| General > General Technical Chat |
| Why is the 741 op amp still produced? |
| << < (20/24) > >> |
| coppice:
--- Quote from: rsjsouza on October 01, 2020, 05:19:25 pm ---Don't forget that evolution usually is also tied to potential revenue. The fact nobody ever refined a process to reach LTZ or LM levels may be due simply to a lack of interest or too narrow of a market to be tapped into. The number of different voltage references available when compared to 30 years ago is impressive, but they are all competing in a segment with more relaxed demands. --- End quote --- I've been in new product discussions a few times where someone proposes something, people look at the numbers, and conclude that any more players in such a small market will mean everyone ends up making a loss. These projects do NOT move forwards. People are concerned about tit for tat move - we screw up their nice little niche, and they'll screw up one of ours. While you don't see new very high stability references appearing, you see plenty of improvement in the medium (5PPM to 20PPM) integrated references reaching into ever cheaper complex ICs. That's where the volume is. |
| David Hess:
--- Quote from: floobydust on September 29, 2020, 10:44:51 pm ---Are we saying there's been no improvements in analog IC technology? We should settle for 5V parts or just eat the 10x price hit to beat the 741's "features"? :rant: I saw National Semiconductor's standardized processes, high voltage (even to 170V) bipolar with better feature size, speed. I think you can do more with less silicon when you take these old parts and move them to modern, standard processes. We could compare 741 die shots to see how much it has shrunk. --- End quote --- The 741 and 324 style operational amplifiers rely on the very high base-emitter breakdown voltage of their PNP transistors to support a high differential input voltage range. New "optimized" processes do not allow that. I am not even sure how they make high differential input range parts on a modern high voltage process; I did a search a couple months ago and they were all limited to 7 volts at most. Low voltage parts of course have no issue with this because they should never see a high differential input voltage. JFET input parts could do it, and this is one of their advantages, but how many of those are made on a modern process? None? |
| David Hess:
--- Quote from: exe on September 30, 2020, 12:30:11 pm --- --- Quote from: David Hess on September 29, 2020, 09:04:26 pm --- --- Quote from: exe on September 29, 2020, 11:17:55 am --- --- Quote from: David Hess on September 28, 2020, 11:13:00 pm --- --- Quote from: Zero999 on September 28, 2020, 07:53:43 am ---Not all '741s are rated to 44V. The LM741C is only rated to 36V maxium. --- End quote --- Nothing stops you from testing and grading your own LM741Cs for a 40 or 44 volt operation. Tektronix used to do exactly this for some reason. --- End quote --- Doesn't this compromise reliability? How to know if they can sustain this for prolonged period? --- End quote --- --- End quote --- Shouldn't parts be designed to withstand a certain voltage? I can try, say, supply a 5V opamp with 10V and it might even work for a while, but I bet it won't last long due to overstress (of course, 100% increase in voltage is on the extreme side.). --- End quote --- For 741s and similar ICs which were originally built on the old NPN only process, it is more of a yield problem. The criteria for failure at a higher voltage was excessive leakage which depended on process variation. As far as I know, there was no separate process for 44 volt parts as this would have been completely uneconomical. The same is true of bipolar transistors; the maximum Vceo varies considerably from part to part and the same part will be graded into multiple part numbers reflecting its tested Vceo. A part designed for 5 volts maximum, like TTL, usually takes advantage of the maximum Vbe of roughly 5 volts which varies much less than the Vceo. 5 volt parts do not have to be designed to consider base-emitter breakdown. These same parts however can usually support much higher output voltages on their open collector outputs. |
| coppice:
--- Quote from: David Hess on October 01, 2020, 05:57:51 pm ---For 741s and similar ICs which were originally built on the old NPN only process, it is more of a yield problem. --- End quote --- When the 741 was designed a 5% yield was considered a good day. Now its close to 100% for simple devices like that. Times have definitely changed. |
| SilverSolder:
--- Quote from: David Hess on October 01, 2020, 05:57:51 pm --- --- Quote from: exe on September 30, 2020, 12:30:11 pm --- --- Quote from: David Hess on September 29, 2020, 09:04:26 pm --- --- Quote from: exe on September 29, 2020, 11:17:55 am --- --- Quote from: David Hess on September 28, 2020, 11:13:00 pm --- --- Quote from: Zero999 on September 28, 2020, 07:53:43 am ---Not all '741s are rated to 44V. The LM741C is only rated to 36V maxium. --- End quote --- Nothing stops you from testing and grading your own LM741Cs for a 40 or 44 volt operation. Tektronix used to do exactly this for some reason. --- End quote --- Doesn't this compromise reliability? How to know if they can sustain this for prolonged period? --- End quote --- --- End quote --- Shouldn't parts be designed to withstand a certain voltage? I can try, say, supply a 5V opamp with 10V and it might even work for a while, but I bet it won't last long due to overstress (of course, 100% increase in voltage is on the extreme side.). --- End quote --- For 741s and similar ICs which were originally built on the old NPN only process, it is more of a yield problem. The criteria for failure at a higher voltage was excessive leakage which depended on process variation. As far as I know, there was no separate process for 44 volt parts as this would have been completely uneconomical. The same is true of bipolar transistors; the maximum Vceo varies considerably from part to part and the same part will be graded into multiple part numbers reflecting its tested Vceo. A part designed for 5 volts maximum, like TTL, usually takes advantage of the maximum Vbe of roughly 5 volts which varies much less than the Vceo. 5 volt parts do not have to be designed to consider base-emitter breakdown. These same parts however can usually support much higher output voltages on their open collector outputs. --- End quote --- I have a batch of transistors whose actual breakdown voltage is astronomical compared to the specification. Does/did it ever happen in production, that the process improved to the point where even the lowest specced parts easily achieved the performance of the (older) highest rated parts - so the lower rated parts effectively became price/performance "bargains" (but not advertised, so nobody would ever know)? |
| Navigation |
| Message Index |
| Next page |
| Previous page |