Accurate SRD spice simulation models

[Leo357]

Hey guys,
I have been unsuccessful in searching for a working, accurate SRD spice model and it seems like many have the same problem too. I was wondering if there has been any progress in the simulation of them or if anyone has something worth sharing. I am mostly interested in the MMD-805-T86 model, but anything working will do.

I tried simulating a sine wave generator with the SRD and expected the wave to become '± a pulse generator, where the negative sine voltage is reduced', yet it's nothing even close to that.

What I have attempted to do already and what problems did I run into:
1. I have used the Metelics Spice Model for Step Recovery Diodes, but it didn't seem to have the effect that's expected.

2. Probably the most popular post on this issue had an expert say this, which makes me not very hopeful: "SRDs are the most nonlinear devices available and toughest to model. I have worked with SRDs for designing the higher order frequency multipliers. I used Teklec SRDs for my designs, only problem is that, you need to provide post fabrication tuning options to make it work. I haven't come across an SRD model, which really works ,i.e simulation results matches with practical circuits."

3. The datasheets provided with most SRDs don't have enough info about them to create a working spice model. I have come across one datasheet which contains all the necessary info, but still doesn't work.

P.S. I am new to electronics, but tried explaining the problem to the best of my ability. If something is unclear, I'll gladly verify it.

 

[biff44]

i have not used spice in over a decade, but i do not remember spice having an accurate charge storage model for SRDs

imagine a simple comb generator circuit, with a series inductance, and a shunt mounted SRD. the large signal input sine wave drives the SRD into forward bias conduction. then the input sine wave changes polarity, and drives the diode into reverse bias. But all that charge is still stored in the actual Diode Intrinsic layer, and it has to slowly drift out of the I region. all the while that drift is happening, there appears to be a large reverse bias current thru the diode. right up until the very last electron leave the I region, and then the current thru the diode abruptly goes to zero.

but since changing the current in the input drive inductor from a large negative value to an abrupt zero value, the inductor's collapsing magnetic field causes a huge voltage spike to happen. That voltage spike is rich in harmonic content, so you end up with a comb generator. but modeling that abrupt large reverse current as it suddenly goes to zero--that is what spice does not model well.

and something similar, but less obvious, happens if you are trying to make a X5 multiplier, for example....there are transit time effects, and harmonics of current, as well as voltage dependent changes in junction capacitance. complex stuff