Hello friends,
I was thinking of making a spud test bed to serve as either preamp or headphone amp, where i can evaluate a lot of different tubes (most interested in DHT's) using a daughterboard, avoiding having to rebuild a new amp each time while also giving excellent performance.
I know its not exactly classical and a lot of people will raise there eyebrows and scoff at the lack of iron, but here are some design requirements:
-Should be able to run headphones with low distortion using even anemic DHT's
-Great performance into light loads (preamp duty)
-No iron, not as choke load or output (such as parafeed) - complicates things in many ways, not the least of which is that its hardly repeatable for others who wish to build, as everyone will find different iron, and doubles the BOM cost
-OTL
-Transparent representation of a certain tube (i.e, not having to have an amp with several stages, mixing and matching various input and output tubes for the amp to perform well, at which point sound depends on interplay of too many things, in addition to each stage bringings its own color, in theory also complicating the harmonic spectrum and so on)
-Run 24/7, for months on end
-No (global) feedback
Naturally a hybrid mu-follower (gyrator) fits the bill and sounds excellent and transparent. The output impedance is 1/gfs of bottom device in the gyrator which we can select for, and does not depend on the tube, which automatically opens a door for driving all sorts of loads. In addition it provides a high impedance load to the tube. Gain is nearly mu. PSRR is very high. A lot has been written on the topic already that can be found elsewhere, Ale Moglia's presentation is an excellent primer.
In addition I also wanted to add an optional buffer output stage, this one using OPA1656.
First jumper JP2 just adds a high value film cap, used when driving headphones. Then the following selection jumper either goes out to output, or to the aformentioned opamp buffer. The opamp does have resistors for adjusting gain just in case (because it would be difficult to tack on later), but the jumper JP3 and JP4 turn it unity-gain. The R10 100k resistor is just so the amp is never unloaded, but if we select the opamp jumper it forms a voltage divider with R13, so that the input is on 1/2 the rail supply, since the supply is unipolar.
You might notice there's no toob, that's the whole daughterboard bit. There's a header J7 and J6 to hold a different board that will have the relevant bits (noval, octal, etc socket). Could make a number of those daughterboard depending on what you want to put on there, but i also made the daughterboards so you can jumper things around if necessary to match the pins. By default, it is pin pefect for ECC88 family of noval tubes, because there's a lot of tubes there you can just drop in with no finnicking.
Cathode biasing is using SiC diode, though you could just as well solder some LED's if you so prefer. Using the jumpers you can select the voltage, each step adding about 0.9V. There's one jumper extra at the top giving the ability to disable the SiC array alltogether, allowing either a resistor to be added in place, or to futureproof if i later decide to have biasing on one of the daughterboards.
On the bottom right there's also a selector for running either 6.3v heaters or 12.6v heater tubes on 6.3v, again selection jumper.
You might also notice JP19, used for when DHT tubes are installed, since the heater is the cathode, then you would unjumper JP17 and put the filament reg on pins 2 and 3 of JP13. The reason JP19 is is 2 way selectable is i dont know yet if it makes any difference to have the biasing on pin 8 or 1-7 (for tubes like 4P1L where it is common to "fold" the cathode for more even heating)
In the bottom left corner the 3 jumper ( JP14, JP15 and JP16) are to run tubes that have 2 triodes in the same envelope (like a lot of novals) in parallel. Since you're wearing out the tube anyway might as well make use of both. Again something to experiment with.
There's probably too many jumpers on this board but i hate when i wished i had put one when i hadnt, more than the other way around.. if there's ever a v2 a bunch of them could be dropped
It is relatively simple and nothing much to say, but since im the kind of person that would fall inside an open manhole i'd like a sanity check and any thoughts, any obvious fails or suggestions...
I was thinking of making a spud test bed to serve as either preamp or headphone amp, where i can evaluate a lot of different tubes (most interested in DHT's) using a daughterboard, avoiding having to rebuild a new amp each time while also giving excellent performance.
I know its not exactly classical and a lot of people will raise there eyebrows and scoff at the lack of iron, but here are some design requirements:
-Should be able to run headphones with low distortion using even anemic DHT's
-Great performance into light loads (preamp duty)
-No iron, not as choke load or output (such as parafeed) - complicates things in many ways, not the least of which is that its hardly repeatable for others who wish to build, as everyone will find different iron, and doubles the BOM cost
-OTL
-Transparent representation of a certain tube (i.e, not having to have an amp with several stages, mixing and matching various input and output tubes for the amp to perform well, at which point sound depends on interplay of too many things, in addition to each stage bringings its own color, in theory also complicating the harmonic spectrum and so on)
-Run 24/7, for months on end
-No (global) feedback
Naturally a hybrid mu-follower (gyrator) fits the bill and sounds excellent and transparent. The output impedance is 1/gfs of bottom device in the gyrator which we can select for, and does not depend on the tube, which automatically opens a door for driving all sorts of loads. In addition it provides a high impedance load to the tube. Gain is nearly mu. PSRR is very high. A lot has been written on the topic already that can be found elsewhere, Ale Moglia's presentation is an excellent primer.
In addition I also wanted to add an optional buffer output stage, this one using OPA1656.
-Modern opamp like OPA1656 in unity gain, <2Vrms, is indistinguishable from a piece of wire (IMHO)
-Takes care of heavy lifting, provides even lower Zo, can use small high quality coupling capacitor instead of 50uf+ film chonkers
-Easy to implement (low component count, low cost)
-Now in addition to rolling tubes, you can also be a snob about rolling opamps, all in the same amp!
-Futureproofs it for use with 32R headphones, and enables use of anemic toobz with no issue
-Takes care of any possible slew rate issues, tube not wanting to run the HP cable capacitance, etc..
-Since this will be powered 24/7, i might also decide to run lower currents and be able to get away with it, in addition to essentially having the front end just be a preamp working into a very easy load, giving low distortion (remember these are not speakers where amp is almost negligible, HD650s are very revealing)
-Any other output stage i tried performed worse than just taking the output off the gyrator, while wasting a lot of power and requiring cooling solutions (i.e MOSFET source follower etc), so if we're gonna have anything at least make it not worse than gyrator alone.
-If we dont want to use it, we dont have to. Its just there as an option
For powering the opamp, the simplest solution seems to be a very low noise, very high PSRR LP5097-4.5 LDO. It will just about accept up to 6.5v input so i can take it from the filament heater, and OPA1656 will just about work with 4.5, so we squeeze through this hurdle. Since it will work rail to rail, means we can eek out 1.5Vrms, which is 107db on HD650's, about the loudness of a car horn, I dont think listen to half as loud.
-Takes care of heavy lifting, provides even lower Zo, can use small high quality coupling capacitor instead of 50uf+ film chonkers
-Easy to implement (low component count, low cost)
-Now in addition to rolling tubes, you can also be a snob about rolling opamps, all in the same amp!
-Futureproofs it for use with 32R headphones, and enables use of anemic toobz with no issue
-Takes care of any possible slew rate issues, tube not wanting to run the HP cable capacitance, etc..
-Since this will be powered 24/7, i might also decide to run lower currents and be able to get away with it, in addition to essentially having the front end just be a preamp working into a very easy load, giving low distortion (remember these are not speakers where amp is almost negligible, HD650s are very revealing)
-Any other output stage i tried performed worse than just taking the output off the gyrator, while wasting a lot of power and requiring cooling solutions (i.e MOSFET source follower etc), so if we're gonna have anything at least make it not worse than gyrator alone.
-If we dont want to use it, we dont have to. Its just there as an option
For powering the opamp, the simplest solution seems to be a very low noise, very high PSRR LP5097-4.5 LDO. It will just about accept up to 6.5v input so i can take it from the filament heater, and OPA1656 will just about work with 4.5, so we squeeze through this hurdle. Since it will work rail to rail, means we can eek out 1.5Vrms, which is 107db on HD650's, about the loudness of a car horn, I dont think listen to half as loud.
First jumper JP2 just adds a high value film cap, used when driving headphones. Then the following selection jumper either goes out to output, or to the aformentioned opamp buffer. The opamp does have resistors for adjusting gain just in case (because it would be difficult to tack on later), but the jumper JP3 and JP4 turn it unity-gain. The R10 100k resistor is just so the amp is never unloaded, but if we select the opamp jumper it forms a voltage divider with R13, so that the input is on 1/2 the rail supply, since the supply is unipolar.
You might notice there's no toob, that's the whole daughterboard bit. There's a header J7 and J6 to hold a different board that will have the relevant bits (noval, octal, etc socket). Could make a number of those daughterboard depending on what you want to put on there, but i also made the daughterboards so you can jumper things around if necessary to match the pins. By default, it is pin pefect for ECC88 family of noval tubes, because there's a lot of tubes there you can just drop in with no finnicking.
Cathode biasing is using SiC diode, though you could just as well solder some LED's if you so prefer. Using the jumpers you can select the voltage, each step adding about 0.9V. There's one jumper extra at the top giving the ability to disable the SiC array alltogether, allowing either a resistor to be added in place, or to futureproof if i later decide to have biasing on one of the daughterboards.
On the bottom right there's also a selector for running either 6.3v heaters or 12.6v heater tubes on 6.3v, again selection jumper.
You might also notice JP19, used for when DHT tubes are installed, since the heater is the cathode, then you would unjumper JP17 and put the filament reg on pins 2 and 3 of JP13. The reason JP19 is is 2 way selectable is i dont know yet if it makes any difference to have the biasing on pin 8 or 1-7 (for tubes like 4P1L where it is common to "fold" the cathode for more even heating)
In the bottom left corner the 3 jumper ( JP14, JP15 and JP16) are to run tubes that have 2 triodes in the same envelope (like a lot of novals) in parallel. Since you're wearing out the tube anyway might as well make use of both. Again something to experiment with.
There's probably too many jumpers on this board but i hate when i wished i had put one when i hadnt, more than the other way around.. if there's ever a v2 a bunch of them could be dropped
It is relatively simple and nothing much to say, but since im the kind of person that would fall inside an open manhole i'd like a sanity check and any thoughts, any obvious fails or suggestions...
Attachments
Last edited:
Ive built a 2P29L pre/headphone amp with Ale's hybrid mu follower load similar to your design . For headphone duty I first tried it driving Sennheiser HD800's direct (via a 100uf DC Link polyprop cap). I thought it sounded good but when I measured it I found I had quite a lot of distortion. I cant put my hands on the measurements I made just now unfortunately. So I tried using a 4:1 step down transformer in parafeed which measured much
better distortion wise and sounded better. Of course gain is reduced but that isnt a problem for headphone use. I also use it to drive a MoFo power buffer direct from the mu follower and this works very well indeed.
better distortion wise and sounded better. Of course gain is reduced but that isnt a problem for headphone use. I also use it to drive a MoFo power buffer direct from the mu follower and this works very well indeed.
Thank you for your experience, it makes sense that 2P29L would have an easier life driving an easier load of the primary and that it would improve things, just like using the MoFo power buffer. In my case same objective accomplished with opamp in unity gain (it was primarily included for these kind of anemic tubes that even with gyrator would not be happy with headphone loads). Of course you could tack on the same step down transformer and other things here and have the same improvement, it is food for thought. I also want to try the 2P29L as well
Spudnik base board, gyrator at top left, opamp buffer top right, SiC diode array in middle, place for daughterboard on the bottom
Here are the daughterboards. All the jumpers look silly but it is necessary to adapt any tube (noval or octal) to the baseboard
Just for the heck of it, DHT filament heater that integrates everything into one board, SMPS with 4th order LC filter with over -140db of attenuation at 60khz and increasing, followed by a good cap multiplier for the filament + bringing even more noise attenuation and stiff CSS for the filament -
Here are the daughterboards. All the jumpers look silly but it is necessary to adapt any tube (noval or octal) to the baseboard
Just for the heck of it, DHT filament heater that integrates everything into one board, SMPS with 4th order LC filter with over -140db of attenuation at 60khz and increasing, followed by a good cap multiplier for the filament + bringing even more noise attenuation and stiff CSS for the filament -
How are you doing with this project? I have Ale gyrator boards and starting to test tubes. Right now with ECC88 tube. Problems I have are that gain is too big and I get easily distortion when output close to 2 V. Rk I am usin green LED and I have tried with two green LED-s and same. Any thoughts how to increase input gain to 1-2 V and not having heavy distortion. how to lower gain and what values resistor or LED-s I should use to get it right? Anode voltage is 90 V.
As I wrote to @vilstone in PM, the most important thing of choosing the right tube (Ua, Ia, gm) for this fuction (gain, driving capacity, estimated output swing) AND the appropriate operating point.
If you ignore the tube's datasheet, the selection of operating point would be the "grasping at straws".
Use "paper" datasheets, or use online loadline calculator.
http://trioda.com/tools/triode.html
https://www.vtadiy.com/loadline-calculators/loadline-calculator/
If you ignore the tube's datasheet, the selection of operating point would be the "grasping at straws".
Use "paper" datasheets, or use online loadline calculator.
http://trioda.com/tools/triode.html
https://www.vtadiy.com/loadline-calculators/loadline-calculator/
Basically what euro21 said, good answer to the question.
And for how im doing with the project, i got the boards, soldered everything, started hooking things up in a case, but along the way i changed jobs and moved living place, things got turbulent and i can only hold so many things on my mental stack so it got knocked off for the time being....i dont even get time to enjoy music these days.. i plan to continue on if nothing else i want to see how it pans out.
And for how im doing with the project, i got the boards, soldered everything, started hooking things up in a case, but along the way i changed jobs and moved living place, things got turbulent and i can only hold so many things on my mental stack so it got knocked off for the time being....i dont even get time to enjoy music these days.. i plan to continue on if nothing else i want to see how it pans out.