Hi Team,
I'm wondering what f=50kHz means when defining GBW. For other products like LMP7715, it specifies 17MHz, which means the multiplication of gain in V/V and bandwidth in Hz results in 17M. For LM7321, it specifies 20MHz or 16MHz depending on supply voltage and it also specifies f = 50kHz condition. Can you help elaborate more on this condition?
Thanks,
Ella
Hi Ella,
Your understanding of GBW is correct in that it is calculated by taking multiplying the gain and bandwidth. The difference in the GBW at different supply voltages can be best explained by taking a look at the typical graphs. At different supply voltages, the open-loop phase gain curves are slightly different which can attribute to the different typical specs (16MHz vs 20MHz). The f = 50kHz defines the condition at which the gain bandwidth product is calculated.
We also have more training information on GBW via Precision Labs.
Please let me know if you have further questions.
Thank you!
Best Regards,
Ashley
Hello Ella,
Yes, there were some odd practices back then inside National Semiconductor. Essentially, they are trying to report the intersection of the Aol 1pole rolloff to 0dB. Often that is quite different than the actual 0dB frequency due to higher poles. This one is showing a couple of things here, The easy way to get a true GBP from a plot like this is to go across the 40dB Aol line and take those points times 100X to get the one pole GBP. If you do that here, you can see quite a lot of variation across the stated condtions.
Here is a 2kohm load sim test. Actually showing about 11.5Mhz in the sim model and the phase is way off - that will give erroneous sim results at lower gains,
I am always perplexed by plots like this that state a Cload. Why is that there, it will bring in even more open loop Zout dependence, but let's try it in sim,
Well the GBP is the same, but the load C gives the expected higher F phase shif now, that is just the open loop Zout rolling off against that 20pF. I wonder if the Cload is to make up for not having higher poles in the intrinsic Aol model, kind of an odd way to fix a model.
Here is a relatively recent article on some of the common errors in GBP,
https://www.planetanalog.com/why-is-amplifier-gbp-so-confusing-insight-12/
And here is this last sim file,
