TMCS1133-Q1: HVAC application

Hello,

I can't tell you what is most important in your end application, but I can make some comparisons between the two devices.

The TMCS1133 is a Hall-effect sensor, and in addition to the noise you point out, hall sensors are inherently noisy. The expected noise from the sensor is about 180mA RMS.

The response time is 120ns.

We also have lower bandwidth hall sensors that change this tradeoff slightly - the TMCS1123 will have 125mA RMS of noise and a 1µs response time

the AMC1303 converter is a delta-sigma modulator, but does not include a digital filter to process the data. most of the tradeoffs will depend on the digital filter settling. The digital filter can reduce the noise in the system as shown in an example plot from the datasheet:

The AMC converters will have a response time that varies with the digital filter and OSR you are using. My knowledge of digital filters is limited, but I believe most digital filters require some settling time to get an accurate output signal, so it may take several samples for a signal to propagate to the controller. This will likely depend on the filter used. the period between samples is 12.5µs with the 80ksps data rate you mentioned.

Let me know if you have any more questions,

Levi DV

Hi!

Thanks for mentioning features. My question maybe more general.

We have motor with high initial torque without position sensors (only 3 wires). Maybe it can be the same advice as for traction motor control. I believe startup moment is most critical part of design.

I see for both applications TI basically recomends AMC13xx family with 80ksps, but I was told sw needs for this startup moment especially faster feedback. I believe if we will be able to proceed with 4channel 80ksps on f28p65 with some simple digital filters it will be already big achievement. And I see this problem is solved on even slower hardware. That is why I seek some general knowledge:

Do you have any application note regarding both SW and HW control in this application? Or maybe there is evaluation I can buy and play SW/HW? 

Hello,

I don't know much about motor drive requirements, so I am going to forward this thread to another team that can answer your question.

Levi DV

Hi Stas,

Thank you for interesting in our solutions and reaching out to us on your questions.

In industrial motor drive applications, delta-sigma modulators are much more preferable for current and voltage sensing in high power motor drive systems because they have much better niose rejection performance and accuracy level. 

Usually a high power motor's windings have rather high time constant (Tau=Lw/Rw) up to a few ms or even higher. The switching frequency of such a high power motor drive inverter is often below 10kHz, and the application BW of the motor current control loop is below a half of that. So the 80 ksps in the current sensing sub-system is enough for these application systems' requirements. Setting the BW too high will only let in too much noise and interfere to the motor control.

In a real application, the latency will not be an issue with the PWM control; and the C2000 MCU integrated digital filter modules (SDFM) have a specific 'Comparator Filter Unit' in each channel for quick SCP/OCP response. Refer to F28P65 MCU TRM section 24.8 for details: TMS320F28P65x Real-Time Microcontrollers Technical Reference Manual (Rev. B) The Delts-sigma modulator and C2000 MCU combanation will make the system structure much simplified while still keep the high performance.

As for the reference desig for EV HVAC compressor drive, I only find this one: TIDA-01418 reference design | TI.com, but it can only output 12A. You may also refer to high power designs like the TIDA-010025 reference design | TI.com for modulators usage.

Best regards,

Jerome Shan

Hi!

Thanks for detailed answer.

After your reply I am thinking about use of ADS127L18 with something like AMC3302 as 3xphase current/voltage sensing.

It may give me bandwidth like 400kHz, and if I don't need it, I can reduce it with digital filter in ADC.

What do you think?

Stas

Hi Stas,

I can't get the reason of the necessity for the 400kHz BW.

If AMC3302 is prefered, the ADS127L18 is not necessary; and an OPA for differential output to single end conversion may be necessary. Refer to figure 14 of this: Three-phase inverter reference design for 200-480VAC drives (Rev. A).  Long tracks on board for the analog output signals to the MCU's ADC input pins may also become a challenge.

Best regards,

Jerome

As I understand working frequency high estimation is 20 kHz and for FOC it is good to have 10 samples for each PWM cycle. 

For high power HVAC compressor driving, the fsw commonly used is 5~6 kHz. Are you going to use 20 kHz for some specific consideration?

Can I somehow calculate these numbers? Like is there any pdf with guideline  how to estimate working frequency and required sensors bandwidth for motor regarding parameters? Also if there is something to read regarding how sensor data transforms into control signals (FOC) especially with TI MCUs?

Regarding frequency:

As I understand 5 kHz is minimum recommended frequency for current ripple tradeoff.

Higher limit is 250kHz as I have specification for motor minimum 2uS between PWM pulses.

Later comes tradeoff for switching losses, sensors bandwidth and FOC control.

As for my knowledge, with higher frequency should be better EMI (simpler/smaller filter, DC Link capacitors).

I think this is the main reason to go for higher frequency.

And here comes a question - how far can we go.

I see 20kHz as reasonable maximum tradeoff we can achieve in system - 50kHz may be too much on switching losses and required 500kHz bandwidth for sensors.

Here is from where comes 20kHz.

So 200KHz bandwidth for sensors - how to measure it best way in TI ecosystem?

That is why I mentioned ADS127L18.

With ADS127L18 I think it is possible to measure analog signal with 16 bit on 200kHz. But is it easy to integrate in F28P65 MCU?

I can anyway use integrated SAR ADC in MCU, but I think 8 bit will be maximum noise free. But maybe it is enough?

Hi!

Thanks for this reference design, it is not isolated, but looks like I thought about.