Are datasheets important?

[ante1]

It is an insulting statement. Ones appearance has nothing to do with ones Electronic knowledge or capability.

MP

Neither has your documented education obviously!

 

[AN920]

As I stated before, I use data sheets as a guideline only. I have saved companies loads of money by designing and redesigning products to use lower grade components. The important part is to understand what parameters may be exceeded under which conditions and to design around the limitations.

The one thing that separates the Technician from the Engineer is the ability to design and evaluate using advanced circuit analysis and mathematics which goes beyond just

 

[audioguru2]

Hi AN920,
If you were designing this 500W inverter, would you use a huge, expensive and poor tolerance non-polar 4.7uF capacitor with a 100 ohm resistor in its CD4047 oscillator, when the manufacturer of the CD4047 recommends using a minimum of 10k for the resistor?
Each IC would make a very different frequency due to the variable on-resistance being much higher than the resistor's value.
Wouldn't a smaller, less costly and good tolerance 100nF capacitor with a 47k resistor be much better?

Would you cut in half the required number of output transistors so that they operate at double their absolute max current rating?

Would you use two quad opamps instead of a smaller and less expensive dual opamp?

 

[AN920]

I would not use bipolar power transistors in a push-pull configuration period! (many hidden problems to the inexperienced, especially at high current levels)

I would not use 50/60 Hz inverter transformers. (too bulky)

The 4047 would be last on my list of choices. (hate RC timing constants, prefer XTAL)

It would we a different design all together. (HF switching)

I would use opamps only for feedback, protection and regulation functions and not for
drivers.

I am saying this based on my experience of designing a commercial 5kW UPS for one of my previous employers.

I designed some 400Hz inverters for aviation applications in the 70's using bipolars but using proper precautions like using clever base drive circuitry, baker clamps, device desaturation detection, transformer saturation detection etc. Today most things are much easier and safer using power mosfets and dedicated drivers.

 

[audioguru2]

Hi AN920,
Of course a HF-switched, crystal-controlled and  Mosfet-output inverter is more modern but I was talking about this 500W inverter project that we have that came from the dark ages.

 

[AN920]

I will say this in a Q & A session:-

Are there questionable design practices involved?
Yes, whether we like it or not. This is just not the way things are done by power electronic designers. This may be due to the fact that this is not the area of expertise of the creator of this design.

Does this flame the designer?
No, he may be brilliant in other areas.
I can write software that works but I suck compared to some other people I know.

Will it work?
Maybe under low power conditions for short periods.

Will it work reliably under the claimed power for extended periods?
Probably not.

Can it be improved?
Like everthing, yes. Here the best way will be to contact the designer and point out any potential problems and work with him towards a solution. Engineers in general are humble people and should not feel offended by this action. Most will welcome any advice to improve a design.

 

[Kevin Weddle]

Maybe the resistor value in the RC common network is a factor. If someone uses 47Kohm and 100nF for the RC, then uses a 1Kohm resistor in the RC common network, then the time constant will be affected. So a small resistor like 10 Ohms might be more accurate if you used a 47Kohm and 100nF.

 

[audioguru2]

Kevin,
The oscillator in the CD4047 uses a single resistor with a single capacitor to set its frequency.
The formula for the frequency is shown in the datasheet and in numerous articles as "1, divided by 2.2 times RC", when the value of R is 10k or more.

If the value of R is less than 10k then the internal resistance of the IC must be subtracted from the value of R for the formula to be correct. The internal resistance is different for each IC.

Our project now shows a 1k resistor and a huge 4.7uF capacitor. If the IC has an internal resistance of 900 ohms then the total resistance is 1900 ohms and the oscillator's frequency will be calculated at 51Hz. Then the frequency of the inverter will be 25.5Hz instead of 50Hz.
The transformer will probably saturate with such a low frequency. Some loads will be damaged.
A low tolerance non-polar 4.7uF capacitor is rare and would be expensive.
It is very wrong!

My circuit used a 47k resistor and a 100nF capacitor. If the IC has an internal resistance of 900 ohms then the total resistance is 47.9k ohms and the oscillator's frequency will be calculated at 47.4Hz which is pretty close to 50Hz. 39k with 100nF produces an output from the inverter of 57.0Hz which is pretty close to 60Hz.
My low tolerance non-polar 100nF capacitor is inexpensive and is available everywhere.

 

[Kevin Weddle]

Interesting point, if less than 10Kohm subtract the internal resistance. The impedance does look too low for MOS. But nothing I design lasts more than 3 hours, so it looks like a plausable design to me.

 

[audioguru2]

Texas Instruments recommends a minumum value for the resistor on a CD4047 oscillator of 10k ohms.
Fairchild recommends a minumum value for the resistor on a CD4060 oscillator of 10k ohms.

These are shown on their datasheets. I think it is important.

 

[Kevin Weddle]

According to your post, you added internal resistance to the 1Kohm for 1900ohm. But you said if it's less than 10K you subtract internal resistance. It's not subtracted.

 

[audioguru2]

According to your post, you added internal resistance to the 1Kohm for 1900ohm. But you said if it's less than 10K you subtract internal resistance. It's not subtracted.

That is correct and is why the 1k resistor now in our project with the 4.7uF capacitor produces an inverter's frequency that is much too low at 25.5Hz.

If you use a 1k resistor then you must subtract the internal resistance of the IC. We don't kow if its internal resistance is 100 ohms or 900 ohms because each one is different. If the internal resistance is 900 ohms then a 100 ohms resistor should be used with a huge 4.7uF capacitor for the inverter to produce 48.4Hz which is close to 50Hz. But if the internal resistance is 100 ohms instead of 900 ohms, then the inverter's frequency will be 241.8Hz which is much too high.

My selection of a 47k resistor makes the frequency always close to being correct because 100 ohms or 900 ohms internal resistance added to it doesn't make much difference.

That is why experts at Texas Instruments and at Fairchild recommend using a minimum value for the resistor of 10k ohms.

 

[Herman the German]

Walid,
Perhaps you have misunderstood the conversation. No one has been disrespectful to audioguru. However, you might not know that audioguru is not an R&D Engineer. In fact, he is not an Engineer. He mentioned in another forum that he did not complete his Engineering course.
In regards to being disrespectful, I find it disrespectful to say someone looks unusual when discussing their knowledge or capabilities. It is an insulting statement. Ones appearance has nothing to do with ones Electronic knowledge or capability.

MP

Hi MP,

I suppose that you know what you are talking about after having been engineer for more than 30 years. I read every argument in this topic and I think that some of your words chosen are very close to disrespect. It is not important to be an engineer. It is important what you do and how you do it. In my special field I was just a technician concerning aerodynamics and air flow, but I had an engineer for breakfast every day. They just talked high tones not knowing their jobs properly.

So please have the kindness not to get PERSONAL in any discussion. To me it appeared like a declaration of war! Since I was soldier for more than 23 years I know something about that "topic" too and I'm proud of having killed 255 communists who wanted my butt but lost the competition.

Just live in peace with audioguru to make me happy.
Herman the German

 

[audioguru2]

Hi Herman,
Nobody was disrespecful to me:
"Texas Instruments and Fairchild recommend a minimum of 10k ohms for the resistor in a CD4047 oscillator, then an inexpensive high accuracy capacitor can be used? Thank-you for pointing it out, Audioguru".

I get thanks like this all the time.

 

[Herman the German]

Hi Herman,
Nobody was disrespecful to me:
"Texas Instruments and Fairchild recommend a minimum of 10k ohms for the capacitor in a CD4047 oscillator, then an inexpensive high accuracy capacitor can be used? Thank-you for pointing it out, Audioguru".

I get thanks like this all the time.

Hi audioguru,

I just saw it that way. Why not take a CD4060? That one can use an Rt > 1K and Rs about 10XRt. Of course, it has an odd outputs (compared with the CD4047),  but can be compensated by setting the correct clock frequency.

Inverter frequencies should be near the nominal main frequency. I tolerate +/- 1Hz. As pointed out before a crystal oscillator will do best and for extra accuracy the crystal can be pulled slightly with a 40pF adjustable capacitor.

 

[audioguru2]

Hi Herman,
The CD4047 was used because it is an oscillator IC that has a digital divider so its outputs are a perfect square-wave, and it has two opposing outputs (one is inverted).

 

[Herman the German]

Hi Herman,
The CD4047 was used because it is an oscillator IC that has a digital divider so its outputs are a perfect square-wave, and it has two opposing outputs (one is inverted).

Hi audioguru,

well, a CD4047 will do the required job. I just thought of the 100Hz cycle it produces internally to be divided by two. It is kind of hard to achieve a stable 50Hz cycle anyway. Higher frequencies are easier and using the appropriate output and just one transistor for inverter will result in the desired function.

I used a 4060 for accurate timing with outputs of 400, 200, and 100Hz and it worked as accurate as a stopwatch.

 

[Herman the German]

P.S.

I purchased a 3000VA inverter in China to be deliverd next month. I'll take it apart and look how that one works. I am well acquainted to the sales manager and she will also deliver a schematic of that device if I join her wedding party.

 

[Steve_hi]

Hi Herman I hope you don't mean that you have to marry her to get a schematic!!! LOL ;D