Sinewave inverter prototype...

[DeepOne]

amended circuit

 

[RETGT]

coool project it is used everywhere in mobile tower application...

 

[DeepOne]

AtMega8 version.

 

[thannara123]

May i Know the specification ?

 

[DeepOne]

this is experimental design, so specification is not definitively defined. i think with this type of upper side drivers output power be not more than 300W with input voltage under 15VDC. some parameters like output voltage and frequency is defined in "defines.c" section

 

[thwcs]

Dear DeepOne!

I'm in the process of circuit layout post # 458 and # 463 for assembly. But I have a few more questions to ask you as follows:
# 458:
We can use many types of quartz (12-16MHz) without having to change the firmware?
# 463:
We can use different types of MCU Atmega8a-pu; Atmega8L-8PU; Atmega8L-pu ... or anything else to note?
# 458 & # 463:
Differences in the 1k and 10k values in base BC817 in the two diagram. that's right or wrong?
3 digital 7 segment LED in the circuit using type common anode or common cathode? example image attached to Choose?


Thank you!

 

[DeepOne]

We can use many types of quartz (12-16MHz) without having to change the firmware?

yes, but it is necessary to prescribe quartz frequency in eeprom as shown in circuit picture. or possible not to install quartz - in this case program is switch itself to use built-in RC oscillator.

We can use different types of MCU Atmega8a-pu; Atmega8L-8PU; Atmega8L-pu

better if mcu officially capable to work at used frequency. but in practice "8L" can work too at 16MHz under 5V.
this scheme also can work with different quartz frequency, after recompilation with new settings in "defines.c"

7 segment LED in the circuit using type common anode or common cathode?

any type - mcu define this by itself.

Differences in the 1k and 10k values in base BC817 in the two diagram. that's right or wrong?

that right but not essential.
_________________________________________

next prototype, not ready.

 

[thwcs]

I've assembled a diagrammatic # 463 but it did not work. Electric current measured before pressing RUN is 0.16A DC. After I review the attached simulation files, I have found the problem:
When I activate the overload protection function (press the button to ain1 = 0 vdc) but I understand that it must be greater than 0 vdc, the overcurrent protection circuit will work. Maybe it has operated contrary to the principle? Then I spent a 1k resistor in ain1 (pin 13 MCU), the circuit was running but was blinking, looking like a police car lights, high electric current when the load is only 25W lamps (9A DC). So far I have not found a cause.
Further, I found a relatively low frequency when running simulations (43 Hz).
Please help me complete the project. I can wait until the end of New Year holidays.
Thank you and Happy New Year!

 

[DeepOne]

Hi, Thwcs.
possible quartz frequency does not correspond to necessary for microprogram frequency. as i remember in .zip archive there was three .hex on differ frequency of the quartz (16Mhz,12Mhz and 13,824Мhz). Please disconnect transformer and check output 50Hz frequency with oscilloscope.
Overcurrent protection operate (shut off bridge) whеn voltage on AIN1 is less than 1,28V.

 

[thwcs]

Thank you for your reply soon!
I chose the correct frequency is 16MHz quartz, on my proteus simulation file also choose 16Mhz.hex.
pin Ain1 connection 1k resistor then to the terminal D of mosfet against polar opposite. When mosfet open, the Vds this mosfet will be very small, I thought it would be MCU smaller than 1.28vdc activate overcurrent protection.

 

[DeepOne]

I chose the correct frequency

anyway it is necessary to check output frequency and signal on mosfet gates.

smaller than 1.28vdc activate overcurrent protection.

when Vds near null V_AIN1 is near 1,6V so for activate protection Vds must be lower than -0,4V (1.6-1.2). that corresponds to 60...80А threshold current.

 

[thwcs]

You mean 0 < Vds < 1,2v ?

I think voltage Vds> 0, the trigger circuit protection. Vds = 0, the normal circuit operation, that is right.

 

[DeepOne]

Vds = -0,4V...0 for normal circuit operation, if Vds<-0,4V<0V that is trigger for circuit protection.

 

[thwcs]

Why when I bypass circuit D to S is not working, then remove the 1k resistor circuit to operate ? (1k connected to the voltage mosfet get Vds to Ain1)

 

[DeepOne]

then remove the 1k resistor

so 5V on AIN1, right? try to adjust 2k2(from PB0 to AIN1) in range 2k2...510. but if overcurrent protection is operate may be there is some fault in circuit.

 

[thwcs]

The good news: The circuit has worked well, I have found the cause USBASP connect to MCU wrong as shown in Fig.
By the way, you can tell me what is the name of the software that you use to draw the diagram. I look at it nice and clear and want to use it.
Thank you & good wishes.

 

[DeepOne]

software that you use to draw the diagram

That is "SPlan 7".
Good wishes and Happy New Year, Thwcs.
_______________
i think also it is better to enlarge mosfet drain - AIN1 resistor to 10k (and PB0 - AIN1 to 22k, accordingly).

 

[D.A.(Tony)Stewart]

Analyzing the Load regulation of this inverter requires knowing the DCR of the primary coil and RdsOn of the full bridge.

Drain-Source On-State Resistance RdsOn = 17mΩ max @ Vgs = 10 V, Id = 20 A, Tj = 125 °C
then x2 for each device in bridge and 1/2 for ground switches and neglecting unknown ESR of battery.
then scaled by core turns ratio 1:28

thus contribution to Zout from (4 active FETS ) RdsOn = 28²*17m*2.5= 33.3Ω where Zout= Rsource to load,ZL

Since Load regulation is ZL(min)/(ZL+Zout)

Thus for 10% load regulation at Pmax, ZL=9*Zout = 300Ω = ZL

Thus if output was intended to be 230V (?) Pmax= V²/ZL = 176W

But if we know Vbat is 14.2V during charging and N=28 then Vout(pk)= 539Vp (?) or 381Vrms if PWM with sine there is plenty of headroom to regulate 230Vrms.

And with 12.3V Vbat at 50%SoC Vout(pk)=344 or 243Vrms
and at 11.8V Vout = 233Vrms -10% for load regulation error.

Thus I see there is enough headroom to regulate 230Vrms with PWM allowing more load regulation error at expense of heatloss.

Reducing DCR , RdsOn is the only way to improve this such that the batter ESR is the limiting case. <5m with a 1000A CA or 1.2*CCA battery with 850CCA at full charge. Then putting batteries in parallel and reducing ESL of cables becomes important using Litz wire.

 

[thwcs]

next prototype. 20W ACIM load test ok.

Dear DeepOne!
I am testing in item # 458 but had little trouble as follows:
- This is my memory in the data set before programming.

- When programming the error, as shown below.

- I read back then noticed a little difference and marked as in Fig.

I do not know why this happened, I do not know what to do anymore.
I do not know how to complete the programming has not ??? to modify anything more?
Please help me!

Thank you!

 

[DeepOne]

that is because mcu is write default values in eeprom (data memory) by themself if contents of the cell at adress 0x4000 not equal to predefined value (0A as i see). so if you want set eeprom content to default drop 0A to 00 and restart mcu, in other case you can modify anything in eeprom, except 0x4000 cell.