PI design

[Anonymous]

I'm working up a PI design to post on my web site for experimenters. The goal is to have a basic PI detector with lots of flexibility, with plenty of knobs to turn so the experimenter can see the effects of different parameters.
The first-cut schematic is done (I think). I found a major timing connection gaffe last night so I thought I would post it before committing to PCBs... I would appreciate it if others could look it over for errors/suggestions.
Power supply & pulse timing
Main circuit
Keep in mind the following:
- Ignore component values, for the most part they have not been entered and there has been lots of copying & moving around.
- The preamp has extra bypass/nulling components so the builder can use a variety of different opamps.
- Some things might look a little wierd (R8a on Q2's collector) because I'm setting it up to build different ways. For example, the timing can be completely inverted to use Q2 as a common-emitter instead of a follower.
Thanks...

 

[Anonymous]

Hi Carl,
This is a great thing you are doing. Love your website!
My comments are the following:
Add a 0.1 uf from pin 5 of IC5 (555) to -SD
The value of R10 (damping resistor) may need to lower depending on inductance of searchcoil.
Q2, R8, and R9 can probably go away letting Q1 drive the gate of the 740 and making R7 100 ohms to turn off the 740 fairly fast. I think Eric's post on circuit showed this long time ago.
R11 and R12 set the gain for the front end amplifier. I would leave R12 1 Meg and make R11 about 2 Kohms for a gain of 500. Also would split R11 into two resistors 1.8 Kohms for R11 now and add 200 ohms between diodes and -input to amplifier to protect it some. This drops the gain from 2000 to 500 increasing bandwidth, but decreasing sensitivity, this gain would need to be made up somewhere (another amplifier?) to have both. There is an advantage with the high gain and using the built in bandwidth limits of the amplifier to cut off high frequency noise. This part depends on which amplifier is being used.
Not sure the purpose of C10 and R14 except to help stabilize a 709?
Might be wise to put in a current limit resistor 5 to 20 ohms power resistor between search coil and Q3. But may not be needed if frequency is kept high enough and the experimenter doesn't get Q3 turned on steady.
I would drop the value of R16 and R17 to 100 ohms giving more current to the amplifier, but the load is not too low, so maybe not important.
I would increase the value of R21 to 100 kohms, and increase R19 to 20 kohms and tie the end of it to pin 3 of IC6. This will work with more types of amplifiers and not use the offset provided on the amplifiers. More range and flexibility, and get the current down through this path.
R25 and R26 should probably be the same value and 470K should be about right.
Dropping the value of C15 and C16 to 0.1 uf will give more sensitivity, but also a bit more noise.
The circuit looks pretty good and probably works just like it is. Anyway, there's my two cents, unless I see something else.
JC

 

[Anonymous]

You might look at putting some filtering in the charge pump section to reduce the ripple generated in this circuit. I feed the charge pump (Eric's suggestion) with an inductor as well to keep the circuit from generating noise back on the power supply line. If the experimentor adds more circuitry and needs more current, a MAX660 will supply more current than the 7660. With all the stuff my charge pump has to supply I am using an LM2662 which is the beefiest charge pump I could find. The only potential drawback for the experimenter is that its only available in an SO8 package.
Thanks,
Charles

 

[Anonymous]

Carl,
I have found that a VCO audio output can cause a lot of trouble by feeding back it's non synchronous noise into the circuit. It is very tough to keep it out of the high gain circuitry.
I like to divide the transmit frequency down to about 300 - 400Hz. I chop the analog signal going to the audio amplifier using an analog switch. The result is good, clean, synchronous, VLF type audio.

 

[Anonymous]

With careful attention to layout, power supply decoupling, etc. the problem of VCO crosstalking to the front end can be eliminated.
I know of one company which solved the problem by driving the speaker with a sine wave using a linear audio amp; however, you don't get much beep per watt that way.
Hope to get a chance to look at Carl's circuit this weekend.
--Dave J.

 

[Anonymous]

<I>Add a 0.1 uf from pin 5 of IC5 (555) to -SD </I>
I don't think I've ever seen this cap make a difference, although most 555 circuits have it. I'll see if I can squeeze it in.
<I>The value of R10 (damping resistor) may need to lower depending on inductance of searchcoil. </I>
As I said in my posting, component values have not been entered and many are grossly wrong!
<I>Q2, R8, and R9 can probably go away letting Q1 drive the gate of the 740 and making R7 100 ohms to turn off the 740 fairly fast. I think Eric's post on circuit showed this long time ago. </I>
True, I threw in the extra components to look at the difference between having the follower and not. Q2 can also be connected in CE configuration so it becomes passive pull-up and active pull-down.
<I>R11 and R12 set the gain for the front end amplifier. I would leave R12 1 Meg and make R11 about 2 Kohms for a gain of 500. Also would split R11 into two resistors 1.8 Kohms for R11 now and add 200 ohms between diodes and -input to amplifier to protect it some. This drops the gain from 2000 to 500 increasing bandwidth, but decreasing sensitivity, this gain would need to be made up somewhere (another amplifier?) to have both. There is an advantage with the high gain and using the built in bandwidth limits of the amplifier to cut off high frequency noise. This part depends on which amplifier is being used. </I>
<I>Not sure the purpose of C10 and R14 except to help stabilize a 709? </I>
No, they are "spares" I threw in to play with. My bench circuit does not currently have C10, R14, or R15. I've seen them used on other PI's but I've never tried adding them.
<I>Might be wise to put in a current limit resistor 5 to 20 ohms power resistor between search coil and Q3. But may not be needed if frequency is kept high enough and the experimenter doesn't get Q3 turned on steady. </I>
Good point, although I guess this could be off-board. Maybe even preferred to be off-board so it can vary depending on the coil.
<I>I would increase the value of R21 to 100 kohms, and increase R19 to 20 kohms and tie the end of it to pin 3 of IC6. This will work with more types of amplifiers and not use the offset provided on the amplifiers. More range and flexibility, and get the current down through this path. </I>
Oooh.... good catch! I meant to put an option on the PCB to do just this, but forgot. I will leave the schematic as-is and add this on the PCB.
Thanks!
- Carl

 

[Anonymous]

Hi Charles,
I wanted to try to stick with the 7660 because of it's popularity. Good point about the inductor, though.
Thanks,
Carl

 

[Anonymous]

This is actually the only part of the design I have not done much with. I originally intended to stop the design (and PCB) at the output of IC8b, then do a separate audio board. In the end, I decided to throw it on and see what happens.
- Carl

 

[Anonymous]

Hi Carl,
You want to put a large reservoir capacitor from the source of the IRF740 to the ground end of the coil. 2000

 

[Anonymous]

Eric,
I will add the coil cap and resistor... however, my grounding scheme is not optimal. I'll post the first-cut layout as soon as I clean it up.
I'm currently using a 1k feedforward on the preamp, and a 300R damping resistor. I better change the damping resistor pads to 600 mils to allow for a higher wattage.
I'll have to rethink the multivibrator scheme.
Thanks,
Carl

 

[Anonymous]

DJ,
The VCO used to be standard practice for a PI. The problem is that our newer designs are going for a nickel at 17 inches. I ran a VCO with the utmost care for layout. Most of the boards I make are for use at 2450MHz - 5600MHz microwave work so a board for a PI was not much of a challenge.
The last time I ran a VCO, I used an opto-coupler to isolate both the audio amplifiers input as well as its ground. I used a seperate battery supply for the headphones. Everything was fine until I plugged a different pair of phones into the circuit. The speaker coils and the headphone cord somehow transmitted enough signal through the air to cause problems.
The detector design I was using was not overly prone to noise pickup. With synchronous VLF audio this detector works fine on the work bench in my lab. There is only a slight increase in the noise floor due to 60Hz and the flourescent lighting and the masses of test equipment that is always on.
One way to make a VCO work is to make it synchronous. Simply use the detectors output voltage to modulate the frequency of the transmit pulse generator! OK, its weird but it works great.
This method may also have some other advantages which are useful in target recognition?
The synchronous VCO method requires more testing. I took the output of the pulse generator and divided it using a CMOS divider. The no target frequency was 1Khz. I used a divide by 4 to provide a 250Hz signal which was used to drive the headphones. A target raises the transmit pulse repetition frequency and the audio synchronously.
It is also possible to chop the detectors output to provide both frequency and amplitude modulated audio. The detector signal modulated the pulse generators frequency by way of a potentiometer so as to control the amount of frequency deviation. All I will say about this is that it sounded different!

 

[Anonymous]

Hi Carl
I've always driven the charge pump from a
division of the main clock. This simple solution
eliminates problems of non-synchronous noise.
Cheers
Malcolm

 

[Anonymous]

Hi Carl,
For the multivibrator, just use one RC timing network, then take a 1N4148 diode from the junction of the R and C to pin 7 on one of the sample monos. Take another diode from the same point and take it to pin 7 on the other mono. The fact that each mono is triggered at a different time allows this RC sharing to work.
I use one dual mono package to develop both sample pulses and the other package to develop both delays. Simplifies tracking on the pcb.
On thing I noticed on the 74HC221 data sheet is that they recommend pins 6 and 14 be externally connected to pin 8.
Eric.