Yepp,
Go to Your local Computer-Repair-Shop,
they probably have a huge box of old CPU Coolers and
might give several of them to You for just a few Dollars.
A 2" X 2" X 1.5" CPU Cooler will dissipate ~10-Watts easily with no Fan,
and ~100-Watts with a working Fan.
You can probably mount 4-FETs on one Heat-Sink,
depending, of course, on the average amount Heat-Dissipation they will be generating,
which should be very little if You use PWM.
The Transistor-God will be pleased.

I'd have to see it to believe it ......... 27-Amp/Hours !!!
Or did You put to many zeros in there ?
If it's actually 27Ah, you won't hurt it with a bunch of small Motors,
but it probably has protection Circuitry built-in that will
limit the max Output-Current so that it won't melt USB Cables and start a Fire.

How are You getting Power out of this "Power-Pack" ???
It's bound to have some type of Current-Limiting-Device.

As per usual, there's no Motor-Amperage-Specs on the Page You Posted.
I have a suspicion that You won't be able to run
more than 1- of these Motors on your USB "Power-Pack".

Keep in mind that PWM will waste much less of your Battery Power than the
cheezy setup you've been trying.
If this is all consolidated into one unit, You only need 1-Oscillator Section,
( that's one Op-Amp, or one 7555-Timer ),
to provide a Triangle-Wave Output to multiple Quad-Op-Amps,
with each Op-Amp driving 1-FET, 1-Motor, all very efficiently.
You can also have 1-Pot that controls the Speed of All, or groups/pairs, of Motors.
If You are interested in doing this, different Op-Amp(s) should be chosen.

You won't have any audible Noise with the Capacitors, the Diode, and the extremely
high PWM-Frequency that was picked in the above Circuit.
If your under ~25, you "might" be able to hear that 15khz,
if You can hear it, just use the formula for calculating the PWM Frequency,
and bump it up to ~25khz,
I'm quite sure You can't hear that high, unless you're actually a Dog.

Not all PWM Circuits are equal, there are a lot of cheezy Circuits on the Internet.

Options ???
I still don't know what you're trying to do ..............
.
.
.

 

Just to emphasize LowQCab's points about USB power banks I think the ah is often untrue. Also when it is true it is the ah rating of the cell which are about 3.7 volts. this is stepped up to 5 volts with a step up converter so that takes the ah rating at 5 volts to 3.7/5 of that value. In practice it will be even less as the step up converter is not 100% efficient. The step up converter will have a limited current output which SHOULD be specified but may not be. (I have just had a look at a 10000 mah (10 ah) usb power bank and it;s rated output current is 2 amps from the USB A socket and 3 amps from the USB C socket.)
One point about using a single triangle wave generator is that it would also prevent any beating between the frequencies of 4 free running oscillators.

Les.

 

That's an interesting point about using the single oscillator with multiple pulse-width stages - it hadn't occurred to me and I'll give it a shot.

The issue isn't that I can hear a 15kHz sine wave, that ship has sailed. But either due to heterodyning with other electronic equipment, strange resonances of cheap motors, or the fact that it's not a particularly clean 15kHz tone - there are much lower frequency artifacts that I've encountered.

I'm wondering how much above 15kHz I can go and still effectively use pulse-width modulation. Is the OPA2743 necessary for ultrasonic frequency or can I get away with TL072s?

 

You can go to around ~100khz with most decent Op-Amps, using them as a Comparitor in this way,
but above 20khz, I like to recommend a proper "real" Comparitor,
and a proper "Gate-Driver" to make sure that the FET-Switching stays hard and clean.

On smaller Motors,
some of the more Powerful Gate-Drivers can drive the Motor directly, with no FET.
But they are a bit expensive in multiples.

Actually, other than maybe Sound, and Electrical-Noise / EMI / RFI considerations,
there's no good reason to have a PWM Motor-Controller running that fast.
Most Motors, (depending on weight/size),
will make more overall Power, and generate better Low-Speed-Torque
running at around ~200hz,
where you can more easily take advantage of "recycling" the Flyback-Spike,
and consequently turn it into higher efficiency operation.

Have You tried running your Motors directly from the Battery ?
they aren't exactly quiet even with zero PWM.
.
.
.

 

Thanks for all the help - this has been very informative.
One last question: with a "current limiting device"/protection circuit, what is the typical failure state when the current draw is exceeded?

 

There are many ways to prevent Over-Current.
Automatically reducing the Voltage until the Current is below a certain Threshold is very common,
but this requires the Circuitry to Dissipate its maximum amount of Heat,
and if You have inadequate Heat-Sinking,
it will still ultimately result in the Magic-Blue-Smoke leaking out.

Dealing with a short start-up Current-Surge is usually survivable,
but if the Circuit goes into a loop, (on,off,on,off,on,off),
it may never supply enough Current, in a short enough period of time,
to get things warmed-up and stabilized.

The Circuitry needs to be sized for "worst-case" conditions,
but in the case of a Chinese Power-Pack, it's sized for the average use case,
and if You start trying to exceed this on a regular basis, or for extended periods of time,
it would not be unexpected for something to fail.

With a Battery-Protection-Circuit, the design goal is to have it "Fail-Open",
in other words, pop like a Fuse.
Otherwise, a Fire could possibly be the result.
.
.
.

 

On the subject of current limiting, an area of confusion is often the relationship between current supply and current demand.

If I have a 10A-capable supply and a motor that takes 0.5A when running then the current that flows under normal circumstances is 0.5A, though on starting the motor might briefly take 1 or more amps to start. Under fault conditions, eg a stall, the current - from the motor's perspective - is only limited to 10A unless there is some other 'intelligence' to recognise the stall and reduce the current. Without that, the current in the motor winding is only limited by its DC resistance which could mean a stall current of several amps. Some motors can withstand a stall indefinitely while others will quickly burn up.

The temptation therefore is to use a lower rated supply, often forgetting that start-up requirement. We often hear the refrain "it worked OK on the bench supply but won't work on batteries" due to a complete misunderstanding of the supply and demand issues around different power technologies.

For example, my wheelchair motors each take around 6A on a nice flat smooth surface. But starting or turning on a soft surface like deep carpet or gravel they can easily pull 100A. And stall current is over 210A if the controller didn't limit it to 120A per motor, rolling back to 90A after 10 seconds or so, and then shutting off after 30 seconds. My 200Ah LiFePO4 batteries will happily provide 500A+ for 20 minutes or so given the chance...

Your 27000mAH (it's nominal C-rating is therefore 27Ah) battery pack, if it has internal protection, will probably have a current limit of typically 3 to 6A (0.1C or 0.2C) though it maybe higher. If you deploy 4 of those Johnson motors at 2A nominal run current it's almost a given that your battery pack will go into current limit if you try to start all 4 at once, and there's a high probability it will do so when all 4 are running (8A nominal).

Then again there might be no protection at all and a Li-ion pack of that size will happily give 5C (100A+) though it's internal wiring and yours isn't going to handle that without burning up.

Can't find specs for the Ravpower unit, but if it's similar to others the usb-c outlet is limited to 3A @ 12v. How are you persuading the bank to go to 12v from its nominal 5v? A so-called trigger cable?

 

No, don't do it, buy ready made PWM driver, they cost 1-2$ since the circuit is so simple, you need an oscillator, pot and mosfet, basically thats it, and you have jumped to another level; you have torq, linear speed control and efficiency !
Google will find the most simple PWM circuits to build or you can easily buy one.