Simple inverter cct.: comments, please

[Skywave]

A simple inverter circuit - found on the 'Net. Drawing is below. The general idea is that the two bipolar transistors constitute a bistable osc. and that drives the two power mosfets. The transformer is a small 9v.-0-9v. primary and a 230-v. secondary (a small mains transformer wired 'back-to -front').
Can someone confirm for me, please, that there are two drawing errors here? Surely the lower 'leg' of each 680 Ω resistor should be shown connected to the gate of the corresponding mosfet?

Any other general remarks about this simple cct. will also be appreciated.
Thank you.

Al.

[Maarten]

Not directly an answer to your question, but it could help in searching experiences of others with the same circuit and is generally a good idea to mention: Where was the schematic found?

A more direct answer to your question: I'm sure the gates should be tied to the relevant 680R/collector. There will probably be some switching losses.

[kalee20]

Where the horizontal traces from the MOSFET gates cross the vertical traces from the transistor collectors to the 680Ω resistors, they should be joined (not just a crossover).

With electrolytic capacitors in the timing, the frequency could be out quite a lot from the nominal - and the mark:space could be quite different from the theoretical 1:1. This has implications that the transformer could see an imbalance, and saturate. The corresponding FET will then be severely stressed. But if the transformer winding has a lot of resistance, it will save the FET - the catch is that efficiency will be compromised.

A better approach would be a divide-by-2 flip-flop to guarantee exactly 50:50 square waves to drive the FETs.

[MrBungle]

Built a few of these. They're not very efficient as the switching is really soggy so the MOSFET sits in the active region for a long time emitting heat. That and the transformer saturates pretty sharpish. They really need to be tuned carefully. Flames come out of the MOSFETs otherwise (I just have killed ten or so building just one of these)

[mole42uk]

I wonder why the circuit is represented as a sort of 'pictogram'? To have any understanding of how it works, or how well it works, I'd have to re-draw it! The errors would then be easy to spot.

[Radio Wrangler]

I strongly agree with Kalee.

The choice of electrolytic timing capacitors is bad. The inverter transformer and Mosfets are very susceptible to problems if the switching waveform is asymmetric.

The two transistor multivibrator circuit has a latch-up isue that allows it to turn on with both transistors on, so that it fails to oscillate. They also can die due to negative-going transients on the base-emitter junctions.

And as Mr Bungle says, the Mosfet gates have quite high capacitance, and the leisurely slow switching they get from this sort of driver wastes a lot of power with the mosfets lingering in a partially on state, spoiling efficiency.

There are driver chips which include an oscillator, a flip-flop and fast driver stages to speed the switching of the Mosfets. They solve a number of problems.

David

[TrevorG3VLF]

"The two transistor multivibrator circuit has a latch-up isue that allows it to turn on with both transistors on, so that it fails to oscillate. They also can die due to negative-going transients on the base-emitter junctions."

The negative voltage on the base-emitter junction will approach the level of the supply voltage. This circuit was OK with old germanium transistors where the permitted base-emitter voltage could be up to 20V. Modern silicon transistors have a limit about 6V. If higher voltages are wanted, then a diode should be inserted in the base circuit to eliminate reverse current.

Trevor

[ionburn]

I would agree with many of the comments above. I have, and occasionally still use, this type of circuit. I tend to have a bit more embellishments to the circuit like additional protection for the mosfets and protection diodes in the emitters of the mutivibrator transistors. I also use more stable capacitors for setting the frequency rather than electrolytics.

Although I have had good success I did start to come across problems when I went beyond fairly low output powers (around 100 to 200V 50mA). This was more of switching high input currents, and may well be due to issues mentioned above. I have had more success with other approaches, rather than higher precision design of this type of circuit.

[Skywave]

Originally Posted by Maarten

Not directly an answer to your question, but it could help in searching experiences of others with the same circuit and is generally a good idea to mention: Where was the schematic found?

The source: http://www.electronicshub.org/12v-dc...converter-circ

Al.

[Skywave]

Originally Posted by mole42uk

I wonder why the circuit is represented as a sort of 'pictogram'? To have any understanding of how it works, or how well it works, I'd have to re-draw it! The errors would then be easy to spot.

Indeed: I wondered why it had been drawn in that fashion too. Yes, I did re-draw it in the conventional style and came to the conclusion as I stated in my OP. So why did I then produce this thread? Two reasons: Just to get a second opinion / confirmation and a general appraisal of the cct. - which has been received - for that, thank you, everyone.

Al.

[ukcol]

Pulling some of the points made together.

A conventional circuit diagram makes things much easier.

The circuit below is 'as drawn.' This shows a bi-stable multivibrator that won't run without input pulses and floating FET gates.



This second circuit shows what we think was intended with its electrolytic timing capacitors and lack of emitter/base reverse bias protection.



Still pretty horrible.

[paulsherwin]

I don't know much about this stuff, but would a simple design like this be adequate at low power levels, such as the 90V 20mA needed for battery valve HT?

The switching frequency seems quite low. Would it not operate better at higher frequencies?

[MrBungle]

Depends really. Sometimes these things literally don't explode due to the source impedance of the power supply rather than any design.

Edit: remove failure mode - misread IRF630 datasheet.

I might build this quick and see what it does ... be right back!

[MrBungle]

I built it, well the oscillator half of it. It works but performance is abysmal to be honest.

Looks like:



On the scope:



I've only got a couple of IRF510's floating around, which I need, so I'm going to skip the sacrificial step of attaching them to a transformer. From the scope trace and the curves of the IRF630 datasheet, this is going to blow up or just burn most of the power off as it's not anywhere near saturation most of the time. Driving nothing, quiescent current is 23mA.

IRF630: http://www.vishay.com/docs/91031/sihf630p.pdf

At least the IRF630 has protection diodes so it won't blow up in that interesting way!

[paulsherwin]

Why is the square wave so poor?

[MrBungle]

Well this is interesting. I had the transistors in backwards above. I'm actually surprised it oscillated at all! May have been related to lunch time

I pulled two of the capacitors I used so it has 1uF of total capacitance which gives around 70Hz.

Corrected trace:



And I shall hang my head in shame at the schoolboy error above.

Edit: I'll add some diodes to square up the wave later. Need to got to the post office

[jimmc101]

In addition to the points already mentioned there is the problem of the slow rise-time at the collectors caused by the 2.2uF capacitors having to charge via the 680R resistor (the other end is clamped to ~700mV by the base emitter diode of the opposite transistor). See attached file. Note SW1 is only there to start the simulation, without it the simulation locks up as described (both transistors on)

There was a long thread about a solid state vibrator here which may be of interest.

Jim

ps Mr Bungle's breadboarding is faster than my simulating!!

[paulsherwin]

That does look a bit better, though the rise time is still disappointing.

[MrBungle]

I can speed that up easily.

Originally Posted by jimmc101

ps Mr Bungle's breadboarding is faster than my simulating!!

Only because my bench is a tip so all the parts were already on it

[MrBungle]

Post office was shut (yay!).

I've added the diodes:



Looking much better:



And good rise time of just shy of 40ns!



I'm going to order some IRF630's from RS (they're only 65p+VAT each) and see where this goes tomorrow