Yes, even though he's dead wrong, and can't read a data sheet.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
I haven't done a crowbar circuit in decades. I do occasionally put a tanszorb on a power rail. But soft triggering an SCR reduces its current handling capacity.
I rarely push a part past datasheet limits; only when it seems safe and there's a huge performance payoff.
Here's an analog computer that models junction temperature and shuts off a mosfet when it looks too high. It's a much better way to push a mosfet than a simple current limit, or even a foldback.
That gets combined with a heatsink temperature sensor and a time constant to make a dynamic model of Tj.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
The fact that the data sheet doesn't mean the same thing to him as it does to John Larkin isn't definitive proof that Fred Bloggs' point of view is wrong.
Krw and John Larkin do think that way - if you can dignify the process as "thinking" - but their conclusions aren't entirely reliable.
If John had gone the trouble of pointing out what Fred Bloggs had got wrong, one might take John's post more seriously, but at this point it sounds more like yet one more squeal of injured vanity.
-- Bill Sloman, Sydney
It applies to both. SCR specs usually include I2t values and one should select a fuse I2t less than the SCR I2t. Or, the SCR should be selected such that its I2t is greater than the fuse I2t.
For an example, see page 3 for I2t ratings:
Neat. That's less circuitry than it would have taken to get the two parameters into an ADC, and since the required accuracy is modest, it's a good fit for the task.
The model is in software, I gather?
I recently did a rough square-rooter to linearize the power dissipation in a small heater--over the relevant range, the change of slope went from 5:1 to +-20%. That helps keep the control bandwidth constant. It works over a pretty wide range of pass transistor temperature, too, as long as it's got a control loop wrapped around it.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
That circuit is used on a laser driver that's all analog, no uP. A Zetex SOT23 nickel RTD picks up mosfet temp which is added to the computed power signal, and that sum is lowpass filtered to approximate the chip mass, then a comparator. If the customer asks for too much power, or doesn't provide enough cooling, we shut him off.
Too bad mosfets rarely provide on-chip thermal sensing or shutdown.
I've done that shutdown in software too, in our NMR gradient drivers. Digitize Ifet and the voltage across the fets, and heatsink temp, and run a model to estimate Tj and shutdown if it looks scary. That's run maybe 1K to 2K times per second. We tested a lot of fets to destruction to estimate the numbers that we used in the models.
I like PWM drive into resistive heaters, when the system can tolerate it. Nice and linear. One other trick is to use a mosfet as the heater; keep the voltage mostly constant and control the current.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Those are dI/dt induced failures (hot spots due to concentrated current densities). I will concede that an unnecessarily large IGT will reduce, but not eliminate, the incidence of failure in the case of dI/dt abuse.
Care to explain how the MC3423 does anything to control dI/dt?
That is nice... Why do you take the difference of the two collector voltages? I've never used a diff pair, but I always see the signal taken from one side. Why no emitter resistors? (Would that extend linearity.. at the cost of gain?)
So a slow enough low pass "time" such that the heat sink temperature has enough time to react to the "new" power level?
The mosfet heater is nice. I've also stuck both the pass transistor and heater resistor together on the thing to be heated.
George H.
Yup, in less sensitive situations I generally use PWM heating too--nice and efficient, no thermal problems with the switch, and so on. PWM is a no-no in this case, though, because both capacitive and inductive pickup are serious issues. I just thought the square rooter was fun. The layout allows me to use it or not use it. I just got the boards stuffed by Beautiful Layout Hunchback, so we'll see how they work. (It's a temperature control experiment that may have considerable practical usefulness.)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
It doesn't, silly. What it does do is ensure enough gate drive for long enough that the full capability of the SCR is actually available. The Zener circuit is like a choke chain on a dog--as soon as it gets started, it cuts off its own drive.
From the 1993 Motorola Thyristor Device Data book, p 1-5-13:
"The di/dt of the current surge pulse is also a critical parameter and should not exceed the device's ratings.... The magnitude of di/dt that the SCR can sustain is controlled by the device construction and, to some extent, the gate drive conditions. When the SCR gate region is driven on, conduction across the junction starts in a small region and progressively propagates across the total junction. Anode current will initially be concentrated in this small conducting area, causing high current densities that can degrade and ultimately destroy the device.
*To minimize this di/dt effect, the gate should be turned on hard and fast such that the area turned on is initially maximized.* This can be accomplished with a gate current pulse approaching five times the maximum continuous gate current I_gt, and with a fast rise time (< 1 us). The gate current pulse width should be greater than the propagation time; a figure of 10 us minimum should satisfy most SCRs with average current ratings under 50A or so."(emphasis added)
Official enough for you?
Like I said, this problem has been known forever.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
That's in fact the signal that we want, approximately the product of Vfet*Ifet.
We don't want linearity! It's a multiplier!
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
The difference voltage goes as tanh(delta Vbe*e/2kT), which has only odd order terms, so the even order nonlinearity in the individual collector currents cancels.
Another approach would be an LM13700, but there are a lot more dual transistors than OTAs in the world, and they're cheaper and smaller, to boot. (I'm a big OTA fan in principle, having grown up on NS linear applications manuals, but I've only ever used them in one-offs.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Ouch, you're yelling. :^) Remember, I am but an egg in the sea of electronics.
Sorry, I'm at home with (appropriately) my first ed. of AoE. Without the emitter resisters, the gain goes as the base-emitter resistance, r-sub-e in AoE notation. Which is linear with the collector current.
I got it..
Re: diff opamp. The signal in each side is equal and opposite You could just take the signal from one side, right? I hope it's OK I ask questions?
George H.
[snip]
I don't think quite. You need to write out the I-V equations for the B-E junctions, the effective TANH transfer function is what yields the multiplication. ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Wow... I think I understand that. Tanks.
OK.. I've heard OTA's can be used as multipliers, but I need to do more homework... it's again got gain control that goes as some current. (right?) I've only used expensive multipliers from AD $5-15. or mixers from minicircuits.
George H.
OK more homework! It's warm ~60's, but raining, here now. I assume I start with the Eber's Moll model for re vs Ic.
George H.
to watch O'Reilly >:-}
Yes. Sort of. Stop thinking _resistance_. ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
jpg
It doesn't have to be.
In my millidegree temperature controller, we use PWM to control the current running into the Peltier junction. The PWM is actually set up as a 10-bit binary number, and the microprocessor that sets the number takes into accou nt the temperature difference across the Peltier junction and gives it a lo t more current when it's cooling hard than when it's acting as a heater.
The response of the junction - in terms of watts transferred per amp of cur rent through the junction - varies by 7:1 over the original design temperat ure range.
Sloman A.W., Buggs P., Molloy J., and Stewart D. "A microcontroller-based d river to stabilise the temperature of an optical stage to 1mK in the range
4C to 38C, using a Peltier heat pump and a thermistor sensor" Measurement S cience and Technology, 7 1653-64 (1996).This makes the transfer function distinctly non-linear, and the microproces sor linearises it for us, allowing us to keep the loop close to critically damped.
The integrated circuit Peltier controllers you can buy from Intel and Linea r Technology are linear, and can't play that particular trick.
Sloman A.W. "Comment on 'Implementing of a precision fast thermoelectric co oler controller using a personal computer parallel port connection and ADV8
830 controller'[Rev.Sci. Instrum. 74, 3862 (2003)]" Review of Scientific In struments, 75 788-9 (2004)."-- Bill Sloman, Sydney
Hmm.. follow the current? I must admit at some level bjt's are still a mystery. (Fet's are easier to understand)
My wife has been a total trooper today, and my must important job right now is to go take care of the kitchen. Tomorrow, George H.
An OTA is a diff pair and a bunch of current mirrors. The bias current (I_ABC in the LM13700 datasheet, for Amplifier Bias Current) is mirrored to become the tail current for the input differential pair. It's really just a glorified version of the circuit John L posted.
The output is basically the collector currents of the input pair, mirrored so that one comes from the positive rail and the other from the negative rail. So all the level shifting is done for you, and you can do whatever you like with the difference current.
They're on the noisy side, due to getting about 3 copies of full shot noise on each current from all the BJT pairs.
The LM13700 also has the Gilbert feature, i.e. diode-connected transistors on each input that you can use to make it into a linear multiplier.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.