explain to me the switched cap frequency controlled gain amplifier (ltc6943)

[SArepairman]

I get that the switch + capacitor works as a simulated resistor where R = 1/Cs*Fs

So I see there he has 1/(0.0000000001*1000) and 1/(0.00000001*1000) (10:1 clock ratio ) =
10 meg to 10K = gain of 1000


I built this circuit up, though I did not have the same parts on hand, so I used a 1uF and a 10nF

1/(0.00000001*1000) = 100k
1/(0.000001*10000) = 100

So the simulated resistances are alot smaller but the ratio is the same.

I used a 10nF integrator capacitor (just like the app note) and TTL clocks using two function generators (1k and 10k Hz).
I am using a LT1677 op amp with this circuit

I also am using the non inverting form of the circuit (swap pins 7A and 6A (datasheet mentions 8A incorrectly, but I cross checked with the LTC1043 datasheet which has it right).

However, this circuit won't work. It just slowly counts up to the + supply rail, the frequency adjust stage does not matter and if I disconnect the 1KHz clock it outputs close to zero, if I connect the 1KHz clock it slowly moves up to + rail on the output.

Why did the original circuit use such high resistances and a JFET op-amp? And how about the integration capacitor ?

*I used all polypropylene capacitors

[SArepairman]

should perhaps pin 12 and 11 be switched on the second LT6943?

I don't really understand the pin switching thing...

My first LTC6943 does act as a SC resistor, I can measure it with the multimeter when frequency changes (50 ohm - Kohm) . the second one is 10meg w/ 1KHz signal, and the resistance does not change with frequency, but it goes down to 1Meg without a square wave....

I also replaced the 2nd LTC6943, I know its not a bad chip. I checked the wiring. Should both of the LT6943 pins (6-7 and 11-12) be inverted? Right now I have 6-7 (chip 1) inverted and 11 12 ( chip 2 ) like in the diagram

THE RESISTance measurement is directional though, like 80 ohms one way and 600 ohms the other way (on the first simulated resistor).

does this make sense? I guess the op amp could be effecting stuff, but it does change at least..

[SArepairman]

oh, the 10megs was the input impedance of my multimeter connected to the circuit.

The first stage goes from .5-10k, the second stage is high impedance

*opamp removed during these tests

[Andreas]

I built this circuit up, though I did not have the same parts on hand, so I used a 1uF and a 10nF

1/(0.00000001*1000) = 100k
1/(0.000001*10000) = 100

So the simulated resistances are alot smaller but the ratio is the same.

Why did the original circuit use such high resistances and a JFET op-amp? And how about the integration capacitor ?

Hello,

you forget that the switch resistance is around 1000 Ohms (divided by duty cycle 0.5).
So in your case you cannot ignore this resistance.

With best regards

Andreas

[SArepairman]

I built this circuit up, though I did not have the same parts on hand, so I used a 1uF and a 10nF

1/(0.00000001*1000) = 100k
1/(0.000001*10000) = 100

So the simulated resistances are alot smaller but the ratio is the same.

Why did the original circuit use such high resistances and a JFET op-amp? And how about the integration capacitor ?

Hello,

you forget that the switch resistance is around 1000 Ohms (divided by duty cycle 0.5).
So in your case you cannot ignore this resistance.

With best regards

Andreas

switch resistance is 300ohms in data sheet, but I see what you mean. But I should still be getting something right?

measuring the SC actually, gives me a range of 800-1000 on the first one.. order of magnitude more reverse polarity. I guessing using a ohm meter on these circuits is not the best idea.

why won't it work at all?

[Andreas]

perhaps your clock does not have full amplitude (-5V to +5V)

with best regards

Andreas

[SArepairman]

whaaaa? this needs a -/+ clock? It says CMOS is fine.
where did you get this idea from?

MOTHER FUCKER I WISH THIS WAS IN THE DATASHEET. this thing has been plaguing me all day.... now to solder in a mid air comparator to translate the 0-5v signal with a -5v/5V signal....

My circuit board is alllllll fucked up from too much probing/solder probe wires (home made, many traces need to be replaced with bodge) but it seems now that it works somewhat as described with just the ohm meter. 

it seems that my nicely made board is going to have a bunch of unplanned additions...

[Andreas]

whaaaa? this needs a -/+ clock? It says CMOS is fine.
where did you get this idea from?

MOTHER FUCKER I WISH THIS WAS IN THE DATASHEET.

Its on page 4 in the picture at the bottom of the side:
The clock has to be with a amplitude of V- to V+.

Edit: the LTC has no ground pin. So it should be clear that his world is referenced by the V- pin.

With best regards

Andreas

[SArepairman]

whaaaa? this needs a -/+ clock? It says CMOS is fine.
where did you get this idea from?

MOTHER FUCKER I WISH THIS WAS IN THE DATASHEET.

Its on page 4 in the picture at the bottom of the side:
The clock has to be with a amplitude of V- to V+.

Edit: the LTC has no ground pin. So it should be clear that his world is referenced by the V- pin.

With best regards

Andreas

 

at least its over