Great post! Insightful and engaging content. Keep up the excellent work. If you want to land your dream job now, visit: https://jobsearchlive.com/jobs

Nice post.

[Typing Master Online Test English](https://www.blog.typingspeedtestonline.com/typing-master-online-test-english/)

[url=https://www.blog.typingspeedtestonline.com/typing-master-online-test-english/]Typing Master Online Test English[/url]

Nice post. I found this is an informative and interesting post, so i think it is very useful and knowledgeable. I am glad to read this post hope your next article is are so useful for me so you are meet in next process for more information click this link below: -

<a href="https://www.blog.typingspeedtestonline.com/typing-master-online-test-english/">test-english</a>

Sweta............

i think that using a more robust model would be more efficient. considering the situations or places of the people who will use this, having a belt clip built in would be useful. i also think that making the box bigger would be a decision that pros outweigh its cons. by having a bigger box, it would be more heavy. but on the flip side you could incorporate more protection a possibly something to cushion the hardware inside. if it was 1/3 of an inch wider, it would still be a handheld device but would be more protected and durable.

@K.C.Lee, "invert PWM polarity": Of course, if this is possible. I would not spend an extra HW-inverter, if it replaces only something like "PWM_counter= 256-voltage".

@K.C.Lee, "PWM is a poor possibility...": I did not really find schematic in your battery charger article. But why is PWM a poor possibility? I suggest it's use (with adequate filtering) just as a DAC. And you can filter it heavily, as in most applications you do not need a fast change of the setpoint of the PSU. The PSU switching frequency is independent of the PWM frequency from the microcontroller and (as I suggested) the regulation loop is also independent from software. You can control the switching regulator in open loop, because it's regulation loop is internally in the LM2596. You do not need extra active components between the µC PWM output and the LM2596. No DAC or even ADC, no analog sawtooth, no comparator or OpAmp. A 10bit PWM counter gives you around 25mV resolution between 5V and 30V. Extra analog stuff only reduces your accuracy and increase cost.

@taciuc marius: Nice test plan - I recognize the standards for testing of automotive ECUs. :-)

Using timer based PWM directly is bad because you have to compromise switching frequency with resolution.  Low switching frequency means physically larger inductors, harder to filter because of the low frequency.

I made my analog PWM to have high switching frequency 300kHz while keeping the resolution high.  I need to have constant voltage or constant current or mixed under full micrcontroller control.  I did played with TL5001 with opamp feedback in first version.

PWM filtering https://hackaday.io/project/4993-dual-channel-battery-chargeranalyzer/log/16159-pwm-part-1
Power circuit: https://hackaday.io/project/4993/log/16890-power-components-more-nickel-dime-time
Sawtooth generator: https://hackaday.io/project/4993-dual-channel-battery-chargeranalyzer/log/18116-sawtooth-generator
Full schematic is here: https://github.com/FPGA-Computer/Battery-Analyzer/tree/master/Ver%202.1%20Rel%200/Eagle%20PCB
there are Eagle files as well as schematic in pdf.

It gets a bit tricky to use a buck converter as SEPIC. (You can do a Zeta, but still that LM part is horrible for my application.)  Also part doesn't work well if I want to run from USB or low voltage source such as discharging the battery.  In my case I optimize for cost not component counts as circuit complexity isn't a bother to me.  Can't beat a 339 for cost.

I suggested to use the PWM only as a DAC, not to drive any power switch (of an SMPS) directly (then you would have to implement the control loop in software). Taciuc suggested to use an LM2596 (a dedicated chip for step down converters), which does the 300kHz (or more) PWM.

I used a much lower PWM frequency, smoothed the PWM out with an RC filter, so no extra inductor - not large, not small - and used this voltage to inject a current to adjust the voltage setpoint.

Of course you can coarsly replicate a switch mode converter chip with OpAmps, Comparators and a roll-your-own sawtooth generator. But why? The simple switchers are really cheap (from China) and they just work. In this  switch mode regulator chips are often some more functions, to ensure loop stability and regulation over a wide range of input and output voltages and currents which are not shown in detail in the datasheet.

I still holds the belief that digital PWM is not ideal as a DAC for controlling power supply.  My integrator DAC has much more linear output, as high resolution as the ADC in the system (and can easily be 12-bit or higher), very low ripples in the sub LSB range.  You still use whatever chip you can as a regulator for fast transient while the microcontroller loop can compensate for droops etc.

Might want to read up on my project.  The charger can run from rated USB range.  LM part starts at 6V, thus unsuitable for my design nor is it in the correct configuration for SEPIC.  The SEPIC is retasked during discharge mode as electronic load.   The firmware could be configured to provide a regulated output on the external load terminal.  The discharge can operate down to 1V or so for NiMH cells in boost mode.  Can't do that with off the shelf switchers.

@K.C.Lee: You are right, in your project, the "Scorpion" solution with the LM2596 would not be useful. I think for a SEPIC converter you need a step-up chip, not a step-down. And the very wide input range may justify the homebrew controller. That's also really true: These standard chips like LM339 are really dirt cheap and circuit complexity does not equal BOM cost in this case.

But a digital PWM also can be 12bit or more. Especially in case of a battery charger/discharger you do not need a high regulation dynamics. Battery voltages don't jump. So you can use a PWM frequency of several 100Hz to several kHz. With 1kHz and 16MHz clock you get ~14bit of resolution. Use an STM32 with >65MHz of clock and you get 16 bit's. It has a reason, that modern digital audio DACs are often "1bit" (mostly sigma/delta). This reason is linearity, it's much more easy to get linearity in time - e.g. with a PWM timer and a filter than linearity over 10.000 of voltage steps.

And re ADC: Many ADCs, like the successive approximation type (*) use internally a DAC, you do it the other way round, using an ADC and an integrator, when you just need a DAC. For me that seems overly complex.

*) The EFM8UB1X CPU you used happens to have this type of ADC.

But nice if your design works and you are happy with it.

EDIT:

I just found the PDF in your GitHub page about the Battery analyzer and here you use a normal filtered PWM from the microcontroller as input to your external 250kHz SEPIC PWM generator. Similar to what I suggested. Although you have to do (can do) the control loop in software. Probably I would have used an integrated step up regulator - just to avoid to have to do this in software. But of course this depends on personal preferences. Mine happens to be in the field of hardware.

Scorpion 3.0 is a multipurpose tool for the masses, it is a cheap product but not a cheap design, of what could be a better way of living for some people. A good solution would be to design a low price adjustable and multi-purpose DC to DC converter power supply that is very simple to use https://www.reecoupons.com/categories/halloween. I hope my idea will be helpful to you. Thanks

Hi Marius,

All looks great!! Keep up the good work! I'm eager to see how it ends up, hopefully it will help a lot of people! Looking forward to seeing new posts about the progress, Cheers!

Things that I plan to add to this project page in the following period

- CAD files and rendered pictures with the design of the case  --> done

- Software update with new menus and English comments --> done (more to come)

- Actual pictures with the case when it will be ready --> done

- Another video with the appearance of the complete hardware design after adding the case  --> done

- Test plan for Electrical Engineering tests to be performed on the unit (according to ISO16750-1 and ISO16750-2 international standards for automotive and car battery connected devices) --> done

- Test report and conclusions 

Topic about the production cost

   After I posted the component list and uploaded a full excel sheet with the component prices, you are gonna ask me where do I buy my components from. The answer is Chinese market (alibaba, ebay and others). Probably you are worried about the quality of the components but I can tell you that so far it's been going good from this point of view. Keep in mind that I'm creating this design for the 3rd world countries and it is supposed to be low cost and affordable. Trolling with a chunked smile on the corner of my mouth, I can say that that's the art and beauty of creating electronics from scratch: being able to make a circuit last even if you use low quality capacitors. Just kidding guys! Don't take me too serious. :P 

   The truth is that it's a pretty robust design and I'm personally satisfied by the way it evolves. After all, I'm planning to use it myself if I get stuck in a village somewhere. I always say "be careful what electronics you create, cause one day, they might save your life. We had a few examples here on hackaday.com.

    So I'm happy that it actually turned out to be cheaper than 9$ to produce. This is just like I initially estimated and it's a good thing. Of course, I haven't got into the added cost of the actual labor, packing, accessory cables and shipping. And above all, the selling share will add to the final price. 

PWM resolution discussion (as promised)

I implemented the following code for controlling my buck converter using this output PMW (as you probably saw)

#define pwm_period 700                  //PWM Period example 700*(1/16,000,000) = 43.75us / 22.857 KHz
...
//--------------------config PWM
P1SEL |= BIT2;                                     // P1.2 Select as TA0 output
TA0CCR0 = pwm_period;                    
TA0CCTL1 = OUTMOD_7;                  // TA0CCR1 reset/set output mode
TA0CCR1 = 0;                                      // TA0CCR1 PWM duty cycle
TA0CTL = TASSEL_2 + MC_1;            // SMCLK, up mode

 

The equation for determining the output frequency of the PWM is the following:
pwm_period*(1/16000000). After this, you come up with the period and you have to divide 1/period again, in order to establish the frequency. 16000000 is the number of oscillator cycles per second that is then internally divided and dealt with. Now, in my case I used the PWM period of 700 (correct me, I know, is actually 701), because I want a frequency that is more than the audio band. I really don’t want it to pollute my audio devices when is functioning and I really don’t want it to make audible noise. It would be awkward. Now the internal timer module, based on the values of TA0CCR0 and TA0CCR1 in this case, actually starts to count until it reaches the value of the TA0CCR0. Which means I have an internal  counting of 701 oscillations and then it starts all over again.
Based on the PWM filtered period, my external optocoupler will have a different internal resistance being inserted in the respective voltage divider and it will act like a potentiometer (a non-linear one). The graph of this PWM potentiometer looks like this:

So the LED of the optocoupler is actually having an external filtering capacitor that filters the PWM duty cycle in a continuous variable voltage. This means that is actually normal to see the optocpupler not opening at values of the PWM that are lower than 200, because the LED has a threshold forward voltage itself. Now, the problem is that, as you can see from the graph, I must set all my voltages from 10V to 25…30V with only 160 leftover units of my PWM scale. I want to be able to set the output voltages from 100 to 100mV and that leaves me with a necessary of at least 200 different PWM values, which I don’t have.

I implemented this piece of code in my A3 firmware and the entire system was oscillating. It was trying to set a certain voltage value using a PWM duty cycle of 615 units and then the read output voltage was too high and when it was lowering the duty cycle to 614, the output reading was too low and it was oscillating like this. 

A decent solution to the problem would be increasing the resolution by lowering the frequency on the LED. This has it’s drawbacks. Besides the fact that the parallel capacitor won’t be able to filter the LED voltage so good, I have a lot of steps to increment the PWM duty cycle register before I hit the right spot using the A3 software and this would take a few seconds to reach the desired output voltage after setting it up.

At this stage I won’t modify the hardware (except the value of the capacitor). I think I will go with a lower frequency and I will have to implement an algorithm that will make the duty cycle to jump to the right spot according to the desired output voltage.    

 
What do you think about this?

Post your suggestions, thoughts, ideas here.

In the above schematic pin 2 and 3 of the optocoupler is connected. So it is quite useless to use it at all. You can use an ordinary NPN transistor and save some component cost.

But I would do it still differently: Take your PWM signal from your microcontroller divide it down to 1,25V of amplitude (nominal feedback voltage of LM2596-ADJ) and feed it into an RC filter (1 to 3 stage). The output voltage is coupled via a resistor into the feedback point of the LM2596. So a current is injected, which makes the regulator think, the output voltage is to high and it regulates back.

The only drawback is that your control curve is inverted: 100% PWM ratio -> 0V out and 0% PWM ratio -> maximum output voltage.

But the advantage is it is quite linear and you don't have the high value capacitor at the feedback point of the regulator, which impairs regulation speed.

I would leave the voltage regulation to the LM2596 and not implement a software regulation loop. Just give him the desired value.

Perhaps you want to do some simulations with the free tool LTSpice as I did, when I designed a control for an LCD backlight using exactly the same approach as described above.

Or simply invert the PWM output polarity - probably a bit in the PWM setting.

This is actually good thoughts with replacing the optocoupler with a transistor. You are right here. The thing with 100% PWM ratio equaling 0V is a bit scary and I tried it. It works in theory, but it can generate some faulty scenarios. Like, for instance I have the mobile phone already connected to the output, I connect the Scorpion 3.0 to the AGM battery and it takes 0.02s for the MCU to boot up, configure, and make the PWM 100%, but meanwhile I have a severe dirac voltage spike at the output.  

PWM is a poor way of controlling output voltage.  Been there done that.  I looked into ways of maintaining high resolution while not slowing the power supply switching frequency.   https://hackaday.io/project/4993-dual-channel-battery-chargeranalyzer

This is how I control the voltage similar to what Martin suggested by injecting a voltage: https://hackaday.io/project/19078-adjustable-linear-bench-supply-in-1k  The resolution is as high as your ADC, very linear and there are no PWM noise to filter.