well, I didn't purchase them, and I know at some point we'll be cleaning up the bench at work. I could make one tube disappear during clean up! :)
I'll get back to you on that.
That's what I thought. Similar to us discrete parts guys with filters where we pull some pre-cooked schemes out of the hat, have Spice piece it together, then stomp on the group delay a bit so it becomes more flat. Except that I don't use PSpice anymore.
Just curious: What happens if the load on VOUTN bucks a bit, dumps back, and X1 gets triggered while M1 is conducting. A muffled boom, solder splattering about? Or is such load behavior not s'posed to happen?
My first buck switcher used an uncompensated uA709 in hysteretic mode, which drove a 2N2219, which drove a 2N3055 as an emitter follower. Since the 3055 didn't saturate, it was pretty fast. Ran at 22 KHz, just about my auditory limit in those days.
Here's my latest switcher, 350 KHz roughly...
John
Very clever use of the REF input. kind of shortens the component count! :) How well does it maintain (-) with respect to the (+) output ?
Very nice. Was this the one that had stability issues a while ago?
I wonder why the Rdson of the LM5112 output devices are so different. Looks like it can really only sink but not source much. Otherwise it might have been a nice chip for small direct drive push-pull converters. However, it's out of league in cost for many of my designs anyhow.
AFter looking at the circuit a bit, I wonder what happens if you place an excessive load on the output? It looks to me that it may stop! just an observation. hope not to be in the wrong state :)
You are sort of correct, but the ultimate rejection if the LDO is designed correctly is the ratio of the capacitance of the pass device (generally a p-fet for a positive regulator) to the load (output) capacitor. It really isn't hard to do such a design, but it requires the insertion of parasitic devices in the chip to achieve a high "passive" PSRR.
I never designed a positive regulator with a N-fet on the output stage (i.e. a follower rather than a pass device), but I suspect much better high frequency rejection could be achieved. Of course, it wouldn't be a LDO but a high drop out regulator.
Just off the top of my head, since many op amp circuits don't need DC accurate supply rails, but just quiet rails. So I wonder if you could get better high frequency rejection just whipping together a zener a N- fet as a positive regulator, with a pile of capacitance across the gate to ground.
As an example,. I've used really crappy shunt regulators inside chips just to achieve some internal PSRR. Not very accurate as far as voltage goes, but it keeps the outside noise out.
Nice. So the -12 is ratioed to (+5 supply + 1 Vbe)
How much current can it handle? A few hundred mA safely?
Best regards, Spehro Pefhany
Yup. Plus the BCX has to pull down the oscillator node, which takes a little extra current. The values shown produce about -12.5 at mid-load. Regulation is plenty good enough to power opamps.
It was designed to supply -12 at up to half an amp, but it's good to about 0.9 before it toasts the dinky 10 mm inductor I'm using.
John
REF input? I don't follow that. The "A" input to the fet driver is just another logic level that I'm not using.
The -12 is referenced to the +5 supply and doesn't much care about the raw +12 coming in. Load regulation is pretty good, a tenth of a volt or so from 35 mA (where it gets continuous) to 500 mA, my max load.
John
If the voltage is more positive than -12, like shorted maybe, Q1 is off and the oscillator free-runs, which turns on the p-fet at about
75% duty cycle. That will soon pump a huge current into the inductor hence into the load.My +12 supply comes from a switchmode wall-wart, and is polyfused [1] as well, so if anything goes wrong in this inverter, or in its load, nothing radical should happen.
John
[1] Two 1206 polyfuses in parallel, actually.
Yup. It's fine once I made the moral compromise to use a tantalum on the output. It's derated on voltage, and I slow-start the +5 reference a bit, so wish me luck.
7 amps/3 amps ain't bad for a SOT-23! This is a cool part because the output rails are sort of independent of the input logic ground, which could be handy in other places.It's 50 cents in moderate quantities!
John
You can use one inductor to make a positive and negative supply, both regulated. I would say this is only possible with pulse skippers (PFM) That is, you need to dump the "charge" from the inductor fully rather than continuous conduction since the inductor needs to be shared.
There are chips that use such a scheme.
Cincon makes some very nice, inexpensive DC-DC converters. The following is an example of one of them:
The LTC1263 does the 5V to 12V conversion for you. Linear thinks it is for programming flashes, but they are wrong. It is really for running op-amps.
The LTC1144 will make a not quite -12V from the +12V.
Nice (although very expensive and quite limited current). No inductors at all, and all ceramic SMT caps should be possible.
Best regards, Spehro Pefhany
Note sure how appropriate my response is with respect to the project at hand:
..but..
(40106 do everything! :-) oscillating, one-shotting, PWM-ing, beeping, blinking etc. -a must for the junkbox)
You may also need a 2N7000 or the like.
s d -V maker--- ---- Op-amps ! ! ! ----- N channel ----- ! GND
Some of these down pumpers get messed up if the output gets pulled above ground. This prevents that problem.
The 555 (even the CMOS versions) can give you a regulated +12V from a
+5 input.I've got no cursor on the the display so I will explain rather than doing ascii art.
The 555 gets wired as more or less a normal oscillator circuit using pin 7 as the feedback. I resistor from the output voltage to the TRIG input prevents the oscillator from cycling if the output is over
+12V. If the output capacitor is large, the ripple is not too big.
A tantalums, oh boy ...
There has to be a reason why the words tantalum and tantrum sound so similar ;-)
Yes, but look at the RDSon for pulling up. Even a cheap npn/pnp follower can do better.
Sure. But my PWM sections usually consist of 1/6th of a Hex-Schmitt plus a npn/pnp follower, to a grand total of just under 10c :-)
Have something to add? Share your thoughts — no account required.
Ask the community — no account required