Why not the LM339 instead of LM311 in AADE LC Meter clones?

[KamalS]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

I am not sure it is very useful. It continually talks about opamps then moves on to simulating with ideal opamps. The TI employee cannot derive a formula for the oscillation frequency.

Well - he does kinda give a formula:

Yes there is a formula. For a maximum error of 10%, the maximum propagation delay is 10% / (2 * max frequency).

For 1 MHz that is 10% / (2 * 1MHz) = 50nS.

I however do not understand from where it comes from - looks like some kind of Nyquist sampling relation.

 

[E-design]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

**broken link removed**

From the manual for this kit.

Accuracy and Resolution
L/C Meter IIB has four digit resolution which for small values of L and C are 1 nHy and .01 pF. You cannot accurately measure values this small. The resolution greatly exceeds the accuracy. You can measure values as small as .01mHy and .1 pF with about 15% accuracy. You generally won't find components this small. For example a piece of wire less than one inch long is .01 mHy. The resolution is, however, relative and can be used for sorting a batch of similar components as it truly does indicate which are slightly larger of smaller than others. Also, for small values of inductance, the leads will contribute quite a bit to the value. Measuring from the ends of the leads instead of next to the body of the component can add up to .025mHy.
For small values the frequency of operation (test frequency) is about 750 KHz decreasing to about 60 KHz at .1 mFd's or 10 mHy's and about 20 KHz at 1 mFd or 100 mHy's.

I seriously doubt if the 393 can provide stable operation at 750kHz

 

[keith1200rs]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

Yes there is a formula. For a maximum error of 10%, the maximum propagation delay is 10% / (2 * max frequency).

For 1 MHz that is 10% / (2 * 1MHz) = 50nS.

I however do not understand from where it comes from - looks like some kind of Nyquist sampling relation.

It comes simply from the fact that the period of 1MHz is 1us. 10% of that is 100ns. There are two transitions of the comparator per cycle so halve the 100ns to get 50ns. So, a very fast comparator would oscillate at 1MHz with a particular set of components. Use a comparator with 50ns propagation delay and it will oscillate 10% slower because the period will be 1us +50ns +50ns. In reality it is more complicated because the propagation delay varies with overdrive and with a relaxation oscillator the overdrive will increase until the output switches.

I seriously doubt if the 393 can provide stable operation at 750kHz

I don't see why it shouldn't in a normal relaxation oscillator. The equation for its oscillation will be way off at that frequency though. It simulates over 1.3MHz at 5V but then you get around 1V of overshoot. The LM311 would oscillate at 2.3MHz with the same values and 0.3V of overshoot.

It is all irrelevant to the original question anyway. I think the circuits which use the LM393 do so because it is adequate for the job.

Keith.

 

[E-design]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

Before testing the circuit on the bench I also ran a simulation. Simulation suggested oscillation over 1.5MHz (which I did not believe to be true). When I studied the plots the simulations indicated that the 393 model was switching way under 100nS which is bogus. I then tested the circuit in practice to verify.

 

[keith1200rs]

My model seems to give 340ns/290ns large signal response, 1.3us/1.2us small signal response using the test fixtures in the data sheet.

Keith.

 

[E-design]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

Maybe another manufacture's 393 may oscillate stable/predictable over 500k as all parts are not created equal. Can't tell the manufacturer from the part I found in my parts collection. The problem is that sometimes you don't have a choice in the part you will get when ordering from surplus discount electronic part stores, like many hobbyists may do. Even when brand stamped you may getting a cloned pirate part. Hard to tell the difference most of the time. Pirate parts are popping up everywhere even from reputable vendors.

Added after 28 minutes:

when we get newer design without need for LM3xx and internal comparator could well be used like that of Phil Rice, why think of alternatives for LM3!!? You could try to improve on 16F628A based design

**broken link removed**

 

[KamalS]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

My model seems to give 340ns/290ns large signal response, 1.3us/1.2us small signal response using the test fixtures in the data sheet.

Keith.

Are you using LTSpice?

Could I have the schematic for the same in that case please?

I also want to read up on overshhot and overdrive in comparators.

Where would I find authentic info?

 

[snake]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

hi all,

in these type of circuit they normaly uses a single device to avoid the crostalk between the internal parts, using single devices, so if you dont take some care in the parts that are not used they can interfere in the device that you are using.
i had some trobles with that some time ago.

thx

 

[dick_freebird]

You can't count on an LM339 to run faster than about 500kHz since
prop delays are about 1uS at low overdrive.

I would consider a low voltage CMOS comparator if the voltage
range is adequate. These will be much faster and have negligible
input bias current.

 

[keith1200rs]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

Are you using LTSpice?

Could I have the schematic for the same in that case please?

I also want to read up on overshhot and overdrive in comparators.

Where would I find authentic info?

No, it is not LTspice - SIMetrix. The model I used was a TI one. The conditions for the two cases - large & small overdrive - are in the datasheet so it is simply a matter of setting up a couple of voltage sources on the inputs with the right characteristics. Looking at a simple relaxation oscillator as you vary the components to increase the frequency is also informative - you can see how the overdrive increases as you try to increase the frequency and the frequency deviates from the theoretical.

Keith

 

[keith1200rs]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

The attached shows the comparator operation at two frequencies. When it is slow the output switches when there is around 60mV difference between the two inputs. When it is fast the two inputs can be 1V apart by the time it switches. You should be able to duplicate the results in LTspice.

Keith.

 

[KamalS]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

Are you using LTSpice?

Could I have the schematic for the same in that case please?

I also want to read up on overshhot and overdrive in comparators.

Where would I find authentic info?

Looking at a simple relaxation oscillator as you vary the components to increase the frequency is also informative - you can see how the overdrive increases as you try to increase the frequency and the frequency deviates from the theoretical.

Keith

I want to read up on overshoot and overdrive in comparators.

As of now, I really don't know what overdrive is and how it factors in opamp applications ( apparently a lot )

 

[keith1200rs]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

I don't know of a link to explain it, so I will have a go myself.

Overdrive in comparators refers to how far the second input is driven above the first. So, take the situation where one input is fixed at 2.5V. If the other input is sitting at 2.4V, for example, and then jumps to 2.505V then that is 5mV overdrive. In other words the second input is 5mV above the first. If it jumped from 2.4V to 2.6V, that would be 100mV overdrive. Small overdrive is the difficult situation for comparators and where they are slowest.

The two examples in the data sheet for the LM393 are 100mV pulse resulting in 5mV overdrive and a TTL logic input. So the 5nV overdrive could be simulated by a pulse from 2.405V to 2.505V, assuming the other inut is fixed at 2.5V.

Overshoot is different and refers to opamps, usually. There is a link here which should explain it:

https://en.wikipedia.org/wiki/Step_response

This may also be useful:

https://www.analog.com/library/analogdialogue/archives/37-04/comparator.html

Keith.

 

[KamalS]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

http://www.aade.com/LCinst/lcm2b.htm

From the manual for this kit.

Accuracy and Resolution
L/C Meter IIB has four digit resolution which for small values of L and C are 1 nHy and .01 pF. You cannot accurately measure values this small. The resolution greatly exceeds the accuracy. You can measure values as small as .01mHy and .1 pF with about 15% accuracy. You generally won't find components this small. For example a piece of wire less than one inch long is .01 mHy. The resolution is, however, relative and can be used for sorting a batch of similar components as it truly does indicate which are slightly larger of smaller than others. Also, for small values of inductance, the leads will contribute quite a bit to the value. Measuring from the ends of the leads instead of next to the body of the component can add up to .025mHy.
For small values the frequency of operation (test frequency) is about 750 KHz decreasing to about 60 KHz at .1 mFd's or 10 mHy's and about 20 KHz at 1 mFd or 100 mHy's.

I seriously doubt if the 393 can provide stable operation at 750kHz

The 750kHz comes from the fact that the AADE LC Meter uses 68uH, 680pF in the tank.

The resonant freq of the 68uH, 680pF AADE LC Osc is ~ 740kHz

What if I choose a 100uH, 1800pF tank?

 

[E-design]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

From the Peter Rice page

Experts may like to adjust the inductor value to raise F1 to near 00060000 to obtain maximum resolution from the meter. An "L" value of 82uH is preferred instead of the specified 100uH

I am not sure how the different software versions of these clones affects the operation. It appears that you need the highest frequency possible to get the best measurement resolution.

 

[KamalS]

Re: Why not the LM339 instead of LM311 in AADE LC Meter clon

In my LC Meter, I would use the "counting method" of finding the freq.

That is, I would, for a fixed period of time, let the exciting signal increment a timer.

For example, I would let the output of the LC oscillator interrupt a pin for 3 secs, and in the ISR for that interrupt, I would keep a track of how many times the interrupt was fired.

Here is the specifications of the AADE meter:

The resonant freq of the 68uH, 680pF AADE LC Osc is ~ 740kHz

For the nominal values of L1 (68 uH) and C1 (680 pF) an increase in L of 1 nH (.001 uH) or an increase in C of .01 pF produces a frequency change of slightly more than 5 Hz ( 5.439923 ).

A 0.2 second measuring period can resolve 5 Hz and therefore .001 uH or .01 pF.

For small values the frequency of operation (test frequency) is about 750 KHz decreasing to about 60 KHz at .1 m F's or 10 mH's and about 20 KHz at 1 m F or 100 mH's.

A 10mH L will be tested @~ 60kHz
A 1nF C will be tested @~ 480kHz

For my version of the meter, I would use the LM339, L = 100uH, C = 2nF.

The sampling period has to be increased to 3 secs instead of 0.2 secs that could have been possible with the LM311, making the meter 10 times slower.

For small values the frequency of operation (test frequency) is about 356 KHz decreasing to about 49 KHz at .1mF's, 35kHz for 10mH's, 15KHz at 1mF, 11kHz for 100 mH's.

A 10mH L will be tested @~ 35kHz
A 1nF C will be tested @~ 290kHz

Does the specs of my meter look good?