Kinetis Microcontrollers中的主题 Re: MKV58 vs. i.MX RT

MKV58 vs. i.MX RT

andreacanepa — Mon, 23 Apr 2018 16:22:47 GMT

The question is simple: in 2018 is it worthwhile developing a new project using MKV58 (kinetis)?

In 2018 the i.MX RT 1050 was officially released, and in a few months i.MX RT1020 will be available, so what is the point now to consider using MKV58?

The i.MX RT use Cortex M7 (like MKV58), they have a price that can be 1/2 or 1/3 compared to MKV58, have lead times of 14 weeks (compared to MKV58 which today has a 39 weeks !!!), they are fresh products.

If we compare i.MX RT1020 with MKV58 they have the same package (144 LQFP), they have about the same peripheral equipment, moreover i.MX RT also has the integrated USB and the interface for SD cards (uSD, SDHC, etc. ) that many of us wanted in MKV58.

i.MX RT1020 has the clock frequency up to 500 MHz, while MKV58 stops at 240MHz.

i.MX RT1020 has the RTC, while MKV58 does not have it.

Perhaps the most negative aspect in i.MX RT1020 is that it no longer contains the intergrated flash, but it could also be considered positive if we think about the freedom to mount the model and the amount of flash needed outside (but we complicate a little life to make the reading of external data safe).

As for the analog part, i.MX RT1020 seems worse than MKV58: i.MX RT1020 has lost the 16 bit ADC vs. MKV58, (but the conversion quality of that 16 bit ADC has never been exceptional, it is very susceptible to internal noise and it is mandatory to run several averages of the signal to get a good result, so it's not a great loss). But the downside is that i.MX RT lost the fast (5MSPS) HSADC 12 bit with the differential inputs that the MKV58 had inherited from the DSC family (56F83xxx and 56F84xxx), replaced by a normal 12 bit (1MSPS) SAR.

i.MX RT also lost the DAC!

My impression is that i.MX RT1020 has lost many "analog" qualities compared to MKV58 but that has gained a lot in digital peripherals: the FlexIO for example is very interesting.

What is my concern? What little by little NXP is trying to "pull" exFreescale (and exMotorola) customers to their processors, implementing the pricing and delivery time policy: how is it possible that two similar microcontrollers can have such a big difference in price and lead time?

Probably if the application does not require the analogue part missing in i.MX RT, the right choice would be to use it in a new project.

I would like everyone to write his opinion about it, even analyzing the comparison between the two microcontrollers, to hear the opinion of NXP regarding my statements, to understand how the situation is now. Why does NXP have many similar components in its portfolio? What is the convenience of having all this?

Andrea.

Re: MKV58 vs. i.MX RT

arimendes — Mon, 23 Apr 2018 20:38:33 GMT

Hi, Andrea,

I remember that the MKV58F1M0VLL24 cost less than $ 8.00 for 100 units in the Freescale direct store in 2016.

I'm sure if NXP had not bought Freescale we would have a plenty of KV5x in stock at Freescale original price for KV5x. And we would probably already have a new generation of Kinetis V-Series with USB, QSPI, SD-CARD, etc.

I think that the I.MX RT1020 can't replace the KV5x. The I.MX RT1050 in an application with TFT display that has to use external memory anyway, makes sense. But I.MX RT1020 + Hyperflash will have a similar price of the STM32H7. And the STM32H7 is far better than the I.MX RT1020. (3x 4.5 MSPS ADC, 2x PGA, 2x COMP, TFT, 1 M SRAM,etc).

Ari.

Re: MKV58 vs. i.MX RT

andreacanepa — Tue, 24 Apr 2018 08:02:41 GMT

Thanks Ari for sharing your thoughts.

As for the price of MKV58F1M0VLL24 I can tell you that today it is sold directly from NXP at US$ 10.56 (100+ pieces) (but we are talking about the 100 pin.) Instead, the 144 pin version is sold for US$ 11.11. For a comparison, the price of i.MX RT1050 (as RT1020 is not yet available) is available for direct sale at US$ 4.49 (price valid for only 1 piece), so less than half of MKV58. The price of RT1020 be less than RT1050 since it has lower characteristics and a lower available peripheral.

I'm sure if NXP had not bought Freescale we would have a plenty of KV5x in stock at Freescale original price for KV5x. And we would probably already have a new generation of Kinetis V-Series with USB, QSPI, SD-CARD, etc.

I think so too: so I do not have much confidence in the long life of MKV58. Even if someone will write that it is part of the "Product Longevity" program, that the production of MKV58 is guaranteed for 15 years starting from June 2016, but what we observe is that even if it will be produced but if the price continues to grow, if the lead time continues to increase, it is not an acceptable product.

The I.MX RT1050 in an application with TFT display

Yes, I agree with you, but I'm pointing to i.MX RT1020 that does not have TFT control or other unnecessary devices for motor control (Motion control, inverters, dc drives, etc.).

But I.MX RT1020 + Hyperflash will have a similar price of the STM32H7.

The target price of RT1020 is $ 2.18 (for 10K pieces) but compared to the RT1050 target price of $ 3.28 (for 10K pieces) means that it will cost about -33%, so we can assume that for 100 pieces RT1020 could cost around $ 3.00 (100 pieces). I do not know how much it costs a HyperFlash, but before reaching the STM32H753ZIT6 price of $ 13.24 (for 100 pieces) the road is long!

However, here the comparison is between i.MX RT1020 and MKV58.

Andrea

Re: MKV58 vs. i.MX RT

arimendes — Tue, 24 Apr 2018 16:45:02 GMT

Andrea,

The HyperFlash starts with U$ 5,00. So RT1020 + HyperFlash = U$ 7,18 (Production price). The future STM32H743VG (1M Flash/ 1M SRAM) target price is U$ 8,70. So the road is not so long and the destination is much better.

NXP Killed the KV5x and all Kinetis series, it is like a dead alive microcontroller. So I'd rather compare it to a living equivalent microcontroller from another manufacturer.

What kill the RT1020 for Motor control and Digital power applications is the ADC, only 2 ADCs with only 1 MSPS is ridiculous for a 500MHZ cortex-M7.

Ari.

Re: MKV58 vs. i.MX RT

andreacanepa — Thu, 26 Apr 2018 06:39:45 GMT

The future STM32H743VG (1M Flash/ 1M SRAM)

Yes, I have compared STM32H753ZI and actually has some interesting features, but availability is poor: I found it only on Farnell and the price is € 14.26 (for 100 pieces). If you know STM32H7 you can tell me if:

Is there an IDE dedicated to it? I had looked on the STM website but found a very long list of third-party IDEs, but none made and maintained by STM.
Is there a hardware configuration help system, such as the NXP MCUXpresso Config Tools (or Processor Expert)?
Do they have a kind of "product longevity life" that specifies the life span of their MCU?

NXP Killed the KV5x and all Kinetis series, it is like a dead alive microcontroller.

I agree with you, and in fact that was my feeling. I am also very amazed that no NXP man has yet intervened to deny (with valid data) our statements.

What kill the RT1020 for Motor control

In fact it is precisely what I had written in my first post: the analog part is very poor compared to the MKV58 features. Besides the already mentioned poor 12 bit ADC, we can remember the lack of: at least one DAC, differential inputs for ADC, PGA (Programmable Gain Amplifier).

So to conclude: if MKV58 can be considered dead (thanks to NXP), and i.MX RT1020 is not a worthy substitute for it, we just have to migrate to another manufacturer.

Andrea.

Re: MKV58 vs. i.MX RT

andreacanepa — Thu, 26 Apr 2018 07:21:30 GMT

Well, looking better I gave myself the answers alone:

Is there an IDE dedicated to it?

No, no IDE is made by STM. There are free IDEs, but I do not know them. If you have any suggestions, let me know.

Is there a hardware configuration help system

Yes: STM32CubeMx.

Do they have a kind of "product longevity life"

Yes: link

For STM32H7 is 10 years starting from 01/2018. Very good!

ADDED: during my research I discovered that there is also a sort of "Freemaster" STM version (STM Studio Monitor), but I have the impression that it is very poor compared to FreeMaster.

Andrea.

Re: MKV58 vs. i.MX RT

arimendes — Thu, 26 Apr 2018 16:52:52 GMT

Andrea,

For the STM32H7 you can use the Atollic IDE, a professional IDE that was recently purchased by ST and made available for free.

I'll make it clear that I think that the Kinetis KV5x is much better for Motor Control, digital power management and general industrial applications than any other ARM, including the new STM32H7 from ST.

If the NXP staff assures us that within a year we will easily find KV5x at distributors, including the low-end ones with 512K flash. I would recommend to start new designs using this chip family.

I found a NXP roadmap in the net including a new KV6x dual core (google Kinetis KV6x) . If they do it would be great because it will show us that NXP will continue the tradition of Freescale in the development of high-end microcontrollers for Motor control and Power Conversion.

Ari.

Re: MKV58 vs. i.MX RT

andreacanepa — Fri, 27 Apr 2018 06:38:15 GMT

Making comparisons between MKV58 and STM32H7 I realize that this last is not inferior to MKV, indeed in some respects it is much higher.

I want to remember that one of the major shortcomings on MKV58 is the lack of USB and SD interface, while STM32H7 has even 2 USB and SD, in addition to a large amount of timers, the core with 400 MHz, 2 MB flash, 2 DAC, 3 ADC 16 bit (or 14/12/8 bit), 2 comparators, 2 opamps, I2C up to 1 Mbps, and above all some digital inputs are tolerant 5V !!

The configuration software is very nice and intuitive, very similar to MCUXpresso Config Tools (what a coincidence ...).

If the NXP staff assures us that within a year we will easily find KV5x at distributors, including the low-end ones with 512K flash. I would recommend to start new designs using this chip family.

I suggest you read this topic: LINK

I have little confidence that the situation will change: in fact, from December 2017 (when that topic was born), now nothing has changed. I am very sorry for this situation, because I have always worked with Freescale products with good results.

Andrea.