Hey all! Newbie here. Been trying to follow a few tutorial videos on running transient analysis on rectifiers and multipliers in ngspice but getting “low time step” errors when attempting to run it in Kicad’s simulator. I feel like I’m just overlooking something obvious, but have tried grounding the output, switching out the diodes, etc to no avail. I’d like to be able to play around with the capacitor values and see how they affect the dc voltage drop and smoothness
Appreciate any help
The ground node for simulation has to be called either GND or 0, nothing else. And it is required!.
Sometimes when simulating with higher voltages, setting some options to relax precision may help. Put
.options abstol=1u chgtol=1e-12 reltol=0.01 method=gear
into a text box on the Eeschema canvas.
If this does not help, please zip your project (including the diode model) and post it here.
I have published a more complex multiplier Hybrid Switched Capacitor converter, 28 to 700 V as a simulation example at Simulation examples for KiCad/Eeschema/ngspice - #4 by holger.
Removed the numerical from my ground and input that parameter, did the trick thanks a bunch! Appreciate all the examples I can look through at that link as well.
As some context, I’d like to use a 12-24V transformer with a 1.5-2a rated secondary and some sort of voltage multiplier topology to get 80V+ to the plate of a 12AX7 vacuum tube (draws <300mA) running in parallel with 12-24V AC rectified to power my op amps and tube heater at 12.6V (inrush ~1.3A until heater filaments warm up). In your opinion, is the Cockcroft-Walton multiplier a viable solution? I’ve seen some people selling pcbs that use a mosfett, 555, and voltage regulator to convert dc-dc for this exact application, but have read you shouldn’t use a dickson charge pump for high voltage applications which leaves me confused.
I am not a circuit designer, so I cannot answer your question solidly.
Concerning the 12AX7, my data book from 1968 tells me that its anode current is 1.2 mA, not 300 mA. Maybe this helps when looking for an upconverter.
No worries, I appreciate your help!
FWIW, typical voltage multipliers can get very inefficient and have high output impedance, meaning that if the load current varies, the output voltage will vary with it. My experience has been with high frequency multipliers, not 60 Hz.
If you can keep the number of stages low, there’s a decent chance of it working at 1.2 mA. At 300 mA, probably not so well.
Some suggestions for simulation: get the datasheet for the caps you are using and find out the equivalent series resistance of the caps. Put this value in series with each cap, and you will get a better idea of the multiplier performance. Then, you should vary the output load resistance and make small plot of the dc output voltage as a function of dc output current or output resistance. This should be done to an output current at least as high as the maximum you ever expect to see. This will give you an idea of how steady the output voltage will be.
A no-load simulation like you show will most likely give you an optimistic idea of output voltage, though that depends a lot on your expected max load current. 1.2 mA is pretty small.
John
Yeah not sure where I was getting 300mA from, it’ll see a max of 6mA on the higher voltage. Not dead set on a transformer but picked up a cheap low profile hammond that has a single secondary rated at 24V 1.65A to at least start tinkering. With two half wave stages that gets me to just under 100V in optimistic circumstances with 4 diodes, 5 caps and a thermistor to try to keep the initial inrush <500mA. Have considered also using a 12V with two secondaries and switching to a full wave that’ll still give me just under ~100V in the same two stages.
I stumbled across this article on picking the proper capacitance based on ripple (https://www.irjet.net/archives/V3/i2/IRJET-V3I2114.pdf) but ended up more confused, so I’ll probably spend a bit of time browsing old posts on stack exchange or some other forum where someone explains picking proper capacitance and power diodes for this sort off topography to come up with some proper values, I just kind of threw in whatever to get an output. If you have any of your own links/insight regarding this I’d definitely appreciate it, otherwise appreciate the help/ideas for making the sims accurate, cheers john
If wanting to get a basic understaning, it might be best to start with a very basic circuit that includes a minimal of items that you can use in more advanced circut knowing (it/they) worked in a minimal circuit.
You can use the Tuning tool to glean some analytical knowledge about it…
Below shows a basic Capacitor circuit in both Kicad/ngspice (and LTspice for comparison) - same results (but, note the difference is Tran statement)
(I use this circuit for identifying Cap values for my circuits and, testing actual Cap hardware produces darn-near exactly as the simulation)
Video shows Tuning tool (I should have Zoomed in for clarity, sorry…)
I can’t offer that much insight. Years ago, I spent a lot of time looking at voltage multipliers and capacitiance sizing for high frequency voltage multipliers (100 kHz to 1 MHz, up to 10 kV. I found a lot of papers with contradictory results, and most of these papers were from the 1950s through 1970s. A lot of the more theoretical ones made incorrect assumptions about the diode conduction intervals. In the end we resorted to simulation and experiment.
At 50/60 Hz it’s a lot easier. Include cap ESR in the model and possibly transformer leakage inductance if you have any idea what value it is, and I think you can get pretty close. Don’t neglect loading the output. Maybe the models are better at this frequency as well, because the parasitic capacitance and inductance matter a lot less.
John
