PIC 18Fxxxx A/D TAD and Fosc

10MHz or

In setting the A/D clock period you want to make it as fast as possible so that you get your A/D reading a short time after you trigger a reading. But you should not set the time any shorter than 1.6 usec. or else the reading may not be accurate. So just look over all your choices and pick one that results in a period slightly greater than 1.6 usec. Your choices are to derive the A/D clock from the processor clock (10MHz or 40 MHz), or to use an independent RC clock.

I have never seen a PIC that didn't have some register that set the A/D clock period.

...sure. Why not?

What specifically don't you understand? The A/D process takes 9.5 TAD times to complete. If TAD is too short, then the switched capacitors in the A/D converter do not have enough time to fully charge up, so the apparent measured voltage may be inaccurate. The A/D clock is derived from the CPU clock by a selectable prescaler, or by an RC oscillator.

Same thing. The UART clock is derived from the CPU clock , except that the prescaler has much more programmability (not just powers of 2).

For more info check out:

Robert Scott Real-Time Specialties

Robert Scott Ypsilanti, Michigan

For starters what exactly "Fosc" is (I stumbled on how to derive the proper setting for the ADCON register. I however have no idea what it means. so it takes 1.6uS to do the conversion and the math works out, but I am still in the dark. What is exactly is Fosc/64?

The uC in question has a setting for Fosc and a setting for TAD. Not to be dense here, but the datahseet lists TAD settings from 2 to 20. The datasheet lists 11 TAD per 10 bit conversion. What exactly does changing the TAD setting mean? What is the correct setting? It would appear that if 11 is the minimum, then setting TAD to 12 (the nearest value) would be the correct choice. But that would beg the question as to why there are available settings of 2, 4, 6, 8, 12, 16, 20.

Not to be rude or redundant, but I needed a laymens explanation, not datasheet speak. I have read a dozen datahseets and 2 dozens websites and forums including the forum you listed. Surely I can post the same question there and only hope I don't get the standard RTFM! reply. I grasp things easily, but am TOTALY new to uC architecture and concepts. Programming is one thing, uC timing and the specifics are not something I am comfortable with.

Thanks for the reply... sorry I am not able to put it all together.

It is the CPU oscillator frequency (10 MHz or 40 MHz). I'm not sure, but I think if you use the PLL, Fosc refers to the multiplied-up frequency (40MHz).

No, it takes 11.5 TAD periods to do a conversion (not 9.5 as I said earlier because I was thinking of a 8-bit conversion and I see you are talking about a

10-bit conversion). The TAD period must be 1.6 usec or more, but otherwise it is whatever you decide to make it by your choice of ADCON bits.

Is the the Focs frequency divided by 64.

It means changing the tap on the prescaler that derives the A/D clock from the CPU clock.

Any setting that yields a TAD period of at least 1.6 usec.

The minimum setting is specified in terms of TIME not in terms of the divide ratio. But you must specify it in terms of the divide ratio. The chip does not know what frequency XTAL you are using, so they offer different divide ratioes and leave it up to you to pick an appropriate divide ratio to make the A/D clock period come out correctly ( > 1.6 usec.) in terms of time.

I have found the Microchip datasheets to be among the clearest in the industry for giving things in as close to laymans terms as is possible without sacrificing accuracy. Just keep reading them. Nothing else is needed other than a general understanding of principles of digital electronics.

Robert Scott Real-Time Specialties

Robert Scott Ypsilanti, Michigan

I goofed. Fosc is the oscillator frequency divided by 4.

Robert Scott Ypsilanti, Michigan

You seem to not understand what I am asking. I have no idea what TAD is. I have no idea what Fosc is. Therefore I have no way to wrap my head around what they do and how they are calculated.

Again this means nothing to me. So we change the "tap" and come up with a different time. That tells me nothing about how to calcualate either number.

Scott, divide ratio of what? What is being divided? Examples and numbers would help.

different >divide ratioes and leave it up to you to pick an appropriate divide ratio to make the A/D clock:

I FULLY understand the concept. I think your missing that point. I DONT understand what numbers to use, what they mean and where they come from. Again examples would help.

I think that is the problem here. I READ THEM and have questions and you are answering the questions with almost direct quotes from the Datasheet. No matter how well YOU think the datasheet is written, I am struggling to put a few of the concepts into perpective. YOU understand your anwers but I do not. If I did understand, then the datasheet would have been all I needed in the first place.

I understand your trying to help and I appriciate that. However, your answers are as cryptic to me as the information on the datasheet. I am no closer to being able to calculate the Fosc and TAD bits for a given uC than I was when I posted the question.

Anybody else out there in PIC land willing to give a newbie some help?

TAD is the time it takes for one complete cycle of the clock that runs the A/D process. The A/D process takes 11.5 of these TAD clocks to complete. So if TAD is 2 usec, then the A/D process will complete in 23 usec. after you trigger a conversion.

Fosc is the frequency of the instruction cycle clock, which is the CPU clock divided by 4. The Fosc clock is used to step through the instructions. It is used in many places in the chip besides the A/D section, such as the input to a timer and the UART.

I just said what they do. As for calculating them, the TAD time is calculated by taking the Fosc period (which is the reciprocal of the Fosc frequency) and dividing by whatever ratio you select in your choice of ADCON configuration bits that to set.

A "tap" refers to a point in a series of divide-by-2 circuits. By picking a different tap you select a frequency that is equal to the input frequency divided by 2, 4, 8, 16, etc. By the way, I don't know if you have mentioned it before, but which particular PIC are you looking at? It would help me to relate to what you are seeing.

The divide ratio is what you specify when program the ADCON registers. The datasheet says which divide ratio goes with which bit settings.

The Fosc clock frequency.

OK. If the external crystal oscillator is 10MHz (no PLL), then the Fosc frequency is 2.5 MHz. If you select a divide-by-16 in the ADCON configuration, then the TAD frequency is 2.5 MHz / 16 = 156250 Hz. Therefore the TAD period is

6.4 usec., which is a valid TAD period since is is greater than 1.6 usec.

different >divide ratioes and leave it up to you to pick an appropriate divide ratio to make the A/D clock:

The choice of what numbers to use is up to you. You have to decide what your design requirements are. For example, most people want their A/D operations to complete as fast as possible, so they program the ADCON registers to generate a TAD time as short as possible, as long as it is not less than 1.6 usec. However the choice of crystal also impacts current drain. If your device is battery powered, then you want to use as low a crystal frequency as you can, consistent with getting the job of your application done. Also, you might want to consider how the choice of CPU frequency might impact the available UART frequencies.

I think I understand where you are coming from. I just went through a difficult learning period on trying to crack the "secret" of programming Nokia Symbian cell phones. I read all the Nokia documentation and asked lots of questions in the Nokia forum. For the longest time it seemed like they were all in a secret society and I was confused because I had never been taught the secret handshake. But I finally found some documentation that got me over that initial hurdle.

Robert Scott Ypsilanti, Michigan

No, it is the oscillator frequency. If you're using the PLL, it's the oscillator frequency times four. The instruction clock is Fosc / 4. Other terms you'll find in the datasheet are Tosc, which is 1 / Fosc, and Tcy (the instruction cycle time), which is Tosc times four.

--
John W. Temples, III

Thanks. I stand corrected.

Robert Scott Ypsilanti, Michigan