Mosfet small resistor on source

[PovTruffe]

Not wrong, just not great IMHO. Things like this, quote "When the
microcontroller detects that the output current has dropped below the
required level, it pulses the MOSFET four times in rapid succession to
boost the voltage output. Four pulses are used to generate more current
flow and to speed up the rise time under load."

That's a sledge hammer method, requires a larger inductor than than a
cycle-by-cycle design would need because it might enter continuous
conduction mode (CCM), saturate and the FET goes *PHUT*. Maybe that's
why they have R7 in there.

I feel so weak in analog electronics :-((
I have read a few things about boost converters but still dont understand
them very well.

Even my simplest and cheapest (discrete) converter designs all the way
back to the early 90's have an inner current mode control loop. It's
state of the art but this one doesn't have it.

OK, do you think you can supply a schematic of one of your "simplest and
cheapest designs" ?

Watch out for that 1N914, 75V is riding it at the limits.

OK about this, thanks. However I dont think such a high voltage would be
required for say 10 mA max through the body (between the two hands).
The resistance depends of course on the electrodes, the body resistance, etc.
I will test all this and hope I will not die from electrocution!

No surprise there. Here is another little issue with this circuit: C2 is
1uF, C6 is only 10nF. So if the load would suddenly become low impedance
C2 rapidly dumps its charge into C6, pin 6 can go above VDD, then into
its parasitic substrate diode with gusto ... *POOF*

I did not seee this one...
Looks like a clamping diode on the input pin would be sufficient ?

I dont think this is the cause of the A/D problem though. It might just be a
register configuration mistake or a simulation issue.

Thanks for all this usefull information.

 

[Joerg]

I feel so weak in analog electronics :-((
I have read a few things about boost converters but still dont understand
them very well.

Don't worry, we have all started that way.

OK, do you think you can supply a schematic of one of your "simplest and
cheapest designs" ?

Unfortunately not, then I'd get into major trouble. Those are now being
mass produced and are within larger circuits. It is the result of
consulting assignments by clients. But I can tell you this much: They
are mostly CMOS Schmitt trigger inverters used as oscillators, and then
these have several transistor-based loops. The inner one is a current
cut-off so you can live with a fairly minimally sized inductor. Then the
voltage loop, and finally a max-current cut-off for protection. Or vice
versa, current controls the converter and there is a max voltage
cut-off. It really doesn't get much cheaper than this.

If you want to learn the ropes you can do what I did: Read the old
Unitrode app notes cover to cover. All of them, several times until they
almost come out of your ears. They are now part of Texas Instruments.
Then build a few converters using their chips. If you are into low cost
designs the next step would be to replace the chip with your own
discrete circuitry.

OK about this, thanks. However I dont think such a high voltage would be
required for say 10 mA max through the body (between the two hands).
The resistance depends of course on the electrodes, the body resistance, etc.
I will test all this and hope I will not die from electrocution!

Be careful. 75V can give you quite a zing. It is, from a safety agency
point of view, not a harmless low voltage.

I did not seee this one...
Looks like a clamping diode on the input pin would be sufficient ?

Yes, but it would need to clamp before the substrate diode comes on. How
to do this depends on how the ADC in there works and that I don't know
because I never used Microchip controllers.

I dont think this is the cause of the A/D problem though. It might just be a
register configuration mistake or a simulation issue.

Thanks for all this usefull information.

Bonne chance et soyez prudent

 

[PovTruffe]

The schematic is ugly.

They never say how much voltage it's supposed to generate.

Their design is not supposed to supply more than a few mAmps.
I supposed that the application note was there just to give a few guidelines.
The voltage then depends on how we use the design: which electrodes (wet
or dry), body resistance, points of entry and exit, etc.

Two of the caps are shorted.

Which ones ?

They never identify the waveforms.

Let it run, then short the electrodes. This is the suicide bomber of
PICs.

I will have to order a few PIC samples and then make some tests.

Just implement their flowchart!

Yes, it is a simple design when it comes to programming. But the analog
part, though also simple is a bit hard for my analog skills.

 

[Joerg]

???

Yesterday there were lots of soccer TV ads on there, today there aren't
anymore.

Hey, we've got genuine Borussia Dortmund soccer beer glasses here. 6000
miles away from their stadium

 

[Martin Riddle]

Datasheet says they are high-impedance inputs as default.
PIC are always that way.

With a soft pull-up.

Cheers

 

[Grant]

Measure some! Be careful to not blow out your ammeter.

Why is it one must keep warning about this? Put a shunt in circuit and
measure voltage across the shunt resistor. Perhaps that my first
electronics job was with power converters I learned this simple one early?

That's why I'll solder a some Rs in parallel for a shunt if I cannot
find a proper xx mV/yy A shunt in a box It's been years since I
used multimeter to measure current direct.

Grant.

 

[Grant]

Not if it's NiMH cells, those can develop several amps of current. But
R7 would be a rather crummy way to limit power. Unless it was a several
watt resistor type it would go phssssoush and leave a nasty smell
behind. As would the FET unless cooled. Also, it would be sort of "We
are not so sure about the quality of the uC code so let's not trust it" ;-)

The proper way to do this is to write code that is reliable and handles
the start-up situation every time, plus a fuse.

And try the circuit out with uC removed, it should fail safe with no
control from microcontroller. Depending on oscillator type and startup
times, could be quite some ms go by before uC is sane enough to control
a circuit.

Grant.

 

[Grant]

http://ww1.microchip.com/downloads/en/AppNotes/01209B.pdf

Ah, thanks. That circuit knows how to kill itself? Anyway, the MOSFET
hardly turns on with 3V supply, much current and it's a current source?

Not nice. I see now some of the issues others raise.

Grant.

 

[E]

With a soft pull-up.

Pull-ups are off by default (/GPPU bit in OPTION register). And GP2 is
configured as analog input on power up.

 

[PovTruffe]

Ah, thanks. That circuit knows how to kill itself? Anyway, the MOSFET
hardly turns on with 3V supply, much current and it's a current source?

I did not see this one either! Looks like I will have a lot of work before I get
this to work (reliably).

However I intended to use a 5V supply because there will be a LCD display
that works with this voltage only. This should take care of this problem.

Thanks. What else?

 

[PovTruffe]

Let it run, then short the electrodes. This is the suicide bomber of PICs.

The circuit is not supposed to switch all the time, only when some current flows
through the electrodes. However I dont know actually if I will be able to find a
reliable and fast reacting way to detect when the electrodes are being used.
I think I can have a reaction time of a few tens of microseconds, but this is
probably enough to kill a PIC...

What do you think about adding a resistor in series with the analog input ?
This will increase the impedance of the analog source and might increase the
acquisition time, but there is probably a good compromise. Then there are
the internal PIC clamping diodes...

Thanks for all the information anyway

 

[PovTruffe]

That is what i saw; French ad..which matched the TLD..

Sorry I did not know this link could contain an advertisement.
I just see the uploaded image but the ad may not be shown to the originator IP.

I really dislike those invading ads. Next time I will upload a file to a personal
web site.

 

[Michael Karcher]

Unfortunately not, then I'd get into major trouble. Those are now being
mass produced and are within larger circuits. It is the result of
consulting assignments by clients. But I can tell you this much: They
are mostly CMOS Schmitt trigger inverters used as oscillators, and then
these have several transistor-based loops. The inner one is a current
cut-off so you can live with a fairly minimally sized inductor. Then the
voltage loop, and finally a max-current cut-off for protection. Or vice
versa, current controls the converter and there is a max voltage
cut-off. It really doesn't get much cheaper than this.

Is the 2-transistor-pseudo-thyristor line-powered flyback SMPS I recently
fixed also your design? Sounds even more cost-optimized than what you
sketch over there (OTOH, I know that flyback primary SMPS and boost
converter is something different).

Regards,
Michael Karcher

PS: I had to replace a broken diode in the secondary, not related to the
minimal primary-side circuit.

 

[Tim Williams]

OK, do you think you can supply a schematic of one of your "simplest and
cheapest designs" ?

Probably looks something like this
http://webpages.charter.net/dawill/tmoranwms/Circuits_2010/RegBO.png
Hmm, this schematic doesn't illustrate the current limiting the actual
model contains.

Tim

 

[Joerg]

And try the circuit out with uC removed, it should fail safe with no
control from microcontroller. Depending on oscillator type and startup
times, could be quite some ms go by before uC is sane enough to control
a circuit.

Yep

The worst one I had on the table here was a line powered circuit. Probed
the uC, that made the oscillator stumble ... phut ... *KABLAM*

 

[Joerg]

Is the 2-transistor-pseudo-thyristor line-powered flyback SMPS I recently
fixed also your design? Sounds even more cost-optimized than what you
sketch over there (OTOH, I know that flyback primary SMPS and boost
converter is something different).

Not guilty, your honor. I didn't do that one. Seriously. No, no, I
really didn't ...

But mine often do evoke utterances of disgust during design reviews.
Comments such as "Eeuw! I think I'm going to get sick".