Basic problem with transistors

[mneary]

The threshold voltage is NOT a useful 'turn on' point. It is the point when the MOSFET begins to leak. (for lack of a better word). Although 3.3V will cause substantial current to flow in an IFR510 and similar devices, it's not turned on enough to be used near its current ratiing.

To be used near its current rating, the older MOSFETs like IRF510, 740, etc. needed more like 7V (we usually used 10V to be safe). Some newer "logic level" devices which cost more are useful with 5V on the gate.

 

[solis365]

I googled H-bridge and there's lots of circuits for it as they have been around for a long time. As for cost its all about finding the right component.

I'm sure someone has built an HB that runs from 3.3V logic with cheap components, you just have to find it.

heres one site:
BJT H-bridge Circuit Details
though it may be designed for 5V.
heres an interesting read about interfacing with a controller..
**broken link removed**

try searching for things like "3.3V logic" "H bridge" "interface" "controller"

What is the power rating of your motors? those IRF 5X0s can handle many amps, but if they are small motors they might only need something like 0.5A, and hence you wouldnt need expensive and more-complicated-to-drive high-power mosfets.


and mneary is right, that bit you quoted from me was erroneous; 5V will probably not turn an IRF5X0 on (I use X since 510, 520, etc are differently rated versions). You WILL probably need 10V which would require a dedicated 10V rail and some level shifting which you probably do not want to do as it will increase cost.

can someone straighten me out on my MOSFETs? I think I've managed to confuse myself. at least confirm that my 3 modes of operation are correct!

 

[giftiger_wunsch]

I'll have to check the rating of the motors, my friend has the robotic arm at the moment; I don't know what the current rating of D-type batteries are, but looking at the circuitry of the controller for the arm, it very simply connects the four D-type batteries directly to the motor(s) being operated. The current must simply be limited by the output of the batteries and the resistance of the motors. The fact that all five motors can operate at the same time with no loss of speed suggests that batteries can produce a current at least 5x that required by each motor, though.

 

[giftiger_wunsch]

This is beginning to seem rather hopeless... I looked through the datasheet for the MOS-FETs available from Maplin, where I buy most of my electronics, but I didn't see any mention of the voltage required to turn the MOS-FET 'fully on'; either way it seems unlikely that any of them can be turned 'fully on' with 3.3 or 5V if the usual value is 7-10V. If I can find such components they will most likely be too expensive for my budget.

Thanks for all the help, but unless anyone has any other ideas, I'll probably be forced to give up this project.

 

[solis365]

try looking for "logic level" mosfets. they can be driven by logic level signals. i havent looked into 3.3V logic level ones but there must be a few around.

generally they cant deal with as much as full-blown power mosfets, but for your D-cell powered motors they should be way more than enough. I don't imagine your motors take more than 100-300mA or so.

Now that I think about it, these are such low power motors that you dont even need to be thinking about power mosfets and such, they would be overkill.

look for some BJT-based H-B circuits, they should be enough for your application. and cheaper.

basically the problem is this: microprocessors can only source a small amount of current, usually around 20mA (for all of their pins combined). BJTs take a small amount of current into the base. there has to be more base current if you want it to drive more emitter current, so for high-power applications this base current requirement can get close to the limit of your microprocessor.

however for such small motors, you will probably be okay.

especially look for BJT H-B circuits that use more than one BJT connected in a darlington configuration, this will lower again the amount of current needed from your uC.

I think you will be fine in this application. However as a precaution maybe put a small heatsink on the uC. This won't give you too much extra headroom but it might give you just enough. You won't be running all of the motors at once for very long periods of time anyway.

I'm just briefly looking at these links im sending you as they might be helpful.
DC Motor interfacing with Microcontrollers tutorial: BJT Based H-Bridge for DC motors : 8051 Microcontroller Projects AVR PIC Projects Tutorials Ebooks Libraries codes


you might not even need power BJTs, depending on the current requirements of your motor you might get away with a darlington configuration of the el-cheapo 2n3904/2n3906. possibly a higher-rated BJT as the second stage of the darlington with a 3904/6 as the first stage, which would still be cheap.

 

[marcbarker]

The posted link 'H-Bridge using bipolars' looks quite good, but to use it much lower than 12 V the losses in the transistors may be significant. But I think that'll drive OK from 3V. BJT Based H-Bridge for DC motors

Question: why is microcontroller @ 3V a problem with using mosfets? (I realise now I suggested driving the lowside fets directly) The fet gates don't need to be driven by the 3V levels.

 

[solis365]

the trouble is that he has to do enough bridges to drive 5 of these motors. generally you would stick in some transistors between the uC pin and the mosfet gate for level shifting/gate drive and it would be no problem, but (a) he's a novice and (b) he's on a pretty tight budget.

I think the entire arm itself cost $35 or so (if i remember from an earlier post) and that he doesnt want to spend more than the full cost of the arm in parts. Furthermore, he's ordering from Maplin which isnt exactly cheap.


you could try ordering from Farnell or something (you are in the UK, right?) shipping might be more, I don't know. but they should be cheaper per part than any "hobbyists" store.


ACTUALLY I found an interesting store you might want to check out. I havent tried ordering from them yet but I plan to. Very cheap basic electronics. You can order kits (contain several useful types) but you might also be able to order individual parts in small quantities. Their shipping to the UK is $10 (under the "shipping" link in the nav bar on the left) so I imagine they are US-based. that shipping cost might be a little high for you, I'm not sure. But look at their kits, they contain useful things (big resistor/cap/transistor kits) and are cheeeeeap!
https://www.vakits.com/

 

[marcbarker]

A completely transistor-free solution?

, if the current is low enough, use a paralleled up 74S941 tristate driver(s)? **broken link removed**
power the IC from the 6V supply, drive the logic directly from the uC, I think the logic levels are compatible enough without adding any transistors.

Or maybe audio speaker driver IC's could be used as H-bridges?

 

[solis365]

ah yes here they are. Transistors : NightFire Electronic Kits
individual transistors for a few pennies each (though you have the shipping problem...)
just pick a cheap transistor and look at the datasheet and see if you can make your circuit around that!

 

[audioguru]

The old IRF510 mosfet is made by International Rectifier, not Maplin. So guess where to go to get the best datasheet??

It is spec'd with a 10V gate voltage, not anything less.

Its threshold voltage is the gate voltage where a few of them barely turn on (0.25mA).

 

[marcbarker]

The old IRF510 mosfet is made by International Rectifier, not Maplin. So guess where to go to get the best datasheet??

It is spec'd with a 10V gate voltage, not anything less.

Its threshold voltage is the gate voltage where a few of them barely turn on (0.25mA).

if the motor current is lower than 1A, 10 V of gate drive is wasted, because the FET channel has become well 'ohmic' with gate voltage @ 5V

**broken link removed**

According to the datasheet, 5 V of gate drive is just good for 1 A motor drive current.

 

[audioguru]

You look at "typical' devices on the graphs so I guess you test each of your Mosfets to make sure they do not have minimum specs that are printed in text on every datasheet.
Or maybe you know somebody at the factory who supplies you with "typical" Mosfets.

The max on-resistance of an IRF510 Mosfet is 0.54 ohms when it is cool and has a 10V gate voltage. Then when it is warm and with 1A of current it has a voltage loss of almost 1V.
Your "typical" mosfet will have a little higher voltage loss when its gate is 5V.
Is a loss of 1V and more acceptable?

 

[giftiger_wunsch]

Okay lots of replies so sorry if I don't respond to everything. I'll just try to summarise some of the main information I've discovered so far:

Motor voltage: 3V (the 6V battery pack is actually divided such that two of the D-Cells power 2 motors and the other two power the remaining three).

Motor current: Measured at 200mA when a single motor is being driven (so can drop to 66-100mA without experiencing many noticable issues).

Regarding bipolar transistors: I originally planned to use these, but marcbarker pointed out some shortfalls. And with the output of the uC being 20mA max, that may be an issue.


I was thinking about the option of putting transistors between the uC and the MOS-FETs, but as already pointed out this would further increase the cost and I'm not exactly experienced yet either; ending up with 40 of various kinds of transistors, as well as a load of resistors, possibly diodes to isolate the ground circuits, etc., sounds extremely complicated... and expensive.

Bipolar transistors I can get for about 12p each from maplin if I order in the quantities which would be required, but I don't know of many other electronics suppliers, and maplins don't do any logic-level MOS-FETs. I'll take a look at the site suggested by someone in one of the posts on here

 

[marcbarker]

spec'd at 10V

In the OP Application, the needed current as understood is much less than 1 A, so it'll be a waste of time driving the gate with "10V" if 6V would still do virtually the same. "Vgs=10V" is preferred of course it is, but 10V isn't available in this application, luckily, with some careful analysis with a low motor current, 6V would work, only just.

IRF510s are old school nowdays early generation devices. There are many lower Rds(on) and higher gm newer generation fets available that's better suited, that a 510 wouldn't be picked, so it's not really worth wasting effort going that deep into the analysis. But I will...

You can see from the IRF510's typical curve I posted, the 'typical' Rds(on) it has (@ Vgs=6V) = 0.6 ohm. 6V is the battery voltage. The effect of parametric spread on the shape of the 'typical curve' will take some imagination, based on worst case datasheet figures, but in practice in the real world it's actually much less when real samples are measured.

At the drain current of less than 1A, the fet is so well enhanced (Vgs=6V) into the 'ohmic conduction' part of the curve, that a fair bit of variation of actual Vgs(th) from 'typical' could be tolerated before the fet starts to drop out of ohmic conduction mode. In practice, Vgs(th) doesn't vary that much from typical, if it does more than 1V that's very bad luck, or the fet is an original 1980's one. Actual perfomance can be better than typical as well as worse than typical, and modern fets often out-perform the (old) datasheet.

What often happens with fets is that people don't understand the curves, temperature dependence and production spread variance properly. They'll get 'bitten' by a project that went wrong, then after that they mis-trust the datasheet blaming 'typical' data as 'mis-representative;, then as this advances, they cover themselves each time by interpreting all of the 'worst-case' data they can find and applying it all together at the same time (which is totally invalid of course, but it does minimise the risk of them ever making a mistake in their design).

Anyway, it's all academic really, because even though an IRF510 would work with 6V of gate drive, it would be better to use instead later generation automotive 1.5V Vgs(th) fets that have milliohm Rds(on), which are the same price.

 

[marcbarker]

Regarding bipolar transistors: I originally planned to use these, but marcbarker pointed out some shortfalls. And with the output of the uC being 20mA max, that may be an issue.

Bipolar transistors seem better than MOSFETS now, in the light of new information: the motor current being so low. Since bipolar drive current= motor current / hFE, the max output current of the uC isn't a concern any more.

A h-bridge made out of npn & pnp transistors seems to me a good option now.

Or maybe utilise a tri-state buffer?

 

[giftiger_wunsch]

Okay, given the current information, I've drawn up another schematic which is hopefully closer to what it should look like now. I'm not sure if I have connected the ground (green) correctly; I tried to connect it such that when PA1 is high and PA2 is low, PA2 acts as ground for PA1, and vice verca. That prevents a short-circuit if PA1 and PA2 are ever both driven high.

Any comments are welcome.


The transistor which seems most appropriate is the BC337-16 model transistor. Maplin doesn't provide a full data sheet, but the Ic is 500mA; I'm not entirely sure how to interpret the rest of the information.

 

[audioguru]

The upper transistors in your H-bridge are emitter-followers with a 0.7V or more voltage loss.
Then the 3V motor gets only 2.1V minus the voltage loss across the 12 ohms resistor.
The diodes do nothing.

 

[Hero999]

Sorry if it's already been mentioned but you need to use an IRL540 because it's cheap , is logic level and has more than the required current drive.

 

[giftiger_wunsch]

The diodes do nothing.

But without them there would be a short circuit between the PA1 and PA2 pins, and virtually no current would flow through the BJTs since that's the more resistive path.

The upper transistors in your H-bridge are emitter-followers with a 0.7V or more voltage loss.
Then the 3V motor gets only 2.1V minus the voltage loss across the 12 ohms resistor.

Is there a way I could avoid this? Or will I simply need to increase the voltage supply?


I had a feeling it was too good to be true that I may have finally found a solution...

 

[marcbarker]

What about this, people?

**broken link removed**