Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Does Solid State Marx Generators actually work?

Status
Not open for further replies.
By trying a while longer I managed to make a workable simulation (in theory). Although this thread has the name Marx generator, the topology (switching capacitors between H-bridges) appears to be a circuit which I first saw named a Nakagome charge pump. Thus the name on my screenshot.

I didn't see the connection between charge pump and marx generator, but now you say it - I can see it, interesting.

There are no rogue or backward current flows.

I want to see your rogue or backward current flows and try your latest simulation, so I wanted to try your simulator at falstad.com, but falstad seems to be down today.

Since high voltage is present, mosfets are not the right device to use.

I don't understand why you say that? There are much more high-voltage high-current mosfet's available than there are BJT's


To provide bias there may be a suitable node which produces the right voltage at the right time. It may be a supply wire, or a ground wire, or another leg of the same transistor. Sometimes a transistor is not needed because a diode can do the job.

An alternate switching device might be SCR's. They continue to conduct as long as capacitors are charging. As charge current declines to zero, SCR's turn off. This simplifies clocking.

It doesn't sound like a reliable bias to me in this application, especially when the load could be a spark gap, then the circuit will resonate like the church bells of Notre Dame when Quasimodo gives them a spin.

Speaking of spark gaps, you say that in order for a Nakagome charge pump to charge it has to be connected to the load, that would mean it will not function as a real marx generator, if that still holds true in your topology.

The problem with SCR's is their slow turn on/off times, so they are not very suitable for fast switching. What you need is something that is half mosfet and half transistor - the IGBT, then you also get your PN/NP junctions.

BTW, I finally found a spice model for my IGBT's that works in ltspice, so now I can make a better simulation when I put it together in ltspice tomorrow I hope..
 

The new PCB's arrived, but I'm not sure it's worth the trouble populating them. I have tried to make a realistic simulation by using the correct models, and now it behaves very closely to the physical circuit. With 3 stages I'm losing 33% of the energy, so this circuit will overheat if running continuously.

How do I reduce the energy loss? In the meantime I will try BradTheRad's excellent suggestion but with IGBT's.

Screenshot from 2019-07-22 14-14-18.pngScreenshot from 2019-07-22 14-13-10.png

- - - Updated - - -

Just to inform you that replacing the diodes with voltage controlled switches added 10V to the load , so you could say that the diodes account for 5% loss.
 

The energy loss happens due to low load resistance. Discharge time constant Rload*Cgenerator is in the same order of magnitude as the output rise time.
 

The energy loss happens due to low load resistance. Discharge time constant Rload*Cgenerator is in the same order of magnitude as the output rise time.

I tried calculate the loss more precisely, with 20ohm load efficiency is 46% and with 200ohm load efficiency is 29%. And both results are based on 100V input.

With 1000V input 20ohm load the efficiency is 16% and not to mention the max output voltage is only 2.4KV (4KV expected).

I added a fourth stage, and then a switch before the load just to keep the trace flat until marx is erected.
Screenshot from 2019-07-23 17-02-21.pngScreenshot from 2019-07-23 17-00-14.png

I'm beginning to think those marx generators are too flimsy and too lossy

And try see https://uspas.fnal.gov/materials/11SBU/PPE_AdvancedTopologies.pdf page 7-9 "2% parasitic capacitance reduces output voltage by 30%"
 
Last edited:

I'm beginning to think those marx generators are too flimsy and too lossy
Efficiency requirements and low load resistance have been introduced rather late to your thread. The present circuit seems to be operated outside the useful load range. I agree that other circuit topologies may be better suited for your purposes.
 

Efficiency requirements and low load resistance have been introduced rather late to your thread. The present circuit seems to be operated outside the useful load range. I agree that other circuit topologies may be better suited for your purposes.

I agree that efficiency requirements are a late addition, but that was based on my assumption that the only loss there would be the igbt switching loss and miscellaneous pettites not exceeding 10%. Unfortunately my assumption was based on my ignorance, so it was wrong, which I see now.

The low load resistance was inferred by the fact that it's a marx generator, but I should have made that clear.
 

There are much more high-voltage high-current mosfet's available

I guess it's a question whether the body diode is fabricated in all mosfets (rated for high or low voltage), and whether it allows current to flow backward up into the positive power supply (the 'rogue' current).

In addition I often see unexpected behavior in the mosfet model in Falstad's simulator.

you say that in order for a Nakagome charge pump to charge it has to be connected to the load

Just in the sense that a current path has to be available to include the capacitor in a loop in order to observe discharging as well as charging. I suppose you referred to a previous thread:

www.edaboard.com/showthread.php?267438-nakagome-charge-pump-doubt

The concept does work to accumulate charges on multiple capacitors.

Falstad's contains a spark gap element. This simulation demonstrates 4 capacitors, analog switches and 200V supply. The spark gap is set (perhaps unrealistically) to break down at 980V.

Marx gen (4 cap Nakagome ana-swi) 200V supply 980V spark gap.png

After a discharge the capacitors may take an extra cycle or two to charge, until voltage is high enough to span the spark gap.

The screenshot was taken as a discharge was occurring across the spark gap.

The diodes make life easier. They prevent current backflow.

The 50 ohm resistors are unnecessary. Without them each transition results in severe current spikes.

- - - Updated - - -

Here is the link which contains the above simulation.

tinyurl.com/y32lfkry

By a click it:

1) Opens the website:

falstad.com/circuit

2) Loads my schematic in the simulator.

3) Runs it on your computer.

Your computer needs to have Java installed.
 

Hi BradTheRad

Here is the link which contains the above simulation.

tinyurl.com/y32lfkry

falstad.com is up and running, but unfortunately when I click on the link it opens a default LCR circuit, maybe wrong link?


I guess it's a question whether the body diode is fabricated in all mosfets (rated for high or low voltage)

Power MOSFET's all come with the body diode, that's (also) the reason why I use IGBT's - you can get them without body diode.
 
Last edited:



I made the simulation in LTspice as I have my own models there.

Screenshot from 2019-07-27 14-56-58.pngScreenshot from 2019-07-27 14-57-54.png

At first sight I notice the energy output is higher that the input, and I assume it's because the caps discharge from 200V to -20 to -60V, so I'm not sure the sim is running correctly.

Also I'm not sure about the way the discharge is working when (c1-) is connected to the 200V during discharge, it seems to be working, but how/why?

I mean it would be reasonable to expect that the load should be connected to (c1-) instead of GND because there are 4x200V in the caps. But when I tried that the circuit worked like crap and the output voltage was a meager 540V, all in all it resembled the other topologies I tried.

I think I will put it together when the new PCB's arrive, it would be interesting to see how it behaves in real life.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top