LM828 voltage converter question

It's probably bootstrapped internally.

Cheers

I'd guess node "CAP-" as substrate, S4 as N-channel, source tied to "CAP-", gate tied to ground, drain tied to "VOUT". ...Jim Thompson

It seems simple enough to me?, S1 and S3 charges the cap and S2 and S4 passes the charge in reverse order to C2 (output).

I don't see where a - supply is needed? C1 when switched on will charge C2 with a - voltage because C1 is being attached in reverse order.

Jamie

It's a flying-capacitor inverter, not actually an H-bridge functionally. Of course that's just definitions, but the fet sequencing is different.

When the right-side switches turn on, the CAP- node jumps to -V. So S4 can be an n-channel fet with positive voltage on its gate; or even ground. I don't see any need for any voltage more negative than Vout.

The gate drives probably swing between Vin (V+) and Vout (V-). S1 and S2 can be p-fets, S3 and S4 n-fets. Can't be sure without a detailed schematic. I think S2 could be an n-ch, too.

You could make one of these yourself with discrete fets or a quad analog switch, an HC4066 maybe, with an external clock. It could bootstrap its own V- supply.

I kinda like the (as Phil named it) Groucho Marx Generator, a charge pump using four or more opto SSRs and big electrolytic caps, switching at a couple hundred Hz. Very low noise.

John

Thanks for your reply. Right, I see now that S4 can have its gate grounded and will operate properly - don't know why I thought it needed a negative voltage. Won't there be a problem with shoot through in a discrete version if there isn't dead time between the first set of switches opening and the second closing?

Yup, some non-overlap in the gate drives is prudent. I think HC4066s do that automatically. So can my opto SSRs.

S4 could have its gate grounded, but efficiency would be better (lower Rds-on) if the gate is pulled up to V+. So they probably do that.

John

I tried playing with a crude version in LTSpice and can't seem to get it to operate properly - when S2 closes CAP- only pulls down to about -1 volt, and there's a 100 amp pulse through S2 as CAP- rises back to 0 volts. I think it needs dead time.

It looks to me like the voltage doubling charge pump version will work without dead time (though there will be efficiency loss from shoot-through) because you're just stacking the voltages and there's no way for the pumping capacitor to discharge other than into the load. But I don't think the voltage inverting charge pump will work without dead time, because you have to swap the ground connections on the pumping capacitor, and without dead time there's a period where both leads of the pumping capacitor are shorted to ground.

Good ole 4066's.. I made a circuit years ago for the C-64 that allowed me to grab a tone from a FSK signal and convert it to a analog slope for the Joy stick input of the C-64. Wrote code in the C-64 using my own assembler software, for various different types of communication modes along with receiving SSTV and FAXes..

I used the 4066 as a gating system for a sample and hold on the outputs in the RC timing ramp.

Triggered the sequence via an edge detector on the squared wave that was processed via a high gain v-comparator from the receivers audio output..

it worked flawlessly..

But for some reason, I remember the low state R(ON), being around 250 ohms, not 50 as in the SN74HC4066 ? Maybe I used some one else's.

Jamie

you mean you think you see some shoot through?

Jamie