Not at all. I just checked my calculated curves and plate-to-grid feedback on a KT88 configured for a mu of 8 (the mu of a triode-connected KT88) and it gets a plate resistance 5-10% lower than the triode-connected case, plus the curves don't lay over nearly as much in the low current/high voltage section of the plots. It is more linear, lower plate resistance, and you can drive the tube to saturation with no grid power.
Now triode is simply all plate-screen feedback (UL) applied that is possible. All other percentages of UL will give you higher rp than triode-connection.
Returning feedback to the electrode that caused the distortions in the first place is superior to applying them to another electrode with different characteristics. It's that simple. Plus, if you leave the screen alone you can enjoy all of its benefits in shielding the other electrodes from the field of the plate like decreased miller capacitance and ability to drive to saturation with no grid power driver.
I'd 'Like' this if I had a FB account...LOL
Just to clarify, I believe this 'saturation point' you refer to is on the pentode characteristics, and thus more current at lower plate voltage than is available triode connection with g1 voltage less than zero. Also, from a practical standpoint, even lower plate resistance is available if you knock mu a bit below the triode connected. hardly a difference between 7 and 8, and I suspect output Z drops a bit more...🙂
cheers,
Douglas
Just to clarify, I believe this 'saturation point' you refer to is on the pentode characteristics, and thus more current at lower plate voltage than is available triode connection with g1 voltage less than zero. Also, from a practical standpoint, even lower plate resistance is available if you knock mu a bit below the triode connected. hardly a difference between 7 and 8, and I suspect output Z drops a bit more...🙂
cheers,
Douglas
Well, triodes saturate at some positive G1 voltage. When I refer to saturation I am referring to a point where more drive from a perfect driver will just cause more current demand from the driver but will give no more output.
The cool thing about plate-grid feedback is that you can take a KT88 and set it for a mu of 5 or so (higher mu than 300B) and get lower plate resistance, better linearity, lower input capacitance, and higher power out than a 300B. Oh, and it is cheaper too.
I built an amp this way and I think it turned out pretty well. Check out the last few posts. I think this is pretty good performance for an open-loop amp. Not as good as my Unity-Coupled amp but still pretty good.
http://www.diyaudio.com/forums/tubes-valves/248376-amp-kt88-push-pull-shunt-feedback-output-via-p-channel-fet.html
It is indeed saturation. In other words it is the point where the positive g1 voltage gets too close to the plate voltage.
Same thing with cathode fb but input impendace is increased at expense of a small complication in the OT and less possible combinations (but still plenty enough) if ideal geometry has to be achieved. Nothing really difficult. Also, the cathode connection partially compensate the DC bias in the OT in SE amps so that a smaller gap can be used getting higher inductance.....
Last edited:
Cathode feedback is series applied voltage feedback and has the same benefits of parallel applied feedback except that high input impedance is maintained, which is a very desirable attribute. However, things get complicated with a pentode if you want to maintain constant voltage between cathode and screen. You would have to make a floating supply unless there is something clever going on like in the McIntosh configuration to maintain constant screen voltage with respect to the cathode.
In principle, I like series applied feedback better but there are practical considerations that sometimes make things difficult. Parallel applied feedback works just find if you drive it right. Like anything in engineering its just a question of managing tradeoffs.
You either need higher voltage swing from the driver, or higher current swing; this is the difference.
In principle I agree however the only thing that separates the cathode from the ground using fixed bias is the resitance of the cathode winding which is small usually (at least in my OT's). In SE mode where the only thing that changes a lot relative to quiescent condition is the screen grid current but is going to impact on the g2 voltage very little, typically less than 1V change.
My client, discussed with me the PP project in which there will be an output transformer for 4 pieces EL509 with separate cathode windings. I convinced him that it was better to take the local feedback directly from the output winding having a pin for the midpoint. Just like in Audio Research VT-130. But I suggested in this case to abandon the UL and apply feedback directly from the anodes to the driver cathode because the EL509 does not like the large voltage on the screen grid. http://www.diyaudio.com/forums/tube...l509-6kg6-quality-question-2.html#post5103849
You can certainly introduce cathode feedback from the secondary if the OT design allows for that. For the PP it is better as the current can change quite a lot especially with that kind of tube.
I will make a transformer in my factory. The design is mine too. It is a toroidal output transformer a core made of nanocrystalline alloy weighing 7.7 kg. Magnetic core will be selected from the party in 20pcs. The client is ready to overpay for this. Power transformers will be made on similar nanocrystalline alloy cores too. https://40.img.avito.st/1280x960/3404276840.jpg
Last edited:
Some OT posts have been deleted or moved. Try to stay on topic, guys...It's obvious that instantaneuosly everything changes. It was referred to average. The anode current will deviate a little bit from quiescent value only when distortion gets high, around 7-10%. Average G2 instead is not constant.
See here on pag.4, for example, the pentode class A in SE mode:https://frank.pocnet.net/sheets/010/p/PCL82.pdf
The anode current increases only by 3-5% @ 10%THD while G2 increases by about 70%! Of course these % will vary from one tube to another but the trend doesn't.
Last edited:
Hi Wavebourn, may I ask you a simple way to achieve this? I guess a zener referenced to the cathode with the support of a bjt, then a large cap between cathode and g2, but there could be better systems.
Thanks