Review of schematic

[ahacking]

Version 6 of the schematic.
saw_v6.pdf (102.5 KB)

Thanks

Adam

[Sprig]

@ahacking

As far as I undertsand the UNL2803 has an internal flyback diode. This is why I removed the previous diodes.

Okay, I did not look up the data sheet for every part you used.

You could probably use the same optocoupler LTV-847 if you put the buzzer on a different driver. Might save some cost and space. Just something for you to think about.[quote=“ahacking, post:103, topic:6507”]
Is this possible to run simulations from KiCAD or do most people draw circuits in Spice and run in that environment?
[/quote]

I know Spice is to be supported, but I have not looked into it. I still use LTSpice separately. You will only get out of Spice the effort that you put into it.

I use it mostly to model transistors to ensure that I am ordering the correct one for the intended circuit design. Designing circuits with transistors is not my day job, so I forget, or get backwards, which specifications matter most for the application.

[Sprig]

@ahacking

Is D1 not required because the inductance of the fans is so small?

I don’t work with electric motors in the electronics I work with.

However, the problem that I potentially see is that one flyback captured can throw off the current in the opposite fan.

I do reserve the right to be horribly wrong. I’ve never had the need to put the leads on a scope to see what is actually going on. Now, doing so is on my list.

@ahacking

Also is it good practice to pull the gate of the transistor to 0V with a high resistance ground connection?

A look at the datahsheet for the PQP30N06L shows the Gate - Body Leakage current to be 100nA Max. One can use Ohms Law to figure the input impedance.

FETs operate on voltage, not current. It is my opinion that the 220Ohm resistor is not needed between the MC and the gate.

However, the 100KOhm resistor to ground is needed to sweep the charges out of the capacitve junction of the FET.

The FET will charge just like a small capacitor. Since the MC also has a high input impedance, there is no place for the charge to be swept out of the FET if this resistor is left out. The result is that the FET will turn off with a very slow fall time; or in fact it could even remain in the “on” state.

Valu

[Sprig]

@ahacking

I see this can handle 60V and 30A and be…

…likely to be VERY expensive!

You only need a device rated for ~200mA depending upon the fans you actually use.

[1.21Gigawatts]

@Sprig

I suspect that that transistor will not fully turn on with a logic level signal.

I suspect you don’t fully understand bipolar transistors. Of course it would turn on. His schematic shows a BC547 but in his post he says he wants to use the TIP50. One is rated for 100ma the other 400V, neither one is really appropriate for the task.

@Sprig

D1 appears unnecessary.

Why is that?

@Sprig

And, I’m confused what you are doing with the optocoupler.

I don’t understand why Adam keeps changing things. Those optocoupler transistors were connected to ground until he changed them to 5V. And where did those resistors between the optocoupler and the Pi come from?

@ahacking

I see the ground was not connected to the cathode (+5V instead) on the output side of the optocoupler

Transistors do not have cathodes! Perhaps do a little research on transistors.

@Sprig

You could probably use the same optocoupler LTV-847 if you put the buzzer on a different driver.

He had the LTV-847 there before changing to the transistor array. Why use the opto if it does not provide enough channels? Not to mention the need for additional discrete components.

@Sprig

Might save some cost and space.

An 18 pin package vs a 16 pin package which requires 4 additional diodes? Both parts cost about $1.

@ahacking

I also see some comments showing some spice integration with KiCAD?

In my personal opinion Spice support in KiCad is a total waste of development effort, especially at this stage of KiCad’s development.

@Sprig

One can use Ohms Law to figure the input impedance.

No, one cannot!

@Sprig

It is my opinion that the 220Ohm resistor is not needed between the MC and the gate.

An opinion that is entirely incorrect.

@Sprig

Since the MC also has a high input impedance,

But if the MC pin is driving a transistor surely it is configured as an ouput.

@Sprig

…likely to be VERY expensive!

Yes, $1.11.

The PQP30N06L is certainly overkill for your requirements. I would suggest something like the Microchip VN2410L.

@ahacking

Temp monitoring would be great but from what I have read that requires a DA converter for the Pi.

Check out the DS18B20 and DS18S20.

And yes, your pullup resistors for the Pi inputs should be to 3V3.

Edit: And if you plan to use a MOSFET for driving the fans change the schematic to show a MOSFET transistor and not a bipolar transistor.

[Sprig]

WOW, NIT PICKY much tonight?

I’m only going to back nit a few of them. The whole point was to NOT shove down a bunch of information and write a book at the same time. Most of my comments were meant as general observations.

D1 appears unnecessary.

1. The 24V line is already reverse polarity protected by the new D1.
2. There are two motors in parallel in the same circuit. Instead of the diode, I’d run 4 capacitors from each motor lead to chassis ground.

@1.21Gigawatts

An opinion that is entirely incorrect.

Not entirely, but I suppose I didn’t mention at expected human pressing button speeds.

@1.21Gigawatts

But if the MC pin is driving a transistor surely it is configured as an ouput.

I didn’t look at every datasheet for every device on every pin . My text was based upon my current understanding that the Pi GPIO pins are internally current limited to source 20mA.

If this is not the case for the Pi GPIO pins, then thanks for letting me know I got it confused with some other product.

[1.21Gigawatts]

@Sprig

WOW, NIT PICKY much tonight?

No, besides it’s only 4pm here.

@Sprig

The 24V line is already reverse polarity protected by the new D1.

Completely irrelevant.

@Sprig

2. There are two motors in parallel in the same circuit. Instead of the diode, I’d run 4 capacitors from each motor lead to chassis ground.

This is the typical approach to reduce emissions (EMI) generated by motors that are driven by a DC voltage. Here we are discussing driving motors with PWM. Those capacitors are just going to appear as an additional load for the transistor driving the motors but will offer no protection whatsoever for the transistor.

@Sprig

Not entirely,

Yes, entirely.

@Sprig

I didn’t look at every datasheet for every device on every pin .

You use this excuse a lot. May I suggest that if you are not familiar with, and not willing to look at the datasheet for, a part then refrain from commenting on the part.

@Sprig

Since the MC also has a high input impedance

@Sprig

My text was based upon my current understanding that the Pi GPIO pins are internally current limited to source 20mA.

[Sprig]

Pi Outputs

If you don’t like it from my point of view, then fine, consider me an idiot and put me on ignore.

16mA, not 20…

[Sprig]

@1.21Gigawatts

You use this excuse a lot.

Yea, in this post YES.

If you would care to purchase the EXTRA DATA on my account, I would happily download every datasheet mentioned. Then I would not have that excuse.

You can happily send me your credit card number so that I can purchase this extra data plan for my internet this month.

Thanks in advance!

[1.21Gigawatts]

The actual value of the GPIO current sink/source is not the issue. In one statement you refer to the pin as having a “high input impedance” and then in the next it’s a XXma current source.

@Sprig

If you would care to purchase the EXTRA DATA on my account, I would happily download every datasheet mentioned.

You’re a funny guy Jim!

I was only suggesting that if you do not have access (for whatever reason) to the necessary information then perhaps refrain from giving technical advise, especially to someone who appears to be willing to blindly follow what could be misleading advise.

[Sprig]

@1.21Gigawatts

The actual value of the GPIO current sink/source is not the issue. In one statement you refer to the pin as having a “high input impedance” and then in the next it’s a XXma current source.

That is exactly what some of the pins on chips can do; the Atmega328 MC and even the stand alone I2C PCF8574 has this capability.

[quote]The device features an 8-bit quasi-bidirectional I/O port (P0–P7), including latched outputs with high-current drive
capability for directly driving LEDs. Each quasi-bidirectional I/O can be used as an input or output without the use
of a data-direction control signal.[/quote]

[1.21Gigawatts]

If you are trying to state the obvious, that some devices have pins which can be configured to be either inputs or outputs, then well done!

But we both know that the above statements were in reference to a pin that was driving the gate of a FET, therefore an output. Making this statement …

@Sprig

The FET will charge just like a small capacitor. Since the MC also has a high input impedance, there is no place for the charge to be swept out of the FET if this resistor is left out.

… total nonsense.

In order to get this thread back on topic I will henceforth reply only to Adam, or to correct any misguided advise he might receive.

[ahacking]

Hi Jim and Rob,

Thanks for the all help and dedication to this thread.

You both clearly have the best intentions but I’m very sorry to you both arguing. From my perspective, a discussion is helpful as I get to see a few points of view and how these are rationalized in terms of the circuit.

I’ll do a better job of questioning changes and checking over the schematic before the next post.

Thanks again and all the best

Adam

[ahacking]

@1.21Gigawatts

DS18S20

Thanks. This looks like a great solution. I’ll look into it.

[ahacking]

@1.21Gigawatts

VN2410L

Datasheet overview VN2410L
Categories Discrete Semiconductor Products
Transistors - FETs, MOSFETs - Single
Manufacturer Microchip Technology
Series -
Packaging ? Bulk ?
Part Status Active
FET Type N-Channel
Technology MOSFET (Metal Oxide)
**Drain to Source Voltage (Vdss) 240V **
**Current - Continuous Drain (Id) @ 25°C 190mA (Tj) **
Drive Voltage (Max Rds On, Min Rds On) 2.5V, 10V
Vgs(th) (Max) @ Id 2V @ 1mA
Gate Charge (Qg) (Max) @ Vgs -
Input Capacitance (Ciss) (Max) @ Vds 125pF @ 25V
Vgs (Max) ±20V
FET Feature -
Power Dissipation (Max) 1W (Tc)
Rds On (Max) @ Id, Vgs 10 Ohm @ 500mA, 10V
Operating Temperature -55°C ~ 150°C (TJ)
Mounting Type Through Hole
Supplier Device Package TO-92-3
Package / Case TO-226-3, TO-92-3 (TO-226AA)

This looks like it supports 190mA continuous drain current so I would need 1 per fan to be safe.

How about using TN0606N3-G

Categories Discrete Semiconductor Products
Transistors - FETs, MOSFETs - Single
Manufacturer Microchip Technology
Series -
Packaging ? Bulk ?
Part Status Active
FET Type N-Channel
Technology MOSFET (Metal Oxide)
Drain to Source Voltage (Vdss) 60V
Current - Continuous Drain (Id) @ 25°C 500mA (Tj)
Drive Voltage (Max Rds On, Min Rds On) 3V, 10V
Vgs(th) (Max) @ Id 2V @ 1mA
Gate Charge (Qg) (Max) @ Vgs -
Input Capacitance (Ciss) (Max) @ Vds 150pF @ 25V
Vgs (Max) ±20V
FET Feature -
Power Dissipation (Max) 1W (Tc)
Rds On (Max) @ Id, Vgs 1.5 Ohm @ 750mA, 10V
Operating Temperature -55°C ~ 150°C (TJ)
Mounting Type Through Hole
Supplier Device Package TO-92-3
Package / Case TO-226-3, TO-92-3 (TO-226AA)

[1.21Gigawatts]

Yes, the TN0606N3-G looks like a good choice. The VN2410L was actually capable of more than 190ma but it would require a higher gate voltage. I also didn’t notice it had been obsoleted.

[straubm]

The case/mounting might be a factor to consider. Rds max. is shown in the datasheet to be 15 (!) Ohms at 3V gate drive dropping to 2 Ohms at 5V gate drive.
Your realistic gate drive voltage will be somewhere in between.

TO-92 packages typically have a thermal resistance around 160°/(W dissipated) when mounted without a heatsink and without any forced airflow.
I do not know how much current your fans draw, but let’s assume 400 mA for both. Let’s further assume a target of 40° temperature rise above ambient. This amounts to 0.25W dissipated. So you can tolerate an Rds on max. around 1.5 Ohm.

[EDIT] Of course that goes for 100% PWM (always on). You have to derate that for the PWM cycle you anticipate. So @50% PWM 3 Ohms is ok, and so on.

@ahacking My intention is to bring your awareness to the issue.

[1.21Gigawatts]

Yes, that is correct. I can’t find it now but I thought he stated, or it was on the schematic at one point, that the fans were 100ma each.

The datasheet gives a figure of 132°C/W, for a 40°C rise at 200ma it could dissipate 0.3W for an Rds(on) of 7.5 Ohms at 100% DC.

But I agree, a little more headroom would be better.

The FQP30N06L that you picked out earlier is certainly a lot more capable and cheap enough. There’s also the PHP79NQ08LT.

[straubm]

I agree, the FQP30N06L might be the way to go: 62.5°C/W, Rds on max. some 0.035 or so, gate threshold max. 2.5V. Perfect. No worries about temperature or anything.

The PHP79NQ08LT is even better regarding Rds on and gate threshold.

Toss the dice …

[EDIT] @ahacking: you might consider using a proper MOSFET symbol for Q1 in your schematic. Check the pin numers against the device you intend to use, however, as there are different configurations out there…

[EDIT 2] @ahacking in the schematic V6 you ask whether the GPIO inputs should be driven by 3V3: yes.

[ahacking]

Thanks. The fans are 0.1 amp each. Will change the symbol to a MOSFET