Just spent a day to read through this thread and I very much appreciate your efforts and sharing of information.
I use NI Multisim 13 at the moment and found a couple of comments that Ayumi's models don't work very well in that software.
Is that a true statement?
Are there any other library of models that work better for Multisim or do I need to go over to LTSpice or some other software to be able to simulate for proper results?
There have been a lot to take in, so I'm a little bit confused at the moment, so any help to get me in the right direction would be helpful.
I use NI Multisim 13 at the moment and found a couple of comments that Ayumi's models don't work very well in that software.
Is that a true statement?
Are there any other library of models that work better for Multisim or do I need to go over to LTSpice or some other software to be able to simulate for proper results?
There have been a lot to take in, so I'm a little bit confused at the moment, so any help to get me in the right direction would be helpful.
Seems like it might be the right way to go.
I made some tests yesterday and everything works fine "as is" but it is very very slow. I estimate it to go a thousand times slower than I'm used to.
Made some tests and found that this line have a very big impact on the speed.
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-10*V(A,K)"
Something with this exponent (in red) causes the speed issue.
I tried to decrease it one step at the time 9, 8, 7...
The speed was equally slow all the time until I reached 3, then it released into full speed as I'm used to. It felt like a on/off switch in speed.
I feel that this equation goes into stall for some reason and have no clue why. Something in the calculation apparently doesn't fit Multisim 13.
It runs, but extremely slow and with wrong result if I put in anything but a 10, so it need to be 10. If I set a 4 it runs approx. 500 times slower instead of a 1000 times.
Any specialist out there that have a clue what's going on ?
The same type of exponential works fine and have no impact on the speed what so ever. These lines works fine with no issues with the exponent:
"BM1 M1 0 V=(0.028826571*(URAMP(V(A,K))+1e-10))^-0.90897681
BM2 M2 0 V=(0.622671*(URAMP(V(GG)+URAMP(V(A,K))/13.089625)+1e-10))^2.4089768
BP P 0 V=0.00087237591*(URAMP(V(GG)+URAMP(V(A,K))/21.021735)+1e-10)^1.5"
I made some tests yesterday and everything works fine "as is" but it is very very slow. I estimate it to go a thousand times slower than I'm used to.
Made some tests and found that this line have a very big impact on the speed.
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-10*V(A,K)"
Something with this exponent (in red) causes the speed issue.
I tried to decrease it one step at the time 9, 8, 7...
The speed was equally slow all the time until I reached 3, then it released into full speed as I'm used to. It felt like a on/off switch in speed.
I feel that this equation goes into stall for some reason and have no clue why. Something in the calculation apparently doesn't fit Multisim 13.
It runs, but extremely slow and with wrong result if I put in anything but a 10, so it need to be 10. If I set a 4 it runs approx. 500 times slower instead of a 1000 times.
Any specialist out there that have a clue what's going on ?
The same type of exponential works fine and have no impact on the speed what so ever. These lines works fine with no issues with the exponent:
"BM1 M1 0 V=(0.028826571*(URAMP(V(A,K))+1e-10))^-0.90897681
BM2 M2 0 V=(0.622671*(URAMP(V(GG)+URAMP(V(A,K))/13.089625)+1e-10))^2.4089768
BP P 0 V=0.00087237591*(URAMP(V(GG)+URAMP(V(A,K))/21.021735)+1e-10)^1.5"
Tried a model from Norman Koren (8 parameter) in Multisim 13 and run into the same slow performance (1/1000 of normal speed) with this line:
"Bp P K I=(0.1377434031m)*uramp(V(P,K)*ln(1.0+(-0.3680995655)+exp((3.754409336)+(3.754409336)*((36.96361454)+(251.1379958m)*V(G,K))*V(G,K)/sqrt((30.10332346)**2+(V(P,K)-(-1.996524827))**2)))/(3.754409336))**(1.222404059)"
Shifted the "**" to "^" and it run at full speed with expected result.
"Bp P K I=(0.1377434031m)*uramp(V(P,K)*ln(1.0+(-0.3680995655)+exp((3.754409336)+(3.754409336)*((36.96361454)+(251.1379958m)*V(G,K))*V(G,K)/sqrt((30.10332346)**2+(V(P,K)-(-1.996524827))**2)))/(3.754409336))**(1.222404059)"
Shifted the "**" to "^" and it run at full speed with expected result.
Sorry for bothering you again. but what I wrote earlier have no value.
Found that the problem are earlier in the equation.
I removed the last part of the equation all together...
From:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-10*V(A,K)"
To:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))"
Both runs very very slow. Both with what appears to be correct result.
But this runs fast at normal speed but with wrong result:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-3*V(A,K)"
I'll take a break from this...
Found that the problem are earlier in the equation.
I removed the last part of the equation all together...
From:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-10*V(A,K)"
To:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))"
Both runs very very slow. Both with what appears to be correct result.
But this runs fast at normal speed but with wrong result:
"BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.00049917061*URAMP(V(A,K))^1.5)))+1e-3*V(A,K)"
I'll take a break from this...
Now I found what caused the speed issue.
By removing the first URAMP command everything is up and running as fast as the tube models that are included with Multisim 13.
Original equation:
BIAK A K I=URAMP(V(IK,IG)-URAMP(V(IK,IG)-(0.0038690245*URAMP(V(A,K))^1.5)))+1e-10*V(A,K)
Modified equation:
BIAK A K I=V(IK,IG)-URAMP(V(IK,IG)-(0.0038690245*V(A,K)^1.5))+1e-10*V(A,K)
This is the only change made compared to Ayumi's original model.
Might be a bug.
Seems like the timestep are set extremely low when the equation starts with URAMP. The trial circuit I have used get the timestep to 7.8125 pS instead of 10 uS as it sould be.
Thats a lot of extra calculations.
I assume you have a lot of experience since you have been in this thread for a long time, so I would like to ask if Ayumi's models are considered to be the most accurate and the closest to the real thing?
...or are there any other libraries of vacuum tube spice models you would recommend to start with.
Wayne Clay aka cogsncogs
Norman Koren (8 parameter)
Rydell
Duncan
Maybe something else?
URAMP is one of the basic functions of SPICE since day one, any program that can not handle it well should be tossed in the bin... As to your question - for triodes, all of the model types work fine - unless there is grid current, then Ayumi is preferred. For pentodes, I only use the Ayumi models unless the Reefman models are available.
Send my fine Multisim to the bin?
Naah...
Thank you very much for your thoughts about which models to go for.
Very much appreciated...
Have attached a picture where I test three different tubes and type of models.
The generic delivered with Multisim do not handle 2nd harmonic correctly so that's why I try to find better models.
Both Ayumi and the 8 parameter Norman Koren (modified by Wayne Clay) get the 2nd harmonic according what I expect it to be.
But there are approx. 10dB difference between them regarding the 3rd harmonic.
Any comments on that?
Which model du you think gets closest to a real world circuit?
VCC = 150 Volt
Anode resistor 16K
Cathode resistor 470
Grid series resistor 1K
Input 1.0 Volt at 10 kHz
Edit:
Sorry, got the wrong color on the generic 6922 in the oscilloscope where it is blue, hard so see, but in the middle of the green and red curves.
Attachments
Last edited:
For some reason, Multisim isn't happy with so many "behavioral" directives. Some of the syntax is a bit different, but can usually be corrected with little difficulty.
For triodes the Koren models work well. If you are using an tube without a model, it isn't difficult to find the parameters using Microsoft Excel's "Solver" function.
Robert Weaver posted (and corrected) a method for solving pentodes. Pretty complex but well thought out: Vacuum Tube SPICE Models
Seems like time step locks into very small steps despite what your set.
Just as you say, I too have found Koren's modell to work fairly well and gives much more "real" results compared to the Generic types included in Multisim. But occasionally it too runs into problems if the circuit have long settling time because it will not run unless it "Start from Zero".
It's a pity because I really like Multisim otherwise.
Just a historical reference -- back when we were using '286s the antecedent of Multisim was almost given away -- something like $70. Then National Instruments bought the company and drove the price up 20x.
NATI did the same thing with "Softwire" -- a program which could have upended their T&M business -- Softwire was free-ware distributed with ADC/DACs -- to write the code you just attached "wires" in a graphical setting. NATI bought Softwire and stopped selling the stuff.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.