Ayumi Pentode LTSpice Model Creation Question
I could not find a "t1T4" file on my PC.
So I am attaching my digitized triode plate characteristics for the 1T4
Also, I am attaching the resultant triode model that Ayumi generates.
So am I supposed to have a t1T4.csv file? I have never seen one, I thought it was just an array name inside of R.
Thanks,
Tom
I could not find a "t1T4" file on my PC.
So I am attaching my digitized triode plate characteristics for the 1T4
Also, I am attaching the resultant triode model that Ayumi generates.
So am I supposed to have a t1T4.csv file? I have never seen one, I thought it was just an array name inside of R.
Thanks,
Tom
Yes, the digitized characteristic should be in CSV format, and it should contain less data points - 3-4 points per Eg1 should be sufficient. I also like to skip the Eg1>0V curves to reduce the error rate.
In any case, attached please find the simplified dataset (be sure to rename the suffix to CSV before using) and here is the command to trace the dataset, try it for yourself to see if it works for you.
View attachment 1T4_T.txt
In any case, attached please find the simplified dataset (be sure to rename the suffix to CSV before using) and here is the command to trace the dataset, try it for yourself to see if it works for you.
Code:
Ip.cal("1T4_T.csv", "t1T4", Cgp=1.0E-14, Cgk=3.6E-12, Cpk=7.5E-12, pentode.chara=c(67.5, 67.5, 0, 0.0034, 0.0015, 250000))View attachment 1T4_T.txt
Ayumi Pentode LTSpice Model Creation Question
Thank you Jazbo8 for the help, I'll let you know how I make out!
I wondered about too many data points. Sometimes too many are too much. Algorithms sometimes have convergence/fitting issues with a flood of data, I understand that.
OK, I am building a simple two tube amp, just for the heck of it. I have attached the schematic. No, it is not an amp that is "crunchy" or hi-fi but it uses unusual tubes, at least they are unusual for this time. So, I like to model these things to see what the voltages might be and fuss around with other components. blah, blah, blah.
If it sounds half way decent, I will build a second one and use it at work for my PC audio. Most of the engineers where I work are so young they have never even seen a tube let alone a working tube amp! Go ahead, touch it!
Question:
I can only find pentode characteristics for the 117L7. (Attached)
1 - How would I derive triode behavior from this data.
2 - What software would you recommend to model this tube?
Thanks!
Tom
Thank you Jazbo8 for the help, I'll let you know how I make out!
I wondered about too many data points. Sometimes too many are too much. Algorithms sometimes have convergence/fitting issues with a flood of data, I understand that.
OK, I am building a simple two tube amp, just for the heck of it. I have attached the schematic. No, it is not an amp that is "crunchy" or hi-fi but it uses unusual tubes, at least they are unusual for this time. So, I like to model these things to see what the voltages might be and fuss around with other components. blah, blah, blah.
If it sounds half way decent, I will build a second one and use it at work for my PC audio. Most of the engineers where I work are so young they have never even seen a tube let alone a working tube amp! Go ahead, touch it!
Question:
I can only find pentode characteristics for the 117L7. (Attached)
1 - How would I derive triode behavior from this data.
2 - What software would you recommend to model this tube?
Thanks!
Tom
Attachments
That's really an odd tube, if you really want to make a model using Ayumi's method, then you need to trace the triode characteristic yourself, or try your luck with Model Paint Tools.
Ayumi Pentode LTSpice Model Creation Question
Jazbo8, you gave some really good advice to synchro27 on 26th May 2016, 12:13 AM
Pentode to triode characteristics - See RDH4, p.34-p.36.
This is interesting to me as well as the 117L7 has no published triode characteristics. Or at least none that I could find.
Hey, I am ready to breadboard the thing and create my own triode curves with real electrons. I hope my sample tube is characteristic of the population...
Jazbo8, you gave some really good advice to synchro27 on 26th May 2016, 12:13 AM
Pentode to triode characteristics - See RDH4, p.34-p.36.
This is interesting to me as well as the 117L7 has no published triode characteristics. Or at least none that I could find.
Hey, I am ready to breadboard the thing and create my own triode curves with real electrons. I hope my sample tube is characteristic of the population...
Ayumi Pentode LTSpice Model Creation Question
Thank you Jazbo8! Your abbreviated data file and Ip.cal statement worked.
Curiously, when I looked at the contents of t1T4, using str(t1T4), [And I guess I just realized the answer to your earlier request of the contents of t1T4...] Your method yielded 12 elements including Ea (the value that threw the error) while my long data file yielded only 10 elements and Ea was not in the list. The error was evident at that point even before the r2spice command was executed. This will take some time for me to reconcile and figure out what triggered my earlier problem. Probably too many points and the zero initial point for each grid voltage. Thanks again!
Yes I will, over the next month, try both methods of generating plate curves and let you know. In the meantime, all my parts have arrived and I am beginning construction! Undoubtedly, manually running the tube would be easier and faster. Project photos to follow.
I will post the models for 1T4 and 117L7 when I finish them for others.
Does this forum have a place to post models from members? I guess moderating that forum would include running the model to make sure it wasn't "obscene"!
Cheers,
Tom
Thank you Jazbo8! Your abbreviated data file and Ip.cal statement worked.
Curiously, when I looked at the contents of t1T4, using str(t1T4), [And I guess I just realized the answer to your earlier request of the contents of t1T4...] Your method yielded 12 elements including Ea (the value that threw the error) while my long data file yielded only 10 elements and Ea was not in the list. The error was evident at that point even before the r2spice command was executed. This will take some time for me to reconcile and figure out what triggered my earlier problem. Probably too many points and the zero initial point for each grid voltage. Thanks again!
Yes I will, over the next month, try both methods of generating plate curves and let you know. In the meantime, all my parts have arrived and I am beginning construction! Undoubtedly, manually running the tube would be easier and faster. Project photos to follow.
I will post the models for 1T4 and 117L7 when I finish them for others.
Does this forum have a place to post models from members? I guess moderating that forum would include running the model to make sure it wasn't "obscene"!
Cheers,
Tom
paint_kit and extractmodel equations
Just an observation
(I am new to this, so forgive me if it has been mentioned before, or, if I am incorrect please let me know):
This thread has encouraged me to try paint_kit.jar to develop Spice models. I have observed that the Spice sub-circuit models output from paint_jar are different to that output by Derk Reefman’s “extractmodel”, which I use to get Spice models from the uTracer.
Specifically, the equation in paint-kit is:
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
Whereas that corresponding line in the model output from ‘extractmodel’ is:
G1 1 3 VALUE={0.5*(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
The difference between the two is the 0.5* highlighted in the’ extractmodel’ equation.
In Koren’s original paper I do not see the 0.5 (or ½) (am I right?) , but Derk Reefman in his 2014 document “Spice models for vacuum tubes using the uTracer” uses the 0.5
I guess it is a moot point as to whether the 0.5 should be in or out: my understanding is that it does not make any difference, since the model parameters MU,EX,KP,KG1 and KVB are optimized to work with their respective equations, but, the point is that, for example, the EX from paint_kit is different to the EX from ‘extractmodel, even if the input data were the same.
Am I correct in this?
PS: why it matters to me: I tried to manually insert the parameter values for MU,EX,KP,KG1 and KVB from paint_kit into the model from extractmodel, and got half the expected anode current.
I use the Koren model embedded in micro controller code to control bias current in my PP amp, and wanted to introduce a ‘reference model’ so one could display the tube emission % compared to that of published data.
Just an observation
(I am new to this, so forgive me if it has been mentioned before, or, if I am incorrect please let me know):
This thread has encouraged me to try paint_kit.jar to develop Spice models. I have observed that the Spice sub-circuit models output from paint_jar are different to that output by Derk Reefman’s “extractmodel”, which I use to get Spice models from the uTracer.
Specifically, the equation in paint-kit is:
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
Whereas that corresponding line in the model output from ‘extractmodel’ is:
G1 1 3 VALUE={0.5*(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
The difference between the two is the 0.5* highlighted in the’ extractmodel’ equation.
In Koren’s original paper I do not see the 0.5 (or ½) (am I right?) , but Derk Reefman in his 2014 document “Spice models for vacuum tubes using the uTracer” uses the 0.5
I guess it is a moot point as to whether the 0.5 should be in or out: my understanding is that it does not make any difference, since the model parameters MU,EX,KP,KG1 and KVB are optimized to work with their respective equations, but, the point is that, for example, the EX from paint_kit is different to the EX from ‘extractmodel, even if the input data were the same.
Am I correct in this?
PS: why it matters to me: I tried to manually insert the parameter values for MU,EX,KP,KG1 and KVB from paint_kit into the model from extractmodel, and got half the expected anode current.
I use the Koren model embedded in micro controller code to control bias current in my PP amp, and wanted to introduce a ‘reference model’ so one could display the tube emission % compared to that of published data.
The current versions can be found at Ayumi's website:
Ayumi's Lab.: Audio
The site is in Japanese but the version numbers are evident and if you hover over the "Windows" links the English file names will appear in the browser's status bar (at least they do for Firefox and Internet Explorer). Note that the current versions (as of this post) are:
pctube_1.21_win.zip
tubemodel_3.20_win.zip
I hope this helps.
Ayumi's Lab.: Audio
The site is in Japanese but the version numbers are evident and if you hover over the "Windows" links the English file names will appear in the browser's status bar (at least they do for Firefox and Internet Explorer). Note that the current versions (as of this post) are:
pctube_1.21_win.zip
tubemodel_3.20_win.zip
I hope this helps.
Hello jazbo8,
2P1P model runs great, thanks!
I have attached 1AD4, 1P24B, and 1T4 triode curves here.
I used Curve Captor make triode models which run good with ltspice.
I wonder do you have time to help me creating pentode/triode models.
I am interested to see what difference between tridoe models created from Curve Captor and from Ayumi's.
Thanks again!
Yes, Google Translate also works fine, probably from within any browser. But my point was that you can always get the latest versions of the Ayumi models from this website.
Do you have the pentode characteristic for the 1P24B? The one I have does not seem to correlate with the triode characteristic that you provided.
Also, I tried making the pentode model for 1T4, but the result was way off from the characteristic shown in the GE datasheet, so it's not really usable.