MiniPuls3 Problems

[porkluvr]

I set-up a quick LTspice simulation and it shows high voltage getting through to U9 pin-8.

Sooo...

I suspect you may have also damaged Q2, U1, and U9 (and c6?). I recommend that after you install the zener, you check those components. (Do not install the LF357 until the power supply components have been verified to function correctly.

My simulation was not entirely accurate, but the picture it painted was very grim, so please heed my warning.

So, remove the old LF357. Install zener diode (make sure to get polarity correct). Check the +5V and -5V circuits. THEN you can try a new LF357 and see what happens.

I don't think Zener diodes are usually specified to be used where fast reaction is expected, but it has got to be better than nothing.

edit: Mouser Electronics (Mansfield TX, USA) does carry the SA5.0CA and they do ship to Greece. Of course shipping charges will be higher than you would probably like but they have a tremendous selection of components. Mouser.com or http://gr.mouser.com/

[porkluvr]

I had to take some shortcuts to make this, but please ignore what is wrong with it and instead concentrate on what is right.

The top trace is the flyback pulse which would be different depending on pulse-width and coil variations, but would be clamped to about -400V by the 1.5KE400CA transorb.

The bottom trace is where U3 gets damaged from leaving out the SA5.0CA.

If you have good +/-5 volts then maybe all you need to worry about is U3.

My simulation predicted that the Q2 base voltage climbed as time went on (whether or not the SA5.0 was installed) and it looks to me like there should be a resistor from Q2 base to +5V, but that is not something I can do anything about.

[WM6]

I had to take some shortcuts to make this, but please ignore what is wrong with it and instead concentrate on what is right.

The top trace is the flyback pulse which would be different depending on pulse-width and coil variations, but would be clamped to about -400V by the 1.5KE400CA transorb.

The bottom trace is where U3 gets damaged from leaving out the SA5.0CA.

If you have good +/-5 volts then maybe all you need to worry about is U3.

My simulation predicted that the Q2 base voltage climbed as time went on (whether or not the SA5.0 was installed) and it looks to me like there should be a resistor from Q2 base to +5V, but that is not something I can do anything about.

Hi porkluvr,

great simulation, thanks.

What about if we replace SA5.0 (in need) with some sort of fast switching diode (like 1n4149, 1n914) serial connected to get Zener U about 5V and total recovery time of about 25nS?

[johnyp]

But doesn't SA5.0CA works bidirectional ? How to achieve that? We will need 2 sets of 7 diodes anti-parallel. No ?

[WM6]

But doesn't SA5.0CA works bidirectional ? How to achieve that? We will need 2 sets of 7 diodes anti-parallel. No ?

Yes, on schematic we can see bipolar one, but as we can see on simulation picture only negative (-180V) pulses are really destroyable, positive goes only to about +12V. In antiparalel coupled diodes we can combine faster 1N4148 serial with slower Zener diode (which suppress on positive side).

[johnyp]

I see now where the trouble is. So i will order 10 supressors ( to have spares ) and problem solved. Thank you all.

[porkluvr]

Maybe johnyp's idea for a zener diode is not a terribly bad stopgap (interim) measure. I am not recommending it, but so long as a zener would turn on and off reasonably fast you might get away with it.

Note that, according to Ohms law, whatever diode that you put there will be subjected to almost 400ma current! Admitted, it is in short pulses, but 400mA is a substantial jolt. I would fear for the long life of your garden variety 1N4148.

A zener is not a "perfect" solution because zeners are not known to be fast acting and I hesitate to recommend it. But sometimes you've got to do what you've got to do.

Expanding on that notion, you might consider something like two 1W, 4.7V zeners, in series, back-to-back.

It's late so I'd better call it a day before I write something crazy.

edit: I just took a look at the NXP 1N4148 datasheet which states 450mA repetitive peak current, so a string of 1N4148 might be ok to use (but I would look for something 'prettier').

[WM6]

edit: I just took a look at the NXP 1N4148 datasheet which states 450mA repetitive peak current, so a string of 1N4148 might be ok to use (but I would look for something 'prettier').

If we use double or triple value of R5 (as current limiter) it might be ok for protected solution in need? What you say?

[porkluvr]

If we use double or triple value of R5 (as current limiter) it might be ok for protected solution in need? What you say?

Note that in my "edit" in post 22 that I had stated NXP's 1N4148 repetitive pulse current spec being 450mA. R6 being 1k would normally limit current to within that value in the Minipuls3 circuit.

Unfortunately... in their 1N4148 datasheet, NXP does not specify exactly what period and duty-cycle meets their definition of "repetitive". (Those DOGS.)

From what I see in the Minipuls3 schematic, I can not know what it takes to make Q2 operate correctly (something is missing, in my view) and I am hesitant to make any suggestion, except that Q2's proper operation needs to be taken into account, NO MATTER WHAT scheme you come up with.

edit: Two series connected 5.6V zener diodes (back to back) might work, but that is not going to be as fast as the original TVS would be

[Rumcajs]

Hi all,

see this link for D4 replacement (producer is FUNAI):
http://elektrotanya.com/files/forum/.../Supressor.pdf

or:
http://elektrotanya.com/?q=hu/conten...-helyettesites

(first comment, supressor.pdf)

brg

Rumcajs

[johnyp]

Please let me understand wow pi works. We send a pulse to the coil that it's feedback is connected to an opamp that reads the time that field still exists. The opamp's output is always 1 or 0 since it can look for very small voltages on it's input and then we are looking for changes in opamp's output frequency which is a pulse with duration according to the coils field collapse time. Am i wrong ?

[porkluvr]

No, you described a digital system (all ones or zeroes).

What goes on inside your AT90S4433 is mostly digital, but because how it does what it does is determined by its program, we cannot see exactly how it functions just from looking at it.

Your preamplifier is analog. Small voltage variations on the input (LF357-2) will get "magnified" and reproduced at the output (LF357-6).

Read Carl Moreland's "Hammerhead PI" article.
http://www.geotech1.com/pages/metdet...ead/HHv1p5.pdf

Hammerhead is a good platform to get a grasp of PI fundamentals.
Google search for "PI metal detector theory" to get some more descriptions of how PIs works.
(Hmmm, maybe you also need to Google "analog electronics" and "digital electronics", too?)

[WM6]

Read Carl Moreland's "Hammerhead PI" article.
http://www.geotech1.com/pages/metdet...ead/HHv1p5.pdf


Hammerhead is a good platform to get a grasp of PI fundamentals.

.. yes, and allways buy some semiconductor spare parts togheter with your DIY kits too. You never know.

[johnyp]

Yes i agree but i see that the opamp works in compare mode and not in follow mode so the output will always be 1 or 0 depending on the voltage in the INV input which has negative voltage till the field collapses between the pulses.This gives a PWM output. Not ?

[porkluvr]

Yes i agree but i see that the opamp works in compare mode and not in follow mode so the output will always be 1 or 0 depending on the voltage in the INV input which has negative voltage till the field collapses between the pulses.This gives a PWM output. Not ?

What makes you say that this is comparator mode??

The U3 in your detector is configured as a -1000 amplifier. Notice the 560Ω input resistor, and 560kΩ feedback resistor connected between output pin-6 and input pin-2. That is how an inverting amplifier is made.

Notice that there is 2mVp-p in, and 1000mVp-p out in my simulation circuit.
And the output is inverted from the input.
I used the LT1037 simulation model because it adds less error than LF357 would and should be less confusing to see the issue I try to illustrate.

If there was not a feedback resistor, then U3 could be called a comparator. But there is, so it's an analog amplifier.

You should read up on some opamp theory. Here you go: http://www.pic101.com/op_amp.htm