Gentlemen,
some of you may know the triode-strapped or UL-op curves I measured and plotted over the years, see here. The measuring setup to do so is trivial for triode(d) plots and only slightly more complex for UL (screen FB) ones.
Now I would like to do some measurements and plots of anode curves with cathode FB applied, but I wasn´t able to come up neither with a measuring setup nor a mathematical approach to transform already taken triode(d) or pentode/BPT or screen FB datasets into CFB ones. So far I suspect that a direct measuring approach couldn´t be set up easily on the bench, IOW would be too complex and not practical.
Maybe a mathematical approach to transform datasets that already have been taken would be easier and more practical, but I must admit this exceeds my current knowledge how to do such a transformation (BTW, I am using Excel for storing and visualizing the data sets).
Any idea would be greatly appreciated, thank you!
Regards,
Tom Schlangen
EDIT/P.S.: Qualitatively it is easy to predict what the application of CFB will do to the anode family of curves: The grid curves will narrow together (since mu will drop) and get steeper (since gm will rise). What is missing is how to derive the quantitative change for a given percentage of CFB...
some of you may know the triode-strapped or UL-op curves I measured and plotted over the years, see here. The measuring setup to do so is trivial for triode(d) plots and only slightly more complex for UL (screen FB) ones.
Now I would like to do some measurements and plots of anode curves with cathode FB applied, but I wasn´t able to come up neither with a measuring setup nor a mathematical approach to transform already taken triode(d) or pentode/BPT or screen FB datasets into CFB ones. So far I suspect that a direct measuring approach couldn´t be set up easily on the bench, IOW would be too complex and not practical.
Maybe a mathematical approach to transform datasets that already have been taken would be easier and more practical, but I must admit this exceeds my current knowledge how to do such a transformation (BTW, I am using Excel for storing and visualizing the data sets).
Any idea would be greatly appreciated, thank you!
Regards,
Tom Schlangen
EDIT/P.S.: Qualitatively it is easy to predict what the application of CFB will do to the anode family of curves: The grid curves will narrow together (since mu will drop) and get steeper (since gm will rise). What is missing is how to derive the quantitative change for a given percentage of CFB...
Tom,
CFB amounts to just changing the reference point for ground, which just shifts the input voltage by %X of the output swing. So just add X% of the plate voltage swing (variance from B+) to the grid input voltage data.
The problem comes up when trying to chart this new data conventionally (constat Vg curves) since the usual constant grid voltage curves aren't constant any more in the adjusted data.
So I would suggest taking finer grid delta Vg curves initially in the grounded cathode setup, then CFB adjust the grid Vg data by adding X% of Vp delta. Then you will have to interpolate through the dataset to compose constant Vg curves.
This would mean finding the nearest two Vg datapoints in the adjusted dataset to the desired const Vg at each Vp point on a curve. So first select a Vp point. Then search the dataset at that Vp for the nearest Vg above and below the desired constant Vg.
So let Q=(desired Vgconst - lesser Vg at datapoint)/(greater Vg at datapoint - lesser Vg at datapoint). Then deltaIp = Q (delta of Ip between lesser Vg and greater Vg datapoints) So the final Ip = Ip of the lesser Vg datapoint + deltaIp.
Clear as mud no doubt.
----------------------------------------
By the way, I am working on modding a Tek curve tracer, and have decided that additional data displays beside the usual Ip versus Vp at constant Vg plots would be greatly desired. I'm hoping to display triode or g2/g1 Mu versus Vg at stepped constant current levels or with a specific plate load versus Vg. (Small HF AC signal injected to grid, frequency selective V or I detector at the plate) Could then easily display grid 1 Gm versus the operating points too.
Everyone is familiar with the usual plate curves, but they don't really display the data a designer would most like to see. Why not produce the directly useful data curves too.
Don
CFB amounts to just changing the reference point for ground, which just shifts the input voltage by %X of the output swing. So just add X% of the plate voltage swing (variance from B+) to the grid input voltage data.
The problem comes up when trying to chart this new data conventionally (constat Vg curves) since the usual constant grid voltage curves aren't constant any more in the adjusted data.
So I would suggest taking finer grid delta Vg curves initially in the grounded cathode setup, then CFB adjust the grid Vg data by adding X% of Vp delta. Then you will have to interpolate through the dataset to compose constant Vg curves.
This would mean finding the nearest two Vg datapoints in the adjusted dataset to the desired const Vg at each Vp point on a curve. So first select a Vp point. Then search the dataset at that Vp for the nearest Vg above and below the desired constant Vg.
So let Q=(desired Vgconst - lesser Vg at datapoint)/(greater Vg at datapoint - lesser Vg at datapoint). Then deltaIp = Q (delta of Ip between lesser Vg and greater Vg datapoints) So the final Ip = Ip of the lesser Vg datapoint + deltaIp.
Clear as mud no doubt.
----------------------------------------
By the way, I am working on modding a Tek curve tracer, and have decided that additional data displays beside the usual Ip versus Vp at constant Vg plots would be greatly desired. I'm hoping to display triode or g2/g1 Mu versus Vg at stepped constant current levels or with a specific plate load versus Vg. (Small HF AC signal injected to grid, frequency selective V or I detector at the plate) Could then easily display grid 1 Gm versus the operating points too.
Everyone is familiar with the usual plate curves, but they don't really display the data a designer would most like to see. Why not produce the directly useful data curves too.
Don
Thank you very much Don,
I think your suggestions show up a viable way to do this sort of plots. Meanwhile I had a serious look at MathCAD and I am quite sure a transformation of already gained data can be done by pure mathematics, too. I also found some interesting hints in one of the old, but groundbreaking books from German Professor Barkhausen for a mathematical approach.
Regardless of methods to be used in the end, isn´t it somewhat disturbing how many people talk about the merits of CFB while seemingly only very few can give hard mathematical facts about its influence on anode family of curves?
Thank you again,
Tom
I think your suggestions show up a viable way to do this sort of plots. Meanwhile I had a serious look at MathCAD and I am quite sure a transformation of already gained data can be done by pure mathematics, too. I also found some interesting hints in one of the old, but groundbreaking books from German Professor Barkhausen for a mathematical approach.
Regardless of methods to be used in the end, isn´t it somewhat disturbing how many people talk about the merits of CFB while seemingly only very few can give hard mathematical facts about its influence on anode family of curves?
Thank you again,
Tom
hey-Hey!!!,
CFB is doing the same thing as U-L unless you do it McIntosh-style and keep g2-k voltage constant. Dealing with the appearant input voltage requirements or the real one( g1-k ) could pose issues. Aside from the McIntosh rigging, I have not seen OPT designed for pentode operation with tap references at the same percentage as the CFB coil.
cheers,
Douglas
CFB is doing the same thing as U-L unless you do it McIntosh-style and keep g2-k voltage constant. Dealing with the appearant input voltage requirements or the real one( g1-k ) could pose issues. Aside from the McIntosh rigging, I have not seen OPT designed for pentode operation with tap references at the same percentage as the CFB coil.
cheers,
Douglas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.