Analog Squarer {Y=X^2} 1 Quardrant

The reduction in VDD has been a double-edged sword. Compressing +/-

power hogs. I need a simpler (+/-5% accuracy [-40C to 105C]) design that will be used in a high reliability product. Input range of 0.05V to 1.70V with a bandwidth of dc to at least 1.0MHz. Simplicity, small size, low power

A single diode,properly biased is fairly decent square law for a small range. Use a matrix with resistors; in the old analog daze (40+ years ago) they were called DFGs.

he

hen

Interesting. He wants 1MHz and 5% accuracy, so the three inverters have to run faster than 20MHz with a 1.7V input Might still be a power hog.

It's tested at +/- 15V, but it works fine down to 2V or so. Other OTA designs (RCA/Intersil's old CA3080) were recommended down to +/- 2V.

design a circuit who's output would be 1/Vin with all my other requirements, that would be perfect. Trimming is not allowed, got to work in production, no tweaking, every time.

Have a look at Widlar's low-drift breakpoint amp, Figure 8 in National AN4,

It uses transistor saturation rather than diodes to form the breakpoints.

Cheers

Phil Hobbs

Use APSK (Amplitude and Phase Shifted Keying).

but that temperature range (400K +/- 20%) may not play well with the required +/- 5% accuracy

the 145 degree range is worth about 290mV of bias

power hogs. I need a simpler (+/-5% accuracy [-40C to 105C]) design that will be used in a high reliability product. Input range of 0.05V to 1.70V with a bandwidth of dc to at least 1.0MHz. Simplicity, small size, low power

What precisely are you trying to do? If you don't need full

power hogs. I need a simpler (+/-5% accuracy [-40C to 105C]) design that will be used in a high reliability product. Input range of 0.05V to 1.70V with a bandwidth of dc to at least 1.0MHz. Simplicity, small size, low power >

RL

Jim, As my question states, I only need one quadrant, the first or third, both inputs tied together. I am trying to keep a MOSFET on for as long as possible, from releasing sm oke on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10 :1 so the "time on" changes 1:100.

Regards, Harry

Would it suffice to do it bolometrically?

Feed the current into a PTC (perhaps a silicon PTC so it has the same tempco!) and monitor its resistance in real time. Turn off when it's too high.

You can of course use the MOSFETs themselves for this, but you need to be able to sense the drain voltage.

Tim

-- Deep Friar: a very philosophical monk. Website:

Jim, As my question states, I only need one quadrant, the first or third, both inputs tied together. I am trying to keep a MOSFET on for as long as possible, from releasing smoke on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

Regards, Harry

inputs tied together.

on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

Oooops! In the fine print, in red yet, I missed the "1".

I'll dredge up my old games and report back. ...Jim Thompson

Harry D a écrit :

both inputs tied together.

on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

A reseted double integrator will do that nicely. Release the integrators reset on your fault condition trig, then compare the output to your wanted threshold...

power hogs. I need a simpler (+/-5% accuracy [-40C to 105C]) design that will be used in a high reliability product. Input range of 0.05V to 1.70V with a bandwidth of dc to at least 1.0MHz. Simplicity, small size, low power

Lookup table in ROM.

Parallel ADC -> ROM Address bus.

ROM data bus -> parallel DAC.

Or...

Parallel ADC -> resistor network, aka custom non-linear DAC.

Cheers

inputs tied together.

on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

Harry, Try this...

Given your loose tolerance requirement you could probably hand select four discrete transistors... just match the Vbe at 1mA. ...Jim Thompson

You may want to monitor the safe operating area, or SOA, which involves both current and voltage. See:

I think it would be most effective to use a PIC to monitor the parameters, but it will not be fast enough to detect destructive transients.

Another analog option may be to use an incandescent lamp (perhaps driven by an op-amp) to provide a signal based on i^2t.

You can also get self-protected MOSFETs:

Paul

both inputs tied together.

on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

My 40 year old automotive chip design tricks are coming back to me. Here's how to do it without an OpAmp; and the added transistors don't have to match those in the MAT-04....

both inputs tied together.

smoke on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

Since there's no arbitrary constants in there, I assume the temperature stability is pretty good?

Best regards, Spehro Pefhany

both inputs tied together.

smoke on overload conditions, so tracking the I^2*T constant of FET and also need a temperature term which is strait forward. The FET current changes 10:1 so the "time on" changes 1:100.

Of course !-) Do the math...

Vbe = KT/q*LOGe(I/Is) where Is is the same for all 4 devices in the MAT-04.

Equate the voltage drops down the two "legs" of the MAT-04, reduce until you lose the LOGe. You're left with...

I1*I1 = IMOUT*I3

No temperature terms ;-)

In critical applications you might want to cascode collector, C4, to avoid Early effect unbalancing the betas.

The 4-device (MAT-04) scheme is pretty robust... you can inject other currents between Q1 and Q2, and between Q3 and Q4 and get various interesting equations. ...Jim Thompson