Quiz Time!

[cadstarsucks]

Here is a real life application that I did that has been flying in small plane cockpits for 20 years... lets see who can figure it out first!

needed: an electronic replacement for a 40 year old "drag cup" instrument.

background: a drag cup is a set of magnets spun by a 3 phase tach generator that drag on an iron disk fighting a return spring.

restriction: the ONLY source of power is a tachgen that puts out from 2Vpk@1Hz to 80Vpk@70Hz and there is no real unit to unit consistency in the voltage so don't be trying that

D.

 

[Sceadwian]

The answer is of course, 42.

 

[Pommie]

What was the question? I can't identify one.

Mike.

 

[cadstarsucks]

The answer is of course, 42.

Thank you for the spit take!

D.

 

[cadstarsucks]

What was the question? I can't identify one.

Mike.

It is not a question so much as it is a challenge.

20 years ago a sales guy ask me why we kept no bidding a particular job. I looked briefly at the spec and said I did not know it does not seem bad. Unusual, perhaps, but not impossible. (there are very few things that are actually impossible, impractical perhaps but not impossible)

I took it to my boss and asked and he said there was no power available to run the circuitry. Once I had picked my jaw up off the floor I replied "But Tom, our signal source puts out 70-100W full scale. We have plenty of power, we just need to figure out how to use it" Once he had picked his jaw up off the floor he agreed and told me to do it. So I did.

I believe the best thing for EE students and rusty engineers is to have a good challenge to wrap their minds around. So here one is. A three phase frequency meter that runs off the signal source from 1-80Hz at 2-80V peak.
I know it is possible because I did it 20 years ago.

D.

 

[House0Fwax]

The way I see it is, there are two criteria;
1. The tachometer itself.
2. ( the big one ) the power supply.

I once took a 'drag cup' tachometer apart and stuck a reed switch near the spinning magnets. This signal I then fed to a very inacurate frequency counter that I'd made out of some 7 segment led displays, some 4026's a 4017 and a 555. Then I tried to calibrate it using a pcb drill.
Utterly futile attempt, but great fun.
... So that's part one sorted.
As for part two, all I can come up with is charging a capacitor and using some kind of regulation.

**Goes to back of class**

 

[Pommie]

Is it not just the same as displaying the mains frequency which is done by many clock circuits? IE rectify it to power the device and feed an unrectified signal to the counter. Maybe I missed something but this seems trivial with todays devices.

Mike.

 

[cadstarsucks]

The way I see it is, there are two criteria;
1. The tachometer itself.
2. ( the big one ) the power supply.

I once took a 'drag cup' tachometer apart and stuck a reed switch near the spinning magnets. This signal I then fed to a very inacurate frequency counter that I'd made out of some 7 segment led displays, some 4026's a 4017 and a 555. Then I tried to calibrate it using a pcb drill.
Utterly futile attempt, but great fun.
... So that's part one sorted.
As for part two, all I can come up with is charging a capacitor and using some kind of regulation.

**Goes to back of class**

HOT ****! I didn't think anyone would have heard of those silly old things, never mind seen one!

Neat idea, but the point was that they were no longer in production and manufacturers needed to support planes in the field.

It actually drove a mil spec meter movement, but you are correct in that it could have been an LCD, although at the time they were quite pricey.

You are on the right track with the power supply... the oddity there is running at both sub 2V AND over 60V...

Extra points for obscure knowledge!

D.

 

[JimB]

OK now that you have defined the indicator as a moving coil meter, my first idea would be to rectify the 3-phase signal, giving a pulsing DC at 3 times the original signal frequency.
That signal can then be used to drive a pulse shaper giving a defined pulse width and amplitude.
Low pass filter the pulses to drive the MC meter and the deflection will be proportional to the frequency (and RPM).

Generating the power for the pulse shaper is a bit more problematic at low speeds.
Does the indicator really NEED an 80:1 turn-down ratio? ie if fullscale is 8000rpm, do we really need a sensible indication at 100rpm? or, can we make the bottom end of the scale say 200 or 300 rpm where we will probably have plenty of volts to use for a supply?
At 2v peak, when we rectify, we will lose two "diode drops" from the supply, so we would have to use germanium or schottky diodes to maintain as high a supply voltage as possible.
So we need to make our pulse shaper (probably a monostable multivibrator type circuit) run reliably from about 1.5 volts.

We could use a step-up transformer to give us some more volts to play with. But the problem here would be with the low frequency, it would need a lot of iron to work at 1hz, but we don't need to transfer a lot of power because our circuit is inherently low power so we may get away with a reasonably sized transformer.

JimB

 

[cadstarsucks]

OK now that you have defined the indicator as a moving coil meter, my first idea would be to rectify the 3-phase signal, giving a pulsing DC at 3 times the original signal frequency.
That signal can then be used to drive a pulse shaper giving a defined pulse width and amplitude.
Low pass filter the pulses to drive the MC meter and the deflection will be proportional to the frequency (and RPM).

Indeed yes. Moving coil was what I was told to use. Low pass filters are far too finicky, and there is afar better way to achieve a defined pulse...

Generating the power for the pulse shaper is a bit more problematic at low speeds.
Does the indicator really NEED an 80:1 turn-down ratio? ie if fullscale is 8000rpm, do we really need a sensible indication at 100rpm? or, can we make the bottom end of the scale say 200 or 300 rpm where we will probably have plenty of volts to use for a supply?

The part was actually a 2" OD cockpit can. Honestly, I am not trying to be obtuse, I just did not count on so many different directions. It actually had to be in calibration at 2-80RPM as it replaced existing units in thousands of cockpits.

At 2v peak, when we rectify, we will lose two "diode drops" from the supply, so we would have to use germanium or schottky diodes to maintain as high a supply voltage as possible.
So we need to make our pulse shaper (probably a monostable multivibrator type circuit) run reliably from about 1.5 volts.

As I recall I did indeed use 200V germaniums since HV schottkys did not exist at the time...

We could use a step-up transformer to give us some more volts to play with. But the problem here would be with the low frequency, it would need a lot of iron to work at 1hz, but we don't need to transfer a lot of power because our circuit is inherently low power so we may get away with a reasonably sized transformer.

JimB

While I did not use any iron... today I think I would put a small boost circuit in to get up a bit. I think I actually had a switched cap doubler that would startup at 1V or so to allow the opamp to run...but nothing would stand up to 100V...

D.

 

[JimB]

Low pass filters are far too finicky,

I was thinking in terms of a simple RC network, probably not needed anyway as the MC meter is inherently a low pass filter.

and there is afar better way to achieve a defined pulse...

Ok then, what is it?

...but nothing would stand up to 100V...

Some zeners to limit the high voltage would be a reasonable idea, but we need to be aware of the series resistors which will steal some of our supply volts at low RPM.
We could use a modified "series regulator" where the pass transistor is biassed hard on at low input voltage and turns off as the supply voltage rises.

JimB

 

[cadstarsucks]

I was thinking in terms of a simple RC network, probably not needed anyway as the MC meter is inherently a low pass filter.


Ok then, what is it?


Some zeners to limit the high voltage would be a reasonable idea, but we need to be aware of the series resistors which will steal some of our supply volts at low RPM.
We could use a modified "series regulator" where the pass transistor is biassed hard on at low input voltage and turns off as the supply voltage rises.

JimB

Hmmm...at 2V in we are losing 0.6V on the germanium bridge, what would the drop out be and how much voltage can it withstand?

D.

 

[cadstarsucks]

Two days with no response...shall we just jump to what went to production for discussion?

D.

 

[JimB]

Yes

JimB

 

[House0Fwax]

Yes please.

 

[cadstarsucks]

Yes please.

Ok...

The 3 phase tachgen puts out 2V peak to 60 or 80V peak from 1 or 2Hz (I do not remember exactly it was 20 years ago) and being aircraft actually did need headroom though I did not think that mattered much considering the complexity I had brought in to the discussion with out that detail.

We have already established that I used 200V germaniums in a half wave three phase bridge since 20yrs ago that was all that was available...even now 200V schottkys are new tech.

The regulator is a bit of a trick as it is a discrete design that uses a somewhat unknown type of part, a DEPLETION mode mosfet, as a pass element. I was expecting the discussion to continue to the "so what the hell did you use" stage on that one. As a normally ON device it allowed me to hang a zener on the drain (circuit) side as a voltage sensor. I think it was a 4.3V zener running at 0.1mA into the base of an NPN. The collector to the gate of the FET with source drain resistor for a 3V low drop out supply. The BSP149 (well the TO92 version of it at least) gave me low drop out with 200V survival.

A 7660 or some related switched capacitor inverter with a guaranteed startup of 1.5V allowed the circuit to start running at 2V peak.

A frequency to voltage converter was also pieced together as there was no such thing that would run reliably at that voltage at that time. More creativity involved.

A 1.2V shunt reverence, remember we are talking military/aerospace, supplied stability.

Some HCMOS gating would run at the low voltages required reliably. A 74HC14 schmidt with a negative feedback resistor to reduce the hysteresis, a resistor bypassing the bridge, and a resistor to the positive rail to pull the zero to the middle gave me my input signal.

The input signal from the HC14 to the select pin on a 74HC4053 analog switch alternately connecting a stable capacitor from the 1.2V reverence to an integrator. Simply an OPAMP negative input with a parallel RC in the feedback and the positive input to ground. The transfer function becomes Vo=Rf*Cref*Vref*Fin.

The movement drive, while not really what I was looking to highlight, was OPAMP current sourced. This was done because over the military temperature range of -55/+71C copper varies quite a bit. I used to know the copper, nickel and platinum R/T curves but that was many years ago. Suffice it to say that it was probably around a 20% change that could not be tolerated in an instrument rated at 3% worst case over temperature.

I will be looking forward to your comments.

D.

 

[cadstarsucks]

What, no comments?

Dan

 

[Hero999]

I don't think anyone noticed that you edited your post.

My only question is what sort of current did your circuit end up drawing?

I'm much too young to remember from first hand experiance but I've been told that old MOSFETs can't dissipate much power and I'm certain the MOSFET in the datasheet you posted wasn't around back then.

 

[cadstarsucks]

I don't think anyone noticed that you edited your post.

Ah that makes sense.

My only question is what sort of current did your circuit end up drawing?

I'm much too young to remember from first hand experiance but I've been told that old MOSFETs can't dissipate much power and I'm certain the MOSFET in the datasheet you posted wasn't around back then.

Actually the silicon was, the part I use was the same die in a TO92 package, but they appear to have discontinued that variant.

The supply current ended up at around 1.5-2.5mA depending on the reading on the 1mA meter movement.

The power dissipation ratings have nothing to do with the age or technology of the silicon, they depend on how large the die is and how well you can get the heat out of the die. You can get a TO220 transistor rated for a 100W but if you do not heatsink it well it is only good for around 2W.

There are better power packages these days like the SSOT23 (super sot) that basically give you much more metal in the lead frame to conduct the heat from the die and the power pad parts that connect the die to a copper slug which then gets connected to some form of heatsink.

Dan

 

[periasamy]

hello sir, i need +5v power supply circuit and 28 to 42.5 volt 2 amps power supply circuits please send this details useful to my project work