Surrendering to the 555 Timer Option

I don't have an NPN or NFET in the BOM, just a PNP and a PFET. I could use the PNP as pair of diodes for a wire OR, but that's going to put a minimu m Vgs of about 0.7 volts which might be too much and turn it on with only a 3 mA load. There is nothing in the data sheet of the HC4060 to make me th ink it can source the 3 mA directly without more voltage loss than I'd pref er. I was going to say I could drive the load directly from the 555 timer, but the CMOS version has more voltage drop than I'd like.

As I think about this, I'm starting to like the HC4060. I can use the spar e section of JK FF to turn the 0, 1, 2, 3 count into a single pulse of each four. Using different taps will allow patterns which might be better than a single beep every four seconds. I am having trouble finding a 74HC4060 spice file. Nexperia has something called hc_tnomi.cir, but it doesn't see m to be a functional model, just timing or drive capability. When I tried to bring it into LTspice it generated a 5 pin model with the pins named 0,

1, 2, 3, 5.

I see CD4060 models mentioned people have created, but the files seem to be hosted on dead sites.

I found files in the LTspice group, but I can't seem to make them work. On e of them is CD4060Bg-model.txt which contains ".SUBCKT CD4060Bg". Should this be renames as a .lib or some other file? The .asy file and a test sch ematic are provided. It can't seem to find the guts complaining about an u nknown subcircuit.

I do find LTspice to be one of the most confusing to use tools I've ever en countered. It's one of those things that you have to stay fresh with or th e details just kill you.

I'm really liking this approach though, so I'm willing to put up with a bit of effort to simulate this circuit.

--

  Rick C. 

  -- Get 1,000 miles of free Supercharging 
  -- Tesla referral code - https://ts.la/richard11209

program

n low

ybe

e an

? I am

oes

ith

case

be

ate

so

here

a

and a

PNP

many

l

I think they are doing all that in the software or at least in the Arduino. I've only been to a couple of meetings and the software isn't talked abou t in much detail.

--

  Rick C. 

  +- Get 1,000 miles of free Supercharging 
  +- Tesla referral code - https://ts.la/richard11209

r

in low

maybe

be

with

s

ight

t

ould

look

at

U and

se

e

e PNP

y low

ow

'll

hey

e for

no

the

Medical equipment has to be proven to not harm the patient. It is not requ ired to otherwise be fail safe. Military gear often has to continue workin g in spite of any single failure. That's a different animal indeed.

I don't know what the restrictions are on this project. It is entirely pos sible it will never produce anything other than a hobby project. There are any number of alternative ventilator designs that simply were never produc ed for who knows what reason. One question I have not asked is even if the y design this device and get someone to produce it, how do they know anyone wants to use it? Have they asked any of the end users what they need? Wh at they require it to do and how it will fit into a medical setting.

Not my problem. I'm just designing an alarm board. They might ask me to l ay out other parts.

--

  Rick C. 

  ++ Get 1,000 miles of free Supercharging 
  ++ Tesla referral code - https://ts.la/richard11209

Can't recall what its function was exactly have to look again, it may have just been setting a PWM base frequency.

The company prolly uses the same blower-board for a number of products; annoying to have to run a clock for that from the processor up to the board and re-program processor-derived clock for different motor characteristics on different products. Slap a 555 on there and change one resistor for a different machine.

How do ya prove a product doesn't harm the patient, anyway? It could become sentient and hit them in the face with a hose out of spite! I mean sure it seems very unlikely but anything could happen, I guess.

Some kind of logical positivism-thing

g

se the PNP as pair of diodes for a wire OR, but that's going to put a mini mum Vgs of about 0.7 volts which might be too much and turn it on with only a 3 mA load. There is nothing in the data sheet of the HC4060 to make me think it can source the 3 mA directly without more voltage loss than I'd pr efer. I was going to say I could drive the load directly from the 555 time r, but the CMOS version has more voltage drop than I'd like.

are section of JK FF to turn the 0, 1, 2, 3 count into a single pulse of ea ch four. Using different taps will allow patterns which might be better th an a single beep every four seconds. I am having trouble finding a 74HC406

0 spice file. Nexperia has something called hc_tnomi.cir, but it doesn't s eem to be a functional model, just timing or drive capability. When I trie d to bring it into LTspice it generated a 5 pin model with the pins named 0 , 1, 2, 3, 5.

be hosted on dead sites.

One of them is CD4060Bg-model.txt which contains ".SUBCKT CD4060Bg". Shoul d this be renames as a .lib or some other file? The .asy file and a test s chematic are provided. It can't seem to find the guts complaining about an unknown subcircuit.

I got the 4060 to simulate, but it is really, really slow. I've been simul ating the entire design for around 150 seconds of simulation time and not e ven noticed the delay. This model is taking around 1 second for a millisec ond of simulation time. Not workable with a clock running slower than 10 H z. Maybe I'll just trust that the clock circuit works and rig up my own mo del. I really only need the bottom few bits.

As I said before, that's the bother of using LTspice. It can be so hard to find workable models.

I am liking the 4060 for this though. I think I can find good ways to deal with the decoding and output drive.

--

  Rick C. 

  --- Get 1,000 miles of free Supercharging 
  --- Tesla referral code - https://ts.la/richard11209

Huh !

Did you look at the model I posted ?

You don't need the diode for 3:1 mark space, it does work on 2V in LTSpice, uses an LTC1441 (dual), can drive 13mA, Iq about 5uA - what more do you want ?

The LTC1441 is a bit pricy but you could use any low power RRIO CMOS comp (or op amp that's happy being a comp.).

(Microchip do MCP6543 at $0.35 which has chip enable) - never had much joy with running microchip models in LTSpice - but that wouldn't bother me - and you are not having any joy simulating LT comps in LTSpice)

MK

r...

tried was a comparator with about half a dozen passives. A positive feedb ack resistor moves the Vth up and down on a voltage divider threshold at th e positive input. To get different duty cycles a diode and resistor are pa ralleled with a resistor for the negative feedback to the cap. This is ess entially the same as the two inverter loop, but instead of two inverters to provide different polarities to the resistor and cap a single comparator w ith two inputs do the same job. Instead of reversing the polarity of the c ap relative to the single input, the comparator uses positive feedback to m ove the threshold.

to simulate properly in LTspice. The cap should be charging to the positiv e input, but at lower Vcc it trips much sooner and mucks up the duty cycle a lot. I can't figure this out at all. I've tried every single comparator in the LTspice library thinking a voltage reference was getting in the way or something. None of them work properly at the low end of the specified voltage range.

Yes, thanks for your design. It does use two comparators with a number of discretes (8 or 9?) even if both parts are in one package, and has the prob lem in simulation I mentioned where the negative input from the capacitor w ill trip before it reaches the positive input level. That happens with eve ry LT comparator I tried which was every one in the library.

My design has a dual comparator with internal reference already for level s ensing of the Vin and the Vbb on the supercap. Not that that means much. I suppose two different comparators could be used on the same design as opp osed to an entirely different chip or the same comparators with a voltage r eference added. They make comparators in quads, right?

The 4060 seems like the best bet of all I've seen so far. Too bad I can't find a decent model. The comparator is a good solution too, but it really bugs me that I can't simulate it properly. There is something wrong that I am missing.

--

  Rick C. 

  --+ Get 1,000 miles of free Supercharging 
  --+ Tesla referral code - https://ts.la/richard11209

I don't see a problem in LTSpice with the LTC1441 - and I carefully checked the inputs at switching time - all looks OK.

The Microchip model for the MCP6541 won't run in LTSpice. It runs in Microchip's simulator - current (average) running = 3uA. Max output current isn't great - you would need a MOSFET.

If you use the MCP6543 with enable the BOM is: 1 chip, 4 resistors and 1 cap, plus the "power" device. Almost no change in F from 2V to 5V, no diodes.

It should work with any RRIO comparator - but the Microchip one looks pretty keen value unless you go Chinese.

MK

wer...

I tried was a comparator with about half a dozen passives. A positive fee dback resistor moves the Vth up and down on a voltage divider threshold at the positive input. To get different duty cycles a diode and resistor are paralleled with a resistor for the negative feedback to the cap. This is e ssentially the same as the two inverter loop, but instead of two inverters to provide different polarities to the resistor and cap a single comparator with two inputs do the same job. Instead of reversing the polarity of the cap relative to the single input, the comparator uses positive feedback to move the threshold.

m to simulate properly in LTspice. The cap should be charging to the posit ive input, but at lower Vcc it trips much sooner and mucks up the duty cycl e a lot. I can't figure this out at all. I've tried every single comparat or in the LTspice library thinking a voltage reference was getting in the w ay or something. None of them work properly at the low end of the specifie d voltage range.

S r

of discretes (8 or 9?) even if both parts are in one package, and has the problem in simulation I mentioned where the negative input from the capacit or will trip before it reaches the positive input level. That happens with every LT comparator I tried which was every one in the library.

el sensing of the Vin and the Vbb on the supercap. Not that that means muc h. I suppose two different comparators could be used on the same design as opposed to an entirely different chip or the same comparators with a volta ge reference added. They make comparators in quads, right?

n't find a decent model. The comparator is a good solution too, but it rea lly bugs me that I can't simulate it properly. There is something wrong th at I am missing.

I simplified the circuit so there just isn't anything to interfere unless I am missing something. A comparator oscillator with the cap (- input) not reaching the threshold (+ input) before the output switches. This is at 3 volts Vcc and it is about 0.1V shy. At 2 Vcc it is much worse, nearly 0.4 V shy. In some simulations I saw the same problem with a lesser degree on the lower part of the cap waveform.

You don't see this in your simulations?

--

  Rick C. 

  -+- Get 1,000 miles of free Supercharging 
  -+- Tesla referral code - https://ts.la/richard11209

y

our

50 in low

d maybe

an be

se

rs with

his

e

might

to

t .

rst

could

id look

d at

MCU and

erse

the

t

the PNP

n

ery low

How

ou'll

they

ame for

uino

r the

$2

re

quired to otherwise be fail safe. Military gear often has to continue work ing in spite of any single failure. That's a different animal indeed.

ossible it will never produce anything other than a hobby project. There a re any number of alternative ventilator designs that simply were never prod uced for who knows what reason. One question I have not asked is even if t hey design this device and get someone to produce it, how do they know anyo ne wants to use it? Have they asked any of the end users what they need? What they require it to do and how it will fit into a medical setting.

lay out other parts.

IF you are considering a microprocessor, you may want to consider the TI MP S4xx. Ultra low power in sleep mode and relatively low power consumption i n normal running mode. Getting a design approved for medical use, as you pointed out, takes a lot of rigor. That same level of rigor (if not more) is also required for any software device that the system should incorporate. If you are not knowled geable in software and system safety design and analysis techniques, you a re in for significant learning curve. Making an assurance case for system s based on software takes a significant amount of time. For medical system s IEC62304 is a good starting reference. If the device is Class C, the ent ire software development process is used to build an assurance case. Look at MIL STD 882E and DO178C for guidance as well. There is a whole lot more to design assurance than just slapping some code together. j

My thresholds are a different proportion of the supply voltage and at 3V supply it looks fine.

But at 2.5 it starts to go wrong.

When I use your resistor values I get the same problem as you.

So due to my different component values it looked OK at three volts - but I agree with you - the model definitely doesn't work properly.

MK

There was some question as to whether this company

Would release the firmware for the ventilator design they public-domained as it might expose them to potential litigation risks, though it seems they were confident enough in it to do so.

The cost of litigation regarding some kind of firmware flaw in a medical device like a ventilator and showing negligence to the standard required even by tort law would seem enormous. Have to do an independent review of the code and while rotten firmware killing otherwise healthy Toyota drivers was somewhat of a slam-dunk once the glaring issues were well-known, figuring out whether some particular bug (all code has them) in otherwise well-written ventilator firmware contributed to the injury or death of an already sick patient or not seems a tough row to hoe.

uly

Your

0.50 in low

he

,

and maybe

can be

case

mers with

this

the

at might

y to

not

it.

first

I could

did look

eed at

o

e MCU and

everse

t the

ent

d the PNP

own

very low

? How

you'll

ue

at they

Same for

ss

rduino

for the

ed

h $2

were

he

g.

required to otherwise be fail safe. Military gear often has to continue wo rking in spite of any single failure. That's a different animal indeed.

possible it will never produce anything other than a hobby project. There are any number of alternative ventilator designs that simply were never pr oduced for who knows what reason. One question I have not asked is even if they design this device and get someone to produce it, how do they know an yone wants to use it? Have they asked any of the end users what they need? What they require it to do and how it will fit into a medical setting.

to lay out other parts.

MPS4xx. Ultra low power in sleep mode and relatively low power consumption in normal running mode.

t of rigor. That same level of rigor (if not more) is also required for an y software device that the system should incorporate. If you are not knowl edgeable in software and system safety design and analysis techniques, you are in for significant learning curve. Making an assurance case for syst ems based on software takes a significant amount of time. For medical syst ems IEC62304 is a good starting reference. If the device is Class C, the e ntire software development process is used to build an assurance case. Loo k at MIL STD 882E and DO178C for guidance as well.

e together.

This design is so simple I can't see using an MCU. I could see using a Gre enPak. But then I backed off of that too for the same reasons. Why requir e someone to program parts if it really wasn't needed? I think this entire design could be done in a single chip and a diode to prevent reverse curre nt from the supercap to the input power rail.

Maybe I should give the GreenPak a second look. It would be pretty interes ting to do this in one chip. I bet I could get that to simulate ok!!!

I say no MCU, but I'm going to take some time today to explore my options I guess. Another one chip solution sounds pretty good. Hmm... might not be one chip since there will need to be a way to program it and adding the US B chip to do that is pretty common these days.

--

  Rick C. 

  -++ Get 1,000 miles of free Supercharging 
  -++ Tesla referral code - https://ts.la/richard11209