Take a look at the NuTube curves:
https://korgnutube.com/wp-content/uploads/images/IaVa_IgVg_No9.jpg
The curves look beautiful until you realise that all the sweetness is happening above the max. allowed dissipation power only.
One could consider to bias the NuTube at 32 VDC / 50 microAmps, which is right at the power limit. This should allow squeezing in a CCS load to the 36 VDC supply, which would help with linearity of the stage. However, the grid bias for this operating point would need to be at +1.5 V, which is too close to the cutoff at about +3V.
Biasing with more headroom for the grid signal might require a PSU voltage higher than 36 VDC, and would move the operating point to the "knees" of the curves, which is asking for trouble.
I don't see how the NuTube would work well for our purposes.
https://korgnutube.com/wp-content/uploads/images/IaVa_IgVg_No9.jpg
The curves look beautiful until you realise that all the sweetness is happening above the max. allowed dissipation power only.
One could consider to bias the NuTube at 32 VDC / 50 microAmps, which is right at the power limit. This should allow squeezing in a CCS load to the 36 VDC supply, which would help with linearity of the stage. However, the grid bias for this operating point would need to be at +1.5 V, which is too close to the cutoff at about +3V.
Biasing with more headroom for the grid signal might require a PSU voltage higher than 36 VDC, and would move the operating point to the "knees" of the curves, which is asking for trouble.
I don't see how the NuTube would work well for our purposes.
If a small DHT is acceptable then 5676 and 6286 triodes should fit well enough (with 0.8-0.9V starved filament for reasonable life).
I would say that those curves are not very helpful for assessing voltage amplifier performance, the "resolution" when used within specs is poor and I would not dismiss it without some experimentation. They certainly show the device is capable of high peak current and so it will like to be operated in class A2 (like the old big transmitting tubes....). I guess that if one repeats the curve-tracing, rescaling the current at lower operating range with more resolution, they can improve quite a lot.
Looking at the grid current characteristics I would say that this device can be easily operated with positive bias and needs 2 conditions:
1) a DC-coupled follower at the input which provides (small) power drive...noting like traditional tubes which can require milliamps....
2) another follower at the output which provides a load in Megaohms range
Both things can be done with normal JFETs, I think. Operating conditions at 36V/28 uA approx. with +0.5V bias suggest it can provide approx. 8.5 Vrms into 1M load with low distortion with 0.7 V rms (plus the small loss of the follower) at the input.
I would try this before anything else.
P.S.
DC-coupling is ESSENTIAL for the driver source follower. If you use capacitor coupling you call for troubles....
Last edited:
1.5% THD is rather high distortion for a preamplifier outputting just 1V.
I am reading it's operated at 12V plate voltage. What's the typical anode current 70 uA?
Nope, that's just a design choice. There is no special operative condition and anything else is automatically worse.....
One just needs to define the goal. Certainly 1Vrms at 1.5% THD is not good enough for anything in my book but that's not the only way to use the NuTube.
In that case I suspect the high distortion is deliberate to satisfy some audiophile requests....
You can see here another example where distortion is just over 1% at 4.2V rms output....with just 220K plate load.
Nutube Balanced Amp
Combining anode CCS (or even better a gyrator) and high input impedance of the following stage I am pretty confident it can be even less THD at 2X output!
Ok, need to be more specific about the design targets: The front end should be low-distortion. 2nd harmonic in opposite phase than that of the output stage would be preferred. [I just expanded the list in the first post accordingly.]
How can a NuTube be biased to swing ±15 V at low distortion?
How can a NuTube be biased to swing ±15 V at low distortion?
The lower distortion is obtained with a suitable load. This is nowadays almost normal practice with tubes too, especially when large swings are needed with limited supply. A CCS or a gyrator as plate load + a high input impedance following stage can provide such load. With this "trick" one can get low distortion even from tubes that are normally labelled as poorly linear. I don't see why the NuTube should be an exception.
36V/28uA with +0.5V bias should be good enough for 8.5 Vrms output at low distortion. 8Vrms, considering some loss at the output stage, will make a 10W amplifier. Input required 0.7 Vrms. Of course 8.5Vrms is not a barrier and the NuTube can go a bit further....
At the input of the amp you need a source follower to screen the signal source and the volume control from the NuTube as this is operated in positive grid field where grid current flows.
Both source followers could be just FETs with resistive source load if enough supply available. There are a number of suitable JFETS and MOSFETs. If a MOSFET it should be chosen with low reverse transfer capacitance Crss as this is what matters in source follower application. There are several types with just 1 pF Crss....actually the LND150 is rated for 0.5 pF.
36V/28uA with +0.5V bias should be good enough for 8.5 Vrms output at low distortion. 8Vrms, considering some loss at the output stage, will make a 10W amplifier. Input required 0.7 Vrms. Of course 8.5Vrms is not a barrier and the NuTube can go a bit further....
At the input of the amp you need a source follower to screen the signal source and the volume control from the NuTube as this is operated in positive grid field where grid current flows.
Both source followers could be just FETs with resistive source load if enough supply available. There are a number of suitable JFETS and MOSFETs. If a MOSFET it should be chosen with low reverse transfer capacitance Crss as this is what matters in source follower application. There are several types with just 1 pF Crss....actually the LND150 is rated for 0.5 pF.
I doubt that the NU tube can cut it.
In the B1 Korg circuit the tube sees an easy load with the JFets. Still, the distortion figures.....
When i wanted to use the B1 Korg for Mofo i wanted more swing and i raised the question on 4th july 2018 in the Forum
"Quote:
Question
What are the possibilities to increase voltage swing?
Plate choke would be great but with a Rp of 300k Ohm for the korg triode a crazy undertaking.
At the same time there are chokes made by SACTHAILAND with 7400 Henries which makes a -3db point at 10 hz possible
Those are grid chokes but the tiny current of the tube could allow for their use....?
Report Post Reply With Quote Multi-Quote This Message
Old 4th July 2018, 05:29 PM #191
Nelson Pass is offline Nelson Pass United States
The one and only
Nelson Pass's Avatar
Join Date: Mar 2001
B1 with Korg Triode
The voltage gain and swing characteristics for the Nutube circuit that I
used were the load-line that I preferred. You can get more or less gain
and more or less distortion by playing with the loading and bias figures,
but it becomes a different animal (one that you might prefer).
If you want to increase the voltage swing I suggest a gain stage after
the Nutube, and while you're at it, you might want to make it inverting,
as that Nutube circuit inverts phase and requires another phase inversion
after it to get the whole thing right. Previously I have just inverted phase
at the speaker when using it.
In the B1 Korg circuit the tube sees an easy load with the JFets. Still, the distortion figures.....
When i wanted to use the B1 Korg for Mofo i wanted more swing and i raised the question on 4th july 2018 in the Forum
"Quote:
Question
What are the possibilities to increase voltage swing?
Plate choke would be great but with a Rp of 300k Ohm for the korg triode a crazy undertaking.
At the same time there are chokes made by SACTHAILAND with 7400 Henries which makes a -3db point at 10 hz possible
Those are grid chokes but the tiny current of the tube could allow for their use....?
Report Post Reply With Quote Multi-Quote This Message
Old 4th July 2018, 05:29 PM #191
Nelson Pass is offline Nelson Pass United States
The one and only
Nelson Pass's Avatar
Join Date: Mar 2001
B1 with Korg Triode
The voltage gain and swing characteristics for the Nutube circuit that I
used were the load-line that I preferred. You can get more or less gain
and more or less distortion by playing with the loading and bias figures,
but it becomes a different animal (one that you might prefer).
If you want to increase the voltage swing I suggest a gain stage after
the Nutube, and while you're at it, you might want to make it inverting,
as that Nutube circuit inverts phase and requires another phase inversion
after it to get the whole thing right. Previously I have just inverted phase
at the speaker when using it.
The whole idea of using the NuTube was to use the 36 V supply without messing with a boost converter. Adding a CCS as a plate load will reduce the plate voltage to something lower than 36 V, so we can't really bias up to 36 V.
Also, biasing the grid to +0.5 V means we can't swing +/-2 V input signal without hitting into grid current / cut-off.
Please take another look at the design targets (first post of this thread). We want more output voltage.
I just ordered a bunch of 3V4 tubes. I will run them through the curve tracer to get the triode curves. Let's see what this gives. Oh, the 3V4 is a lot cheaper than the NuTube!
You forgot one requirement; no HV circuit failure can wreck the VFET. I get the feeling that there are no more available... 🙂
IOW, good luck with that...
cheers,
Douglas
IOW, good luck with that...
cheers,
Douglas
Again. It's several times we are going back on this. The JFET followers as driver and buffer at next stage are NOT a guarantee of high plate load.
The effective plate load is the parallel combination of the value of the plate resistor and the input impedance of the next stage. If the plate resistor is, say, 200-300K and the JEFT follower at the next stage provides 1M input impedance then the actual plate load is 170-230K, respectively.
In this case, it's not surprising at all that distortion is poor because the effective plate load is smaller than the internal plate resistance of the NuTube which is rated at 330K.
With such a low ratio between load and internal plate resistance ANY triode will perform poorly if large ouput signals are required!! The only application where it works fine is a phono preamplifier....
If the plate resistor is a proper CCS or Gyrator with several megaohms impedance then the 1M of the next stage becomes the dominant load. That makes a lot of difference. If more even better...
With traditional vacuum tubes a ratio of 5 is generally the minimum to get good results with intrinsically linear devices.
A ratio of 2 is typical for the power tubes in the output stage for maximum power output!
Last edited:
P.S. A gyrator is an electronic choke and is better than any actual choke given the crazy values in play, IMHO.
Why do you want to swing 2V peak? That would mean 17V rms out. The VFET cannot deliver 36W output power. It will go into wild clipping before the NuTube hits that limit.....
Instead 1.4V peak at the NuTube grid will result in 12V rms out and 18W power output which is the onset of clipping, looking at the Pass article.
I know the 3V4 is a lot cheaper, that's why I suggested it. I never advice expensive tubes. 🙂
However the 3V4 has other disadvantages....you need a socket with careful mounting to limit microphonics and a dedicated filament supply. At this point I would consider using a proper transformer which would not be too big or heavy. A good Antek toroidal or R-core of suitable rating will be good.
To cover the dropout of the CCS / gyrator plus the pk-pk voltage swing, seems that 36V rail would need a boost. Using a dedicated B+ supply would open up a world of possibilities.
If you wanted to try to use the 36V rail and give the NuTube a better load, perhaps this actual choke from Lundahl? A grid choke, LL1670, rated at 540H at 0.8mA 4.8K DCR. $62 USD each though, so not cheap.
https://www.lundahltransformers.com/wp-content/uploads/datasheets/1670.pdf
If you wanted to try to use the 36V rail and give the NuTube a better load, perhaps this actual choke from Lundahl? A grid choke, LL1670, rated at 540H at 0.8mA 4.8K DCR. $62 USD each though, so not cheap.
https://www.lundahltransformers.com/wp-content/uploads/datasheets/1670.pdf
540H is too small...!!!😀 The internal resistance of the NuTube is more similar to that of typical signal pentodes at 0.33M. Nelson Pass had considered using 7400H grid choke...
That was the biggest one I could find on a quick search 🙂 looks like SAC Thailand makes a 7kH grid choke, maybe this is the one NP was looking at.
SAC Thailand
SAC Thailand
And what frequency does that wonderful 7400Hy number begin to behave like a capacitor? So to say, what is its self resonant point( L and C_winding)? I am going to bet it is below 1kHz...so everything above that is being loaded by a capacitor. Not exactly brilliant IMO.
Also, from my first warning, a 'failure' can be just the warm up variability causing some unexpected stuff. Try something simple, and build Salas' DCG3 linestage and hybridize it with tubes without popping a transistor of any sort first...then go play with the unobtainium.
cheers,
Douglas
Also, from my first warning, a 'failure' can be just the warm up variability causing some unexpected stuff. Try something simple, and build Salas' DCG3 linestage and hybridize it with tubes without popping a transistor of any sort first...then go play with the unobtainium.
cheers,
Douglas