Uinp is an amplitude. You can change the value od Uinp using the yellow arrow in the upper part of the graph. The default value of Uinp is not the optimal value. It is selected automatcally by the program. You should select appropriate value of Uinp by yourself.
Oops! Missed that!
Thank you very much for the clarification, and sorry for my mistake! 🙂
Yes. As I wrote it is the amplitude. This is not peak to peak or RMS value. Of course I can add results for P-P and/or RMS (for sinusoidal signal), but I do not know, whether this brings new valuable information to the program.
Take into account that the yellow indicator changes one "half" of the signal, the other is adjusted automatically in such a way that of the non-linearity of the tube is taken into account. Therefore, the distances between yellow lines and white line are not the same and depend on the operating point (nonlinearity).
Take into account that the yellow indicator changes one "half" of the signal, the other is adjusted automatically in such a way that of the non-linearity of the tube is taken into account. Therefore, the distances between yellow lines and white line are not the same and depend on the operating point (nonlinearity).
I received signals about some strange behavior of the simulator consisting in the fact that when changing the slope of the loadline using the blue external indicators on the current or voltage axis, the value of calculated second harmonic h2 did not change.
I explained this phenomenon by the fact that when such a change takes place automatically changes the operating point. This change is related to the fact that the lengths of sectors taken into account in the 3-point method of h2 calculation changes proportionally, so that the value of the second harmonic remains practically constant.
Since, however, questions about the accuracy of the calculation were repeated I decided to explore this phenomenon in more detail.
So I changed the less accurate 3-point method for a more accurate, but required more computation 5-point method.
And here's what it turned out.
1 The calculated values of second harmonic h2 using 3-point and 5-point method differ only slightly. Here is a comparison:
2. The theory of keeping the proportions causing small changes of h2 confirmed in a 5-point method. However, it is more sensitive to the 3-point method, so some changes can be observed.
3. Even when h2 is almost constant, h3 and h4 changes. Since h3 and h4 were not calculated, watching only h2 user of the simulator had the impression that the distortion did not change.
The current program is completed by the calculation of h3, h4 and the resulting coefficient htot.
I verify the algorithm for correctness. It seems that the calculation of the distortion are correct.
In the previous version it was not an error, but the calculations were less sensitive.
I explained this phenomenon by the fact that when such a change takes place automatically changes the operating point. This change is related to the fact that the lengths of sectors taken into account in the 3-point method of h2 calculation changes proportionally, so that the value of the second harmonic remains practically constant.
Since, however, questions about the accuracy of the calculation were repeated I decided to explore this phenomenon in more detail.
So I changed the less accurate 3-point method for a more accurate, but required more computation 5-point method.
And here's what it turned out.
1 The calculated values of second harmonic h2 using 3-point and 5-point method differ only slightly. Here is a comparison:
Code:
Tube H2 (3-point method) H2 (5-point method)
2A3 5.354 5.456
300b 2.658 2.736
5687 3.655 3.657
6N16S 6.103 6.876
6N16B 4.020 4.009
6N1P 8.629 8.479
6N6P 1.780 1.782
6S33S 4.220 5.309
6S33S (1 cathode) 6.361 7.700
6S4S 1.762 1.817
6SN7GT 2.346 2.342
6Ż9P (triode mode) 2.198 2.197
E182CC 6.144 6.169
E92CC 8.922 8.933
EC86 6.375 6.330
EC91 2.267 2.263
ECC81 5.789 5.784
ECC82 6.283 6.275
ECC83 7.876 7.785
ECC85 14.863 14.939
ECC88 2.764 2.757
ECC91 10.717 10.725
ECC99 8.933 8.8742. The theory of keeping the proportions causing small changes of h2 confirmed in a 5-point method. However, it is more sensitive to the 3-point method, so some changes can be observed.
3. Even when h2 is almost constant, h3 and h4 changes. Since h3 and h4 were not calculated, watching only h2 user of the simulator had the impression that the distortion did not change.
The current program is completed by the calculation of h3, h4 and the resulting coefficient htot.
I verify the algorithm for correctness. It seems that the calculation of the distortion are correct.
In the previous version it was not an error, but the calculations were less sensitive.
The problem still exists.
After moving voltage and current cursors, you need to move (upper) white cursor.
If this is not done, the THD value is not the same than after the movement.
What I mean with the "movement" of the white cursor is that it is moved to any direction and then returned back to original position. This operation activates the THD calculation.
After moving voltage and current cursors, you need to move (upper) white cursor.
If this is not done, the THD value is not the same than after the movement.
What I mean with the "movement" of the white cursor is that it is moved to any direction and then returned back to original position. This operation activates the THD calculation.
I corrected the bug with the refreshing the Uinp value.
Thanks again for your help to reduce errors and I would be grateful for further assistance 🙂
Hello again 🙂
I 've just added some more triodes. Popular in high-end audio 6N30P-DR (6Н30П-ДР) and very interesting triode 6S45P-E (6С45П-Е).
I also added tube 5670 which is the subject of discussion in one of neighboring threads.
I 've just added some more triodes. Popular in high-end audio 6N30P-DR (6Н30П-ДР) and very interesting triode 6S45P-E (6С45П-Е).
I also added tube 5670 which is the subject of discussion in one of neighboring threads.
gsmok, I can provide parameters of other lamps but which? That you would bring them in the program.
It depends on you 🙂.
I will add any triode whose parameters you provide. To do this I need a text file with the parameters read from the anode characteristics of a tube (the file format I gave in post number #20), or the parameters of the Koren's model for triode.
By the way, can anyone know, how to change the thread title? I have just expanded the program to simulate pentodes and I do not want to create a new thread. This one according to the current title refers only to triodes.
I will add any triode whose parameters you provide. To do this I need a text file with the parameters read from the anode characteristics of a tube (the file format I gave in post number #20), or the parameters of the Koren's model for triode.
By the way, can anyone know, how to change the thread title? I have just expanded the program to simulate pentodes and I do not want to create a new thread. This one according to the current title refers only to triodes.
Last edited:
I was hoping for this from the beginning, when available, it will left the Broskie software behind
Thank You so much
I added a triode 45 (it was mentioned earlier as a tube worth adding to the simulator).
I'm all the time verifying the best model for pentodes. It turns out that I have problems with this
. I checked a few models and during the simulation there are errors with curve fitting. Even an improved pentode model called "Derk" described on "uTracer" site, for the parameters indicated there (EF86 pentode) doesn't agree with the characteristics of real pentodes. For the measurements listed on mentioned page I received the same results, but for other values of second grid voltages results of simulation and manufacturer data does not overlap (calculated currents are lower than specified by the manufacturer).
May be, this is because model from "uTracer" site was build using unusual second grid voltages (250V for EF86).
Of course I can use simple Korens model and finish pentode simulator, but I would like to have more reliable results (for example take into account the secondary emission efect etc.).
I'm all the time verifying the best model for pentodes. It turns out that I have problems with this
. I checked a few models and during the simulation there are errors with curve fitting. Even an improved pentode model called "Derk" described on "uTracer" site, for the parameters indicated there (EF86 pentode) doesn't agree with the characteristics of real pentodes. For the measurements listed on mentioned page I received the same results, but for other values of second grid voltages results of simulation and manufacturer data does not overlap (calculated currents are lower than specified by the manufacturer).May be, this is because model from "uTracer" site was build using unusual second grid voltages (250V for EF86).
Of course I can use simple Korens model and finish pentode simulator, but I would like to have more reliable results (for example take into account the secondary emission efect etc.).
Last edited:
The problem with "one-off" models from the uTracer as well as the short-comings of the various pentode models are discussed in the Vacuum Tube SPICE Model sticky thread. Bottom line, it is not an easy problem to solve especially if you start with the Koren models... I think the Reefman models with a "fudge factor" added would be the way to go, since it appears that most of the differences between the trace-derived models and the datasheets have to do with the relative magnitude of the plate and screen currents. Really looking forward to what you can come up with...
Last edited:
Thanks a lot. The title is just right 🙂.
As a basis for the simulation I use tubes datasheets. Measurement data can be used too, but I prefer averaged data published by manufacturers.
As a basis for the simulation I use tubes datasheets. Measurement data can be used too, but I prefer averaged data published by manufacturers.
Off the chart...
An externally hosted image should be here but it was not working when we last tested it.
