0
\$\begingroup\$

I have a device designed by a more experienced peer of mine. The peer left, and the problem lands on me. enter image description here

The Mosfet Q1, often got killed (failed to switch on), and the fix was simply replacing it till now.

Recently I smelled the Mosfet overheated and the pcb got toasted in the Drain 1 area. My guess is that the 10k resistor R4 is too high, result Q1 not fully open/close, and the internal resistance got very large, it began to self-charge/discharge... enter image description here

The ENable is on by default, when off, 18v turns the gates on/off...

  1. So how do I find a right gate resistor? I found this formula: Rgate=Vgate/(Imax) I assume Vgate=18v, but which one is Imax in the datasheet? the Gate resistance in page 2, is that mosfet internal resistance or recommend external Gate resistor I'm looking for?

  2. should there be a discharge resister similar to R3 (for Q3) that discharge any voltage build up by Q1 internal capacitor?

Thank you all, still learning

\$\endgroup\$
7
  • \$\begingroup\$ First, what is the load in this case? If it is inductive (such as a motor), then it's all too possible that the MOSFET is failing due to induced back-EMF. Second, that circuit is far from ideal for driving a power MOSFET. One edge of the control signal will create a relatively fast fall on the gate voltage - fast switching. Then, on the other edge, Q3 closes and Q1 gate charges solely through R4 - slow. You would want to use a dedicated CMOS driver IC in this case. \$\endgroup\$
    – DELTA12
    Commented Feb 16 at 11:23
  • 1
    \$\begingroup\$ Is there a large capacitor on the load side? Is there a situation, where the 18 V supply is not stable? \$\endgroup\$
    – Jens
    Commented Feb 16 at 13:04
  • \$\begingroup\$ @Jens it is a 6ohm resistor.... \$\endgroup\$
    – jjrfadpammtpiagkjq
    Commented Feb 18 at 7:21
  • \$\begingroup\$ @DELTA12 it is a 6ohm resistor....not sure which number I should be looking at, doesn't N & P type close/open at different speed? if so, I that a wise thing to have R4 control close&open of the 2 mosfet? \$\endgroup\$
    – jjrfadpammtpiagkjq
    Commented Feb 18 at 7:38
  • \$\begingroup\$ Q1A is a source follower, so you need +18V at its gate, to have the source reach full battery potential. Installing the resistor you propose would break this behaviour, and possibly cause Q1A to dissipate a lot of power. The resistances are all fine, the problem is somewhere else. What is maximum battery voltage, and do the +12V and +18V rails actually have those potentials? \$\endgroup\$
    – Simon Fitch
    Commented Feb 18 at 13:01

1 Answer 1

Reset to default
0
\$\begingroup\$

R4 is fine with 10K (you can go as high as 50K). The problem is with R5. Resistors to ground should be at least 100K. In this case Q3 Gate is not fully on because too much voltage drop through R5. Some current remains at Q1A and Q1B Gates which are not fully turned off. When MOSFET 's are not fully on or not fully off, they tend to heat a lot because their resistance Rdson is much higher.

Another possible problem is that 18V is too high for the gates. Usually gates can support up to 20V but maybe not with these models or the 18V is not exactly 18V all the time. I would reduce the voltage down to 15V to make sure.

There could other reasons. But without the schematic of the entire circuit, it's hard to say which ones.

EDITUM:

Blockquote Should there be a discharge resistor similar to R5 (for Q3) that discharges any voltage build-up by Q1 internal capacitor?

There should be a discharge (pull down) resistor if "18V" could become high impedance, e.g. physically disconnected. If "18V" is always either 18V or 0V (open drain, connected to GND...) then it's not theoricaly necessary. Still even in this case it can be useful to add one to make sure the gate is stable at start up or in case of malfunction. Having the gate floating, in itself, will not damage the MOSFET but could cause over heating because the gate will be between on and off many times, and it will wear it out.

With R4 at 10K and a pull down resistor of 100K, the voltage will be reduce by 10% and will be safer to use on the gate.

\$\endgroup\$
7
  • 1
    \$\begingroup\$ Hmmm…only way R5 could be a problem is if the EN signal is from an open drain output with a relatively high pullup resistor. It seems very unlikely, but like you said it’s hard to determine without the full schematic. \$\endgroup\$
    – Ste Kulov
    Commented Feb 18 at 10:12
  • 1
    \$\begingroup\$ Or an output with 3V and a high internal resistance. \$\endgroup\$
    – Fredled
    Commented Feb 18 at 13:04
  • \$\begingroup\$ Ohhh. Gotcha. The unknown origin of the EN signal from the missing schematic could have a large series resistance. I see what you mean now. Thanks! \$\endgroup\$
    – Ste Kulov
    Commented Feb 19 at 3:56
  • \$\begingroup\$ I build a simplified circuit on a protoboard, and connect it to a dc load set to 6ohm, 12v psu and 18v psu. no heat on D1. I found heat occurred only when 18v drops to 12v, I'm going to add Q3 and R5 to my protoboard to see if Q3 will open in full. Enable comes from MCP23017 (controlled by Arduino?), based on datasheet it should be 5-0.7=4.3v (will need to verify this on the real device)....18v was boosted from 12v, should be constant... \$\endgroup\$
    – jjrfadpammtpiagkjq
    Commented Feb 22 at 7:10
  • \$\begingroup\$ R4 is fine with 10K (you can go as high as 50K) >> can you teach me how to find the gate resistance? seemed like my peer doesn't calculate the value at all, but using some common value instead....(100ohm for digital control, 10k for power...) \$\endgroup\$
    – jjrfadpammtpiagkjq
    Commented Feb 22 at 7:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.