TCA0372 shortage?

Though why it should be that way if it is, is also mysterious, the internal schematic doesn't look particularly different than the TCA0732, except that the output stage as drawn in the TCA0732 datasheet doesn't make sense, I'm assuming it actually looks more like the ST part inside. :)

It may just be an anomaly of how ST do their datasheets

but max might happen, less than 10mV is better than usually 1mV occupationally 15mV, you have to design for it unless you want to start sorting

Incidentally it's cool how the latest revision of the TCA0372 datasheet was like literally yesterday and it seems no one has noticed the output stage as drawn doesn't make sense...

It's a cheap part and seems like there are few other parts like it at that price; if the outliers show up rarely enough that it makes economic sense to sort rather than use some more expensive part in a particular application then f*ck it, sort

But again it may just be an anomaly of how two different companies do their datasheet and the ST part is actually as good or better in practice, beats me. Just a thing I noticed :)

tor

razza%$&@!!

tages

I did notice the output on the block diagram made no sense the way it is drawn, I just assumed some drew an arrow wrong on a transistor and no one cared :)

it seems it been that way at least back to the Motorola data sheet I found dated 1996

Even if they drew the upper stage as a regular Darlington it's probably not a very good representation of the output stage, no ballast resistors in the emitters and three diode drops but they describe "zero-deadband crossover distortion." Is the slew-rate symmetric?

There may be some sliding-bias thing going on they don't want to elaborate on.

The ST part uses a current mirror arrangement as the upper driver which might explain the lower supply rating and offset voltage spec discrepancy if it exists.

Beasts like that are (IME) used for things like driving servomotors or making floating ground/supply rails (my usual application). The motors are inside some outer FB loop, and nobody cares about a 15-mV offset on a supply rail.

For the occasional precision application, wrap a chopamp feedback loop round it.

Cheers

Phil Hobbs

butor

hs!

afrazza%$&@!!

oltages

doesn't look symmetric on fig7

three diodes should match an darlington and compound pair no?

Yeah but that puts it in class B and there will be no crossover distortion if all the transistor junction drops are always matched to those of the diodes, how are the drops matched when the output devices heat up. maybe the diodes and transistor junctions are matched enough because they're on the same die and well-coupled enough to the output devices?

Don't know enough about doing power amps on an IC to know how they accomplish this.

The dual-Sziklai totem pole clearly wouldn't work as drawn, but it does suggest the actual performance of the TCA0372--it pulls closer to the positive rail than your old-school Darlington op amp output. (The lower limit is much more Darlington-ish.)

Cheers

Phil Hobbs

negative feedback can improve crossover distortion but can't eliminate it and it'll get worse as frequency increases and open loop gain drops. The blurb seem to imply it has zero (or close enough) crossover distortion, everywhere...

it may actually be more like the ST part with the current mirror inside and the output stage as drawn on the ON part is just (for whatever reason) sanctioned BS!

Good grief, a drop-in replacement! Almost.

The voltage rating is interesting. 50 volts for 50 milliseconds.

Other strange specs are the single negative supply (0 volts) and the noise (10 uV).

Good to know about.

ributor

nths!

my

zzafrazza%$&@!!

voltages

eet

put

rs

so that bottom should be a darlington and the top Sziklai ?

Can't do a monolithic power PNP AFAIK so the bottom can't be a P-darlington I don't think.

stributor

months!

nd my

razzafrazza%$&@!!

ly voltages

sheet

utput

bly

tors

nd

er

but if the top is a Sziklai it is the equivalent to a PNP then the bottom should an NPN

Rochester is showing >120k pieces of the TCA0372 in PDIP, so in a pinch we could make a reverse DIP breakout board to fit the SO16-wide footprint. ;)

Cheers

Phil Hobbs

or if you have room for it add both footprints overlapping

I could also make my own mouse-bite baby board, with an MSOP dual opamp and some discretes. I really don't want to do that. I am using the wide surface-mount package.

One other possibility is to parallel a LOT of small opamps. I just did five RRIO amps in parallel, as the bias supply for a lithium niobate modulator.

I had a beautiful pulser output stage, then the fast PNPs and Avago phemts both died. My GaN version is beautiful, but I need this opamp.

Actually, the TCA seems to be an active part. We'll buy a couple k more just in case.