Split H-Bridge: Will this work?

With the HIP408x, all four of your H-bridge MOSFETs should be N-channel, I think you do realize that, right?

I'm rather fond of those ICs.

No, no, only one supply for the HIP4080, with a diode and flying capacitor for each high-side's gate driver. That supply will need to be about 12 volts.

Yes, you can nicely accomplish your goal with a HIP408x - the two power supplies you're thinking of will be +2V for one high-side MOSFET drain and 35V for the other high-side MOSFET drain. Note, the HIP4080 doesn't know anything about the voltages on the MOSFET drains, until it turns one of the high-sides on, then it has to be able to handle the voltage, which it can, for anywhere from 0V to 80V.

You can put the current-limiting resistor in series with each diode under test - it won't have any effect when the diodes are seeing the reverse voltage.

If you use a humongous low-Ron type for the high-current- carrying FETs, you could run all 30 diodes (with their individual series resistors) in parallel, at about 1A each, but I might be inclined to use two or three MOSFETs for the job, each running 15 or 10 amps maximum. You'll be using individual gate resistors for each MOSFET anyway, so this means one HIP4080 chip can drive all the MOSFET bridges.

Right.

Make your drawing and analyze each state. BTW, I assume you ordered one of the HIP408xA types?

I'm curious, what kind of diode would find this test, 1A and 35V, difficult or stressful? A small-signal type?

Are you going to log the voltages across the diodes under current by looking at the resistor-diode nodes? I see that the voltage swing there is only from about 0.0 volts (well, 10A * Rds(on) = under 50mV for a MOSFET with Rds(on) under 5 milliohms) to about 1.3 volts, so you could use 74hc4051 multiplexer ICs (another fav of mine) to scan the voltages.

You wouldnt appear to need a H bridge, just a half bridge would do. Use 2 supplies in series with the diode between the supply common and the transistor common.

I'd go discrete as well here but all in parallel sounds quite brazen. It might be a good idea to place a blast plate over the FET ;-)

--
Regards, Joerg

http://www.analogconsultants.com/

Half bridge sounds better. Wire diode anode in series to sense/ballast resistor to ground. Use a small P MOS to pull up to +30V and a large N MOS to switch the diode on to -2.0V. Piece of cake. Gives you 30V to supply your control electronics. Harry

[ snip ]

:-) Actually, modern MOSFETs are pretty amazing. The 55-volt parts meant for the automotive market are especially attractive, because some have very low Rds(on), and are quite inexpensive. I'm looking now at a table of 40 or so parts we stock to choose from when working on a design. There's a fair rate of migration for these parts, as the manufacturers work to balance performance cost and profit, so not all the parts are still readily available. But there are great choices. For example, Fairchild's huf75344g3

which costs $2.13 at DigiKey and they have 600 in stock. The '75433 has a maximum Ron of 8 * 1.3 = 10.4 milliohms with Tj = 80C, which means it would drop only 0.1 volts at 10A, and dissipate 1 watt. Since RthJA = 62C/W, these numbers show these MOSFETs could work at 10A with no heat sink at all. But I'd use a small clip-on heatsink.

There are lots of other good choices besides the huf75344.

I agree, sketching various possibilities, I don't see any disadvantage to a half-bridge over an H-bridge. That's because the 2V supply voltage is so low, compared to 35-volts reverse, boosting the 2nd voltage to 37 volts.

Right.

No,no. The HIP will only get one 12v supply, the bridge uses two.

Very good, that's exactly what I woke up thinking. I did it this morning using a kinda half bridge.

Thanks to everybody. I did try a half bridge this morning, worked fine. At least on one diode. 1 amp If, 35v Vrevrse. Next thing: 30 diodes. Yes. I plan to use 3 parallel MOSFETS, like 50 amp ones running at 10 amps each.

Tell us more about the diodes, and what you hope to accomplish with your tests. Sounds interesting.

You say a 1 ohm in series? I assume since you want to test 30 at one time, you'll have a 1 ohm in series with each diode?

--
"I\'m never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

I don't have the diodes numbers or spec here with me, Monday I'll get all the data and post it. We screen parts for S class reliability.

yes.