Hi all,
I have modelled a little 6080 amp with 5687 drivers and all is well in LTSpice.
However when analysing my results it is obvious that the 5687 is a bit marginal to get full power out of the 6080's, so I decided to try substituting a 6DJ8 for the 5687.
At this point the model refuses to run failing with "Time step to small". How can two valve models behave so differently in exactly the same circuit ?
Shoog
I have modelled a little 6080 amp with 5687 drivers and all is well in LTSpice.
However when analysing my results it is obvious that the 5687 is a bit marginal to get full power out of the 6080's, so I decided to try substituting a 6DJ8 for the 5687.
At this point the model refuses to run failing with "Time step to small". How can two valve models behave so differently in exactly the same circuit ?
Shoog
Hi some LTspice models are better than others, chances a different 6DJ8 model will work fine, if the "amplifier" bursts into RF oscillation the sim may hang. Try going into simulate then edit simulation command select start external DC supply at zero voltage and also set the maximum timestep to 1u , chances are you have already tried this.
Ken K
Ken K
Time step too small means that there are large changes in current or voltage in a very short time.
It may be that the model is more accurate.
You should look at the LTspice thread here on how to fix this.
A quick try could be to use the alternate solver, or one of the other integration methods. That often works for me.
See Tools|Control Panel|SPICE.
Jan
Try this:
*
* Generic triode model: ECC88
* Copyright 2003--2008 by Ayumi Nakabayashi, All rights reserved.
* Version 3.10, Generated on Sat Mar 8 22:42:43 2008
* Plate
* | Grid
* | | Cathode
* | | |
.SUBCKT ECC88 A G K
BGG GG 0 V=V(G,K)+0.34001426
BM1 M1 0 V=(0.009343174*(URAMP(V(A,K))+1e-10))**-0.49661195
BM2 M2 0 V=(0.75127268*(URAMP(V(GG)+URAMP(V(A,K))/26.621288)+1e-10))**1.9966119
BP P 0 V=0.007130155*(URAMP(V(GG)+URAMP(V(A,K))/35.434921)+1e-10)**1.5
BIK IK 0 V=U(V(GG))*V(P)+(1-U(V(GG)))*0.0041180199*V(M1)*V(M2)
BIG IG 0 V=0.0035650775*URAMP(V(G,K))**1.5*(URAMP(V(G,K))/(URAMP(V(A,K))+URAMP(V(G,K)))*1.2+0.4)
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)
BIGK G K I=V(IG)
* CAPS
CGA G A 1.4p
CGK G K 3.3p
CAK A K 1.8p
.ENDS
*
*
* Generic triode model: ECC88
* Copyright 2003--2008 by Ayumi Nakabayashi, All rights reserved.
* Version 3.10, Generated on Sat Mar 8 22:42:43 2008
* Plate
* | Grid
* | | Cathode
* | | |
.SUBCKT ECC88 A G K
BGG GG 0 V=V(G,K)+0.34001426
BM1 M1 0 V=(0.009343174*(URAMP(V(A,K))+1e-10))**-0.49661195
BM2 M2 0 V=(0.75127268*(URAMP(V(GG)+URAMP(V(A,K))/26.621288)+1e-10))**1.9966119
BP P 0 V=0.007130155*(URAMP(V(GG)+URAMP(V(A,K))/35.434921)+1e-10)**1.5
BIK IK 0 V=U(V(GG))*V(P)+(1-U(V(GG)))*0.0041180199*V(M1)*V(M2)
BIG IG 0 V=0.0035650775*URAMP(V(G,K))**1.5*(URAMP(V(G,K))/(URAMP(V(A,K))+URAMP(V(G,K)))*1.2+0.4)
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)
BIGK G K I=V(IG)
* CAPS
CGA G A 1.4p
CGK G K 3.3p
CAK A K 1.8p
.ENDS
*
Also you could try just using DC rails instead of modelling the power supply (if it's the same file you posted on the other thread).
I tried stripping the PSU but that didn't work.
Have already tried the standard library model and the NH6DJ8 model, both the same.
Will try a few other things.
Shoog
Have already tried the standard library model and the NH6DJ8 model, both the same.
Will try a few other things.
Shoog
Its just off the diyaudio LT spice sticky post I think. I did have a problem with and LTP and a CCS to ground. I had to add a 10Meg resistor to ground to get the DC to initialise.
Timestep too small means the system is attempting to progressively use smaller and smaller time steps and found that it didn't converge.
High frequency or dI/dt (such as SMPS) modelling seems to be quite a regular thing. Check the frequencies being produced for the model too. As others have said being more specific with convergence parameters such as resistances for grounds, in series resistances for caps and voltage/current sources or sinks to ensure that suddenly you don't see a meteoric rise or fall of current. One other thing that I've seen before is using the startup option - this starts from 0 rather than X and so there's no big starting current that some models had issues with.
Also you can setup a standard parameter to add in series resistance (ie 10Meg etc) to ground which also helps - under the options tab.
High frequency or dI/dt (such as SMPS) modelling seems to be quite a regular thing. Check the frequencies being produced for the model too. As others have said being more specific with convergence parameters such as resistances for grounds, in series resistances for caps and voltage/current sources or sinks to ensure that suddenly you don't see a meteoric rise or fall of current. One other thing that I've seen before is using the startup option - this starts from 0 rather than X and so there's no big starting current that some models had issues with.
Also you can setup a standard parameter to add in series resistance (ie 10Meg etc) to ground which also helps - under the options tab.
Last edited:
It looked like the 6DJ8 model was taking to long to respond to the applied voltage. Appeared like it was fighting a ringing voltage line as the PSU was coming up. Fine tuning the chokes might help to stablize things.
Yes it complains about the LTP CCS without a parallel resistor, but adding it in send the input stage into wild oscillation. it runs better without.
Shoog
Yes it complains about the LTP CCS without a parallel resistor, but adding it in send the input stage into wild oscillation. it runs better without.
Shoog
Last edited: