Hey Gary,
Haven´t downloaded the sim yet but will do.
Have you simmed it with different 6C33 models?. Think I have three or four giving different simmed results.
Also have you checked THD with GFB removed and also the amount of feedback?
I´m also puzzled as gain is only 6dB.
Haven´t downloaded the sim yet but will do.
Have you simmed it with different 6C33 models?. Think I have three or four giving different simmed results.
Also have you checked THD with GFB removed and also the amount of feedback?
I´m also puzzled as gain is only 6dB.
Last edited:
Hey again Gary,
Have now simmed the circuit with your models and with normal 20dB gain and get clipping at ca 12W. THD at 2W 0,3%, THD at 6W, 0,8%.
With your original schematic with ca 6dB gain, 0,08% at 2W. Also unsymmetric clipping at ca 4W. Overloading the input tube?
Open loop gain 27dB.
Maybe something drawn wrong in you schematic? All DC voltages OK though. Haven´t changed it except for the FB resistor when running 20dB gain.
Have now simmed the circuit with your models and with normal 20dB gain and get clipping at ca 12W. THD at 2W 0,3%, THD at 6W, 0,8%.
With your original schematic with ca 6dB gain, 0,08% at 2W. Also unsymmetric clipping at ca 4W. Overloading the input tube?
Open loop gain 27dB.
Maybe something drawn wrong in you schematic? All DC voltages OK though. Haven´t changed it except for the FB resistor when running 20dB gain.
Last edited:
This statement as a generality is incorrect. OTLs don't 'overload' output tubes any more than a transformer coupled amp does as long as one pays attention to the circuit at hand. I have built fairly small OTLs that are quite low in distortion, without feedback.
It is true that there is an economy of scale- the bigger the OTL, the more efficient. So the charm and challenge for many, the holy grail so to speak, is the small OTL.
Lately I have been building some guitar amps that use 4 6AS7Gs total, with zero feedback. The output is a Circlotron, but uses an input transformer with a single-ended voltage amplifier and driver tube. There is no intention to use this amp with an 8 ohm load. I think that is where people get into trouble with small OTLs.
Like any technology, there is appropriate use. One thing I have learned over the years is that even if you have a SS amp, the fact that it can drive 4 ohms does not mean it sounds its best on 4 ohms; if you are focused on the best sound then seek a higher load impedance. Tubes in general prefer to drive higher impedances which is why 16 ohm speakers were common in the 1950s. In the case of my little guitar amp, I planned for 32 ohm load.
With 20 6C33s as depicted in the original circuit of this thread, I would expect over 200 watts with vanishingly low distortion at only 20 watts. But that assumes that the circuit is set up right... I think there is a problem with the voltage amplifier not being able to drive so many 6SN7s and I would like to see a simulation of the driver circuit performance.
Practically speaking there needs to be bias adjustment for each power tube.
IMO/IME you want a direct-coupled connection between the driver and the output tubes. If you don't, you run into all kinds of problems getting bass response, bias stability and overload recovery to all happen at the same time.
Just for fun I simmed the outputstage alone. Feeding it with 100V ptp resulted in 12V ptp out, which puzzles me. THD at this level, just over 2W, is in the ballpark of 0,15%. But how good are the 6C33C models?
Very much points in the direction that the low sensitivity of the outputstage together with an incapable driver is what ruins it.
Very much points in the direction that the low sensitivity of the outputstage together with an incapable driver is what ruins it.
Hmmm....
That suggests there's something fundermentally wrong with the O/P stage...
The driver tube arrangement to me just looks 'wrong' somehow...
Seems to be LOTS of local NFB in the driver.....
Be better IMO to replace with a simple CF DC coupled to the G of the 6C33C
Hey Alastair,
Tried to remove all local feedback from the CF driving the output tube but almost nothing happened gainwise but the THD doubled.
The driver stage has 46dB gain when checking differential out and off course 40dB when checking each output.
Simmed THD at 50V peak(not ptp) from one of the outputs is around 2%.
Tried to remove all local feedback from the CF driving the output tube but almost nothing happened gainwise but the THD doubled.
The driver stage has 46dB gain when checking differential out and off course 40dB when checking each output.
Simmed THD at 50V peak(not ptp) from one of the outputs is around 2%.
What happens if you try to drive only a pair of drivers and output tubes instead of the whole thing?
Also, the fact that the top tube bank is driven by the voltage amplifier applying its inputs to the grid and cathode, where in the bottom bank the signal voltage is applied to the grid and *plate* might be making a difference....
Ah--Yes, Sorry bout that!.....
What I cant quite understand is just Why he's made it SO complicated!
Each stage with its own gain stage driver, seems a little OTT to me, when a simple single gain stage per phase, followed by the CF's for each O/P tube would be easier and realise the full potential of the available power The potential for distortion could be lower too....
It would be easy then to implement bias setting for each O/P tube as you advise too....
--Summit pretty essential for a 6C33C....
Err--I'm pretty thick.....
Can someone describe just What the Bottom of the driver triodes is actually doing....?
Seems to me, its amplifying the O/P tubes signal, and effectively cancelling out the drive to the O/P tube grid....
--Or perhaps I'm being More stupid than normal!
Can someone describe just What the Bottom of the driver triodes is actually doing....?
Seems to me, its amplifying the O/P tubes signal, and effectively cancelling out the drive to the O/P tube grid....
--Or perhaps I'm being More stupid than normal!
Looks like a mu follower to me. It appears to take some voltage input from the output tube cathode as feedback.
We are going to see a reduction in the AC voltage swing off of the cathode of the power tube as opposed to its grid, but then the bottom driver section has gain, and I don't see a divider at the grid of that bottom tube, so I think you are right- a lot of the drive voltage will get canceled.
On top of that, I'm not all that happy with that 47K resistor in the top tube's grid circuit. The more I look at it, it seems to me that as the cathode of that top tube goes up and down, the 47K follows it. A tube needs a difference between its inputs; how in the heck do you get any drive to the top tube? Wouldn't it be more effective if that 47K was tied to some sort of virtual ground? I've not run any simulation, but intuitively it looks like a lot of drive is being lost in the driver.
"Can someone describe just What the Bottom of the driver triodes is actually doing....?
Seems to me, its amplifying the O/P tubes signal, and effectively cancelling out the drive to the O/P tube grid...."
Sure looks like you are correct. I had just assumed this was acting as a counter modulated CCS to help the top driver triode cathode output, but its phasing is wrong. Only thing I can figure might be a distortion cancelling dynamic load for the top cathode follower (or both CFs). When the gm of the top cathode follower increases, the load (bottom triode) increases slightly to keep a constant voltage across the CF. This would require some fine tuning of the local feedback. Would be observable in the simulation if the voltage from the top CF grid to either it's cathode or the output tube's cathode stays constant versus signal swing. (or maybe was just tweeked for minimum distortion in the final output) Dunno if it's setup that way actually. Would be much more economical to combine all these separate drivers into one. Too many coupling capacitors around too.
The floating 47K grid resistor is just a bootstrapped grid bias R to make the top triode input Z higher.
Seems to me, its amplifying the O/P tubes signal, and effectively cancelling out the drive to the O/P tube grid...."
Sure looks like you are correct. I had just assumed this was acting as a counter modulated CCS to help the top driver triode cathode output, but its phasing is wrong. Only thing I can figure might be a distortion cancelling dynamic load for the top cathode follower (or both CFs). When the gm of the top cathode follower increases, the load (bottom triode) increases slightly to keep a constant voltage across the CF. This would require some fine tuning of the local feedback. Would be observable in the simulation if the voltage from the top CF grid to either it's cathode or the output tube's cathode stays constant versus signal swing. (or maybe was just tweeked for minimum distortion in the final output) Dunno if it's setup that way actually. Would be much more economical to combine all these separate drivers into one. Too many coupling capacitors around too.
The floating 47K grid resistor is just a bootstrapped grid bias R to make the top triode input Z higher.
Last edited:
The resolution is not that great, but it looks like he is using 6SN7s. If so he could use a grid resistor 10X what he is without any issues.
That bottom triode looks like a current-limiting device, kinda common in SS but here its not making sense. It sinks more current from the CF as the output current increases, acting a bit counter productive. Or what am I missing? Perhaps this linearizes the pusf-pull of the upper and bottom output tubes. Kinda like forcing the upper/lower pair to have more equal gain?
It looks to be certainly Lowering the gain, but I dunno about equalising it....
Maybe the O/P is into SS rather than Vacuum!😉
Certainly 'Looks' SS in its sorta format to me....
- Status
- Not open for further replies.