Another death in the family has slowed down the progress. Sherri (my wife) will be away for at least a month, to help her mother deal with everything.
I got this setup ready for testing last week. It has produced no sound yet, let alone any measurements. If circumstances permit, I may get some experiment time this weekend.
I got this setup ready for testing last week. It has produced no sound yet, let alone any measurements. If circumstances permit, I may get some experiment time this weekend.
Attachments
Hi John,
We just had two solid weeks of Spring weather here finally, 60's to 80 deg. F, I got a sunburn now from biking and hiking! So has been a little slow on experimental progress here, sorry, but I have most of the experiment set up on the bench. I had to fix a power supply too. Supposed to rain tomorrow, so I should get around to some testing finally.
The more I have thought about the UL distortion minimum, the more I suspect it is being caused by screen current distortion countering 3/2 power law distortion at heavy output loading. Zload = Zout typically.
If so, then even triode wired pentode might benefit from using a MOSFET to control the screen grid at some DC voltage below, or even at, the plate V, and returning the screen current through the MOSFET drain to B++ instead of the plate.
(In UL, returning the current to the 43% tap causes a 57% Screen current "missing from the output" effect. Connecting the MOSFET drain to a boosted B++ will cause an even bigger 100% Scrn current missing effect. This may lead to a different optimum (higher) drive V tap %, maybe even to 100%. Varying the Scrn DC level versus plate DC will also change the amount of Scrn current effect. This deleted screen current will effectively round off the knee of the plate curves on the left side to varying degrees.)
So your original thoughts on lowering the screen voltage below plate V may be quite relevant after all. Clearly the triode wired pentode case needs to have a look-see at the screen voltage drop from plate effects and at screen current versus plate current. Same experimental setup to variable UL too, just one case.
And as a follow on to all these tests, one could also consider extending the experimental setup to eventually include an Op Amp to control the screen V (via common source MOSFET driver) for minimum error in the plate output (versus input drive), just to see what the optimum signal on the screen would like to be. But this will be a ways off yet to get to.
Don
We just had two solid weeks of Spring weather here finally, 60's to 80 deg. F, I got a sunburn now from biking and hiking! So has been a little slow on experimental progress here, sorry, but I have most of the experiment set up on the bench. I had to fix a power supply too. Supposed to rain tomorrow, so I should get around to some testing finally.
The more I have thought about the UL distortion minimum, the more I suspect it is being caused by screen current distortion countering 3/2 power law distortion at heavy output loading. Zload = Zout typically.
If so, then even triode wired pentode might benefit from using a MOSFET to control the screen grid at some DC voltage below, or even at, the plate V, and returning the screen current through the MOSFET drain to B++ instead of the plate.
(In UL, returning the current to the 43% tap causes a 57% Screen current "missing from the output" effect. Connecting the MOSFET drain to a boosted B++ will cause an even bigger 100% Scrn current missing effect. This may lead to a different optimum (higher) drive V tap %, maybe even to 100%. Varying the Scrn DC level versus plate DC will also change the amount of Scrn current effect. This deleted screen current will effectively round off the knee of the plate curves on the left side to varying degrees.)
So your original thoughts on lowering the screen voltage below plate V may be quite relevant after all. Clearly the triode wired pentode case needs to have a look-see at the screen voltage drop from plate effects and at screen current versus plate current. Same experimental setup to variable UL too, just one case.
And as a follow on to all these tests, one could also consider extending the experimental setup to eventually include an Op Amp to control the screen V (via common source MOSFET driver) for minimum error in the plate output (versus input drive), just to see what the optimum signal on the screen would like to be. But this will be a ways off yet to get to.
Don
Hi Tubelab,
Sorry to hear about the death in the family there. I guess we are all getting up there, sad world. When its my turn, someone is going to get an earful about suggested mods to the world order:
The vacuum tube hobby will hence-forth act like anti-oxidents, and extend happy lifetimes indefinitely. My two cents contribution to the Universe.
And no more taxes or tax forms.
Don
Sorry to hear about the death in the family there. I guess we are all getting up there, sad world. When its my turn, someone is going to get an earful about suggested mods to the world order:
The vacuum tube hobby will hence-forth act like anti-oxidents, and extend happy lifetimes indefinitely. My two cents contribution to the Universe.
And no more taxes or tax forms.
Don
Hi Tubelab,
Sorry to hear about death in your family - hope all is well with the rest of your nearest & dearest. Looking forward to your mearurements & report.
Smoking-amp,
Indeed, it does seem my innocent remark about screen voltage may turn out to have some bearing. This is exciting - using an opamp to optimise a tubes performance. I know for some that mixing SS & tube in the one circuit is sacrilegious. To them this will be nothing short of a hanging offence but has this sort of use been attempted before?
Keep up the good work.
John
Sorry to hear about death in your family - hope all is well with the rest of your nearest & dearest. Looking forward to your mearurements & report.
Smoking-amp,
Indeed, it does seem my innocent remark about screen voltage may turn out to have some bearing. This is exciting - using an opamp to optimise a tubes performance. I know for some that mixing SS & tube in the one circuit is sacrilegious. To them this will be nothing short of a hanging offence but has this sort of use been attempted before?
Keep up the good work.
John
jkeny said:
Hey-Hey!!!,
My input/phase splitter of choice, the differential cascode works and souds best when built with a properly matched pair of triodes on the bottom, and a pair of MOSFET's on 'top'. The MOSFET's are source driven/grounded gate and thus avoid a lot of the wacky gate-drain capacitance issues normally encountered. Their being a PP pair further aids dealing with gate capacitance issues.
cheers,
Douglas
Firstly Tubelab, my sincere condolences with the death in your family. As you may recall, tomorrow it will be 8 months that I lost my wife, and the loss is still severe. My very best wishes for your family.
Brandersnatch is right, but I still have not convinced myself that there is a definite place for a combination of technologies. I have had enough success with semiconductors to keep the technologies apart - but that is a personal stance.
Like Tubelab (his post #95) I am also inclined not to dwell to much on overload and g1 cirrent drawn, in the sense that that constitutes going outside the borders of linear operation and should simply be disregarded. Apart from what others may have done, I would define maximum g1 swing within the bias magnitude as the limit, and output power achieved there is simply maximum. Such are the relevant graphs seen by me.
I am now a little behind in this excellent thread, and will just say that, as far as I can recall, I miss the factor Mu (g2-anode). That is the mu between the screen (as input) and the anode. If I have overlooked it I will realise that soon enough, but at prersent I am trying to construct a set-up where that can be measured. It would seem logical that that has something to do with the different optimum value for g2 taps between tube types. My measurements would include a 6L6 and 6V6. Hopefully some further information would come from that.
Regards.
Brandersnatch is right, but I still have not convinced myself that there is a definite place for a combination of technologies. I have had enough success with semiconductors to keep the technologies apart - but that is a personal stance.
Like Tubelab (his post #95) I am also inclined not to dwell to much on overload and g1 cirrent drawn, in the sense that that constitutes going outside the borders of linear operation and should simply be disregarded. Apart from what others may have done, I would define maximum g1 swing within the bias magnitude as the limit, and output power achieved there is simply maximum. Such are the relevant graphs seen by me.
I am now a little behind in this excellent thread, and will just say that, as far as I can recall, I miss the factor Mu (g2-anode). That is the mu between the screen (as input) and the anode. If I have overlooked it I will realise that soon enough, but at prersent I am trying to construct a set-up where that can be measured. It would seem logical that that has something to do with the different optimum value for g2 taps between tube types. My measurements would include a 6L6 and 6V6. Hopefully some further information would come from that.
Regards.
Hey Johan,
It is simply the only way I can reasonably build the circuit. A pentode eats up too much of the voltage headroom. The B+ is only fed to the two stage amp through the OPT centre tap. B+ is limited by the Class A bias on the finals. The phases or sides of the diff amp are connected to taps on the OPT's primary. This is usually the U-L tap, but with multiple taps available on my custom Iron it can be almost anywhere I want.
So, I have some NFB tied into this circuit, and it's all short path. A pentode has the disadvantage to the MOSFET of needing a heater and g2 supply, in addition to having a lower input impedance( little bits of grid current, of the type that gives us the maximum grid circuit resistance figures ). The pentode has a far nicer input capacitance profile( and quantity ) than the MOSFET. The pentode also needs its cathode heated...
As with any design challenge, it is just one of the possible solutions. Understand the weakness of all the parts and components and use them where these weakness don't make a big deal. The design spec went something like this:
no loop NFB.
short path NFB
simple power supply design
minimum number of stages
class A bias on the PP finals
Output Z & stability sufficient to tolerate parallel Quad ESL 57's
It was fun, and I am finishing the latest of this design right now. Radial pentodes from Eimac...
cheers,
Douglas
It is simply the only way I can reasonably build the circuit. A pentode eats up too much of the voltage headroom. The B+ is only fed to the two stage amp through the OPT centre tap. B+ is limited by the Class A bias on the finals. The phases or sides of the diff amp are connected to taps on the OPT's primary. This is usually the U-L tap, but with multiple taps available on my custom Iron it can be almost anywhere I want.
So, I have some NFB tied into this circuit, and it's all short path. A pentode has the disadvantage to the MOSFET of needing a heater and g2 supply, in addition to having a lower input impedance( little bits of grid current, of the type that gives us the maximum grid circuit resistance figures ). The pentode has a far nicer input capacitance profile( and quantity ) than the MOSFET. The pentode also needs its cathode heated...
As with any design challenge, it is just one of the possible solutions. Understand the weakness of all the parts and components and use them where these weakness don't make a big deal. The design spec went something like this:
no loop NFB.
short path NFB
simple power supply design
minimum number of stages
class A bias on the PP finals
Output Z & stability sufficient to tolerate parallel Quad ESL 57's
It was fun, and I am finishing the latest of this design right now. Radial pentodes from Eimac...
cheers,
Douglas
Douglas yes ....
I gave my immediate reaction, but sat down and thought about it afterwards (weakness of us researchers). It does make sense; I read that you use it in the tube way (with output transformer), not low voltage direct output as per more usual. Interesting .... and I cannot find fault with that (not that one is mainly here for that kind of reaction).
So, who knows ... I might yet become an infidel at my age! (Combining glass and sand?! Uhg! Should not happen. Stands to reason!) . Fortunately I am not yet a tweed; will have to see what mosfets are available here.
Regards.
I gave my immediate reaction, but sat down and thought about it afterwards (weakness of us researchers). It does make sense; I read that you use it in the tube way (with output transformer), not low voltage direct output as per more usual. Interesting .... and I cannot find fault with that (not that one is mainly here for that kind of reaction).
So, who knows ... I might yet become an infidel at my age! (Combining glass and sand?! Uhg! Should not happen. Stands to reason!) . Fortunately I am not yet a tweed; will have to see what mosfets are available here.
Regards.
Johan,
If there is something you want but can't get, drop me a note...
As to mixing sand and glass( actually made from sand ), it all depends on what your Holy Text writings contain I think. I am beginning to think that there as many ways to build a bad tube circuit as there are to build a bad SS one...
I think as with all else when one follows the fanatics, trouble waits in the long grass. I suspect that there are no ways to measure how a circuit will sound save to listen to it...no matter if it is SS or tube.
cheers,
Douglas
If there is something you want but can't get, drop me a note...
As to mixing sand and glass( actually made from sand ), it all depends on what your Holy Text writings contain I think. I am beginning to think that there as many ways to build a bad tube circuit as there are to build a bad SS one...
I think as with all else when one follows the fanatics, trouble waits in the long grass. I suspect that there are no ways to measure how a circuit will sound save to listen to it...no matter if it is SS or tube.
cheers,
Douglas
I have my test setup up and running. No parasitic oscillation, so looks good. I have only run a simple test so far, sweeping between pentode and triode mode, using a separate B++ supply for the screen (MOSFET drain actually) current return.
Monotonically decreases in power toward triode mode but distortion also monotonically decreases, no minima apparent in between. 2nd harmonic is dominant at -20 dB triode mode to -17dB pentode mode. This lack of a dist. minima may be the expected case when using a separate B++ supply for screen current return. (this is not the normal UL setup this way) About 1 to 2 Watts max out so far, using 40 mA on the 6L6 tube, I may push it further later after some operating experience is gained. I am attaching the test rig schematic. Not shown on the schematic is a floating 250V supply available to connect from the plate to the B++ drain supply point or the 50% primary tap, (instead of a fixed ground referenced B++ as shown). This will allow for more normal UL operation modes with screen current returned thru the xfmr.
Seven power supplies in action, 2 of them for the P-P current nulling CCS. 1 for filament. 1 for screen DC program. 1 for screen DC power to MOSFET drain. 1 for B+. 1 for Vg1 bias. What a job to get it up and running with all those adjustments!
The output Xfmr is a 5K P-P Edcor CXPP100-SP1, but is reconfigured for 3200 OHM load on the 6L6 tube. CCS runs on the rest of the primary with 4X current of the tube to balance DC in Xfmr.
Don
Monotonically decreases in power toward triode mode but distortion also monotonically decreases, no minima apparent in between. 2nd harmonic is dominant at -20 dB triode mode to -17dB pentode mode. This lack of a dist. minima may be the expected case when using a separate B++ supply for screen current return. (this is not the normal UL setup this way) About 1 to 2 Watts max out so far, using 40 mA on the 6L6 tube, I may push it further later after some operating experience is gained. I am attaching the test rig schematic. Not shown on the schematic is a floating 250V supply available to connect from the plate to the B++ drain supply point or the 50% primary tap, (instead of a fixed ground referenced B++ as shown). This will allow for more normal UL operation modes with screen current returned thru the xfmr.
Seven power supplies in action, 2 of them for the P-P current nulling CCS. 1 for filament. 1 for screen DC program. 1 for screen DC power to MOSFET drain. 1 for B+. 1 for Vg1 bias. What a job to get it up and running with all those adjustments!
The output Xfmr is a 5K P-P Edcor CXPP100-SP1, but is reconfigured for 3200 OHM load on the 6L6 tube. CCS runs on the rest of the primary with 4X current of the tube to balance DC in Xfmr.
Don
Attachments
"Most interesting will be to see if there's an odd order null."
Yes, there is some 3rd harmonic. I will do some tests and post spectrums for varying %UL, varying load and the connection point for the screen current return.
"Where is the Jacob's Ladder and Wimshurst machine? "
Picture attached of Warp Controls from a damaged flying saucer, notice the obvious distorted space-time continuum visible in the background at Warp 54.
Don
Yes, there is some 3rd harmonic. I will do some tests and post spectrums for varying %UL, varying load and the connection point for the screen current return.
"Where is the Jacob's Ladder and Wimshurst machine? "
Picture attached of Warp Controls from a damaged flying saucer, notice the obvious distorted space-time continuum visible in the background at Warp 54.
Don
Attachments
Listen,
You earthlings should stop making jokes about things you know nothing of. Pityful enough that you still use reproducing equipment to experience "sound sensations", as you call it in this funny language, which cannot even make up its mind as to pronunciation - I mean "daughter" and "laughter" - no telepathy, teleportation, nothing (backward people).
Alien
(For the moment acting under assumed ID)
You earthlings should stop making jokes about things you know nothing of. Pityful enough that you still use reproducing equipment to experience "sound sensations", as you call it in this funny language, which cannot even make up its mind as to pronunciation - I mean "daughter" and "laughter" - no telepathy, teleportation, nothing (backward people).
Alien
(For the moment acting under assumed ID)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.