TIP120 Question

[EasyGoing1]

Originally Posted by eccerr0r

Yep, cathode to V+ to reverse bias it to be correct. Connecting cathode to the transistor, the diode will just heat up and the inductor won't work. Indeed you said in post #7 that you drew it wrong. Bad me for not scolding you on post #5

BTW need to be CIVIL folks, voltage leads current for inductors... Capacitors do inrush, and stalled motors do inrush (this is a mechanical "capacitance" problem).

So are you saying that this is the correct way to connect the diode, that WILL provide fly-back protection? Cause the way it looks to me, it will permit current to flow into the collector.



And what about this configuration, wouldn't this increase transistor protection all the more?

[eccerr0r]

first one is sufficient, D2 is superfluous.

[EasyGoing1]

Originally Posted by eccerr0r

first one is sufficient, D2 is superfluous.

Originally Posted by redwire

.

Originally Posted by stj

.

Originally Posted by lti

.

Pictures of final ... umm ... er.... "product" - if anyone's interested.

[sam_sam_sam]

What are you keeping in the box inquiring minds want to know

[EasyGoing1]

Originally Posted by sam_sam_sam

What are you keeping in the box inquiring minds want to know

I don't really know ... a friend of mine asked me if I could make this so this is the proof of concept... you can unlock it with a UDP datagram with the word "open" in it ... WiFi-controlled lockbox.

[stj]

you put drugs in it and link it to an rfid reader so you can supply your "customers"

[EasyGoing1]

Originally Posted by stj

you put drugs in it and link it to an rfid reader so you can supply your "customers"

Yeah, but wouldn't the first "customer" in line simply empty the box?

[eccerr0r]

Refill every after customer? They should only be able to open it if it's their turn...

[PeteS in CA]

I suggest adding a resistor base-emitter to reduce sensitivity to leakage and noise. This is normal practice with regular transistors, and I would think the high beta of a Darlington would make doing so even more important.

[redwire]

The TIP120 has them built in, otherwise it would be even slower switching. Plus you get a free diode too.

[PeteS in CA]

I sit partially corrected about the TIP120 incorporating B-E resistors. However I think those are to prevent leakage or noise from turning on the Darlington. IIRC, it's a C-B resistor that keeps a transistor or Darlington pair from saturating. Saturating would slow the turn-off time.

[redwire]

As I know, the E-B resistors are mainly to dissipate stored charge and thus speed up turn-off times. If you build a Darlington on the breadboard out of two transistors, they are super slow switching without them. But they also work for keeping the transistors off due to leakage or an Arduino pin floating during RESET.
I just noticed the datasheet switching time test is a cheat - they apply a -4V off-bias and +12V pulse to get (force?) the ~1usec rise/fall times.

Darlingtons are slow, and lossy because of their high VCE sat. 2V at 4A and 4V at 5A means they dissipate much more heat that a mosfet.
I think OP is fine using them for the solenoid driver, but PWM with an LED or running the TIP120 at a few amps it would just run hot and need a heatsink.

I haven't seen a C-B resistor used to help with saturation, or a Baker Clamp. I tried making SMPS with Darlingtons and it was awful compared to mosfets.

[eccerr0r]

Compare a TIP120 C-E drop with an IRF620 D-S drop at 5A load ...

Yes these things are slow but there have been plenty of SMPS with BJTs, though I can't say, other than the MC34063, many Darlington designs... and it too wastes power to speed up the Darlington.

[EasyGoing1]

Originally Posted by redwire

If you build a Darlington on the breadboard out of two transistors, they are super slow switching without them.

I'm curious ... define "slow"...

[PeteS in CA]

redwire is correct that I was confusing a C-B resistor with a Baker Clamp, which uses a diode to reduce storage time. Lo siento.

Basically, transistors' rise, fall, and storage time combine to limit the practical switch frequency of a transistor-based SMPS to less than ~50KHz. Above that practical frequency a transistor is either off or in its dissipative linear region for too much of the potential "On" time to allow the necessary input voltage range for an off-line SMPS.

[redwire]

TIP120 switching times are about 0.5usec on and ~3usec off with a -ve voltage drive to force it off faster. If you're not doing that or saturating it, it's much slower I would say up to 10usec to turn off in the real world. You could not use it to pulse an LED at 1usec say. Note the datasheet complicated test circuit is using +8V and -12V drive, which is unrealistic.
Power MOSFETS are 100x faster and less losses when fully on, so less heat. This is why they are king in SMPS use.

TIP120 is OK as a power switch as long as you are OK with the losses/heat VCE(sat)=2V and not switching really fast.

[eccerr0r]

IRF620 is only around 10x faster than the TIP120

BTW, this is specific to Darlingtons. I'm pretty sure you can build custom BJT circuits faster than off the shell Darlingtons, after all people have used BJT output transistors for RF circuitry -- note that the gain per stage is very low and separate from each other to keep the speed up.

[redwire]

IRF620 switches in nanoseconds, not microseconds. FAST.
I've tried building SMPS using Darlingtons and it was basically terrible to get past even 30kHz out of them. It's because their gain is so high, bandwidth is low. Turn-off takes forever. I suppose you could add circuitry to lessen charge storage. But the last power transistor having a hFE of 10-20 is no fun either.
Back in the old days of the first SMPS i.e. Apple and others in the 70's didn't even use them as I recall.
Today, it's a single BJT with transformer drive in SMPS.

[PeteS in CA]

Originally Posted by redwire

TIP120 switching times are about 0.5usec on and ~3usec off with a -ve voltage drive to force it off faster. If you're not doing that or saturating it, it's much slower I would say up to 10usec to turn off in the real world. You could not use it to pulse an LED at 1usec say. Note the datasheet complicated test circuit is using +8V and -12V drive, which is unrealistic.
Power MOSFETS are 100x faster and less losses when fully on, so less heat. This is why they are king in SMPS use.

TIP120 is OK as a power switch as long as you are OK with the losses/heat VCE(sat)=2V and not switching really fast.

In addition to limiting the realistic maximum switch frequency, during those turn-on and turn-off times, the device is in its linear reason, i.e. dissipating power. As those turn-on and turn-off times become a greater percentage of the switch frequency period, the device becomes hotter. That's the reason devices that could give decent input voltage range at 40KHz or 50KHz often are still used at 30KHz. It's 3+ decades, but I participated in making that decision in a few designs.

The faster switching times of MOSFETs took the practical maximum switch frequency way above 100KHz, which shrinks magnetic components, and probably were what made zero-voltage-switching quasi-resonant phase modulation designs practical (add in synchronous rectification, and welcome to >90% efficiency).

[PeteS in CA]

Originally Posted by redwire

IRF620 switches in nanoseconds, not microseconds. FAST.
I've tried building SMPS using Darlingtons and it was basically terrible to get past even 30kHz out of them. It's because their gain is so high, bandwidth is low. Turn-off takes forever. I suppose you could add circuitry to lessen charge storage. But the last power transistor having a hFE of 10-20 is no fun either.
Back in the old days of the first SMPS i.e. Apple and others in the 70's didn't even use them as I recall.
Today, it's a single BJT with transformer drive in SMPS.

Back in the late 70s and into the 80s, transistors were used rather than Darlingtons. At Boschert, we used 2N654# and MJE1300# transistors (TO-204 and TO-220, respectively) from Motorola. That was Mot's Switchmode series of transistors. Astec, who built Apple II+ power supplies, used television horizontal flyback transistors. At the power level of the Apple II+ power supply, 43W IIRC, Astec and Boschert both used a self-oscillating discontinuous flyback topology. This reduced component count and total cost. The typical efficiency was 65%-70%.

The "big" goal when I was at Boschert (Bob Boschert was still there, though as the VP of Engineering; Ray Noorda was the CEO) was to get their flyback models down to or below $1 per watt and efficiency above 70%. At 150W and above, Boschert used a two stage design, a primary side buck regulator followed by a square wave inverter. The controller for the buck regulator was a 723C, a linear regulator adapted to be a switching regulator. Boschert did current mode designs. The then-popular SG3524 series was voltage mode, and the current mode UC384# series was not yet available.

There still were occasional wooly mammoth sightings in our part of Sunnyvale.