I'm trying to locate triode curves for a vintage B424 triode tube. All I found were curves with the grid voltage versus anode current. I would rather like to have a set of grid curves with anode current versus anode voltage. As this is more easy to work with for me. So I decided to give ChatGPT a try. After some initial discussions with ChatGPT on how to get the curves printed, ChatGPT suggest to download and install GNUPlot. This can be downloaded for free from https://sourceforge.net/projects/gnuplot/files/gnuplot/
In order to get accurate triode curves it turned out that it is most helpful to give ChatGPT a list with triode data points derived from the odd graph on the old tube data sheet.
First you have to tell ChatGPT that you need it to produce the code for GNUPlot. Then you tell it to use the emperial triode model formula.
Next you need to supply all the available data of the tube. My data for the B424 looked as follows:
The triode properties are S=3.0 mA/V mu=24 Ri=8kOhm for the curves max anode voltage 250V, max anode current 10 ma and negtive grid voltage between 0 and -8 V in steps of -0.5V. The measurements are as follows at grid voltage 0 volt at anode voltage 50V 0.8 mA - 100V 3.2 mA - 150V -8.3 ma. At grid voltage -2V at anode voltage 50V 0.1ma - 100V 0.9 mA and 150 V 3.5 mA. At grid voltage -4 V and anode voltage 100V 0.1 mA - 150V - 1.0 ma. At gid voltage -6 V and anode voltage of 150 V 0.1mA anode current. Can you calculate a similar triode curve model for me ?
ChatGPT summarised the data as follows and then started calculating the optimal parameters for the best fitting curves:
Got it! Let's proceed by analyzing the new tube's data and refining the triode curve model for it.
Where:
Let’s compute these values next!
The optimized parameters for the new tube are:
This file was cut and pasted into Microsoft Notebook and saved as b424.plt in the documents folder.
You then open GNUplot and type at the command prompt load 'b424.plt' and press enter.
A window with the plotted curves will open.
You then check if the data points you took fit the curves.
The fit was pretty accurate.
If GNUplot does not deliver the required output then feed the error code, or remarks on what happens (or not happens) to ChatGPT and it will gennerate a new code for GNUplot to try again.
Love to hear your experiences on this subject.
In order to get accurate triode curves it turned out that it is most helpful to give ChatGPT a list with triode data points derived from the odd graph on the old tube data sheet.
First you have to tell ChatGPT that you need it to produce the code for GNUPlot. Then you tell it to use the emperial triode model formula.
Next you need to supply all the available data of the tube. My data for the B424 looked as follows:
The triode properties are S=3.0 mA/V mu=24 Ri=8kOhm for the curves max anode voltage 250V, max anode current 10 ma and negtive grid voltage between 0 and -8 V in steps of -0.5V. The measurements are as follows at grid voltage 0 volt at anode voltage 50V 0.8 mA - 100V 3.2 mA - 150V -8.3 ma. At grid voltage -2V at anode voltage 50V 0.1ma - 100V 0.9 mA and 150 V 3.5 mA. At grid voltage -4 V and anode voltage 100V 0.1 mA - 150V - 1.0 ma. At gid voltage -6 V and anode voltage of 150 V 0.1mA anode current. Can you calculate a similar triode curve model for me ?
ChatGPT summarised the data as follows and then started calculating the optimal parameters for the best fitting curves:
Got it! Let's proceed by analyzing the new tube's data and refining the triode curve model for it.
Step 1: Data Summary
Tube parameters:
- S=3.0 mA/VS = 3.0 \, \text{mA/V}
- μ=24
- Ri=8 kΩ= 8 \, \text{k}\Omega
- Max voltage: 250V
- Max Current: 10mA
- Grid voltage range: 0V to −8V-8V
Step 2: Refining the Model
We’ll use the same empirical formula:
Ia=Is(Vb+μVgVp)αI_a = I_s \left( \frac{V_b + \mu V_g}{V_p} \right)^{\alpha}Where:
- Is is the scaling factor.
- α controls the curve shape.
- Vp is set at 100V.
Step 3: Calculating Parameters
We’ll need to fit the observed data to this model to find the optimal values of IsI_sIs and α\alphaα for the new tube.Let’s compute these values next!
The optimized parameters for the new tube are:
- Is≈3.37 mAI_s \approx 3.37 \, \text{mA}Is≈3.37mA (scaling factor)
- α≈2.21\alpha \approx 2.21α≈2.21 (exponent controlling curve shape)
Step 4: Updated GNUplot Script
Now, let's update the triode current equation in the GNUplot script for this new tube:This file was cut and pasted into Microsoft Notebook and saved as b424.plt in the documents folder.
You then open GNUplot and type at the command prompt load 'b424.plt' and press enter.
A window with the plotted curves will open.
You then check if the data points you took fit the curves.
The fit was pretty accurate.
If GNUplot does not deliver the required output then feed the error code, or remarks on what happens (or not happens) to ChatGPT and it will gennerate a new code for GNUplot to try again.
Love to hear your experiences on this subject.
Attachments
This is the code ChatGPT generated to draw the triode curves for the B424 tube. You can use this file to test your GNUplot version. You can also upload this (cut and paste form Notebook) to ChatGNU and tell it that you like a triode curve model comparable with this earlier generated code. Might save time to get the required results for other triode tubes.
You could also consider to as ChatGPT to generate SPICE compatible code. I have no experience with SPICE, but surely worth a try.
The curves all look like the same curve?
How about having ChatGPT design the whole amplifier with some given tubes and specs?. Or present a crude design and have ChatGPT tune up the design.
How about having ChatGPT design the whole amplifier with some given tubes and specs?. Or present a crude design and have ChatGPT tune up the design.
Last edited:
I tried that a while ago, it was a disaster, just for a simple full wave rectifier circuit. But I did not spend a lot of time, just a couple of hours, and it was the free version.