Mosfet driver capacitors? (again)

While designing my motor driver on shoe string budget I ran to a sort of contradicting information.
Attached picture is from the datasheet of a driver similar to IR2104. I find the given capacitor values a bit baffling since the Cbs is 10 bigger than Cdecouple.
Should I follow the generally accepted 10-20 times Qtot for Cbs and 10 times Cbs for decouple or should I stick with the values in datasheet?
I'm kinda leaning towards the general rule so that I can change to IR2104 in case these super cheap alternatives are bust.

Use the datasheet. That’s why the manufacturer publishes one. To ignore the datasheet is akin to shooting ones self in the foot.

Self inflicted wounds are entirely preventable and well, just plain dumb.

PS: posting links to datasheets helps others to be able to help you.

Have you checked the App Notes and Design Tips that International Rectifier hints at?
I'd start there.

I am a bit biased about the data sheets, they were written by engineers that know the part, general rules in part are by people not wanting to read the data sheet or do not understand it. The parts are guaranteed to the data sheet, not to what somebody thinks it should be.

The flip side of that is, how much do you actually want to trust the datasheet for a cheap Chinese part is properly engineered?

You'd be surprised at the quality of Chinese copies of products (now why they still don't seem to be able to get to the stage of "copy and improve" as the Japanese did so well in the 1980s is another matter). I don't think the OP is trying to use this part at anything near 600V, or near any of its other rated limits. For hobby level voltages they are likely to work just fine.

For the capacitors: you have to look in more detail at the actual workings of the part to know how they are being used. The values suggest that C1 is just for power supply decoupling, while C2 is the reservoir used to switch both transistors.

I do see an interesting difference in the schematics posted in #1 and #3. In the first capacitor C2 appears to be connected to the midpoint of Q1 and Q2. In the second there is no such connection. That makes for a huge difference in overall behaviour.

The second schematic also has a third capacitor across the two output transistors.

1 Like

The problem with this comment is that currently, nearly all of the electronics we purchase is manufactured in China and in order to do so, they were given - gratis - all of the intellectual property to manufacture them. Absurdly bad strategic planning by American tech companies! :roll_eyes:

Point is, they are not actually "copies" in the first place.

Have you heard of the CH340 IC? It is discussed here quite often. :grin:

CH340 is not carried as an individual chip by the big distributors - Digikey, Mouser, Arrow, Avnet, Sager. Why is that?
Mouser shows them as a Sparkfun sourced item, with a dribble of Rev E due in this month (155 pieces)
And Sparkfun shows just 7 10-packs in stock.

I just checked a 45 item order I placed with Digikey recently.
There are items from China, but only 18 of 45. Simple resistors, diodes, IC sockets, discrete transistors, USB sockets. ALL of the ICs, capacitors, LEDs came from other countries.
So "nearly all" is quite a broad overgeneralization.

Just as well! :laughing:

I can't figure out how to quote so I'll do without
"quality of chinese products"
The main reason I'm willing to try these eg2104 chips is my past good experiences albeit with products other than ic's. In my day job I use chinese carbide end mills with one of those $100 high freq spindles to machine AISI316L, have done so daily for past 10 years with great success. I'm on my 2. spindle because I crashed the 1. one.
I find it sort of difficult to believe that LCSC would carry these in stock if they were 100% BS

Drive voltage: You are right my design drive voltage is 48VDC with bus overvoltage trigger at 50-52VDC.

I think it is evident by now that I'm total newbie to electronics and coming from family of plumbers I tend to picture schematic to same analogy ie check valves, valves and such.
I know that many see a tin foil hat with big propeller at this point but acc. to google I'm not alone.

The way I read these schematics is that the C2 connection is the same, the EG2104 just draws it on the right hand side of Q1&Q2. Both still have a connection from VS to midpoint of halfbridge where the lover leg of C2 also connects. Both schemas also connect C2 top to VB and diode.
I don't know if there is big difference but I guess I'm going to find out.
I understand the C3 in IR2104schema as the power input buffer caps.

I have found them reasonably close, generally a little on the conservative side. It is surprising how many US parts are made in china in the same fab.

The chip's decoupling cap is what recharges the bootstrap cap, so it needs to be 10+ times larger than the bootstrap cap to avoid too much supply droop.
The bootstrap cap is what charges the MOSFET's gate capacitance, so it needs to be 10+ times larger than that to avoid too much bootstrapped supply droop. It also has to hold
enough charge to cope with gate leakage current over the period of PWM invoved.

This is all basic physics so you don't have to worry about whether the datasheet is right
or not, physics is always right for these things!

For instance if the effective gate capacitance is 3nF, 33nF Cboot and 1uF Cdecouple ought
to work, although larger might be better (especially for slow PWM).

The effective gate capacitance is best calculated as total gate charge / 10V, since this
is the worst-case demand for charge.

This topic was automatically closed 120 days after the last reply. New replies are no longer allowed.