TCA0372 (or other)

Much better thermally, especially the SO16 wide. No thermal pad, but four V- pins. There are some no-connect pins that we also use to transfer heat out. A giant topside copper pour under the part to hit 8 pins, then a lot of vias to inner layer or bottom pours to transfer heat.

You could get crazy and add epoxy or a gap-pad on the underside of the part. But half a watt is tiny.

Looks OK if you heat sink the leads to lots of pcb copper. Dips seem so last century.

Probably, but the TCA is an amazing bargain.

Specs say 137C/w. So 1/2 watt would take the junction to 94C in a 25C ambient. Operating junction max is 150C.

Note that the dip is available in quantity for now but designated for EOL.

I could find no difference in the data sheet between the two. My personal inclination (given a supply choice) would be to select the B version.

Thanks, Trolling DK I found the OPA551/2. (60 V, 200 mA) But the TCA's are burning a hole in parts bin.

George H.

Thanks John S, The B version has more supply current... that's all I could find.

I didn't see the 'last time buy' flag. Thanks.

George H.

You shouldn't have any issues at 0.5W for the DIP-8 package. I agree that a power pad would be better. Weird about the B version--the only difference shown in the datasheet is higher typical supply current.

It would be nice if they had a version with lower V_OS.

Cheers

Phil Hobbs

AoE, Table 4.2b, page 272, lists a dozen possibilities. "Monolithic Power and High-Voltage Op-Amps" with output currents from 0.5 to over 10A, and voltages to 120 volts.

But then, you have TCA0372 burning a hole in your pocket.

It would be fun to see how much heat comes out which leads.

Well I'm using it in a thermal loop so V_os is not so important. George H.

Thanks Win, Hey, no TCA on that page? (you need a column showing which have a thermal pad.. maybe in the x-chapters. :^)

George H.

Negative power lead(s)?

OK, so this worked fine today, but it's a dang V^2*R heater, (and I'm not sure where to put the gain. (thermal loop)) And I was thinking of a single break point, SQRT approximation. (A diode(Zener?) and R in the feedback, in parallel with the linear R)

So say a voltage dynamic range of X where should I put the break point? Sqrt (X) seems too low to me.

George H.

I drew a picture, sqrt X looks about right. (sorry) ~5 volt zener.. how mussy are they? GH

Probably. An xray would help.

Is your loop analog? I've done a full square root in the loop, but that was uP code. In fact it included supply voltage compensation, so I had a subroutine that knew V+ and knew Rheater and made heater power linear on input. Later I got brave and made NMR heater controllers PWM and gambled that the PWM frequency wouldn't show up in the spectra; it didn't.

A single breakpoint should make the transfer function as linear as possible at all points in the operating range (no, I haven't done the math). The sqrt function gets wild near zero, but your heater probably doesn't have to work at zero power. Probably you can ignore the low part of the operating range.

There must be some papers somewhere on optimal placement of breakpoints.

Right full analog. A uP is the way to go for most of us. A diode and the right resistor. It mostly hangs out at the 1/4 power point (1/2 voltage). As you say there's a nasty bump at low voltage... which is what causes the problems at higher powers. (linear-wise)

George H.

You won't have much choice on package soon, looks like, Mouser lists the TCA0372 in 8-pdip as EOL.

All DIP-8 TCA0372s appear to be EOL at this time

The schematic in the TCA0372 datasheet is still wrong after all these years

Yes, and Yes.

Right, (see John S above.) Hey I noticed Rochester Electronics has about a million in stock.

George H.

Well that's good at least. The DIP package was the best! (because I burned them out)