
First of all sorry Kay for giving a nearly identical reaction you gave but I saw yours only after I had posted mine.
Jaytor, why make a dc supply for all the heaters? A power amp with ac on the filaments can be completely silent .
Jaytor, why make a dc supply for all the heaters? A power amp with ac on the filaments can be completely silent .
You can run the 12ax7 CCS from the negative bias supply and save yourself the -12V supply. Consider some servo bias boards. You will need a more accurate model of the transformer if you want to avoid HF stability issues with the NFB. Include the leakage inductance (coupling factor) and the effective primary capacitance to the OPT model. You may need to add a dominant pole to the design. So I would provision series RC between plates of U1 and U2 and parallel C across R26 and R27. You may wish to use 6550's but if not servo bias the grid leaks need to be 47K. I think having the coupling caps after the mosfets will be fine although with 1uF and 100R you could get blocking distortion. Looks pretty good.
Last edited:
Thanks for the feedback. In my OP, I mentioned that I was planning to use the Tent Labs Bias supply. This has separate bias outputs and separate current sources for each of the four output tubes. I'm not sure exactly how this works to adjust bias, but they claim that it keeps bias between the four tubes very close. I assume some form of servo circuit.
I realized after my post that my OPT model was wrong and have been doing some simulations with a corrected model.
I have also been considering switching to a single pair of KT150s instead of two pair of KT88s. Anyone tried both options?
Thanks again.
I realized after my post that my OPT model was wrong and have been doing some simulations with a corrected model.
I have also been considering switching to a single pair of KT150s instead of two pair of KT88s. Anyone tried both options?
Thanks again.
I looked at KT150 - you need more B+. I think the linearity of the 2 off KT88 is better. You could use E88CC/6DJ8 cathode follower rather then mosfets if you want more valve but that's up to you. I've used the Tent Labs in the past - make sure your negative supply is filtered as there's little rejection. Other than that they will work fine for you. You don't necessary need balanced drive on the 12ax7 the CCS tail will do this for you saving you an transformer. You can use the other side for the NFB which can then come from the speaker.
Below is my version of the servo bias - you can play in ltspice. J1 is B+, J3 is power up mute to keep bias negative until tube is warmed. J4 is grid, J5 is cathode including 10R cathode resistor.
Below is my version of the servo bias - you can play in ltspice. J1 is B+, J3 is power up mute to keep bias negative until tube is warmed. J4 is grid, J5 is cathode including 10R cathode resistor.
Last edited:
@baudouin0 - thanks for the info. That's helpful. The Tent Labs product I'm looking at includes it's own power supply. Are you saying that additional filtering is required?
I've been vacillating between using an input transformer and not - still undecided. I kind of like the idea of having the fully symmetrical but that's mostly from a circuit aesthetics perspective 🙂. Not sure if it would make a positive (or any) difference in sound quality.
I'm going to have to put this project on hold for a little while but will get back to it later this summer.
Thanks everyone for your insights.
I've been vacillating between using an input transformer and not - still undecided. I kind of like the idea of having the fully symmetrical but that's mostly from a circuit aesthetics perspective 🙂. Not sure if it would make a positive (or any) difference in sound quality.
I'm going to have to put this project on hold for a little while but will get back to it later this summer.
Thanks everyone for your insights.