FET's in series

It's just a couple tenths of a volt, not enough to matter.

It looks fine to me. C1 could be bigger, but is probably OK. The voltages drops across the fets aren't perfectly balanced, but is pretty good. Technically, you'd want the source voltages to be evenly spaced, not the gate voltages, but that's a quibble. What, me quibble?

Fets make nice heaters, good heat conduction path, linear loop, often cheaper than resistors.

My condolences about the musical.

Just a reminder George, Chapter 9 is a free sample:

You didn't say what was wrong. When I run your simulation it seems to work. The current is 200 mA and there is a small ramp on the gate drive when the current is shut off. I suppose the cutoff voltage is around 3.5 so the gate is not driven very hard below that. What is it not doing?

The gate drive for M2 and M3 are not quite matched to M1. To make it perfectly symmetrical the gate of M1 would need to be grounded and R! needs a different drive. You can tie the bottom lead of R3 to M1-gate. With 100 ohm drive that shouldn't be a problem. It helps in the simulation. The remaining voltage difference is from R1 being tied to the M3 drain rather than the M4 (non-existent) gate. Using a 750 K resistor for R1 gets to well within a volt match.

It's the ratio of Vgson to the HV rail that gives the offset. R1 down 3Vgson/V3, R3 up the same ratio. If it were a KV, it wouldn't matter.

You can avoid current spikes by slowing down the op amp a bit. What's the hurry? It's only a heater.

To reduce turn-on and supply effects, decouple across each of the R1-R3string with equal cap values.

RL

Yeah, I shouldn't have posted and run. I figured out (sitting in the audience before the show.) that it's the gate voltage. Giving me the difference voltage. (The guy next to me was looking at my paper full of scribbles, but he didn't ask.) If I used a FET with a lower Vgson is there more leakage current at Vgs=0?

The lower current loop.. Yeah I just threw that together to the "get" the voltages across the FET's. As JL says more C1.

RE C's across fet's. (see figure 9.111 in Win's link, with diode's and more. I don't quite get the diodes..) The fet's are all part of a low temp heater, so I'd like to keep it as simple as possible.

George H.

Oh I want to keep the voltage drop across all the fets equal, then the power is equal. I'd like to place them symmetrically around the probe to reduce thermal gradients. The other options a ring of resistors, linear with power loss on the surface, or a single fet and some R's.

Oh, I want to keep the voltage drop across all the fets equal, then the power is equal. I'd like to place them symmetrically around the probe to reduce thermal gradients. The other options; a ring of resistors, (linear with power loss on the surface), or a single fet and some R's.

George H.

I was able to get the voltage (and therefore power) equal in M1, M2 and M3 by moving the lower end of R3 to the gate of M1 and changing the value of R1 to 787 kohms. I didn't change your simulation to see if that still works for other currents, but at the current you set it for it seems to work fine. The simulation shows the powers matched to within 3 mW. With a 750 kohm resistor for R1 the mismatch on M3 is 160 mW. I expect in a real circuit variations in the FETs will make more of a difference.

I asked Legg about Fet's with lower on voltage. How much does Vgson change with temperature?

The musical was Awesome! (I won't bore you with details, but I'm still humming. 42nd street. My son played tenor sax in the pit band. My daughter was not lead, but played the older diva.)

George H. (All musical talent comes from my wife.)

If this is about the 50C heater, why not use several NPNs with emitter resistors, in parallel? Base voltage controls the heat, of course. It's a lot easier to control than series connection, if your application isn't using HV.

Just tweak the divider resistors a bit to about equalize the drop across all three fets. The small errors result from the gate turnon voltages of the two upper fets. No big deal, but you can improve it a bit.

Logic-gate fets have nice low turn-on voltages. At your currents, 2 volts or less maybe.

Generally a couple of mV per deg c, which isn't significant in your circuit. It will change more as the current changes.

I'm one of those rare people who don't like music.

Leakage currents? In real life at modest temps, these are zero.

Yeah, I'm not sure how much thermal gradients will matter. So just a single fet and R's could be fine. (just letting you know this may not get built.)

Resistor ratios: I was hoping to leave the supply voltage as a variable. (and I still have to find out how much Vgson changes with T.) But the circuit doesn't care at all about leakage of the upper fets, they could have a much lower on voltage.

(with two "more equal" heater's I could play games with the position.)

George H.

Yes, in real life, with everything at equilibrium, or changing slowly, the diodes don't do anything. But imagine an output suddenly-shorted to gnd or a rail. We worry about what might happen.

Note the diodes are zeners. First, a diode protects against -Vgs limits. Next, with a quick enough reaction, a FET can protect itself against a +Vgs limit. With a zener we play it safe.

Yeah, so think of R1,2,3 and M1,2,3 as all one heater with three inputs..(and a local ground return for bottom of R3..

You could get compulsive and use three opamps and three current sense resistors, and run each fet across the full supply.

Use a quad opamp and some r-packs.

A couple as in ~2mV? like a diode? Don't answer that!

Huh, Have you read CS Forrester?

Horatio Hornblower? Many times. It think the Aubrey/Maturin novels were partly cribbed from Forrester, with the music thing reversed.