1
\$\begingroup\$

I try to design a (relatively) low cost H-Bridge on N channel MOSFETs. I need to drive 8.5A peltier device with 12V and (roughly) 20kHz PWM.

(after reading some of the comments I changed this question to use P-MOSFETS for high side)

I plan to drive only low side MOSFETs with PWM, so I don't care too much about capacitance of high side MOSFETs. I was able to find quite cheap P-MOS transistors with an Rds value of about 0.02-0.06R.

My circuit would look like this:

schematic

I could probably use bipolar transistors for driving high side MOSFETs, but I have some spare N channel, I hope that using MOSFETs there is fine too.

Now the question is:

  1. Is it a good circuit design?
  2. What resistance values I should use for driving high side?
\$\endgroup\$
  • \$\begingroup\$ Have you considered using PMOS transistors for your high side driver? It would completely eliminate the need for you to have a step-up converter, as the sources on the PMOS transistors will be at +12V. This means you could turn on your high side by simply driving the gate to low. \$\endgroup\$ – justinrjy Mar 28 '15 at 8:53
  • \$\begingroup\$ How expensive is too expensive? Because I'm finding half-bridge gate drivers (with charge pumps) on digikey for a buck. \$\endgroup\$ – Dan Laks Mar 28 '15 at 8:54
  • \$\begingroup\$ @justinrjy I'd rather stay with NMOS transistors, because of better efficiency \$\endgroup\$ – Piotr Sarnacki Mar 28 '15 at 8:58
  • \$\begingroup\$ @DanLaks I'm not from the US, so I'm not sure what would be shipping costs. Most that I found where about about 5-6 USD. Also a lot of them required PWM with duty cycle lower than 100% \$\endgroup\$ – Piotr Sarnacki Mar 28 '15 at 9:01
  • \$\begingroup\$ Since I'm not familiar with distributors in Poland, I'll take your word for it on cost. In that case, I would reconsider justinrjy's suggestion. In the past, it was justifiable to pass over PMOS in favor of NMOS for high-side switches. But PMOS technology has improved in leaps and bounds in the passed several years. You can pick up a PMOS with tens of mOhms now for almost as cheap as an equivalent NMOS. If cost if your primary consideration, I'd definitely consider that route. \$\endgroup\$ – Dan Laks Mar 28 '15 at 9:15
4
\$\begingroup\$

I think your circuit looks good. Although you made the common mistake of placing the PMOS symbols upside-down. The source of the PMOS should be on top for this kind of circuit. Also, put a bypass capacitor on the VDD pin of the IC.

If you won't be switching Q1 and Q3 very fast, R6 and R8 can be fairly large. 10k is probably fine. They'll turn on fast, but turn off slowly. The gate resistors on Q5 and Q6 aren't strictly necessary, but don't hurt and may prevent large current spikes from the uC pins. A small value (100 or less) is sufficient.

A few other suggestions:

  1. Assuming you're controlling everything with GPIO pins from a microcontroller, Q5 and Q6 will need to be logic-level FETs compatible with your logic voltage (5V, 3.3V, whatever).

  2. Put pull-downs at HI-A and HI-B. That'll prevent the gates of Q5 and Q6 from floating during initial power-up and accidentally turning on Q1 and Q3 momentarily.

  3. Similarly, put pull-downs on the gates of Q2 and Q4 in case the TC427 floats its pins momentarily on power-up.

  4. It's a good idea to put a capacitor (0.1uF is enough) between the 12V rail and the gates of Q1 and Q3. When the 12V first comes up, the gate capacitance of Q1 and Q3 will hold the voltage on the gates low until enough current has flowed through the resistors and into the gate. That may cause Q1 and Q3 to turn on momentarily. The capacitor will force the gate voltage to follow the 12V rail as it comes online and prevent that from happening.

  5. MOSFETs do not turn off instantaneously. So if you, for example, turn on Q1 and turn off Q2 in the same instruction cycle in your firmware, Q2 may stay on for a few extra milliseconds and cause a shoot-through condition. Therefore, write your code with small delays between switching instructions. Better yet, use a micocontroller that has an H-bridge driver peripheral with built-in delays.

\$\endgroup\$
  • \$\begingroup\$ Thanks for help! I updated circuit to add capacitors as you mentioned and I flipped PFETs. Regarding your suggestions: 1) I have IRLR2703 so it seems I'm good with ~5V 2) Good point, I added them 3) In such case can I remove pull downs on LO-A and LO-B inputs? 4) I added it to the circuit 5) I have an arduino uno, I'll look for such capabilities, but I'm confident with programming, so that shouldn't be an issue, I'm just a total noob in electronics :) \$\endgroup\$ – Piotr Sarnacki Mar 28 '15 at 11:11
  • \$\begingroup\$ Also, another question - is it OK to use 1/4W resistors in this circuit? I think it should be enough with small currents that arduino can generate, but I'm not 100% sure. \$\endgroup\$ – Piotr Sarnacki Mar 28 '15 at 11:21
  • \$\begingroup\$ For 3), you still want those pull-downs in case the pins on the arduino float. \$\endgroup\$ – Dan Laks Mar 28 '15 at 11:22
  • \$\begingroup\$ Yes, 1/4W resistor will be fine. You could go quite a bit lower, even. None of those resistors will have any appreciate amount of current for any real length of time to overheat them. \$\endgroup\$ – Dan Laks Mar 28 '15 at 11:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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