# TINA-TI // “Algebraic Triode Modeling” revisited

www.lynx.net/~jc/Maillet-GlassAudio98.PDF

pleased to see my original Generalized Algebraic Triode models converging in TINA-TI’s mixed-mode circuit simulator // … adding to a list of recent number crunching tests and design tweaks using this newly discovered piece of software

TINA-TI offers a very impressive list of device models, but its ease in allowing the creation of custom models via import of Pspice-style .subckt text files is perhaps what matters most to me … TINA-TI seems to support the typical set of functional control source equations as Pspice does // and thus paves the way for mathematically characterized (purely empirical) modeling and simulation, as I’ve done with my Triode modeling work … I’m pretty sure ICAP/4, LTspice and AIM-spice all support general forms of algebraic modeling, but it’s on TINA-TI that my models finally seem to be working without a hitch (more to come on this …)

to confirm Tina could handle hefty circuits, I was recently able to simulate an entire Acoustic 361 power amplifier circuit by importing modern power device models (MJE340, MJE350, MJL21194) // … better yet, this week I got Tina to reproduce the original Triode simulation results I first saw in ’95 on a Macintosh Quadra640 running Intusoft’s Pspice5.0 and Matlab …

Here’s Tina in action displaying the classic DC-Transfer characteristics of my data-derived RCA 12AX7 vacuum tube Triode model :

… Tina in action displaying the classic DC-Transfer characteristics of a Common-Cathode gain stage running on my data-derived GE 12AX7 Triode model :

… Tina displaying typical “class-A” TRANsient (real time) overload characteristics of a Common-Cathode gain // **with Capacitive bypass @ 5.2v pk-pk input sine-wave** :

… Tina displaying typical “class-A” TRANsient (real time) overload characteristics of a Common-Cathode gain // **without Capacitive bypass @ 9v pk-pk input sine-wave** :

Hi there,

My name is Viet, I am currently trying to get involved in audio community. However, I am stuck with the model of 12AX7 implemented on TINA (the mathematical model is based on Ayumi’s method). Fortunately, I found out about this article which is suitable for my need and I want to ask if whether or not I can request a sck or tsm file for my TINA TI 9?

Thank so much

HI VIET, THANK YOU FOR ASKING … MY MODEL IS AN ENGINEERING MODEL AND NOT A HOBBY ONE, WHICH MEANS ONE NEEDS TO UNDERSTAND HOW TO USE INITIAL CONDITIONS PROPERLY OR THE MODEL WILL NOT SIT OR START WELL IN A SIMULATION … I’M AFRAID THAT RELEASING THE MODEL “AS IS” WILL CAUSE A FLOOD OF QUERIES AND DON’T HAVE THE TIME FOR IT … I’M HOPING THAT RONALD DEKKER’S uTRACER WILL ONE DAY HAVE ITS OPERATING RANGE EXTENDED ABOVE THE CURRENT 450VDC LIMIT SO I CAN DO A NEWER AND MORE COMPLETE MODELING OF THE 12AX7 TUBE … FOR THE TIME BEING, I’M KEEPING THE MODEL TO MYSELF UNTIL I HAVE TIME TO DO A WRITE-UP ON ITS PROPER USE OR DERIVE A NEW HOBBY-GRADE MODEL

Hi JCM,

I am doing a Math model of your model. Puzzled because you specify a 4th order poly (a + b*x + c*x^2 +d*x^3 + e*x^4) for curve fitting to plate IV @Vg constant data points. This has 5 unknown coefficients, but there are only 4 data points for Vg = 4.5V and 5V. Not enough givens to find unknowns.

HI THOMAS,

GOOD OBSERVATION AND THANKS FOR ASKING … THE “LISTED” TABLE OF CHOSEN DATA VALUES ARE FOR POINTS THAT LINE UP WITH PUBLISHED DATA … IN ORDER TO INTERPOLATE DOWN TO THE ORIGIN AREA I HAD TO FUDGE IN ARBITRARY DATA VALUES ON SOME OF THOSE LAST DATA SETS, SINCE THEY ARE NOT DRAWN IN ON THE ORIGINAL GRAPHS … IF YOU NOTICE IN FIGURE 6(C) THE DATA IS STRETCHED OUT IN A LOG-LOG PLOT AND SHOWS HOW THE ARBITRARY CHOICES I MADE ON THOSE LAST 6 SETS PROVIDE A DIFFERING RESPONSE PROFILE HEADING TOWARDS THE ORIGIN … AS A RESULT OF THIS ANOMALY, THE MODELING WILL POSE A PROBLEM IF STARTUP IS ALLOWED TO COMMENCE AT (0,0) THUS REQUIRING “INITIAL VALUES” SUFFICIENTLY REMOVED FROM THE ORIGIN FOR ALL CONTROLLED SOURCES TO CONVERGE PROPERLY … THIS RESULTS IN AN ENGINEERING MODEL – TRADING OFF FULL CONVENIENCE FOR ACCURACY IN THE INTENDED AREAS OF OPERATION … I DID ALL THIS MANUALLY, SO LIMITED RESULTS ARE TO BE EXPECTED … A GOOD PROGRAMMER SHOULD BE ABLE TO DO MUCH BETTER THAN I DID, ALTHOUGH THERE IS (STILL) THE QUESTION OF HOW PRECISELY TO DEAL WITH NEAR ZERO VALUES WHILE PRESERVING CONCAVITY (PRIMARY STARTUP CONSIDERATION), AND MORE … GOOD LUCK!