MS20 / WM2 gain adjustment

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Last night I was playing with an MS20 PIR board to determine the effect of reducing the gain of its first amplifier stage.  Room temperature was about 80 degrees, experiments were all done inside.  The "gain" is the multiplication factor used by the first amplifier to increase the signal level from the PIR detector before it is passed on for further processing.

The gain of MS20 / WM2 PIR board is controlled primarily by the ratio of R13 to R12.  From the factory, R13 is 1.5 Megohms and R12 is 5.1 Kilohms; so, the factory gain is 294.  I chose to modify the gain by replacing R12, the 5.1K resistor, with a 100K potentiometer (variable resistor, or "pot" ).  The higher the value of R12, the lower the gain.

As I began, the pot was set all the way to 100K (decreasing the gain to 15)  At this setting the device was totally unresponsive.  Gradually the resistance was decreased, and when I got to about 85K (gain = 18), I was finally able to trigger the sensor by moving my hand back and forth six inches away.  It wouldn't detect me walking back and forth 18 inches away.

Further reducing the setting of the pot caused caused the sensitivity to increase as expected.  The most interesting and probably most useful result began as I approached about 25K (gain = 60) .  At this value, I could trigger the device about half the time by walking back and forth across my doorway through which the PIR was "looking", about 12 feet away.

At a setting of 15K (gain = 100), the device would detect me every time I crossed the doorway, and usually would detect me as I walked into another room, still line-of-sight to the PIR, 20 feet away.

In conclusion, at least in the environment in which I tested and for the particular MS20 under test, it appears likely that reducing the gain by a factor of 3 (using a 15K value for R12) could reduce the problem of "falsing".

Hopefully this will provide a starting point for MS20 / WM2 users wishing to decrease the sensitivity of these devices.  Again, if the 15K value did not provide enough gain to achieve desired, operation, R12 should be decreased toward the 5.1K value used by the factory;  if lower gain is still needed, R12 should be further increased.

Regards,
Jon

(Edited by Jon5ja at 7:47 am on June 15, 2002)


(Edited by Jon5ja at 7:50 am on June 15, 2002)
 

Archilochus

Well-known member
Joined
Mar 27, 2001
Messages
1,285
Reaction score
0
Hi Jon,
I don't think this would matter too much in most cases, but changing those resistors will alter the "pass-band" of the amplifier (both high & low cutoff) - have not done the math to find out just how much.

I'm not sure that this would work - (I don't have an MS-20 to try it out on) - but maybe this idea would work without disturbing the pass-band....
At the point where R9 connects to C4 & C7, maybe use an appropriately scaled Pot and resistor to ground (R=100k Pot=1Meg??).  Disconnect the (+) side of C7 from R9 and hook it to the "wiper" of the Pot.  Hook one leg of the Pot to R9, the other to the 100k resistor, the other end of which then goes to ground.  Turning the Pot "towards" ground should result in a smaller signal at the (+) input of the op-amp.  The "R=100k" would set a minimum sensitivity.
This is similar to what I use on my "home-brew" sensors - but the MS-20 is a bit different in that it uses the built-in FET of the sensor element as the first stage amplifier - so I'm not certain this will work out.

If this works, it would only allow for decreasing the sensitivity - but I can't imagine who would want to increase the MS-20s sensitivity anyway!
It might need an "audio" or "log" Pot to work right.

Archilochus


<<<<EDIT>>>>
This is sorta what I had in mind....




(Edited by Archilochus at 7:39 am on June 15, 2002)


(Edited by Archilochus at 10:32 am on June 15, 2002)
 

kmitch

Well-known member
Joined
Dec 11, 2001
Messages
429
Reaction score
0
I did the same thing with R13. I used a 2.5 meg pot in place of the 1.5meg fixed. I didn't track the resistance against the detection distance but was able to reduce the detection distance of the PIR. If the gain was increased too much it would go unstable. Sounds like R12 is a better answer. Perhaps keep the 5.1 K in the circuit to establish an upper bound on gain and use a 25K pot in series.

I think this is an excellent improvement. I think the key to fewer landscape pictures (pictures with no game) is matching the PIR detection range to the field of view of the camera and the flash range of the camera. The dection range changes with the ambient temp so for example, during the summer the gain must be inceased to achieve the same detection range you would get in the winter months.

Keith
 

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Arch is right, of course!  Rather than analyzing the first stage filter/amp, I had just given a cursory look at the reactances of C9 and C10 and assessed that X_C9 was sufficiently low and X_C10 was sufficiently high that it wouldn't be critical, although I anticipated that the lower cutoff would move down, but I didn't anticipate that would be a problem.  I thought I could ignore the upper cutoff because it seemed high enough as to be irrelevant ... unfortunately, that last assumption was based on a missed decimal point in my cursory look!

Looking again, I see X_C9=R12 at 0.95 Hz, and that X_C10=R13 at 2.26 Hz, so the filter is much sharper and more critical than I had mistakenly assumed.  Actually, now that I think about it, I noticed the sharp frequency response as I was messing with the circuit, and a light should have come on.

My objective in all this was to find a single component to vary without having to add considerable circuitry (simplicity was my objective in the little timer mod too... I want to make some changes to add some hysteresis to make it behave more consistently SOOO BAD, but I keep telling myself that it'll serve the purpose like it is, and I don't want to start telling people to cut up and redesign the basic uint).

Just as I was looking at Arch's schematic adding the pot, which would definitely work (other than you might have to make C7 non-polar :) it occurred to me that the FET in the PIR sensor, with its gate grounded and the source allowed to pull up to "equilibrium" through R10, is basically a constant-current device, +/- the signal from the PIR thermopile (or whatever it uses);  it should pull constant current (a hair over 9 uA in my case) regardless of how much we decrease the value of the pull-up resistor, R9.  I verified this by shorting the 100K resistor, and the drain current was not measurably changed.  Since the FET in the PIR will vary drain current based on its input signal, the output voltage should be proportional to the value of R9.  Right?

So... I should be able to leave the filter components alone, and just replace R9 with a 100K pot to linearly adjust the signal level (apparent gain) from the PIR.

Am I missing something, Arch?

Thanks for checking up on me... BTW, guys named after Greek poets who can analyze an active filter, know that Zener and Fresnel are people's names (heck, you can probably even pronounce them), can spell, quote Dr. Strangelove, and build motion-activated cameras for hunting (I presume) are getting pretty rare :)

(Edited by Jon5ja at 6:31 pm on June 15, 2002)
 

bat

Well-known member
Joined
Jan 12, 2002
Messages
4,354
Reaction score
0
Put this gizzamo over here to this thangamyjiggy, hang this doda over the whatamycallit. hehe....man I wish I was "smart"  Boy you gurus are good!
But put me on a blood trail and I'll beat any bloodhound!   :smile-big-blue:
 

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Sorry, guys...  Arch and I waded pretty deep and I really should have e-mailed Arch directly instead of cluttering the board.  I'll do that next time.

I tried what I last described -- just replacing the 100K resistor R9 with a 100K pot so I could vary it's resistance between 0 and 100K -- and it appears to work fine to adjust the gain.  I'd have saved myself some embarassment if I'd seen that at first!

After I did the low-power mod, this one idled at 81.7 uA on 9V, and after I changed R9, it's still idling at 81.7 uA, so it looks like this will be the preferred way to set the gain.

Jon
 

Hill Hopper

Banned
Joined
Mar 10, 2002
Messages
1,570
Reaction score
0
Please continue to post, rather than just e-mail Jon. Some of us are a little behind the curve, but this is the way we learn. I hadn't fooled with electronics any since I was an ET in the early 70's, or circuits since Physics II in engineering school a long time ago. I have spent more time trying to get somewhat back up to speed than building cams. I like to understand what I'm doing and why. You have been doing some great work, and placing excellent posts. Thanks for all the input.
 

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Thanks.  In thinking about it while I was at church (during the singing, not the preaching :) I remembered that there is a 0.47 uF capacitor, C4, shunting the output of the 100K/PIR part of the circuit.  Sure enough, at 1.5 Hz, the reactance of that component is something like 225K, so it's doing some of the low-pass filtering too, and although it might still work fine to just replace R9 with a pot (since you have the active filter following), changing the driving impedance would affect the frequency response of that first part of the circuit.

So... something like Arch suggested or replacing R9 with a pot and just hanging C7 off the wiper may be the way to go.

Jon

(Edited by Jon5ja at 1:42 pm on June 16, 2002)
 

bat

Well-known member
Joined
Jan 12, 2002
Messages
4,354
Reaction score
0
Hey Jon, I wasn't trying to step on your toes, just had that funny feeling reading all that "Good Stuff"....man don't mind me I'm just a dummy when it comes to that kind of stuff....keep it coming man I maybe can catch on to it one day....after all last year this time I couldn't even build a deercam :shades-smoke: now I already built 5 with two more in the works.  bat
 

Brian

Inactive
Joined
Jul 30, 2001
Messages
1,194
Reaction score
0
Humm.  Remings me of when I was working on the GLolab circuit some time back.  Yep had to go pull out the old Filter design books and scratch the cob webs out.

Some really good stuff.
 

Tinhorn

Well-known member
Joined
Mar 13, 2001
Messages
3,516
Reaction score
0
Where can I add my 2¢,  Oh, I know.....

When I was playing with this area on the MS20 about a year ago, I noticed the 1 minute warmup period of the PIR is apparently determined (or at least affected) by the value of C9.  I also seem to remember that the oil needed changing in my car about that time and figured it was easier to figure out which wrench fit the drain plug than what did what on that circuit    OUCH

Tinhorn
 

Archilochus

Well-known member
Joined
Mar 27, 2001
Messages
1,285
Reaction score
0
Hmmmm... gets more interesting every time I check.
Might be best just to go with a Pot for R9 and replace C4 with whatever gets closest to a good compromise for low-frequency cutoff.....

.... Tinhorn..... thanks for the reminder..... looks like the old Honda is just about ready for a tune-up (every 100,000 miles sound about right??  ;-))

Archilochus
 

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Well, I thought I was wrong, but I was mistaken...

When I was playing with the gain of the first amp/filter stage, I had assessed that the upper cutoff of the bandpass filter was way too high to worry about... then I recalculated and decided it was a tight filter.  Well, today, I was looking at various application notes, and recognized that they were using an order of magnitude larger capacitors across similar-sized feedback resistors... so I revisited the MS20, and my original calculations were correct!  The upper cutoff is 22 Hz!

So, after all the discussion here I've decided it's better to do the gain adjustment in a manner that's independent of the filter (I'm still working on what I think is best and most easily implementable), but now I'm wondering -- why so high a cutoff frequency?

Arch or others, what frequency range do you use for your input bandpass?  I wonder if this high cutoff could explain why the MS20 / WM2 seems to be more susceptible to "falses"... with that high a cutoff, a bird could probably fly through the field of view and trigger the device.

Any comments from you experimenters?
 

Archilochus

Well-known member
Joined
Mar 27, 2001
Messages
1,285
Reaction score
0
Hi Jon,
Let's see... I didn't check the sums recently, but I think my "home-brew" sensor cuts off at just under 10Hz.
I've seen other PIR sensor schematics with down to 4Hz upper cutoff.  22Hz does sound mighty high - can't figure why they did that.  So fiddling with the feedback resistor to set gain might make the cutoff even higher! - that could be a real problem with birds, small mammals, and artificial lighting (not that there is much of that in the woods).
I'll have to check out the specs on the RS sensor.

Archilochus
 

ArkansasElkHunter

Well-known member
Joined
Jan 19, 2002
Messages
1,748
Reaction score
0
I don't know about the cutoff but I do knowa bird flying in front of the sensor will set it off,  even a little one.  I have sensors that will pick up a hand wave at 30'
 

TmKarren

Well-known member
Joined
Dec 26, 2001
Messages
274
Reaction score
0
I was just wondering what you three came up with??  What would you replace to make so you get more false pictures?  and would it help for in the hot summer??  What is a POT???  Thanks !!
 

Jon5ja

Well-known member
Joined
Jun 6, 2002
Messages
722
Reaction score
0
Well, I've never quite finished working out something I like, but I think I'm going to try replacing the 100K resistor (R9) with a 100K potentiometer ("pot", or variable resistor).  

Instead of what I was discussing earlier, though, where I was considering just making the 100K "variable", I think I'm going to try putting a 100K pot with the two "legs" of the pot (I don't know what you call the two leads that aren't the wiper) in place of R9, and I'd connect C7 to the wiper of the pot.  This would leave C4 looking at the design impedance.

My only reluctance to doing it this way (and one I'll probably have to test to see if it's justified) is that when the wiper is moved toward the +5V end of the pot, any electrical noise that's on the +5V line will be coupled into the amplifier.  I don't know if it will be a problem or not.  I'll try to do some more experimenting over the next few days and see if this suits me or if I can come up with something better.

One other thing to which we've only alluded is that the upper cutoff frequency of the amplifier may be too high -- possibly we could reduce falses (from birds flying through, etc.) by increasing C10 to 0.01uF or maybe 0.02uF.  I haven't tried anything in that regard, though.

Jon
 

Brian

Inactive
Joined
Jul 30, 2001
Messages
1,194
Reaction score
0
My cutoff is 10Hz and I adjust the gain of the signal before the last amplification stage just before it hits the comparator stage.
 

Latest Posts

Advertisement



Top Bottom