Truth on HC908QC16 and Freescale policy

[Encoder]

Starting from a recent enquiry about HC908QC16 which opens again a window on this "ghost", I would know someting more on this matter and on the Freescale policy for the 8-bit future developments.

May someone explain me the strange story of the HC908QC16 which was announced 6 months ago, for which it is easily available an Evaluation Board, but was never added to the official Freescale documentation nor is still available to the public?

It was also deleted from the latest 8-bit Product Summary of 2006-3Q.

While it really seems to me it would cover a void niche beetween the smart 16-pin  908QB8 and the bigger 908AP, perhaps Freescale eventually wisely decided not to put another HC908 on the market pushing accelerator instead on the much more modern and powerful 9S08 architecture. The recently lowered price of 9S08AW16 (1.95$/1kpcs for the 44 pin) may be a proof of this.

I was deeply impressed by the 908QC16 when FreescaIe firstly announced it in the 2006-1Q 8-bit Product Summary and I was really upset by the unavailability of the part at the time. Now I really hope in a future lower pin count of the 9S08AW 5-V tolerant family to cover that void niche instead of increasing the "old" and already too broad 908 family, facilitating the migration to the newer and smarter 9S08 & BDM architecture. In the meantime I prefer by far to pay a little more for the far more powerful, bigger, smarter, faster 9S08AW16 than a slightly better JL16 (1.61$ for the cheaper 28 pin version) at the same price. Who wants to come back to the involved programming interface of the older HC908 series after having tested the smarter and simpler BDM interface?

What the truth? What do the other Freescale 8-bit aficionados think of this matter?

[Encoder]

Not bad at all from an enquiry on the freescale policy.

I didn't realized there were no emulators for 9S08. This do not bother me as much as Rocco. I worked with MMEVS with 705B16 without never investigate on the addresses, so perhaps it is not so important for me. I expect that BDM is far less powerful than MMEVS but adequate (at least I still hope it!) for the moderate complexity of the projects I manage.

It is a pity and a definite drawback if such higher level tools will continue to be not available for higher complexity systems. For the moment anyway I think I will not come back to HC08 other than for very simple tasks and very low cost projects.

I still wait for an equivalent of HC908B8 o HC908QC16 with the 9S08 5V tolerant technology: 9S08QG are greats but only 3.3V....

[bigmac]

Hello all,

For low pin count applications such as for the 9S08QG8, the use of BDM for debugging means one or two fewer pins are available for the application (I have never been completely sure about the use of the "reset" pin).  This can be a serious restriction for many applications, and would be even more severe for the RS08 devices.  Using the MMEVS to debug a HC08 device provides availability of all I/O pins.

Another consideration for low and medium volume applications (say fewer than 300-400 units per run) is the non-availability of DIP packages for the 9S08 (except for the QG8), and certainly nothing capable of 5 volt operation.

A further reason to stick with the HC08 at the present time may be that it is more readily available than the 9S08 devices  (certainly in the antipodes).

Regards,
Mac

[tonyp]

My feeling is that with the current (and increasing) portfolio of HCS08 members, there is no reason to use a HC08 in any new designs.  The cost of HCS08 is practically the same or lower for equivalent pin/memory parts, while the advantages over the HC08 are manifold.

Besides the higher speeds, BDM, richer peripheral mix, etc. even the fact that the HCS08 has a slightly enhanced CPU instruction set (mostly the LDHX/STHX/CPHX directly on stack) is enough to justify the switch as it makes coding significantly more compact and readable (assembly coder here). C coders (except for compiler writers) probably won't appreciate this small but significant difference.  Now, if they only added a second index register...

We have been replacing all HC08 parts with HCS08 equivalents with every new PCB revision, in the recent months.

[rocco]

---

tonyp wrote:

My feeling is that with the current (and increasing) portfolio of HCS08 members, there is no reason to use a HC08 in any new designs. . .

---

I would normally agree, except that the HCS family has one fatal flaw for our products: The BDM architecture does not allow for a conventional In-Circuit-Emulator.

We currently design on the HC08 family using MMDS and MMEVS development systems. These emulators give us access to the CPU's address bus, and we have the address bus permanently wired to our analyzers. Without the address bus, we cannot debug our system.

As a result, we have come to the conclusion that we cannot use HCS08 processors in any of our products.

Message Edited by rocco on 2006-08-03 01:43 PM

[tonyp]

---

rocco wrote:

Without the address bus, we cannot debug our system.

---

The address/data bus being internal and out of your immediate reach, I take this to mean you cannot debug the software part of your system.

Well, I can't argue against how one prefers to do their work except by saying there are more than one way to debug software.  I've never used address/data bus probing for fixing software related issues, so I don't really miss this option.  (Hardware is another matter but with single-chip mode HC[S]08's, is there a possibility to bother fixing unlikely internal bus "attitude problems"?)

[Alban]

Yes, there are fundamental differences between the principle of an emulator and a debug module.

One of the first point is, if you use DBG on S08 for debugging, it won't be available to your application (memory window protection, code execution keyed...).

Also, I not only expect but REQUIRE an emulator to behave EXCATLY the same way as the silicon stand alone, whereas the debugger is not supposed to (there is a certain degree of intrusion).

On MMDS/MMEVS/FSICE, the only module not being used from the MCU is the Flash which is emulated by on-board RAM. Other than this, all modules used are the one from the MCU.

Just to compare to S12: both emulators and BDM tools are available. Of course emulators are much more expensive.

Alban.

[rocco]

Hi, Alban:

Maybe I have bad information, but this is what I have been told by Freescale sales:

No HCS08 processors are supported on the MMDS/MMEVS/FSICE platform, nor will any be.

There are no emulators made by anybody for the S08 family. The family simply can't support it. I was told this after I failed to find emulation modules for the GTxx series.

I would be pleased if you can tell me that I am wrong.

[Alban]

Rocco,

It's not bad information but this thread is titled HC908QC16 you are also mentioning you bought an MMDS or MMEVS for HC08. I simply understood you were saying Freescale was not supporting HC08  dev tools anymore, which I disagree with.

There is therefore no contradictory information with what your salesman said.

Cheers,
Alban.

[rocco]

---

Alban wrote:
There is therefore no contradictory information with what your salesman said.

---

Actually, the salesman's point was that although the MMDS was expensive (in relation to an EVM), it should be considered a long term investment, since it is capable of handling future processors with the simple addition of an emulation module.

A few years later, no new Freescale processors have emulation modules.

[rocco]

Looking back at my first post in this thread, I found that not only did I not state my concern very well, but I also degenerated into a rant. I'll try again.

Here is what I see as an important issue when selecting an 8-bit processor:
You cannot get an In-Circuit-Emulator for any member of the S08 family. From anybody. All that is available for debugging is the BDM.

I was told that it is a byproduct of the S08 core architecture. A step backward, in my opinion (oops, watch those rants!).

[rocco]

Hi, Tony:

Yes, you are correct with your distinction between hardware and software. The software issues could certainly be addressed with the BDM.

But our system contains a large amount of synchronous logic inside of an FPGA. We don't see how we can visualize the relative timing between the various FPGA engines and the HC08 firmware without the address bus. We bought the MMDS and MMEVS systems specifically to address this problem. The Motorola salesman convinced us to spend big bucks on these, and now Freescale won't even support them.

I find this to be a problem within Freescale lately: They treat their chips as if they are the system, rather than a small component within the system.

Message Edited by rocco on 2006-08-03 04:02 PM

[Alban]

I have to show an objection on the fact that Motorola sold you an MMDS and Freescale is not supporting it !

The FSICEBASE was created for this effect and is a Freescale part. MMDS/MMEVS are FORBIDDEN of sale because of RoHS and lead free considerations.

Developping a brand new emulator platform under Linux for a family a product is not really what I call not supporting.

S%it loads of boards are being re-manufactured (HC08GZ among others...) to especially accomodate this new legal requirement.

Alban.